Sep 12, 2924 - Oprah Got AI WRONG! (My Trip to Lobby Congress)

Description

Chapters

Transcript

0:04 champ you want to come
0:05 sing you want to come sing buddy
0:09 [Music]
0:44 you're loudy you're loudy that dog is
0:47 loudy all right here's the deal tell all
0:51 your friends to come it says there's one
0:54 person in here but I see more of you in
0:55 here you're tapping away all right
0:58 [Music]
1:00 so it is possible we can we can stream
1:03 the Oprah thing but uh apparently ABC's
1:07 got some weird [ __ ] up thing I thought
1:09 we could watch it on Hulu and it said
1:11 streaming on Hulu tomorrow um so I just
1:15 signed up for a trial for some [ __ ]
1:18 TV network that's ridiculously expensive
1:21 so hopefully this works I'm on ABC right
1:24 now I'm apparently on the channel that
1:26 it's going to be on but it's possible it
1:28 switches to my local news in instead of
1:31 the Oprah special so this could be a
1:34 bust welcome welcome to janky
1:38 [Laughter]
1:42 TV all right I'm back from Washington my
1:45 back is still [ __ ]
1:47 up it's hot I'm cranky I'm stressed out
1:53 Kyle sign up on YouTube TV free trial
1:56 okay yeah maybe I should do that too um
1:58 hang on yeah it's a good idea
2:10 let's
2:14 see
2:17 subscriptions let's see
2:21 YouTube
2:26 TV sign up and try YouTube TV
2:31 10 days for
2:36 free now you guys got to help me
2:39 unsubscribe to all this
2:41 [Laughter]
2:45 [ __ ] start free
2:49 trial base
2:53 plan probably in the base plan
2:58 no more info
3:04 of course they're not alphabetical of
3:06 course they're not
3:25 alphabetical all plans include local
3:28 channels
3:33 $65 a month what happened
3:36 to everything streaming being
3:40 cheaper
3:42 unbelievable next check out
4:01 did you watch the debate okay Taylor
4:04 Swift did kamla hara's begging for
4:09 money let's
4:15 see can we
4:21 search let me go back
4:24 to fubo TV
4:34 choose your favorites and we'll Rec oh
4:38 skip
4:40 dismiss all right ABC World News Tonight
4:43 okay so there's
4:45 ABC Denver 7 popular now okay I think we
4:49 have
4:51 it I still have your card
4:54 number you can skip this here we go what
4:58 time is it 58
5:00 good
5:02 Lord it doesn't need to be this
5:05 stressful people look my hair's all
5:08 tall what's happening what's
5:14 happening all
5:16 right let's go over
5:18 to fubo
5:21 TV that says it's got three minutes
5:26 left hair lock on the spacecraft so all
5:29 four members of crew were exposed to the
5:31 vacuum of space while that hatch was
5:33 open closed vertical translation the
5:37 crew is expected to come home early
5:39 Sunday splashing down off the coast of
5:42 Florida and tonight they say they are
5:44 now one step closer to eventual bases on
5:47 the moon hey everybody what's happening
5:49 an Murphy history made and what a
5:51 beautiful view of Earth thank you for
5:53 watching here on a Thursday night I'm
5:54 David M we'll see you right back here
5:56 tomorrow from all of us here good night
5:59 thank you for World News Tonight with
6:01 David mure America's most watched
6:06 newscast op right now at 6 a brush fire
6:10 in Northern Colorado shuts down a busy
6:12 Interstate for hours dry air and Gusty
6:14 winds fuel fire danger on this weather
6:17 action day hot temperatures and Gusty
6:19 winds a fire weather warning continues
6:22 until 8:00 p.m. plus disturbing
6:24 testimony in the boulder King sup trial
6:26 I stood there all right what's that
6:30 let's
6:31 see live now on Fox wait what do I do
6:34 Brandon on YouTube t TV click live and
6:38 find your local ABC affiliate
6:43 live local ABC affiliate see this says
6:48 Denver 7
6:49 News this Mike Nelson joining us with
6:52 the first of the forecast we hit 93
6:55 today that's the 61st day of 90 or
6:57 hotter so far this summer and
7:02 way above the
7:07 average
7:11 so I thought it was on 8:00
7:15 p.m. change your time
7:18 zone oh my God how do I change my time
7:21 [Laughter]
7:27 zone change your location Kyle great how
7:31 do I do
7:33 that let's see manage your account
7:37 location Denver
7:39 area current playback area let's see
7:45 update update home
7:48 area allow us to access your location so
7:52 we can serve the
7:54 right verify your current area
8:02 started I
8:07 know is there any truth to the Aurora
8:11 thing I don't [ __ ]
8:13 know no Central Time how do I change it
8:17 area Let's see we go
8:26 here CC
8:31 oh this is going to be a [ __ ] wamp
8:32 wamp lie Kyle lie uh I don't know if I
8:37 can it's saying let's see update current
8:40 playback area update
8:44 location see it's just getting it from
8:46 my
8:53 GPS go back and click on AI future of
8:58 Everything YouTube TV no
9:05 thanks can I
9:08 search
9:11 search AI
9:17 future got
9:22 it here's today at 7 so this an hour
9:26 from now
9:32 looks like it's going to be in an hour
9:33 go back and click on AI future of
9:36 everything I just had an interview today
9:39 I'm scared of interactions like that
9:41 okay I got it now on fubo okay someone
9:43 got it on
9:45 fubo
9:46 Jesus here we go play
9:50 channel it's not it's not playing it
9:54 damn
9:58 it it's on now I believe oh I believe
10:01 that too then you must be in EST maybe
10:05 you're delayed in time zones in
10:10 Denver Live on YouTube TV
10:15 where
10:17 where
10:19 um
10:21 oh
10:25 ra
10:28 AI future of us
10:31 says airs today at
10:34 7 this is why I can't live in Denver
10:37 anymore this is
10:42 unbelievable the hell happened to
10:47 TV where can I watch Opa Oprah live well
10:50 apparently you're going to watch it
10:53 here are you going to stream it I can't
10:55 get it to [ __ ]
10:58 work invite inv Brandon to share his
11:01 screen and join your live all right we
11:03 can do
11:05 that or we can watch it in an hour we I
11:09 finally found it on my local channel oh
11:12 my God this is so
11:15 frustrating well I guess we'll do it in
11:17 an hour it's on cable TV no it's well
11:20 it's on yeah it's on cable TV if you
11:25 have networks and you're not in Denver
11:28 apparently yeah I know it's on now
11:31 Source Camp I'm very well aware of that
11:34 this is
11:37 a yeah Hulu's streaming it tomorrow
11:40 that's what I found out at half an hour
11:42 ago it's on ABC I know that but you
11:45 can't watch ABC anywhere that's my
11:47 problem um the summon overs we'll be
11:51 here in an hour all right I don't have
11:53 enough
11:55 subscriptions my TV provider is blocking
11:58 it due to Disney this is is unbelievable
12:00 this whole [ __ ] fragmentation of of
12:03 the media is just it's going to be
12:05 convenient you can watch whatever you
12:07 want unless you want to watch
12:11 something oh my God all right
12:22 well all
12:24 right I'm in California and we'll be
12:26 sure we be watching it from you my TV
12:28 provider is blocking it due to Disney
12:31 yeah I just subscribed to $150 worth of
12:34 subscriptions so I could watch it but
12:36 they're not letting me watch it for an
12:38 hour so we can watch it we'll watch it
12:40 at the top of the hour
12:43 um yeah there you go all right
12:46 fantastic
12:49 um so I'm trying to think what I want to
12:53 do oh this sucks I I'm I I am so tired
12:58 of capitalism I really
13:02 am deregulate it it'll be great for
13:05 everyone
13:07 uh-huh search a
13:11 BC uh
13:14 ABC
13:15 Oprah play channel
13:20 news I got to see this
13:25 [Laughter]
13:28 numbnuts she's in interviewing Sam right
13:30 now thanks Dr
13:36 J uh tell us about Washington all right
13:40 I guess I guess we'll call that a punt
13:42 um we'll get the dog singing all right
13:45 so let me tell you about Washington so I
13:47 went to Washington Tuesday night had a
13:50 nice dinner at a Georgian
13:54 restaurant like Georgia the country
14:00 I don't know how to describe the food
14:02 it's kind of like you know Mediterranean
14:04 food where you got all the fresh veggies
14:06 and olives and hummus and [ __ ] remove
14:10 the olives and hummus add in more
14:13 pickled
14:16 cabbage that's pretty much
14:18 it a lot of
14:23 vegetables
14:25 um and
14:27 then so the
14:30 so what we did was on
14:33 Wednesday there were 12 of us there 12
14:36 AI
14:38 entrepreneurs and they basically had
14:40 schedules I got paired up with a guy
14:42 named George who was awesome he's lives
14:46 in St Louis and he's got a company that
14:48 does AI
14:50 robotics lawnmowers for like big grassy
14:54 areas like airports and you know
14:57 Municipal parks and [ __ ] like that where
14:59 you've got many many acres that need to
15:01 be mowed and basically no one no one
15:04 wants to do the jobs he found out that
15:06 there's like basically a labor
15:08 shortage um of people that don't just
15:10 want to mow the lawn for eight hours you
15:13 know in the blasting Sun uh who who
15:17 knew and
15:20 uh so that's his business he's cool he
15:23 had a PhD in robotics and mechanical
15:28 engineering
15:30 he's like three years into that business
15:32 sounds like it's going pretty good um
15:35 and so we went to the first the first
15:40 office we went to was Senator Bud out of
15:43 North
15:45 Carolina and we met with with with all
15:48 of these we met with a
15:50 staffer so far all political okay
15:54 interesting yeah actually you know what
15:57 don't tell me what's going on in the
15:58 Oprah thing I'll just watch it and
16:00 pretend
16:01 like pretend like it's
16:03 [Laughter]
16:08 live um but what we could do so um open
16:12 AI did launch today
16:16 strawberry um it's called 01 so you have
16:22 GPT
16:24 4.0
16:26 gp40 and this no one new one is called
16:30 01 I I don't know what they're [ __ ]
16:32 doing anyway back to Washington um so we
16:36 can play with I played with it a little
16:37 bit I got it about an hour before I left
16:39 work um and played with it a little bit
16:42 and it's interesting what it does
16:46 um so I have some thoughts
16:50 already um okay so with all of these
16:53 people we met with staffers and meeting
16:54 with staffers is really interesting
16:56 because sometimes you get staffers that
16:59 are
17:00 educated about your topic so so the the
17:03 the group that helped organize this they
17:05 set up all these meetings this is what
17:07 they
17:09 do um so they know everyone and they try
17:11 to get sort of the highest most relevant
17:14 staffer you know for for these
17:16 meetings um and the first one we went to
17:20 was was uh a North
17:23 Carolina thing um it was George and I
17:27 and Kathy from the rxn group and then
17:29 this woman I forget her name who was
17:32 from North Carolina who was also had an
17:34 AI startup and so we just introduced
17:36 ourselves so the constituent the one
17:38 from North Carolina went first and she
17:40 said you know here's my here's my thing
17:43 and then George talked about his thing
17:44 and then I talked about my thing and we
17:46 all kind of had our message um for for
17:49 the uh you know for the staffer and the
17:51 basic idea is you're just trying to
17:53 educate them on what message you want to
17:55 get out so the message the message that
17:57 I wanted to get out was
18:01 um there's a couple couple of major ones
18:04 one is a lot of the legislation right
18:07 now is legislating the
18:09 technology as opposed to the use of the
18:11 technology and so Kathy actually had a
18:14 really good analogy to this so if you
18:17 talk about cloud
18:20 computing we don't regulate cloud
18:22 computing but if you're storing Health
18:25 Care information in the cloud that's
18:28 covered under HPP regulations if you're
18:30 doing Financial
18:32 Services um data that's covered under
18:35 Financial Services regulation but you're
18:37 not actually
18:39 regulating the tech what's happening
18:42 with AI is they're regulating the tech
18:45 and it's sort of sweeping all the all
18:47 the use cases up into it that might not
18:49 be relevant for that piece of Regulation
18:51 that's kind of what happened in
18:53 Colorado so part of it was
18:56 just be mindful of how your writing the
18:59 regulations because it's sweeping up
19:02 small businesses and it's sweeping up
19:04 use cases that might might not be
19:06 relevant so the protection is good but
19:09 the execution is really bad um I also
19:14 wanted to talk about the difference
19:17 between generative Ai and AI as we've
19:19 known it before right so if you think
19:21 about ai's been around for decades and a
19:25 lot of the perception is just AI is just
19:27 this one thing
19:29 and what I talked about was you know
19:31 this community so I was there we were we
19:34 were the the salon was the
19:37 representative organization that was
19:39 going around to these groups and so I
19:42 was talking about the salon and all of
19:44 you and saying you know this group was
19:46 started the week after chat GPT came out
19:49 and I get to witness daily people
19:51 Reinventing themselves and Reinventing
19:53 their businesses and I mentioned you
19:56 know Jim Ross he said on a on a call the
19:58 other day
19:59 he feels like he's functioning like a
20:01 10-person company and he's a he's a
20:04 solopreneur um that statement got people
20:07 to sit up they were like oh that's
20:09 really
20:10 interesting
20:12 um the the staffer from Bud's office was
20:16 he was smart he was taking a lot of
20:17 notes he was asking questions he was
20:20 basically saying things like we hear you
20:22 this is really important issue um he
20:25 kind of said there's not going to be
20:27 anything coming from the Fed
20:30 um for probably 2 or 3 years and um and
20:36 I to to which I said so are we going to
20:38 have a patchwork quilt of nightmarish
20:41 state-by-state
20:42 laws and he goes yeah
20:46 probably I said well that sucks so could
20:48 you pass that
20:51 along um oh this was really fascinating
20:54 so I
20:57 um on the flight out there I was sitting
21:00 next to a
21:01 woman
21:03 um and she was the executive director
21:07 for a
21:12 um a what's it called
21:15 Foundation that supports civil
21:18 engineers
21:21 and I said are you using chat gbt you
21:24 know I did I did my obnoxious AI
21:28 evangelism [ __ ] [ __ ] that I
21:31 do she said she had tried it but not
21:34 really and she she said well I do a lot
21:35 of like writing and press releases and
21:37 things like that and I don't want um
21:40 it's got to sound like me and so I said
21:42 well here's some ways you can make it
21:44 sound like you and I told her about take
21:46 some of your writing and put it in a
21:47 document and upload it to chat GPT and
21:49 it'll start writing like you and I was
21:52 given her ideas and she was taking
21:54 notes and then we started I talking oh I
21:58 said you know what are you going to
21:59 Washington for and she was doing some
22:00 lobbying and I said oh I'm doing some
22:02 lobbying and and
22:04 uh and I said I actually testified
22:07 against the Colorado Bill and she goes
22:09 oh that bill yeah she goes I didn't know
22:11 where you stood on that I didn't want to
22:13 say anything but she says because so to
22:17 give you some context of the Colorado
22:19 Bill the Colorado bill passed and was
22:23 signed into law by governor
22:26 polus with a signing statement that
22:29 effectively said this law is a piece of
22:31 crap um I have grave concerns that it
22:35 you know sweeps up small businesses so
22:37 please fix it before it goes into effect
22:39 in 20 late 2025 early
22:43 2026
22:45 and
22:47 um she knew the
22:50 law and she said to me I've made a
22:54 strategic decision not to use chat GPT
22:59 in any
23:00 way because or any generative AI in any
23:03 way
23:04 because the way that law is written I
23:07 don't want to put in a year or two of
23:09 learning how to do something for my
23:11 business changing my processes and then
23:14 ending up with a regulatory burden
23:16 sitting on top of that two years from
23:19 now so I had a a an an actual specific
23:24 example of the chilling effect of that
23:27 kind of legislation that this is you
23:30 know this is someone who's got a big
23:32 Statewide foundation for all the civil
23:35 engineers in the state and is actively
23:39 not using chat GPT because of that law
23:41 or or or you know not not using AI
23:44 because of that law um welcome to the AI
23:47 learning lab we'll be watching the Oprah
23:48 interview live with Kyle at 7 because I
23:51 live in [ __ ] Denver and all of the
23:54 channels have been deregulated and
23:56 decoupled and Disney is an ass ho there
23:59 you go there you have it
24:02 [Laughter]
24:05 um what if I want to watch the the show
24:08 oh sorry you can't but you can you can
24:11 stream it live tomorrow on Hulu no I
24:13 want to watch it live now oh yeah oh hey
24:17 yeah we we uh we have a response to that
24:20 maybe you could go on ahead and go [ __ ]
24:22 yourself does that work for
24:24 you so yeah so we'll be watching we'll
24:27 be watching Opie
24:29 you can have ai and you can have ai and
24:31 you can have ai we'll be doing that in
24:34 45 minutes 42 minutes okay um so that
24:39 was the
24:40 bud the bud
24:43 meeting um the next meeting I had with
24:47 was with Josh Holly of Missouri um
24:51 George who I was with is from Missouri
24:54 so we had a different Handler for the
24:55 Josh Holly thing um I was
25:00 I'm not a fan of his but we didn't meet
25:03 with any of the Senators so it doesn't
25:04 matter we're just meeting with staffers
25:06 um but we who we met with was someone
25:08 who was counsel so an an actual lawyer
25:11 um from the Holly office um and he was
25:15 wicked smart like he knew the issues he
25:18 knew the laws like when when I would say
25:21 things like you know small companies get
25:23 swept up in this overly broad
25:25 legislation he was like well how is it
25:26 too broad tell me specifically so he was
25:29 really digging in on specific
25:32 things um and so there was probably 10
25:35 minutes of
25:37 just harsh cross-examination let's say
25:43 um and I you know I think we did good I
25:45 think I think we navigated the the
25:47 cross-examination well and he was a bit
25:51 more pragmatic he said um I think he was
25:54 the first one that said to us um
25:59 he said yeah the the AI stuff was really
26:02 hot for a while and it's not now right
26:05 because you know we've experienced that
26:07 on this channel that there's this kind
26:08 of
26:10 lull um in in AI news and excitement um
26:14 well it it translated to Capitol Hill
26:17 and so basically if it's not a Hot Topic
26:20 um things tend to take longer if it's a
26:22 Hot Topic they rush it through so so
26:25 that was kind of another indication that
26:27 we're probably not going to see
26:29 anything we've got two weeks left in the
26:31 legislative session now then we go into
26:34 um election season then you go into lame
26:36 duck so like nothing's going to really
26:39 happen until
26:40 2025 um and if AI is not a big Hot Topic
26:45 then probably nothing's going to happen
26:46 for 2025 so that that was some more of
26:49 that um we got in we got in some of our
26:52 message to him he he took notes he was
26:55 he was super engaged he was super
27:02 um like he he was appreciative that we
27:05 were there he didn't really want to be
27:07 there like it was it was he's like I got
27:10 lawyer [ __ ] to do but you know go ahead
27:12 this is part of my job you know he had a
27:14 little bit of that
27:17 attitude uh so that was that let's see
27:20 I'm trying to think who else we who else
27:22 we went
27:23 to we can all watch it with you okay
27:26 let's see um
27:31 so those two were Senators then I think
27:33 we took a lunch
27:40 break
27:42 yeah and then my next meeting was with
27:47 um Diana
27:49 deette who's a congressman so we were in
27:52 the Senate building which is on one side
27:54 of the capital and then the house
27:56 building is on the other side of the
27:58 capit capital and it's like you look
28:01 across you can look across and see the
28:03 other building but it's like a long walk
28:05 and it was hot and so we were taking
28:07 cars so you would basically take a car
28:09 basically around the block but it was a
28:11 big long
28:12 block
28:14 um and so what's fascinating is so if
28:17 you if you haven't ever been there I I
28:19 haven't ever been in the bowels of of
28:21 these
28:23 buildings the Senate cuz the Senators
28:26 have more power right cuz there's only
28:28 two per
28:29 state their offices are a little
28:32 swankier and you know Swanky is not
28:35 quite the right word but you know very
28:36 lawyerly and big
28:38 and um the house offices are a little
28:43 janky so so we're in we're in D's office
28:48 and some other Congressman came in and I
28:50 was shooting the [ __ ] with him and he
28:52 was apparently he was a little anti- aai
28:55 he was a little he was a little nasty
28:56 but he was fine uh
28:59 um but the person we were supposed to
29:01 meet with was in some other meeting and
29:04 then there was a whole group of people
29:06 in this teeny little office and they're
29:08 like okay you know we can meet with you
29:11 now but there's no there's nowhere for
29:13 us to meet so can we just meet in the
29:15 hallway so like this gaggle of like
29:18 eight people sort of scurries out of the
29:21 office and they're holding the meeting
29:23 just outside the
29:25 door um and then we're still sitting
29:27 there we're supposed to be me like a
29:29 half hour's gone by like our our meeting
29:31 has sort of come and gone and this very
29:34 pleasant woman comes out and she says
29:36 sorry the staffer you're supposed to
29:38 meet with is still busy but I'll meet
29:40 with
29:41 you and she goes we don't have anywhere
29:44 to meet so can we just meet in the
29:45 hallway so we're on the hallway there's
29:48 this gaggle of out PE eight people
29:50 screaming there's people walking by
29:52 there's other meetings happening in the
29:54 hallway and it's you know it's
29:56 government buildings so there's no
29:57 carpet it's just like you know hard
30:00 concrete walls and
30:01 Halls very noisy
30:05 um so she gave us a solid 10 minutes
30:08 that was
30:10 fine um so that was Diana
30:14 deette then we met with I forget who it
30:17 was a staffer
30:20 from was that in the
30:26 house yeah
30:28 it was it was a congress person from
30:33 Minnesota and we got there late because
30:36 we were late from that other thing and
30:37 we couldn't find it and when we finally
30:39 got there he had already kind of started
30:40 the meeting he was really [ __ ] nasty
30:45 he was older he was smart anytime he he
30:49 sort of took me he latched on to me to
30:52 like he came at
30:54 me
30:57 um and you know when I would say things
30:59 like you know I would sort of say
31:01 platitudes about you know how bad the
31:04 legislation was and how it affected
31:06 people well exactly how does it affect
31:07 people what specifically about it
31:09 affects people and he was just like
31:11 coming at me he's
31:13 like what we what we what we sort of put
31:17 together after because what what was
31:19 fascinating is he was he was coming at
31:21 me and I was I was just this side of
31:23 breaking into a flop sweat cuz you know
31:25 how when you're trying to say something
31:27 and someone's like
31:30 you get a little
31:32 like and then I thought I don't give a
31:35 [ __ ] if this guy you know so so I I
31:38 avoided flop sweat Panic so I just kind
31:40 of held my own for a while and then he
31:43 and then he went on and talked to other
31:44 people and then at some point he
31:47 actually made the connection he was
31:49 really smart he really knew his
31:51 [ __ ] Kathy from rxn was talking about
31:54 the cloud storage thing and he goes oh
31:58 oh he
32:00 goes you're just talking about not
32:03 legislating the technology but
32:05 legislating the use
32:07 case and we said yeah and and then at
32:10 that point his guard came down what we
32:12 put together after the meeting is that
32:14 he was I think he thought that we were
32:17 anti-regulation that we were like we
32:19 don't want to regulate AI at all and
32:21 we're not we're like just regulate it
32:24 smart regulate it in a sophisticated way
32:26 that doesn't paint like Frontier Model
32:29 companies like open AI with small
32:31 businesses with four employees in them
32:33 right cuz they don't have the same
32:34 resources to deal with your stupid
32:36 [ __ ] regulation so like be smart
32:38 about it right like make a grid and a
32:40 matrix and understand the nuance and you
32:43 know regulate the the use case rather
32:46 than the
32:47 technology um and so once he got that
32:50 his tone completely changed and he was
32:53 really engaged and he was taking notes
32:55 we gave him some ideas so that was that
32:56 one
32:59 um then we went to Senator Bennett's
33:02 Office from
33:04 Colorado and the staffer we were
33:06 supposed to meet with wasn't there the
33:09 staffer that we did meet with was a
33:12 lovely young lady she was maybe I don't
33:14 know 20 wait did you show cranky old
33:17 Mana no I didn't I didn't I didn't do uh
33:20 Kyle from the AI learning lab didn't
33:22 really show up in Washington it's you
33:26 got to be political you know it's it's
33:29 if ever there's a place where you got to
33:31 be political it's DC right you you can't
33:34 you can't burn Bridges and do weird [ __ ]
33:38 so although you
33:41 know apparently you can do really if
33:44 you're shit's weird enough you can do
33:45 any of it
33:47 um so she met with us and she was okay
33:51 she didn't really know the issues all
33:52 that much she was like a year ago there
33:55 was an AI series of summits
34:00 um and
34:03 uh and she kind of sat she was a little
34:06 cross-armed and and she just sort of
34:08 gave us Pleasant nods and things like
34:11 that um so she was
34:14 okay it it was that meeting was just I'm
34:19 taking notes I'll pass them along I
34:21 don't really give a [ __ ] was kind of the
34:22 tenor of that meeting um did you say you
34:25 can make money with chbt I didn't
34:27 although the last me so the last meeting
34:29 was the best meeting so we then went to
34:33 hickenlooper's office and the staffer
34:36 there was named Sam I forget Sam's last
34:38 name but Sam had like his master's
34:41 degree in computer science um he was a
34:45 software developer he was super
34:47 passionate about this he goes this is
34:48 what I do like I'm you know I do Ai and
34:51 I do software and I do this and he had
34:53 he had like four or five different
34:54 things that he did but AI was in it he
34:56 cared a lot about it
34:59 um he asked really good questions he
35:03 completely got um all the stuff we were
35:06 talking about and then he even came back
35:09 and said hey if we were to approach
35:12 regulation this way would that work um
35:15 and and I forget how he was saying
35:18 it oh one of the things we were talking
35:21 about was
35:25 um there's different actors is is how he
35:29 put it which I thought was a really good
35:30 term right that you want to regulate the
35:33 right actors right so you you've got the
35:35 frontier models that are building the
35:38 these core models and then you've got
35:40 people that are just using the models
35:42 wholesale then you've got people that
35:44 are using the models and fine-tuning
35:46 them right and then you've got some
35:48 people that are making their own models
35:50 and right and so there's all these
35:51 different actors in the
35:53 ecosystem and so he was talking about
35:55 like how do we determine the actor right
35:58 and and he goes you know should it just
36:00 be based on like number of employees or
36:02 number of Revenue and I said I don't
36:04 think that can actually work because you
36:06 might have a company that's a single
36:09 person company that's acting like a
36:11 thousanders company because they're
36:13 leveraging the [ __ ] out of AI right like
36:16 and then you might have a 1300 person
36:18 company that doesn't fall under
36:19 regulation because they're using it in a
36:21 really limited way that falls outside of
36:23 that so he actually got that
36:26 Nuance um
36:28 one of the other messages that I talked
36:30 about how we doing on time one of the
36:31 other messages that I talked about was
36:36 um I said when you're thinking about
36:40 regulation
36:44 don't be very careful to write the
36:47 regulation in a way that doesn't
36:51 assume that the technology we have today
36:55 is how it's going to look a year or two
36:57 from now right right because if you say
36:59 hey based on this model or this
37:01 capability of this model here's the
37:04 regulation a year from now it's not
37:06 going to look like that so I said you
37:08 run the very real risk of writing
37:10 regulation that will um be obsolete by
37:15 the time it passes and Sam really got
37:18 that he really got that um I was really
37:22 impressed with him he he talked about
37:26 um understanding the Nuance between big
37:29 companies and small companies right he
37:31 understood the the difference between
37:34 traditional Ai and generative AI um he
37:38 had this whole concept of different you
37:40 know sort of layers and sizes of of
37:42 companies and it at least you know
37:44 figuring out a way to do that um so so
37:48 that one was really good so and then I
37:50 just went from there to the airport
37:53 um the the last meeting was was actually
37:56 really promising you know he similarly
37:59 said probably not going not going to be
38:01 anything from the feds um anytime soon
38:06 but
38:09 um I kind of feel
38:13 like so I I think I met with six
38:17 people I feel like two of them were
38:21 really good the Josh Holly lawyer was
38:23 really
38:24 smart the the hick and Looper staffer
38:27 was really smart I have a feeling the
38:29 Bennett staffer had we met with the
38:31 actual staffer would have been fine so I
38:33 feel like Colorado's in a good place oh
38:36 all of the Colorado people I said I said
38:40 I I met with this executive director
38:42 that is
38:43 not you know pursuing AI because of that
38:46 law they all took Furious notes on that
38:49 one like the fact that I had a specific
38:53 anecdotal piece of data from from the
38:56 day before like flying in of someone
38:59 where there was a chilling effect
39:00 because of this law and we you know one
39:03 of the things George and I were talking
39:05 about was like we want the US to remain
39:08 competitive and if you put us in a in a
39:13 cage where we can't do our [ __ ] we're
39:15 not going to be competitive so when I
39:17 had a specific story of here's a company
39:19 you know where the executive director
39:21 made a strategic decision not to not to
39:24 learn AI not not to do anything with it
39:27 because of the legislation that you
39:29 wrote that doesn't go into effect until
39:31 2026 already having a chilling
39:34 effect crazy all right so that was
39:40 Washington
39:42 um let me I'll see if you have any
39:44 questions because I I've just been
39:45 talking and not looking at the
39:48 screen AI wants to know if you notice
39:51 bias towards or against AI along party
39:55 lines um
39:58 I don't I haven't been looking for it
40:00 but I I did see I did see a Tik Tok
40:03 today that addressed that
40:05 specifically she took a transcript of
40:08 the
40:10 debate and had chat GPT analyze it and
40:13 then she took the transcript of the
40:14 debate and um de deidentified it took
40:18 out the specific names of the players
40:20 just said candidate a candidate B and
40:24 she said it got she got a very different
40:25 result from that so that makes sense to
40:27 me good Kyle sounds like you made some
40:29 Headway yeah I I'll tell you like the
40:33 the Minnesota
40:35 Congressman the the the one that he
40:38 actually turned in the meeting where he
40:40 went from being very combative to being
40:42 you know really listening to us that
40:44 felt like we made progress the senator
40:47 Bud dude I I think he was fine I think
40:49 he'll pass along the
40:51 notes um Josh Holly's Council
40:55 um he took it seriously
40:59 the deette Diana deette one where we had
41:02 the hallway meeting I think that was
41:03 just a you know a nice to have the the
41:07 Bennett one was like e that'll that'll
41:09 go along but but Sam from hickenlooper's
41:11 office I think we made a a real impact
41:14 there he was the first one Sam was the
41:17 first one so so one of the things that
41:19 we said to all of them
41:21 is have small business at the table when
41:25 you when you're you know thinking about
41:28 this legislation and drafting this
41:30 legislation have small business in the
41:32 room right so that you're representing
41:36 the users of this technology not just
41:38 the manufacturers of this
41:40 technology and um Sam was the first one
41:44 I really got the sense that you know I'm
41:46 a constituent and um I really got the
41:50 sense that you know they could
41:51 potentially reach out at some point and
41:53 say hey we want to talk and here's what
41:55 we're thinking about doing and what's
41:56 your feedback on that so
41:58 um what Kathy from rxn told me was that
42:02 I said you know what's the what's the
42:04 end goal here is is the end goal that we
42:06 you know affect any specific change she
42:09 said no she
42:11 said she said there's no specific
42:13 legislation that we're lobbying for or
42:16 against um she said this is relationship
42:19 building right this
42:23 is get the concerns into the
42:26 conversation
42:28 so even if there's no specific action at
42:30 some point in the future there might be
42:32 just something that sticks in their head
42:33 you know we should probably make sure
42:35 that we're separating big companies from
42:37 small companies or you know so so
42:39 Concepts get through and then she said
42:42 what what this really is is that over
42:43 time you keep having you know these
42:45 meetings and you know the staffers get
42:48 to know you and the staffers are the
42:49 ones that that do all the work and so at
42:52 some point if they want to bring in a
42:54 voice um for that particular issue
42:56 they'll have record of all this and
42:58 they'll they'll call you call you out
43:00 and you know bring you back out there
43:01 and things like that so thought that was
43:03 fascinating um yeah so that's
43:06 that uh Kyle for president his slogan is
43:09 embrac the Jank four more years no My My
43:12 My slogan for 2028 is is it can't
43:16 possibly be
43:20 worse where where the AI is actually the
43:23 president and I'm just the fancy little
43:25 spokes model
43:34 Opie overwhelming all at once so I've
43:38 gathered top experts in the field like
43:40 Bill Gates Sam Alman and more to help us
43:44 understand what it is this is crazy to
43:47 me this is the biggest technical advance
43:50 in my
43:51 lifetime Ai and the future of us an
43:54 Oprah infre special tonight an Oprah
43:56 infree special that if if you're in the
43:58 Mountain Time Zone go [ __ ]
44:00 [Laughter]
44:04 yourselves there's a pre-show in mods
44:08 FYI what's a
44:18 pre-show oh I see GPT 401 demo Hume
44:24 update new voice model oh yeah I saw
44:26 that I haven't played with it yet Gemini
44:29 live now available for everyone okay
44:31 yeah we can so we could demo in the next
44:35 20 minutes we could go look at the new
44:38 01 model I can't believe they called it
44:41 [ __ ]
44:42 01
44:44 gp40 and then
44:46 01 not 01 no
44:52 no Kyle want me to come up and share
44:56 some legal stuff I learned learned I
44:57 would love that yeah hop up here Ann
45:00 Murphy
45:06 um oh I want to show you also the new
45:10 notebook
45:11 LM so notebook LM is something I haven't
45:14 really talked about on this channel and
45:16 I it it need it needs to be talked about
45:19 now um I probably should have talked
45:22 about it a while ago but holy [ __ ] did
45:26 they just add a it a crazy new
45:29 feature um so yeah hop up here Ann
45:32 whenever you're
45:34 ready oh remind me how oh I don't know
45:38 you go
45:39 to
45:41 somebody can somebody tell an how to
45:44 click on the button which button where
45:46 it is cuz I
45:48 can't can you please volunteer to handle
45:51 their marketing oh can we please
45:53 volunteer to handle their marketing
45:54 honest to God seriously you've got a new
45:57 thinking
45:58 model you can't call it like GPT
46:02 cognition or GPT thinky
46:06 mington GPT or it's not even called GPT
46:10 it's just called
46:12 01
46:14 really
46:19 really right by the comment box ah there
46:21 she
46:25 is an muy
46:28 okay hey can you guys hear me uh wait my
46:31 volume's too low but go ahead yeah I can
46:33 okay I'm G to start driving in a second
46:35 so I'm just GNA be off camera is that
46:37 okay yeah that's fine okay so I just
46:40 finished Paul reter's um conference oh
46:44 yeah yeah it was amazing we had so many
46:49 like AI luminaries there and then a
46:52 thousand people who kind of spanned the
46:56 Spectrum it's all it's all marketers
46:58 some like agency people some like
47:02 contractors some like you know
47:05 entrepreneurs blah blah blah nice mix of
47:08 people y um but one of the sessions I
47:12 kind of hated because what we learned is
47:17 that the
47:19 copyright cases are definitely going to
47:22 go to the Supreme Court yep and because
47:27 of that and the additional scrutiny that
47:29 will go that will you know go on listen
47:33 to what we are supposed to do our very
47:36 own selves the recommendation because if
47:39 we ever get sued because believe it or
47:43 not even if we taught someone and they
47:47 go and do something
47:49 stupid um or if we if it's like one of
47:52 our clients or
47:55 ourselves listen to what we're supposed
47:57 to do we are supposed to be keeping
48:00 track of all of our search all of our
48:05 conversations with Chad
48:07 GPT so that we can prove that AI wasn't
48:11 part of it I mean sorry we can prove
48:14 what was Ai and what wasn't and what
48:17 human contribution yeah yeah yeah you
48:20 you know about this and that we're
48:22 supposed to be able to yes produce
48:27 why we did it how we did it yeah the
48:30 prompts and all of that that if we were
48:34 to like do it on the up and up so
48:37 everything everything you do you would
48:39 have to be keeping track of which just
48:41 absolutely blows my mind yeah no it's
48:44 it's absolutely ridiculous and it's and
48:47 and it's also just like I I've got this
48:49 new book coming out called collective
48:51 intelligence and like a lot of it was
48:53 written by AI but a lot of it wasn't
48:56 right and it's also like there like most
48:59 of the concepts in there are [ __ ] that
49:01 Kevin and I have been talking about for
49:03 a year right yes exact prompting right
49:07 and exactly exactly and so and there's
49:10 even there's even things like the um I
49:12 don't know if you saw the thing where
49:13 timeland like his his the the rapper his
49:17 his producer took a little beat that he
49:19 made and put it Ino so it started with a
49:22 beat that he made theno basically
49:23 finished the song like where's the
49:25 boundary like what that is and what of
49:28 that isn't like it's it's absolutely
49:30 ridiculous so I'm with you so I don't
49:33 know so that that was a little bit
49:36 stunning and made me feel like oh
49:38 [ __ ] great now I got to get like some
49:40 kind of attorney and blah blah
49:42 blah yeah I don't know I've got I've got
49:45 bad attitude on that because I'm just
49:47 like well I do just because so many
49:50 people are going to be using so much
49:52 stuff like and and so much of the work
49:55 is going to be hybrid human and Ai and
49:59 not just like single AI like multiple
50:02 systems are going to be used to go into
50:04 stuff right so it's like the ability to
50:07 track all that's like literally
50:09 impossible it's literally impossible to
50:12 do that at that point it would be so
50:14 cumbersome you'd have to be like okay
50:16 that's okay I guess I'm just going to go
50:18 back to doing things the old way yeah or
50:22 something um so that was a big learning
50:26 the other one wanted to share is that if
50:28 there's a lull in AI right now not one
50:32 single PE person of that thousand people
50:35 have any idea that there's a low like
50:38 they
50:40 are I mean they're marketers what was
50:42 what was the M what was the mix of you
50:45 know I I'm here and I don't really use
50:47 it but I think I might start to people
50:49 that are actively using it I assume that
50:51 conference it was a lot of people using
50:53 it but what yeah I would say there was
50:56 probably only like 5% of people who
50:58 aren't using it and then probably 30% of
51:02 people who are just like dinking around
51:04 and want to figure it out and then
51:07 everybody else is the the one of the
51:10 takeaways is that like the big big gap
51:13 is nobody's having it's not been easy
51:17 for anybody to go from like some of the
51:20 things you would do on an AI road map
51:22 and actually scaling it like all
51:24 companies are having a hard time scaling
51:26 it so that was interesting that makes
51:30 sense to me I don't think the tools are
51:32 ready to scale quite
51:34 frankly well that yeah that's part of it
51:36 is people get burned and then the you
51:40 know the people who make the stuff or
51:43 who are like the Brokers for the people
51:46 who make the stuff like it's their brand
51:48 reputation and stuff so they don't want
51:50 to give gadgets to people and be like
51:54 well let's just cross our fingers and
51:56 hope that everything works out okay but
51:58 the problem is is that like with
52:01 co-pilot in
52:04 Microsoft it's not even there's no
52:07 effort involved with scaling like it's
52:09 it's going to but there people won't
52:12 have any training they won't have any
52:16 governance like no one talked about
52:18 governance at all nobody not even a
52:21 single like hallway
52:24 conversation not a a Breakout out I mean
52:27 they had we had like I don't know like
52:30 there was like 60 speakers there and if
52:33 it was if it came up I would I mean it's
52:36 not it wasn't in the write up on any of
52:38 the sessions so that was
52:40 surprising what was did you have what
52:43 were there any sort of breakthrough
52:45 moments that you had or anything that
52:47 got you really excited because yeah yeah
52:50 thing I get but what what was on the
52:51 other side yeah okay
52:55 um one of the things that I was really
52:57 excited about was how excited everybody
52:59 everybody there was like that seemed so
53:02 promising everybody has all this pent up
53:05 energy and they want to do things and
53:08 that's really cool um Liz grenan so
53:12 Connor grenan interviewed Liz grenan and
53:15 she talked a lot about stakeholder trust
53:19 and that's a topic that I really care a
53:20 lot about and there were some like good
53:25 ideas about how you start working with
53:28 people like you would should really only
53:31 work with the people who are ready to be
53:33 on an adventure with you you shouldn't
53:35 work with anyone like the the some of
53:38 the advice was like if you're in a space
53:40 where people you have to teach people or
53:43 convince anybody just leave the room
53:46 like we're not doing that yeah this is
53:48 really good this is this is that that
53:50 article that I wrote rise of the agile
53:52 human that that the people that you want
53:54 to that you want to invest time into are
53:56 the on on that are curious and hungry
53:58 and excited and flex right you know
54:02 brave brave enough to break some [ __ ]
54:04 right like like yeah I I I that's that's
54:08 good reinforcement because yeah yeah
54:11 this stuff is to the the the all the
54:14 aite tech right now is too malleable to
54:17 to have people in there that want
54:19 predictable [ __ ] right yeah exactly
54:23 exactly the other thing I really loved
54:26 was Rachel Woods um was there and you
54:30 know I'm in her community and I've done
54:32 courses with her and all that kind of
54:34 stuff but she shared so much about how
54:40 you build your playbooks for AI and
54:43 using AI to do it so that basically what
54:47 you're doing is you're creating these
54:48 assets for your business where you're
54:50 taking these like groupings of tasks and
54:53 you're just like you know totally making
54:55 it routine and then you just um upgrade
54:59 it like incrementally as you go and then
55:01 the goal is of course like those become
55:03 the assets for your business and then
55:05 you can hand off one thing after another
55:08 after another to other people and get
55:11 yourself so you're only working on the
55:13 highest level stuff so she's doing like
55:16 a four week course on it soon I think
55:20 she's waiting to see how many people are
55:22 into it but um yeah so to teach us all
55:26 how to do that which I like for me
55:29 that's money in the bank yeah that's
55:31 really cool that reminds me so Paul rer
55:33 last week released um problem GPT he's
55:37 got like gpts now yes it's so good yeah
55:41 it seems that seems like it's in the
55:42 same neighborhood it so so so so good
55:46 and and uh the other thing he shared I
55:49 don't know if Source Camp is is in is on
55:52 the live she was earlier yeah I'm pretty
55:54 sure she's so this would be particularly
55:57 relevant to Source Camp she he gave us
56:01 well he gave the people who bought his
56:03 scaling AI course um a 29o assessment
56:07 tool where what you do is you go through
56:11 the company that is your client and you
56:13 learn about their business operations
56:15 and everything but each task you
56:17 determine what is the exposure to the
56:20 positive impact of AI so then you can H
56:24 you can prioritize things so you're like
56:26 okay this one this one you know has a
56:29 70% likelihood of being better with AI
56:34 maybe you know you go ahead and you do
56:36 that and so then you begin to be like
56:40 cherry-picking you know the the projects
56:43 that you work on so that the client gets
56:45 like value really quickly yeah um so
56:49 that was really cool and then I met a
56:51 bunch of people who solve a problem that
56:54 I don't have anyone to solve for which
56:55 is when people people's data is a hot
56:57 mess and you start going in there and
56:59 trying to do stuff and it's just like
57:02 you know I can't even help you because
57:04 your data is a hot mess so I I met so
57:07 many wonderful people there and like
57:10 especially collaborators so great I hope
57:13 you poed up the salon get him in the
57:16 salon oh I would love to do that he is
57:20 like he is such a cool cat who Paul rer
57:23 is yes yes yeah he's cool were you there
57:27 when he he spoke at the salon were you
57:29 there that night no he was in he was
57:31 already in the salon yeah well he spoke
57:34 he came and spoke at one of our one of
57:36 our meetings oh man well the other thing
57:40 I wanted to mention because people might
57:42 be interested in this kind of thing for
57:44 future people there were walking around
57:48 with rfps printed out to try to find
57:51 someone to train their teams to yeah to
57:56 build [ __ ] for them I mean literally I
57:58 walked away with people with rfps where
58:02 people had broken down like a Playbook
58:05 and said this is exactly what we want to
58:06 do all we all we need to know is how
58:08 much it costs and how how long it's
58:10 going to take wow that's amazing W so I
58:13 gave it to Josh already so Josh had you
58:16 know my friend Josh Houston he has it in
58:19 his hands
58:20 already and he so yeah I mean there's
58:24 money by going to these conferences like
58:27 oh my God I made so I met so many
58:29 potential clients and stuff it was so
58:31 cool well that's great that's really
58:33 great awesome wanted share Co I forget
58:38 how to I forget how to kick can you kick
58:40 me off so I go back down I don't know
58:42 how yeah I'll get you out of here this
58:44 this just leave an just
58:47 go hie thank you
58:53 bye all right
58:56 that was very cool I forgot she was at
58:58 that at that conference yeah that's
59:00 super cool um
59:03 awesome fantastic all right so we are
59:07 did you get my memo about the RFP cover
59:09 sheets we're gonna we're going to need
59:11 you to all right what was that
59:13 conference called please that was called
59:15 it was the uh uh makon maai marketing AI
59:19 Institute ma AI n and it's it's the the
59:23 marketing AI
59:25 Institute um
59:27 they have the annual their annual
59:29 shindig in
59:30 Cleveland Paul rater is the is the I
59:34 guess the CEO of that organization and
59:37 the main you know sponsor of that event
59:39 um I got to go get some more water so
59:41 entertain yourselves and then we'll be
59:43 back into
59:44 Opie we will do Opie I'll be right back
59:48 chiy everybody
1:00:25 after e
1:01:04 for
1:01:55 spe
1:02:25 e e
1:03:20 all right we're
1:03:23 back we're back
1:03:27 tonight on the AI learning lab Oprah
1:03:30 Winfrey AI for everyone you want
1:03:34 some all right what's happening
1:03:36 everybody what's shaking what's going
1:03:39 down um we got three minutes left so
1:03:42 I'll flip over
1:03:44 here time for the AI
1:03:47 party Jim Ross is in the house hey Jim
1:03:51 you're um I I invoked I invoked your
1:03:54 name and your story a fair amount
1:03:56 um in DC the the thing that you said
1:04:00 about you feel like you you're you're
1:04:02 operating like a 10p person company you
1:04:05 know as a solo preneur um that went over
1:04:08 really well like I tried a number of
1:04:10 different things of specific stories um
1:04:14 the two that went over the best were
1:04:15 that one and then the woman that I that
1:04:17 that I met on the plane who said she's
1:04:19 not going to she's not going to
1:04:22 incorporate AI into her business at all
1:04:24 because of the Colorado law even though
1:04:26 it doesn't go into effect until 2026 she
1:04:29 was just really pragmatic she's like I
1:04:30 don't want to put time into that if uh
1:04:34 you know if if I'm just going to get
1:04:37 regulated you know it's going to create
1:04:38 this regulatory burden I was like damn
1:04:41 laws actually have an
1:04:43 [Laughter]
1:04:47 impact so yeah I know it was really cool
1:04:50 it was very cool I told him I told him
1:04:53 the one guy that was really interested
1:04:55 when I said that he was like oh that's
1:04:56 amazing I said I said yeah he's he's got
1:04:59 this this practice where every morning
1:05:02 for an hour he uses AI for his business
1:05:04 he was like taking notes it was pretty
1:05:08 cool all right let's
1:05:11 see scripts we're doing okay let me let
1:05:14 me resize my window so that we got most
1:05:18 of the screen on for
1:05:20 y'all oh that's good we can just pop it
1:05:23 right
1:05:24 there and and
1:05:28 right do that oh that's good look at
1:05:32 that we're going to get full Opie full
1:05:36 Opie all right how are we going to know
1:05:40 how they going to know if they
1:05:41 incorporate AI into their
1:05:44 business
1:05:47 well what what she was saying is like
1:05:49 she doesn't even she doesn't know right
1:05:52 she's not using chat GPT enough to know
1:05:54 how she would incorporate it but what
1:05:56 she said to me is when they passed that
1:05:59 law she made a conscious decision to not
1:06:03 pursue it to not learn it to not create
1:06:06 an AI committee to learn how they might
1:06:09 do [ __ ]
1:06:11 better because she's like I don't want
1:06:13 to put time into something if it's just
1:06:15 going to create a headache down the
1:06:17 line was that the DC trip yeah that was
1:06:20 I I've been talking for the most of the
1:06:21 last hour about the the DC trip yeah no
1:06:24 I know it it was really
1:06:26 what was amazing was she and I were
1:06:28 talking about chat GPT for for a long
1:06:31 time um and then
1:06:35 um and then she uh once I talked to once
1:06:40 she found out I I testified against that
1:06:43 law she was like I didn't know where you
1:06:45 stood on that law so I didn't really say
1:06:46 anything but once she knew where I stood
1:06:48 on that law then she kind of admitted to
1:06:50 me that that was a choice she
1:06:53 made so bonkers bonkas bonkas bonkas all
1:07:04 right
1:07:09 Dion Oh let me cover up my keyboard
1:07:12 where's my paper where's my black
1:07:17 papier all right here she comes is this
1:07:21 her no that's an ad councel
1:07:27 thing on social media the first thing
1:07:31 that kind of comes to mind is just like
1:07:33 the Terminator movies we've always been
1:07:36 interacting with some sort of artificial
1:07:39 intelligence it's just much more in our
1:07:40 face right now what scares me about AI
1:07:44 is that the technology seems to be
1:07:47 advancing so quickly one day we weren't
1:07:50 talking about it and then the next day
1:07:52 it wasn't just that we were talking
1:07:54 about it it was there for to use without
1:07:58 really even understanding it AI is
1:08:00 happening it's going to happen the
1:08:03 question I would have for anyone who has
1:08:06 power over the development is how are we
1:08:08 going to make this work for us I don't
1:08:10 know what the agenda is with the people
1:08:13 who's even creating AI to me it's like
1:08:15 what is really the point why do we need
1:08:19 AI there you
1:08:21 go yeah this is very much designed
1:08:24 for introducing this to the world
1:08:27 fascinating I've been living on this
1:08:30 planet for seven decades I've seen a lot
1:08:34 of things change but about a year ago I
1:08:37 saw the first change I didn't fully
1:08:40 comprehend it was a demonstration of the
1:08:43 capabilities of artificial intelligence
1:08:46 and I was
1:08:47 bedazzled fascinated curious and also
1:08:52 concerned but there's no doubt it is
1:08:55 here artificial intelligence is here and
1:08:59 life for all of us is about to be very
1:09:03 different over the next hour we're going
1:09:05 to talk about what is coming with a
1:09:08 brilliant group of people they have
1:09:10 different backgrounds and opinions about
1:09:12 the kind of future racing toward us but
1:09:15 they all agree the stakes are too high
1:09:20 for anyone to sit this out now if you're
1:09:23 worried this is going to be too
1:09:24 technical I promise it is easier than
1:09:26 you think yep that's great this is the
1:09:29 notion a computer that works like the
1:09:31 human brain has been around since the
1:09:32 1950s this is good the first chatbot
1:09:35 arrived in
1:09:36 1966 but computers at the time weren't
1:09:39 powerful enough to really
1:09:41 engage well it's been around for decades
1:09:44 forever hooked but but it hasn't been
1:09:46 accessible to to Everyday People until
1:09:50 data to learn from most recently a group
1:09:54 of computer scientists taught machine
1:09:56 there's the Transformer find and process
1:09:59 words more easily so they can understand
1:10:02 how we naturally communicate that new
1:10:05 discovery is powered by these computer
1:10:07 chips called
1:10:09 gpus combine smarter programming with
1:10:12 more powerful chips and you get AI
1:10:15 products like chat GPT there's your 2022
1:10:18 date that's the beginning of the
1:10:19 revolution in my opinion so far beyond
1:10:21 our capabilities that it can seem like
1:10:24 reality itself is
1:10:28 warping Sam Alman is the co-founder of
1:10:32 open AI which makes chat GPT I promise
1:10:36 everybody watching that this was not
1:10:38 going to be so technical that they
1:10:40 wouldn't understand you will do it all
1:10:42 right okay what is artificial
1:10:47 intelligence it's really the next
1:10:49 chapter of computing we have figured out
1:10:52 how to make computers smarter to
1:10:54 understand more to be able to be more
1:10:56 uh intuitive and more useful can you
1:10:58 tell us in basic terms how chat GPT
1:11:01 works you know at the most basic level
1:11:04 we are showing the system a thousand
1:11:06 words in a sequence and asking it to
1:11:09 predict what word comes next and doing
1:11:11 that again and again and again so we've
1:11:13 seen this on our phones where it
1:11:15 predicts you know what the next word's
1:11:17 going to be exactly but to do that as
1:11:19 well as possible think how much you have
1:11:22 to understand about each word that comes
1:11:24 before and the relationships every word
1:11:26 and every word yes and by teaching the
1:11:29 model to look at a bunch of text and
1:11:31 predict what word comes next and doing
1:11:33 this incredible scale again and again
1:11:36 the system learns to predict and then in
1:11:39 there it learns the underlying concept
1:11:40 so that's what's going on behind the
1:11:42 screen yes how fast is this
1:11:45 moving well four years ago most of the
1:11:48 world if they thought about AI thought
1:11:49 about self-driving cars or some other
1:11:52 thing or whatever it was only 2022
1:11:56 was the first time people said okay this
1:11:58 GPT thing like this is computer talks to
1:12:01 me now that's new and then since then if
1:12:03 you look at how much better it's gotten
1:12:05 it's been a pretty steep rate of
1:12:07 improvement just take me to
1:12:11 2035 um I think it's hard to say where
1:12:14 all this can go without sounding like a
1:12:16 crazy
1:12:17 person I really believe the technology
1:12:20 is going to be that
1:12:22 powerful I hope that in 10 years the
1:12:25 dramatic music of us we are limited by
1:12:26 what we can
1:12:27 imagine we have this AI that can help us
1:12:30 figure out anything we have abundant
1:12:32 energy that can help us like reconfigure
1:12:34 stuff in the world and if you want to go
1:12:37 build like a beautiful house you can
1:12:40 dream with the AI you can figure out oh
1:12:42 I want it to look like this or do you
1:12:43 have any other ideas and architect this
1:12:45 amazing thing and then can you invent
1:12:48 some robots that can that can actually
1:12:50 build this this house that we just
1:12:52 imagined
1:12:53 together so the machine that just
1:12:55 becomes a part of your family the way
1:12:57 the phone is a part of our Lives now it
1:12:59 becomes a tool for you to do whatever
1:13:01 you want for your family do you think
1:13:04 it's moving too
1:13:06 fast my current belief is no but I could
1:13:09 imagine things that accelerating would
1:13:12 change that quickly and I would say okay
1:13:13 this is moving too fast like what what
1:13:15 would change that if we made an AI
1:13:18 system that was better at doing AI
1:13:20 research then all of open
1:13:22 Ai and you could push a button and just
1:13:24 say improve really fast
1:13:26 that's an example this I think we be
1:13:27 moving too fast if we're talking about
1:13:30 AI benefiting everyone the industry is
1:13:34 dominated by a lot of people who look
1:13:37 like you male and white and very few who
1:13:40 look like me bias this thoughts about to
1:13:43 address that yeah obviously we want
1:13:46 everybody to see themselves in our
1:13:48 products uh we also want the industry
1:13:51 Workforce to be much more diverse than
1:13:53 it is um and there's slower than we'd
1:13:56 like progress but there is progress
1:13:58 there the most important thing that will
1:13:59 happen here is everyone in the world
1:14:03 will be able to create at a level that
1:14:05 is still hard for us to imagine this is
1:14:07 going to be an enabler of human ability
1:14:12 to create to flourish to make new things
1:14:14 to create new companies and services
1:14:16 like we've never seen and we want
1:14:18 everybody to get to do that not just the
1:14:20 white dudes that's good yeah what keeps
1:14:23 you up at night
1:14:27 the rate of change in the
1:14:29 world from this we haven't seen it yet
1:14:33 we'll do because we haven't seen that
1:14:34 yet doesn't mean we can get
1:14:37 complacent let's say we can make the
1:14:38 technology super safe we can't kind of
1:14:41 do anything about the rate of diffusion
1:14:43 through the economy like people want
1:14:45 more and better and there's sort of some
1:14:47 strong forces at play here if you were
1:14:50 in charge of
1:14:53 regulations what kinds of regulations
1:14:56 what kinds of boundaries what kind of
1:14:58 restrictions do you believe should be
1:15:00 put in
1:15:03 place a partnership between the
1:15:05 companies developing this technology and
1:15:07 governments is really important one of
1:15:09 the first things to do and this is now
1:15:11 happening is to get the governments to
1:15:13 start figuring out how to do safety
1:15:15 testing at these systems like we do for
1:15:17 aircraft or you know new medicines or
1:15:19 things like that and then I
1:15:22 thinkr now we'll have an easier time
1:15:24 figuring out exact what the regulatory
1:15:26 framework is
1:15:28 later who do you speak with in the
1:15:30 federal government we speak with a lot
1:15:32 of people in the executive branch um and
1:15:34 then uh dozens of people fairly
1:15:36 regularly in Congress um there's a new
1:15:39 AI safety Institute in the US that we
1:15:41 work with I personally probably have
1:15:43 some conversation with someone in the
1:15:45 government every few days every few days
1:15:50 wow and what are y'all talking
1:15:52 about what the US needs to do to lead
1:15:55 here
1:15:56 how we're going to build enough data
1:15:57 centers how we're going to deal with
1:15:58 chips and where those chips are produced
1:16:00 and make sure politically we have a good
1:16:02 strategy there how to think about safety
1:16:05 for AI and how the government should
1:16:06 test it how to think about collaboration
1:16:08 with other governments um what the
1:16:11 economic impact that this is going to
1:16:13 look like how the government should be
1:16:14 using this all that the bottom line for
1:16:18 I think most people is whether or not
1:16:21 you can trust the people who are in
1:16:24 charge the the um I know you went
1:16:27 through a phase where your own board
1:16:29 said we don't trust you and then you got
1:16:33 yourself another board how do we know we
1:16:36 can trust you
1:16:38 so the bar on this is clearly extremely
1:16:42 high um the best thing that we can do is
1:16:45 to put this technology in the hands of
1:16:47 people talk about what it is capable of
1:16:50 what it's not what we think is going to
1:16:51 come what we think might come give our
1:16:56 best advice about how Society should
1:16:59 decide to use this um say when we think
1:17:02 it's important to not release something
1:17:05 which we also might get wrong and build
1:17:07 up that trust over time but it is clear
1:17:11 that this is going to be a very
1:17:13 impactful technology and I think a lot
1:17:15 of scrutiny is thus super warranted I
1:17:18 actually saw a headline that said you
1:17:21 were the most powerful and perhaps most
1:17:24 dangerous
1:17:25 man on the planet and I'm wondering how
1:17:29 that sits with
1:17:30 you
1:17:34 I it's it's definitely strange to hear
1:17:37 you say that it doesn't I don't feel
1:17:39 like the most powerful person or
1:17:41 anything even close to that like I feel
1:17:43 the opportunity responsibility in a
1:17:45 positive way to get to nudge this in a
1:17:49 direction that I think can be really
1:17:51 good for people and that is like that's
1:17:53 like a serious
1:17:59 exciting somewhat nerve-wracking thing
1:18:02 um but it's something that I feel very
1:18:04 deeply and I and I and I and I realize I
1:18:06 will never get to touch anything it's
1:18:08 important started 10 minutes ago just
1:18:10 started just what we seen thus far of
1:18:13 people with nefarious ideas misusing it
1:18:19 um causes me a lot of concern and yet
1:18:21 when I sit with you you're so
1:18:23 calm you're so
1:18:26 relatable it seems like everything's
1:18:29 okay is that how you really feel I don't
1:18:31 want to give a false sense of security
1:18:34 here uh I think I'm I'm tremendously
1:18:37 excited for the good I think the good
1:18:38 will be incredible but there will will
1:18:40 be bad too and we'll mitigate as much as
1:18:43 we can we the whole field um but this
1:18:46 technology will be misused like every
1:18:48 other technology before it and we need
1:18:52 to be cleare eyed about that even though
1:18:55 even though I'm so convinced that the
1:18:56 upside will be so tremendous and I think
1:18:59 we're all so excited about it um we do
1:19:02 have to be responsible for the whole
1:19:06 [Music]
1:19:08 package yeah really
1:19:11 interesting really interesting I mean
1:19:13 you know listen he he's this is a you
1:19:16 know this is a this is an edited piece
1:19:19 right um I would assume
1:19:23 that this this was likely architected by
1:19:26 the pr team at open AI is my guess
1:19:30 um you know not quite a puff piece but
1:19:34 pretty close to it so far um this but
1:19:37 this is
1:19:39 definitely this is definitely feeling
1:19:42 designed for the people that haven't
1:19:44 heard of chat GPT did you notice at the
1:19:46 beginning when she said you know Sam
1:19:47 Alman the the you know from open AI the
1:19:50 creator of chat
1:19:52 GPT she said it like she was talking to
1:19:55 second graders right
1:19:58 um this is definitely designed for that
1:20:02 the the downside of it so far like while
1:20:06 while he's explained it okay it still
1:20:09 seems like nebulous like it's a machine
1:20:13 like when he talked about it's a thing
1:20:16 that predicts the next word people don't
1:20:18 know what the [ __ ] that means if he said
1:20:21 you ask it a question and it answers
1:20:24 you uh
1:20:26 like it feels like he could have been a
1:20:28 bit more clear about what it actually
1:20:30 does right like like I like if if people
1:20:34 didn't know what chat GPT was the fact
1:20:36 that it predicts the next word over and
1:20:38 over and over again what does that
1:20:40 [ __ ]
1:20:41 mean well you know what that means is it
1:20:44 can write you a poem it can write you
1:20:46 you know a chapter of a book it can
1:20:48 write you a computer program oh and it
1:20:51 can just write it you know that like
1:20:53 that that's a bit missing right now but
1:20:55 we'll see I assume who who she probably
1:20:58 goes to next is Bill Gates so what Bill
1:21:01 Gates does for
1:21:03 this is you know because he was sort of
1:21:07 the the poster child with Steve Jobs for
1:21:10 the for the opening of the
1:21:14 um for the beginning of the PC and then
1:21:17 when the internet came along you know
1:21:20 Microsoft was really slow but when they
1:21:21 decided to go they went right and so and
1:21:25 then and then because of the Gates
1:21:26 Foundation he's got a Global Perspective
1:21:28 so he's got Gates has got the ability
1:21:32 to I I'm sure he'll talk about the first
1:21:34 time he used chat first time he saw chat
1:21:37 GPT I think Sam Al he Sam Alman came to
1:21:40 Bill Gates's house and showed it to him
1:21:43 he'll probably talk about that and talk
1:21:44 about how significant it is so you know
1:21:48 Sam's Sam's there to be the calm voice
1:21:50 of it's going to be okay people the
1:21:52 robots aren't going to kill you and I
1:21:54 think he he does that pretty well and
1:21:57 then um and then Bill Gates is going to
1:22:00 say this is important and it'll it'll be
1:22:02 interesting to see what the other what
1:22:03 the other folks
1:22:07 do um by the way if you're wondering you
1:22:10 know where this is so I'm in Denver so
1:22:12 we're an hour delayed so this already
1:22:14 happened on the East
1:22:16 Coast so we just started at the top of
1:22:19 the hour so we're 15 minutes into this
1:22:22 into this show
1:22:27 I use perplexity a lot okay she's
1:22:29 definitely asking alarmist questions
1:22:32 yeah I mean
1:22:34 it's I I mean it's the media right like
1:22:38 I I mean even when when I was in DC like
1:22:41 a lot of what I talked about was you
1:22:43 know I think part of what we're swimming
1:22:45 against here is Hollywood has told us
1:22:47 for 50 years that the robots are going
1:22:49 to kill us and every time I said that
1:22:51 they were like yeah
1:22:53 yeah um
1:22:56 I asked three people the direct question
1:22:59 do you use chat
1:23:01 GPT no one is no one is it's back on
1:23:05 well it's not back on here oh yeah it
1:23:08 is sorry about
1:23:11 that holy [ __ ] two holy [ __ ] oh Maris
1:23:15 browny marz brownley have 20 million
1:23:18 subscribers who all rely on you for all
1:23:22 things Technical and this is a moment
1:23:25 we're having tell us what's going on
1:23:28 yeah look I I've been in the tech World
1:23:30 a while and a big part of being in this
1:23:32 world is trying to figure out when this
1:23:34 new thing is a big deal YouTuber Tech
1:23:37 expert I can say confidently it does
1:23:39 feel like a big deal I think it's a very
1:23:41 big deal yeah I want to show you what an
1:23:43 AI generated video one year ago looked
1:23:47 like oh this is great okay and it's this
1:23:49 this is great this is the rock eating
1:23:51 rocks okay it's kind of a meme it's it's
1:23:54 a pretty funny video it's obviously not
1:23:55 a real video obviously the a
1:23:58 generated now you can tell that and you
1:24:01 look at that and you say okay that's a
1:24:02 pretty ridiculous looking video yeah
1:24:04 this was one year ago yeah right now now
1:24:09 this is what AI generated videos look
1:24:11 like what do you mean this is not a real
1:24:14 person not a real scene no no real
1:24:17 cameras were used this is completely
1:24:20 generated by an AI generated video tool
1:24:24 now you can still kind of look at pieces
1:24:26 of this and something's not quite right
1:24:29 no I can't but that looks like a
1:24:32 video uh here's another one this one's a
1:24:35 little more interesting it's a a
1:24:37 handheld cinematic shot of a child on a
1:24:39 beach holding an ice cream con that's
1:24:41 what you ask for that's what you typed
1:24:42 in that's what I typed in I'm assuming
1:24:44 that AI had dist stored somewhere the
1:24:47 picture of the Child AI has millions of
1:24:52 photos and videos of millions of
1:24:55 children stored that it's trained on
1:24:57 okay are they real children is this a
1:24:59 real child it's not a real child but it
1:25:02 looks like one because it's trained on a
1:25:04 ton of real
1:25:05 children Is that real ice cream sure it
1:25:09 looks like it so and this is again like
1:25:11 if you pay super care that's really it's
1:25:13 really
1:25:16 interesting she what she's asking right
1:25:19 there is basically is this thing copying
1:25:22 and pasting real people like is this
1:25:24 compositing real people together she's
1:25:27 not quite grocking yet that this is
1:25:31 generating a original content out of all
1:25:34 of the training data which is that's
1:25:36 where everyone's going to be with this
1:25:38 right everyone everyone that hasn't used
1:25:41 this and hasn't really looked into it is
1:25:42 thinking oh it's copying and pasting
1:25:44 other people's work it's not how it
1:25:46 works so that's that's really
1:25:49 fascinating AR really any rules to it so
1:25:52 everyone who makes tools like this of is
1:25:55 coming up with their own rules on the
1:25:56 Fly oh I'll add a watermark to the
1:25:59 corner or I'll add some metadata to the
1:26:02 file so if someone downloads it then
1:26:04 they we supposed to know where to look
1:26:06 for a watermark it's a couple layers
1:26:08 deep so the point is there isn't really
1:26:09 any good set of rules yet I know this
1:26:13 has happened to you it's happened to me
1:26:15 is to have people take an image of you
1:26:18 put words into your mouth that you
1:26:19 didn't say you're selling things you
1:26:21 don't even know you're selling I mean I
1:26:22 can't even keep up with all the things
1:26:24 I'm about to show you actually is a
1:26:26 voice generator yeah this is actually a
1:26:28 real one right now that's trained on my
1:26:30 last video yes and you can ask it to say
1:26:33 whatever you want and it will sound just
1:26:35 like me is there anything you would
1:26:37 you'd like it to say I'd like to say
1:26:39 Peter Piper picked a pack of pickled
1:26:42 peppers Peter Piper picked a pack of
1:26:46 pickled peppers Peppers let's see if we
1:26:48 can do it okay generate speech 11 labs
1:26:51 to the untrained ear this sounds
1:26:53 basically exactly like a recording of my
1:26:55 voice and even to the trained ear it's
1:26:58 pretty convincing Peter Piper picked a
1:27:00 pack of pickled
1:27:03 peppers oh my God so in a matter of
1:27:07 seconds this tool is generating really
1:27:11 high quality recording level this is
1:27:14 good this isn't good so that's what I
1:27:16 mean so yeah she's very much in the
1:27:18 alarmist place fasinating but it leaves
1:27:22 me a little bit more concerned
1:27:25 you don't seem to be as
1:27:27 concerned I'll say I've seen this story
1:27:31 before cars in general went through this
1:27:34 smartphones of course went through this
1:27:35 computers went through this the internet
1:27:37 went through this where at their very
1:27:39 formative years they're really really
1:27:42 confusing and potentially amazing but
1:27:44 potentially horrible 80% of users just
1:27:47 swore off AI forever humans sort
1:27:50 themselves out to lay the rules down in
1:27:53 a way that we can actually trust that
1:27:55 will be better than it is bad so I've
1:27:58 seen the story before and I'm hoping it
1:28:00 plays out the way it's played out
1:28:02 before thank you no problem next we have
1:28:05 Tristan and ASA who here to talk about
1:28:08 the incredible speed in which this is
1:28:11 happening it's moving really fast um we
1:28:14 had experience with another technology
1:28:16 that was introduced a little bit
1:28:17 recklessly and that was social media and
1:28:20 we're here because we don't want that to
1:28:21 happen again with AI what was your
1:28:23 experience I made infinite scroll you
1:28:26 know that thing where you scroll and the
1:28:28 phone just keeps loading more I invented
1:28:30 it before social media and I invented it
1:28:33 to make people more efficient and it's
1:28:36 now used to waste over half a million
1:28:39 human lifetimes every single month
1:28:42 addiction Doom scrolling polarization
1:28:45 these are all consequences and what I
1:28:48 learned the hard way is that my good
1:28:50 intentions as an inventor just wasn't
1:28:53 enough and is moving way faster and is
1:28:56 way more powerful in its impact on
1:28:58 society than social media I thought
1:29:01 marquez's example uh actually caused me
1:29:03 to have an aha when he was talking about
1:29:07 all technology has gone through a
1:29:09 similar phase where people question he
1:29:12 invented infinite scrolling so he
1:29:15 invented you know how you scroll and it
1:29:17 loads the the extra things at the bottom
1:29:20 so so you can do that
1:29:24 well you say what to that it took time
1:29:28 it took decades to go from the invention
1:29:31 of the car to adding seat bels and
1:29:34 driver's licenses and speed limits
1:29:37 that's right we had to figure it out so
1:29:39 the problem with the AI is that we don't
1:29:42 get that time so explain to me why AI is
1:29:47 now moving so fast in 2017 there was a
1:29:51 huge breakthrough at Google and research
1:29:54 Searchers there discovered a way that AI
1:29:57 could start to learn a kind of universal
1:30:00 language where it could speak the
1:30:02 language of anything and then suddenly
1:30:05 the same lines of code could learn the
1:30:08 language of DNA and learn how to make
1:30:10 drugs could learn the language of images
1:30:12 and could learn to do all those things
1:30:14 that Marquez was showing and so the rate
1:30:16 of innovation went through the roof and
1:30:19 of course once AI can do anything people
1:30:22 start to use it to do everything that's
1:30:24 what's so confusing about AI is it's
1:30:27 going to give us all these benefits at
1:30:29 the same time that it's undermining the
1:30:30 foundations that we depend on and you
1:30:33 can't separate the promise from the
1:30:35 Peril so I think instead of getting
1:30:37 caught up in the question is AI going to
1:30:39 be good or is it going to be bad it's
1:30:41 about is it happening at a pace that our
1:30:44 society can respond appropriately but
1:30:46 guys it's that that that's really
1:30:48 important I'm interested to see what she
1:30:49 says here but it's like that the whole
1:30:51 point of this channel is to judge
1:30:55 whether it's good or bad I don't give
1:30:57 too much of a [ __ ] about whether it's
1:30:58 good or bad what I care about is it's
1:31:01 [ __ ] happening whether we want it to
1:31:03 happen or not and what this channel is
1:31:06 about is like if it's happening whether
1:31:08 you want it or not don't have it
1:31:10 happened to
1:31:12 you like have you happened to it
1:31:17 right can respond appropriately but guys
1:31:20 it's here that's why I started this
1:31:22 program by saying it is here so I I the
1:31:25 argument isn't whether or not it's going
1:31:28 to continue or whether the speed's going
1:31:30 to continue it it because it is already
1:31:32 here it's already here some of it's here
1:31:34 but even more powerful forms of AI are
1:31:36 coming open ai's stated goal is to be
1:31:39 able to build systems of AI that are
1:31:41 even smarter than humans because there's
1:31:43 trillions of dollars to be made on the
1:31:45 other side and so now is the time for us
1:31:48 to say what are the choices that we want
1:31:50 to make about how this technology comes
1:31:52 out what do you want to see happen now
1:31:56 open AI Google Microsoft they need to be
1:32:00 for the har they would create so that
1:32:02 they're all incentivized to pull back
1:32:04 and go at the pace that they can help
1:32:06 Society be prepared and not be
1:32:08 overwhelmed and who is holding them
1:32:09 accountable we need
1:32:12 loss yeah interesting interesting
1:32:17 interesting
1:32:19 um yeah this is definitely
1:32:27 what what they haven't quite done yet
1:32:30 and they may not right this is this is
1:32:32 mass media what they've successfully
1:32:35 done is said holy [ __ ] it's coming it
1:32:38 could do really bad [ __ ] nothing you can
1:32:41 do about it is essentially what the
1:32:44 message has been so
1:32:45 far um what they haven't done is kind of
1:32:50 the the the optimistic view of well
1:32:53 here's the good things it's done here's
1:32:54 the good things it's doing here's the
1:32:57 accessibility it's providing to people
1:33:00 um you know here are people that are
1:33:03 using that video tool in in positive
1:33:06 ways um things like that so the AI
1:33:09 version of when sharks attack yeah this
1:33:11 is this is Shark Week AI shark
1:33:14 [Laughter]
1:33:17 week uh you know what's actually
1:33:19 fascinating
1:33:22 is I wonder if viewers ship dropped off
1:33:25 after the after he showed the voice
1:33:28 thing and she kind of freaked out a
1:33:29 little bit shark NATO yeah exactly AI
1:33:33 shark
1:33:34 NATO
1:33:40 fascinating I just
1:33:43 saw an
1:33:47 email let's
1:33:50 see where's it posted
1:34:01 where am there I
1:34:03 am um it looks like the um the AI Salon
1:34:08 got um we sent out a press release today
1:34:12 from my trip to Washington and it looks
1:34:14 like we got picked up in a political
1:34:15 article so I I'll look for that in a
1:34:18 second actually maybe I wonder if it's
1:34:19 over here
1:34:41 [Music]
1:34:53 but
1:34:57 D people love a good train
1:35:02 wreck I bet you can fast forward since
1:35:05 you paused yeah I probably can
1:35:07 [Music]
1:35:09 boom yeah it's back I've been in a
1:35:13 terrible accident and I am so so sorry
1:35:16 and a man gets on the phone and he says
1:35:20 is this chast's mother she interrupted
1:35:24 deal that I was doing and I lost $900
1:35:27 because of her so I want my $900 back or
1:35:31 I'm going to kill her so I'm driving
1:35:34 like 100 mph down the interstate and
1:35:37 just as I got ready to get off the
1:35:38 interstate a text come in and it was my
1:35:41 daughter and she says mom what is going
1:35:44 on I'm at the doctor's office my
1:35:47 immediate reaction was how the heck did
1:35:49 they get my voice um how did they know I
1:35:51 call my mom mama I'll tell you what I I
1:35:54 learn from it I will not ever take
1:35:57 another phone call unless they're in my
1:35:59 contact
1:36:00 list wow more scary AI is going to kill
1:36:04 you that's robots are going to kill us
1:36:06 bad guys okay this this is this is the
1:36:09 robots are going
1:36:10 to christop want talk about threats from
1:36:15 Ai and other aspects of law in order do
1:36:19 you remember the first time that someone
1:36:21 showed you evidence of AI being used to
1:36:26 commit a crime what was your reaction
1:36:28 yeah if I if I said a bad word we might
1:36:31 get rate limited or they might kick long
1:36:33 time even before I became FBI director
1:36:35 but one of the first memories I have of
1:36:37 dealing with it in this job uh was I was
1:36:41 in a conference room and a bunch of our
1:36:43 folks got together to show me how AI
1:36:46 enhanced deep fakes can be created and
1:36:49 they had created a video of me saying I
1:36:54 had never said before and would never
1:36:57 say uh and I was staring at this video
1:36:59 of myself and I found it incredibly
1:37:02 convincing and believe me it caught my
1:37:04 attention I kep say wait that's not me I
1:37:06 never said that what is this you know oh
1:37:09 the lady on the video right now ai is
1:37:11 most useful in taking what I would call
1:37:14 JV criminals and making them
1:37:19 varsity and that's a problem sure we had
1:37:22 a a case not that long long ago with a
1:37:25 child psychiatrist in North Carolina who
1:37:28 took images that he was able to get
1:37:31 online and then manipulate them using AI
1:37:34 to make them into child pornography one
1:37:37 of the things that made this particular
1:37:39 example uh so egregious and horrifying
1:37:43 is that the child psychiatrist took
1:37:46 photos from decades earlier from first
1:37:49 day of school a school dance that kind
1:37:52 of thing and essentially was able to
1:37:54 manipulate those images to show them uh
1:37:58 naked
1:38:00 wow there's so much out there now on
1:38:03 social media and on the internet
1:38:05 well-intentioned things that people post
1:38:07 about their lives but in the wrong hands
1:38:11 enhanced by
1:38:13 AI people are
1:38:16 vulnerable so you said more and more
1:38:19 types of crimes so that would be one
1:38:21 another one is something called
1:38:22 sextortion which more and more often
1:38:25 somebody posing as a peer targets uh a
1:38:30 teenager typically more and more it's
1:38:32 young boys so you're a teenage boy you
1:38:36 get a reach out from someone who claims
1:38:39 to be a friend of a friend she sends
1:38:42 compromising pictures of herself and
1:38:45 invites you to send pictures back except
1:38:48 it's not a girl those images are
1:38:51 enhanced and in fact it's some guy
1:38:53 behind the keyboard board in Nigeria and
1:38:55 then once they have the images they
1:38:57 threaten to Blackmail the kid and say if
1:38:59 you don't pay up we're going to share
1:39:02 these images will ruin your life
1:39:04 unfortunately this has real word
1:39:06 consequences there have been at least 20
1:39:09 teenagers who committed suicide in these
1:39:12 situations and it's just
1:39:15 heartbreaking do you think we have the
1:39:18 laws or regulations in place now to keep
1:39:24 Americans safe from all that's coming
1:39:27 all the AI threats I would probably
1:39:29 leave legislating to the legislators but
1:39:32 what I will tell you from an FBI
1:39:33 director's perspective um is that this
1:39:37 is a type of technology that we see
1:39:40 manifesting itself in more and more
1:39:43 situations more and more types of crimes
1:39:45 more and more types of threats uh and
1:39:47 there's a degree to which over not up to
1:39:49 speed that can be done without a with
1:39:52 technology
1:39:54 you sure yeah sure what that's a good
1:39:57 point I mean
1:39:58 listen what what AI does both good and
1:40:01 bad
1:40:04 is while yes you can do all of what he
1:40:07 just described without AI you have to
1:40:09 have pretty significant technical skills
1:40:13 to do it with AI it democra so AI
1:40:18 democratizes all
1:40:20 skills right and it's going to
1:40:22 increasingly democ democratize in you
1:40:25 know increasingly sophisticated skills
1:40:27 are going to get democratized so
1:40:29 everyday people can now do things he he
1:40:31 talked about like going from junior
1:40:33 varsity to Varsity that's the thing
1:40:36 that's happening with AI is more and
1:40:37 more people are going to be able to do
1:40:39 this
1:40:46 stuff and the elite bad guys are doing
1:40:48 what so the elite bad guys are the ones
1:40:51 you know for example conducting the most
1:40:53 phisticated cyber
1:40:56 intrusions to me no country no country
1:41:00 represents a broader more severe more
1:41:03 comprehensive threat to American
1:41:06 innovation American ideas so this stuff
1:41:09 they're showing here like this is this
1:41:11 is relatively old convolutional Network
1:41:15 Technology so they're showing China here
1:41:16 they're basically saying China's using
1:41:19 AI to watch all of their people and
1:41:21 we're going to be able to do that here
1:41:24 we are already doing that here right
1:41:26 like this is old technology what things
1:41:30 like segment anything from meta does is
1:41:33 it makes that technology available to
1:41:35 everyone else you can Rel with
1:41:37 relatively you know trivial amount of
1:41:40 programming skills and now with
1:41:41 something like repet um with their
1:41:43 autonomous agent where you can create an
1:41:45 application like I did last week without
1:41:47 any programming skills individuals are
1:41:50 going to be able to do this this kind of
1:41:52 sophisticated you know person tracking
1:41:54 and identification and things like
1:41:58 that American ideas to our Economic
1:42:01 Security and ultimately our national
1:42:03 security than the Chinese
1:42:05 government China has a bigger hacking
1:42:08 program
1:42:09 already than that of every major Nation
1:42:13 combined and has stolen more of
1:42:16 American's personal and corporate data
1:42:19 than every nation our data is already
1:42:21 gone it's long gone the scale of the
1:42:23 threat is significant if you took the
1:42:27 FBI's program and just said forget
1:42:29 Russia forget Iran just do nothing but
1:42:32 China the Chinese government's hacking
1:42:35 program they would outnumber us 50
1:42:37 to1 does that mean the race that we all
1:42:40 feel that we're in now the race to be
1:42:44 first to be better is to our
1:42:48 advantage I do think we need to win the
1:42:51 race I think we are ahead in the race
1:42:53 but I would also say that we shouldn't
1:42:56 in the course of trying to win the race
1:42:59 disregard the importance of building in
1:43:01 safeguards and controls and that sort of
1:43:03 thing to help protect abuse so speed yes
1:43:08 recklessness
1:43:10 no we've got an election coming do you
1:43:14 think this election cycle will be
1:43:19 compromised by
1:43:22 disinformation I fully expect to see um
1:43:27 disinformation at by Foreign
1:43:30 adversaries the the 50x behind thing was
1:43:33 just on the on the hacking um so I think
1:43:36 he's saying we're ahead in the in the
1:43:38 development of the core technology but I
1:43:40 don't know and I you know if you look at
1:43:42 what China's releasing with their video
1:43:44 generation tools there's there's three
1:43:46 major video generation tools from China
1:43:48 right now that are at least in the
1:43:50 neighborhood of Sora so so I don't think
1:43:53 we actually know where we stand relative
1:43:56 to China right
1:43:58 now finding all too often that something
1:44:02 on social media that looks like Bill
1:44:04 from Topeka or Mary from Dayton is
1:44:06 actually you know some Russian or
1:44:09 Chinese intelligence officer on the
1:44:11 outskirts of Beijing or Moscow what
1:44:14 should we be on the lookout for all
1:44:16 Americans should try to be you know more
1:44:19 Discerning consumers of information it
1:44:21 is incumbent on everyone in America to
1:44:24 bring an intensified name the video AI
1:44:27 China has they've got cling there's one
1:44:30 that came out last week and then there's
1:44:32 um cing is a really good one cing uh
1:44:35 it's cing clling ai.com I think or king.
1:44:39 just look up cing that's the one that's
1:44:41 farthest along um but then one came out
1:44:44 last week and then another one came out
1:44:45 like two weeks ago or three weeks ago so
1:44:48 there there's a number of them out there
1:44:49 now um some of them you can't sign up
1:44:51 with if you don't have a Chinese um
1:44:54 phone um cing made it possible I think
1:44:57 it's clling ai.com you can sign up with
1:45:01 an American email for that one then you
1:45:03 can use
1:45:04 it
1:45:06 caution to uh the use of AI and how AI
1:45:10 can be used by bad guys against all of
1:45:12 us but not to panic I don't think this
1:45:14 is a time for
1:45:18 panic let's talk for 15 minutes about
1:45:20 all the reasons you should panic and
1:45:22 then we'll have Christopher say this
1:45:23 isn't a time for
1:45:26 panic that's radiation or have
1:45:30 system ah TV isn't media the
1:45:39 best they're watching you they're coming
1:45:42 to get you China's better than us except
1:45:44 we're better than China because
1:45:45 America's number one
1:45:48 uh but but you know you know what you
1:45:51 shouldn't do you shouldn't worry about
1:45:52 the robots coming to kill you just cuz
1:45:54 we told you the robots are coming to
1:45:56 kill you why would you do that don't you
1:45:59 worry your pretty little head go get
1:46:01 some more
1:46:03 Fritos no point in panicking just
1:46:08 panic I don't think that FBI guy knows
1:46:11 completely what he's talking about he
1:46:13 sounded
1:46:14 special um you got to imagine that we're
1:46:16 hacking them too like again all that
1:46:19 stuff has been there forever AI is going
1:46:21 to take a lot of people's jobs I like
1:46:23 making analogies of the Industrial
1:46:25 Revolution we found other places to
1:46:27 specialize I have a robot in my Pharmacy
1:46:29 that fills drugs all day long for me I
1:46:32 did not like it when it first got there
1:46:34 but I couldn't imagine being without it
1:46:36 now I hope that AI will be able to
1:46:38 change Health Care in a way that you I
1:46:42 bet they go to Bill Gates right now so
1:46:44 that that information would go from the
1:46:47 doctor to the patient uh in a more
1:46:49 efficient manner I think that made me
1:46:52 Shi to nursing school
1:46:53 because basically I'm always going to be
1:46:57 needed I'm trying to Future proof myself
1:47:00 if am I going to be successful is a real
1:47:05 question AI in the future of us for
1:47:08 health education and
1:47:10 jobs Bill Gates so Bill Gates you've
1:47:13 been on this Tech rocket ship for quite
1:47:15 some time in the 70s this is the it's
1:47:18 bigger than anything it's going to be
1:47:19 okay the positive outweighs the good I
1:47:21 bet the internet and now I'm just
1:47:24 wondering how you think this AI moment
1:47:28 Stacks up to where we've been to where
1:47:30 we are right
1:47:31 now ai is going to have more impact than
1:47:34 even the personal computer the phone the
1:47:38 internet right this what I said right
1:47:40 this is what he's known for in this
1:47:41 phenomenal way but this is the biggest
1:47:44 technical advance in my lifetime what
1:47:47 does
1:47:47 the I mean that's important for people
1:47:50 of AI in 10 years I'd Sayed ucation and
1:47:54 healthare will have changed dramatically
1:47:56 y so in medicine there's the research
1:47:59 part the speed of scientific inovation
1:48:02 across all diseases even the neglected
1:48:05 ones like malaran and HIV that will be
1:48:08 accelerated but perhaps even more
1:48:11 profoundly whenever this guy's stolen
1:48:13 everything that made him what he is he's
1:48:14 no Visionary yeah but but he was a guy
1:48:17 that figured out how to take all that
1:48:19 stuff that other people created and
1:48:21 scale it open
1:48:24 I a lot of what they're doing is taking
1:48:28 things like the transformer technology
1:48:31 and they figured out how to scale it
1:48:33 better than anyone else so just because
1:48:36 he didn't invent anything doesn't mean
1:48:38 he didn't innovate what he innovated in
1:48:41 was leveraging that technology that had
1:48:43 been there and scaling it right Steve
1:48:47 Jobs You could argue was the inventor
1:48:49 and and innovated Bill Gates was taking
1:48:52 things and scaling it I would argue that
1:48:54 open AI right now is taking technology
1:48:57 and scaling it I think I actually got
1:48:59 that idea from Pate who I just saw
1:49:01 posted things will change
1:49:03 Wow E easy there with the wild
1:49:09 speculations you almost can think of the
1:49:11 AI as like a third person sitting in
1:49:13 doing the transcript suggesting the
1:49:16 prescription and then you have this
1:49:18 whole session has been transcribed and
1:49:20 so instead of the doctor facing that
1:49:22 computer screen they're engaging with
1:49:24 you in this software is making sure
1:49:27 there's a really good
1:49:29 transcript one of the things it's
1:49:31 exciting to me is how this is going to
1:49:34 impact kids and their personal learning
1:49:38 yeah yeah the dream that you could have
1:49:39 a tutor who's always available to you
1:49:42 and understands how to motivate you what
1:49:44 your level of knowledge is this software
1:49:47 should give us that you know when you're
1:49:49 outside the classroom that personal
1:49:51 tutor is is helping you out encouraging
1:49:54 you and then your teacher you know talks
1:49:57 to the personal tutor and says okay
1:49:58 what's your advice about how I do things
1:50:01 in class the next day they won't have to
1:50:02 spend time grading homework they'll be
1:50:05 immediately know how that went um oh my
1:50:08 God they're going to have so many
1:50:09 mistakes can't replace a nurse or ma
1:50:12 Medical Associate I assume
1:50:15 um do
1:50:17 not this is just my opinion do not
1:50:20 assume that the inaccuracy of the models
1:50:23 of today are going to persist for that
1:50:25 much longer what open AI released today
1:50:29 is a version of its of its language
1:50:33 model that now thinks before it answers
1:50:36 these things are going to get smarter
1:50:37 and smarter and smarter so the current
1:50:39 technology I would agree with you but I
1:50:41 don't think that lasts for that
1:50:43 long we know how that went does this
1:50:46 mean you eventually don't have teachers
1:50:49 and everybody just has their own AI
1:50:52 teacher
1:50:53 we're trying to help teachers uh you
1:50:56 know that personal engagement motivating
1:50:59 a kid uh meeting them where they are you
1:51:02 know the AI in no way uh is a
1:51:04 replacement for that but the fact that
1:51:07 as the teacher saying okay what should
1:51:09 you work on it's working at your level
1:51:11 that personalized notion you know we can
1:51:14 see it's already working pretty well I
1:51:17 talked to Sam Alman and he was pretty
1:51:19 direct about ai's impact on the work
1:51:23 force a lot of jobs may go away do you
1:51:25 agree and how should we start preparing
1:51:30 ourselves to navigate that yeah
1:51:32 particularly as you get in this not the
1:51:34 5 to 10e time frame but beyond that or
1:51:38 the AIS not only are they very good at
1:51:40 kind of White Collar things but also the
1:51:44 robotics embodiment is taking a lot of
1:51:47 blue collar type things whether it's
1:51:49 warehouse work or factory work 3D
1:51:52 printed house rooms cooking things like
1:51:55 that and so that's not only just a
1:51:58 threat to people's livelihoods but you
1:52:00 know people gain dignity and a sense of
1:52:02 purpose and meaning from work that's
1:52:04 right we've organized our society around
1:52:07 the fact that if you you know get a good
1:52:10 education if you get a unique set of
1:52:12 skills that you'll be rewarded for that
1:52:14 if we get 20 or 30% more productive
1:52:17 that's all to the good because there's
1:52:19 so many unmet needs but if you get you
1:52:22 know 300% more productive yeah in a way
1:52:26 and say well that's only good because
1:52:28 then we can make more stuff and we can
1:52:30 have more leisure time but the the way
1:52:33 that we organize it and maintain
1:52:34 people's dignity will require a lot of
1:52:38 Engagement and you know picking leaders
1:52:41 who are attuned to taking that access
1:52:44 and and making sure we're all sharing
1:52:46 fairly people this is a big deal the the
1:52:48 the thing that I think is going to be
1:52:50 the biggest transition for people is is
1:52:53 not necessarily the loss of jobs I feel
1:52:55 like we're we're going to find a way to
1:52:56 navigate that I think that that concept
1:53:00 of of people having their self-worth
1:53:03 tied to the things they do the skills
1:53:05 they have the [ __ ] they learned in
1:53:07 college that they now have been doing
1:53:09 for 20 years in their job and then all
1:53:11 of a sudden that goes away and they have
1:53:13 to they they have to learn to do
1:53:15 something else that's not where they
1:53:18 have their their core you know personal
1:53:20 dignity and value that's the thing I
1:53:22 think is going to be from a mental
1:53:24 health standpoint the the biggest
1:53:26 transition I don't have a good answer
1:53:28 for that one I think that's just going
1:53:31 to be time heard you speak be a lot of
1:53:33 morning yeah in the long run you know as
1:53:35 we get past that 10 years work won't be
1:53:39 as important and you know that uh you
1:53:42 know I think is work won't be as
1:53:43 important thing yeah what are the jobs
1:53:46 yeah it's that feels right to me that
1:53:49 it's we're we're in the middle of a 10
1:53:51 we're at the beginning of a 10e
1:53:54 transition
1:53:56 where the value of work is completely
1:54:01 redefined completely
1:54:05 redefined yeah people people will find
1:54:07 their creativity again that that the
1:54:09 value of creativity and empathy and
1:54:11 personal connection I think that goes up
1:54:13 in value and that maybe the job of of
1:54:17 humans becomes to engage with one
1:54:19 another and let the machines do all the
1:54:21 shitty work that's that's my
1:54:24 hope that you believe are going to be
1:54:26 undisturbed what skills are going to
1:54:29 persist in these in this new world yeah
1:54:33 that's very hard to make a prediction I
1:54:34 mean be aspects of creativity there'll
1:54:37 be aspects of
1:54:39 appreciating people you know social work
1:54:42 the teacher really engaging that student
1:54:45 in the rich way even that I can't
1:54:47 guarantee that software won't eventually
1:54:49 get good at that but you know for the
1:54:51 Next Generation
1:54:53 the more human the more engaged you are
1:54:57 with other people that is a a skill set
1:55:00 that is in such short supply whether
1:55:02 it's me the more engaged you are with
1:55:04 other people right I think that this is
1:55:07 why joining the salon joining
1:55:11 communities of people who are engaged
1:55:14 and generous and empathetic and curious
1:55:18 and adaptable all that sort of stuff
1:55:23 feels to me like it goes up dramatically
1:55:25 in
1:55:26 value but that's the the change is is
1:55:30 going to be profound for a lot of people
1:55:32 that where they they put their value in
1:55:34 their skills and their work um there's
1:55:37 going to be a lot of loss and
1:55:40 mourning in such short supply whether
1:55:43 it's mental health or education uh that
1:55:47 we can never have must soul I want to
1:55:49 join but I'm
1:55:50 weird we are weird friend do you have
1:55:53 none do you have no I have significant
1:55:56 fears about the risks the this is the
1:55:58 first technology that is happening
1:56:02 faster than even the Insiders expected
1:56:05 it's happening faster than you guys
1:56:07 thought absolutely do you think that
1:56:10 your company uh that that Microsoft and
1:56:13 other companies that are going to make a
1:56:15 lot of money off of AI have an
1:56:18 obligation to help us manage and
1:56:21 navigate through this change
1:56:23 absolutely the fact that there will have
1:56:25 to be regulation that the way we think
1:56:27 about taxes will have to change quite s
1:56:30 Papa we're not weird we're irregular
1:56:32 with the government then at least we can
1:56:35 craft something that's that's not just
1:56:37 profit driven but what did we miss with
1:56:40 the internet and social media that we
1:56:43 can use to apply now to this moment of
1:56:47 AI well I I would say speaking for
1:56:51 myself personally and I think many
1:56:52 others that we were a bit naive yeah so
1:56:56 Sandy Hook is a synthetic completely
1:57:00 fake we thought the internet the
1:57:02 availability of information would make
1:57:04 us all a lot more factual the fact that
1:57:06 people would seek out kind of a niche of
1:57:09 misinformation we were a bit naive and
1:57:11 and bringing those lessons of you know
1:57:14 let's get ahead of this let's get the
1:57:16 politicians to be users and and have a
1:57:19 dialogue about uh maximizing the
1:57:22 benefits while uh minimizing the
1:57:25 negatives that's that's actually really
1:57:28 interesting the the
1:57:30 um if the value of the past 10 years was
1:57:33 to teach us how susceptible we are to
1:57:36 misinformation and media
1:57:39 manipulation
1:57:42 um maybe it does set us up to be a bit
1:57:45 more Savvy I I mean I think as scary as
1:57:48 it was to hear from the woman that was
1:57:49 like you know my daughter called me and
1:57:51 told me you know she I need
1:57:53 $900 as scary as that is I think
1:57:56 creating an awareness that okay we've
1:57:58 seen what we can do to each other with
1:58:00 just you know basic media manipulation
1:58:03 over the past decade um and now we've
1:58:06 got tools that make that way more
1:58:07 sophisticated I mean may maybe that's at
1:58:11 least prepared us to be ready for what's
1:58:16 coming and the second root word of it is
1:58:19 which is to be loved are we going to
1:58:21 lose that aspect of human
1:58:24 let me go
1:58:25 back good question humans we have a soul
1:58:29 we SPS book when it comes to your
1:58:32 morning news we got you covered from Lis
1:58:34 with your family BR you the news in
1:58:37 Colorado that matters most we're out
1:58:39 here in the community telling your
1:58:40 stories and making a difference start
1:58:42 your day the right way get Denver 7
1:58:44 Morning News weekdays starting at
1:58:46 4:30 yeah awareness is really important
1:58:48 right now y I'm broke my family thinks
1:58:52 I'm a l
1:58:53 love life
1:58:58 is is this being streamed on YouTube
1:59:00 what's the name of this program with
1:59:02 Bill and Company Bill and Oprah it's
1:59:04 Oprah AI in the future of us on ABC
1:59:07 human is to
1:59:09 be boy that's a good question humans we
1:59:12 have a soul we went through different
1:59:14 levels of life from newborns toddlers
1:59:19 children teenagers human in Arabic is
1:59:21 insan and it translates to two things
1:59:23 one is forgetfulness and that's a
1:59:25 quality that all of us share and the
1:59:27 second root word of it is UN which is to
1:59:28 be loved are we going to lose that
1:59:30 aspect of humanity I think that gets
1:59:32 Amplified I think that as humans there
1:59:35 is nothing that brings us greater
1:59:37 discomfort than lack of knowledge and I
1:59:40 think that AI in many ways comes out of
1:59:44 that drive know all the answers to be in
1:59:49 control one of the things that I reflect
1:59:51 on is be careful about Comfort because
1:59:53 too much comfort is actually going to
1:59:55 make your life worse not better because
1:59:57 it'll decrease your overall capacity
1:59:59 this this is a
2:00:00 Wellcraft that I locate the Divine in
2:00:04 relationships relationships between
2:00:06 humans and the environment to me that's
2:00:09 God and that's the Divine and that's
2:00:11 something that technology can never
2:00:13 stand in for and can never replace
2:00:16 totally agree with that to that's that's
2:00:19 really good messaging and I think
2:00:20 counterintuitively that's where AI
2:00:22 pushes us I am so delighted to be able
2:00:25 to talk to Marilyn Robinson one of
2:00:28 America's greatest novelist who writes
2:00:31 of all the messiness and all the grace
2:00:35 of Being Human so what do you make of
2:00:39 this moment that we're in and these two
2:00:41 words artificial intelligence well you
2:00:45 know uh it's a certainly a departure
2:00:48 from any sense of intelligence we have
2:00:50 had up to this point you know
2:00:53 mulling cumulative knowing and
2:00:55 interpreting the world that has been
2:00:57 called intelligence because we've always
2:00:59 thought of them as uniquely human traits
2:01:02 exactly the impulse behind it seems to
2:01:05 be to eliminate the human hand the human
2:01:08 eye in the making of the reality that we
2:01:11 inhabit to to the farthest extent
2:01:14 possible with the effect of of U
2:01:18 concentrating power in the sense of the
2:01:20 control of this strange
2:01:23 fantasy uh within very few people's
2:01:26 hands and consolidating masses of
2:01:29 capital and masses of Natural Resources
2:01:31 devoted to this very dubious project
2:01:35 it's just a a human impulse to escape
2:01:38 from its Humanity which is something
2:01:41 that happens over and over again
2:01:42 historically the impulse to say uh that
2:01:46 there is nothing intrinsic about a human
2:01:50 person that cannot in theory be
2:01:52 replicated by something that frankly
2:01:54 Works cheaper Works faster I don't know
2:01:58 why it is but it's part of our human lot
2:02:02 to be uncomfortable with our
2:02:04 humanity and to mechanize it you know
2:02:08 out of existence in some
2:02:11 [Music]
2:02:13 way we're not enough in love with our
2:02:15 own existence and we're not enough in
2:02:18 love with everyone else's existence and
2:02:21 I think that's very d
2:02:22 [Music]
2:02:24 dangerous I think that's just so
2:02:26 powerful we're not I vehemently disagree
2:02:29 with it by the way no surprise there but
2:02:33 hasn't that been the nature of who we
2:02:35 are as human beings always seeking the
2:02:38 next the more the better the
2:02:41 other we generalize from the behavior of
2:02:44 people who tend to dominate historical
2:02:47 narrative and are very frightening
2:02:49 people in many many cases not the kind
2:02:52 of people we would have wished to be
2:02:54 making great decisions what is the thing
2:02:57 that concerns you the
2:02:59 most about
2:03:02 AI I'm afraid that there will be so much
2:03:05 investment of every kind into this
2:03:08 project that it will have it this
2:03:11 project that
2:03:13 like her idea that there's this there's
2:03:16 this single entity that's doing this is
2:03:19 just bizarre to me it it's like this
2:03:22 project is not a project right there's a
2:03:24 technology that was invented in 2017
2:03:27 that millions of developers are pursuing
2:03:31 and a handful of companies are
2:03:33 exploiting better than others
2:03:36 but
2:03:39 fascinating that it will have its
2:03:41 consequences before it really deserves
2:03:44 them and so there will be all kinds of
2:03:46 cutting back on all kinds of employment
2:03:48 and so on anticipating this insurgence
2:03:53 of of AI and then it will turn out that
2:03:56 it's a it's a bad project you know that
2:04:00 it's more dangerous than it is valuable
2:04:02 and so on and so we will have a a
2:04:06 disrupted economy on one hand and an
2:04:09 unusable technology on the other yeah
2:04:12 that's horseshit it it like like what
2:04:15 what she's basically saying is we'll
2:04:17 realize oh it was a big mistake and and
2:04:21 the way we used to do things things
2:04:22 doesn't exist anymore and now we just
2:04:24 have this failed experiment she
2:04:26 completely is missing the fact that we
2:04:28 as humans will evolve right as as these
2:04:31 tools change things we will step into a
2:04:34 different
2:04:35 role right yeah Source Camp she's
2:04:38 clearly never used generative AI you
2:04:40 guys she represents hundreds of
2:04:42 thousands of people be nice no listen I
2:04:45 I don't this is not about nice or not
2:04:47 nice this is this is
2:04:50 about so well whatever let's let's just
2:04:52 keep listening to her cuz she she's
2:04:55 clearly
2:04:57 very smart and articulate and thoughtful
2:05:00 about this
2:05:01 but she's also got a very limited view
2:05:04 of what AI is and its
2:05:08 role but it's also going to no question
2:05:12 about it it's going to make life a lot
2:05:16 easier for a lot of people so what's
2:05:18 wrong with
2:05:19 that American schools as for a lot of
2:05:22 written work from students and that's a
2:05:24 very good thing that's an autonomy
2:05:27 creating exercise you know but what we
2:05:30 don't do we we don't tell them that this
2:05:33 is a discipline like an athletic
2:05:35 discipline that they have undertaken to
2:05:37 strengthen themselves not just to bring
2:05:40 themselves over some arbitrary line of
2:05:42 sufficiency that it's the discipline of
2:05:44 doing the work and the thinking process
2:05:47 exactly and students say writing is
2:05:50 difficult which of course it is it is
2:05:52 but difficulty is the point and so if
2:05:55 you were in charge of this whole AI
2:05:59 movement what would you be doing she
2:06:01 turn it
2:06:02 off I would advise that it be used
2:06:06 for you know the kind of research that
2:06:08 produces uh drugs and immunizations and
2:06:11 so on I would uh make a fantastic
2:06:18 Library well that's what it is it's the
2:06:21 world's greatest Library instantly
2:06:23 accessible very very judiciously to
2:06:26 replace work that is drudgery without
2:06:30 the arrogance of replacing the work of
2:06:33 people who you know who are of very
2:06:36 great value you know just because they
2:06:38 are the substance of our ordinary
2:06:40 interactions with the world but you know
2:06:43 I also age as I'm growing older and I
2:06:46 think as you grow older there's always
2:06:47 the risk of trying to hold on to what
2:06:50 was and
2:06:53 when there are currently now children
2:06:58 who are born and who will be born who
2:07:01 will know nothing but this generation of
2:07:04 AI so there will be no looking back this
2:07:07 hour what used to be or the or the way
2:07:10 it was and that's always been the case
2:07:12 but I I get the sense that you somehow
2:07:15 feel that this moment is different than
2:07:19 others in past Generations why
2:07:23 well you know the the thing that occurs
2:07:25 to me to compare with it is the
2:07:28 Manhattan Project that was carried on
2:07:31 among very few people in deep secrecy
2:07:34 these things are talked about afterward
2:07:36 as if they're Collective decisions but
2:07:38 they're not they're scientific
2:07:40 thresholds that are crossed only by
2:07:42 people yeah this is this makes me nuts
2:07:45 right the comparison to the Manhattan
2:07:47 Project is right that was a handful of
2:07:50 people creating you know a machine right
2:07:53 now you got millions of people
2:07:55 developing these Technologies The open-
2:07:57 Source Community I mean one of the
2:07:58 things that meta did here by open
2:08:01 sourcing their powerful models is that
2:08:03 they're giving access to these tools to
2:08:05 to lots of different people
2:08:08 the the democratization of access to
2:08:11 this technology as it's being developed
2:08:14 is completely different than the
2:08:16 Manhattan Project was which was just
2:08:19 designed as a weapon of Destruction
2:08:29 right she's only talking here about the
2:08:33 downside of this not the
2:08:36 the the humanity impact of it my
2:08:40 experience with with these tools is that
2:08:45 the benefit I get from them is as is as
2:08:48 if I'm tapping into the collective
2:08:51 intelligence of humanity not the
2:08:53 artificial intelligence of a machine
2:08:56 that's here to replace us that that this
2:08:58 is giving us access she said it she said
2:09:01 this you know she wants us to build the
2:09:03 ultimate Library that's what large
2:09:05 language models [ __ ] are they're the
2:09:08 world's greatest Library where we get to
2:09:09 tap into the collective output of
2:09:13 humanity so um this particular point of
2:09:16 view makes me a little
2:09:18 cranky CU it's it's just not the same at
2:09:21 all
2:09:22 sequence and there's no going back from
2:09:25 nuclear fishing but uh that doesn't mean
2:09:28 we should put ourselves in the same kind
2:09:30 of position
2:09:32 again yeah that that was a very lopsided
2:09:36 segment didn't like that at all cond
2:09:40 over and only one was named the
2:09:44 September
2:09:48 7 did she ask to speak to ai's manager
2:09:52 yet it's very very CL thank you to
2:09:54 everyone who spoke with us tonight I
2:09:56 hope that they have left you with a
2:09:57 clearer picture of what is coming
2:10:00 artificial intelligence is still beyond
2:10:03 our control and to a great extent our
2:10:06 understanding but it is here we're going
2:10:08 to be living with technology that can be
2:10:11 our Ally as well as our rival it is
2:10:14 something no other generation has ever
2:10:19 experienced this moment requires a
2:10:22 different level of alertness awareness
2:10:26 about who we are and where we're going
2:10:30 what's real and what's not what's
2:10:33 artificial intelligence and what is
2:10:35 human wisdom authentic intelligence
2:10:38 earned through study and experience we
2:10:41 are this planet's most adaptable
2:10:44 creatures we will adapt again but keep
2:10:47 your eyes on what's real the stakes for
2:10:49 all of us could not be
2:10:52 higher don't disagree with that the
2:10:55 authenticity is a big one here's here's
2:10:57 where I see this special like parts of
2:11:00 this special are
2:11:01 fine parts of this special are not going
2:11:03 to age well here's what this feels like
2:11:05 to
2:11:06 me
2:11:19 um um
2:11:27 this is what this feels like 56 pass I
2:11:29 wasn't prepared to translate that as H
2:11:31 doing that little tease that little Mark
2:11:33 with the A and then the ring around it
2:11:36 at see that's what I said MH um Katie
2:11:39 said she thought it was about yeah oh
2:11:41 but I'd never heard it i' never heard it
2:11:43 said I'd always seen the mark but never
2:11:45 heard it said and then it sounded stupid
2:11:47 when I said it violence at NBC
2:11:52 the lunch there it is violence at NBC
2:11:55 geom
2:11:57 GE Allison should know what internet
2:12:00 anyway what is internet anyway internet
2:12:02 is uh that massive computer network the
2:12:05 one that's becoming really big now what
2:12:08 do you mean that's how does one what do
2:12:10 you write to it like mail no a lot of
2:12:12 people use it and communicate I guess
2:12:14 they can communicate with NBC writers
2:12:15 and producers Allison can you explain
2:12:17 what internet is no she can't say
2:12:19 anything in 10 seconds or less oh Alison
2:12:23 will be in the studio shortly Frank such
2:12:25 a dick computer network made up made up
2:12:27 of uh started from Oh I thought you were
2:12:29 going to tell us what this was the it's
2:12:32 not an it's it's it's a computer
2:12:35 billboard but it's Nationwide and it's
2:12:36 it's several uh universities and
2:12:38 everything all joined together right and
2:12:39 others can access it right and it's
2:12:41 getting bigger and bigger all the time
2:12:43 it came in really handy during the Quake
2:12:44 a lot of people that's how they were
2:12:45 communicating out to tell family and
2:12:47 loved ones they were okay because all
2:12:48 the phone lines were down I mean that
2:12:50 Oprah special that that that what this
2:12:52 feels like I think we'll look back on
2:12:53 the over special at some point like it
2:12:56 it didn't have any really egregious
2:12:57 things in it but it but it was
2:13:01 definitely with the exception of Sam
2:13:03 Alman a little bit of the Bill
2:13:05 Gates um that was
2:13:08 definitely
2:13:13 um it it was definitely
2:13:16 reinforcing the Hollywood the robots are
2:13:18 going to kill
2:13:20 us Ze gist right so you know um I don't
2:13:26 think that special did AI any favors and
2:13:31 I don't think it [ __ ] matters I you
2:13:33 know again the whole purpose of this
2:13:35 channel is regardless of whether you
2:13:38 know that woman feels like you know AI
2:13:43 is a project that is designed to replace
2:13:47 people or not doesn't change the fact
2:13:50 that this shit's coming and it's coming
2:13:52 fast
2:13:54 right my push back on the woman talking
2:13:57 about the humanity thing is that I don't
2:14:01 think she's used generative AI like I
2:14:04 don't think you can use it and not feel
2:14:07 a profound sense
2:14:09 of tapping into
2:14:13 humanity
2:14:16 and this community the community that
2:14:18 hangs out here a reminder of how
2:14:20 important our empathy is that's that's
2:14:22 it
2:14:22 an my my experience with generative AI
2:14:27 is I feel more profoundly connected to
2:14:29 People by a
2:14:31 lot than I ever did with social media
2:14:34 social media
2:14:37 cut you know it's an oxymoron right
2:14:39 because it makes us
2:14:42 antisocial I feel like when you tap into
2:14:46 these tools and you feel that profound
2:14:48 sense of like holy [ __ ] the fact that I
2:14:51 can say write me a computer program and
2:14:54 it writes a computer
2:14:56 program that is the knowledge of
2:14:59 millions of people that have contributed
2:15:01 code to the
2:15:04 internet for years and years and years
2:15:07 that makes that possible like there's a
2:15:11 profound for me there's a profound
2:15:17 realization of the value of humanity
2:15:20 when I use these tools not the opposite
2:15:22 of that that was my my the thing I had
2:15:25 the most trouble with I mean okay fine
2:15:29 Oprah didn't get it and Oprah's got a
2:15:31 bit of a point of view on it that's like
2:15:32 that's fine and that that I actually
2:15:35 expected that woman in
2:15:38 particular like she's
2:15:41 missing the magic of AI is that it is
2:15:47 the world's best Library
2:15:50 already it is a reflection of humanity
2:15:55 already it's not there to replace us
2:15:58 right it is part of us it is collective
2:16:02 intelligence not artificial intelligence
2:16:04 but
2:16:05 whatever um I bet Oprah's going to
2:16:07 launch something with AI classes more
2:16:10 specials she has a profit
2:16:13 somehow well I listen I think Oprah's
2:16:15 smart Oprah is like this is going to
2:16:18 change everything I like I at least want
2:16:20 to have a voice voice in the
2:16:23 conversation I mean I
2:16:26 don't like I don't even with the woman I
2:16:28 disagree with I don't hold it against
2:16:30 her that she hasn't used this stuff I
2:16:32 don't hold it against Oprah that what
2:16:36 she's that what she framed up was this
2:16:39 shit's
2:16:41 scary CU it is
2:16:44 scary and the more you use it the less
2:16:48 scary it becomes right so I see just
2:16:51 like the Katie KK piece here I see this
2:16:54 as one of those things that's looking at
2:16:56 AI through a lens of not having used it
2:17:01 that's where everyone starts and so the
2:17:04 only the only information they have is
2:17:07 what have people said about it well
2:17:09 they've said that the robots are going
2:17:10 to kill us oh so I should be scared
2:17:13 afraid of it oh [ __ ] they deep faked
2:17:15 someone I should be afraid of it right
2:17:16 there's there's all the reasons to be
2:17:18 afraid of
2:17:19 it the
2:17:22 Brandon
2:17:23 tid in a couple of hours making a
2:17:26 picture book for his kid and being able
2:17:30 to publish it with his cute little Pixar
2:17:33 kind of version of
2:17:35 himself they're not talking about
2:17:38 that that doesn't get people to watch
2:17:42 TV but that changes lives right so
2:17:52 like did did this did this show advance
2:17:56 advance the cause at all I don't know
2:18:00 people over 60 are the best source of
2:18:02 knowledge on AI we all know this yeah
2:18:04 exactly pay I you
2:18:07 know
2:18:09 exactly um AI does hallucinate
2:18:11 information for now it does but so do
2:18:14 humans so do humans humans get [ __ ]
2:18:18 wrong humans hallucinate
2:18:22 humans don't apply logic to the
2:18:26 situation right I mean the past
2:18:29 decade has been that just people sort of
2:18:32 getting in camps and going you suck no
2:18:35 you suck no you suck well here's facts
2:18:38 yeah we have alternative facts
2:18:42 like humans are worse hallucinators than
2:18:44 ai ai is at least getting
2:18:47 better
2:18:49 um nobody slowing down I agree with
2:18:54 that AI is our mind already I have
2:18:56 developed another part of my brain than
2:19:00 uh that talks AI thinks AI yeah AI is
2:19:03 part of our
2:19:05 brain there's another Oprah special on
2:19:07 AI on
2:19:09 916 is it a repeat or a new program
2:19:14 what's today the 12th if it's on the
2:19:16 16th I doubt it's
2:19:22 unless it's a follow on where they're
2:19:24 going to announce some some of the new
2:19:26 [ __ ] that open AI announced I doubt
2:19:32 it I I I I don't know I don't I don't
2:19:35 know the answer to
2:19:36 that Amelio's
2:19:40 wife we humans are AI we were created
2:19:45 well we'll
2:19:47 see are we in a simulation I you know I
2:19:50 personally don't think so but you know
2:19:52 it's
2:19:53 possible I found a free AI from Opa
2:19:56 under my seat anyone else you get AI and
2:19:59 you get AI yeah you know the the the
2:20:02 thing if if I were to sort of criticize
2:20:05 one thing about that that I feel like
2:20:06 was missed in that
2:20:08 special was let's hear from someone
2:20:11 who's used it that that did something
2:20:13 really remarkable with it like we've the
2:20:17 the salon and this channel is full of
2:20:20 people and and some of the victories are
2:20:22 very very small like like the kids book
2:20:25 and what Jim Ross is doing with his
2:20:26 business and what source Camp is doing
2:20:29 you know educating people and Anne
2:20:30 Murphy in fundraising like I
2:20:33 I there are so many there are so many
2:20:38 things that could
2:20:39 have rooted it in reality as scary as it
2:20:43 is here's someone that did something you
2:20:45 know amazing or cool or powerful um but
2:20:49 listen again when when this Katie kurick
2:20:53 thing came out so this was
2:20:56 1994 so I had just started Urban desires
2:20:59 I just started agency.com and the media
2:21:03 was just like what is this internet
2:21:06 thing who are these weirdos right
2:21:09 like it
2:21:11 was people were discombobulated and most
2:21:14 people hadn't heard of it they weren't
2:21:16 paying attention to it and so I I think
2:21:19 we're at a very natural
2:21:22 state that that that's special to me is
2:21:26 further Evidence so there's there's a
2:21:28 there's kind of
2:21:30 a um a Trope going around right now that
2:21:35 AI the AI hype has peaked and and we're
2:21:39 coming down now and it's all over and
2:21:40 these tools have plateaued and they're
2:21:42 never going to get better
2:21:45 um and and they equated where we are
2:21:49 with AI to the internet bubble of of
2:21:52 2000 where right where the it the hype
2:21:55 got so big and then everything burst
2:21:57 I've argued and this this special with
2:22:00 Oprah reinforces it for me the hype
2:22:03 cycle hasn't even begun yet not enough
2:22:06 people know about AI not enough people
2:22:08 are using it the technology is still too
2:22:12 janky I think we're probably two years
2:22:14 out from where the real hype begins and
2:22:17 it really gets blown out and Amplified I
2:22:20 think that happens two years from now
2:22:24 um and I think I there probably will be
2:22:27 a hype cycle where it it gets so
2:22:29 overblown and then bursts but I think
2:22:31 we're not even close to it I think we're
2:22:33 closer to 1994 metaphorically than we
2:22:36 are to
2:22:38 1999 Tyler Perry just invested millions
2:22:40 in a movie producing use using AI that
2:22:43 that's it oh did he that's interesting
2:22:45 so Tyler Perry also cancelled an $800
2:22:48 million Studio expansion in Atlanta
2:22:51 because of Sora so that's interesting
2:22:54 that he's investing in it he's like well
2:22:55 [ __ ] it I was going to spend $800
2:22:57 million to build a studio why don't I
2:22:59 spend aund million and make an AI movie
2:23:02 I think that's pretty smart
2:23:07 actually for people who have never used
2:23:10 AI how will they respond to this show
2:23:12 prediction be scared I agree with that
2:23:14 Dan I
2:23:17 mean yeah the the
2:23:23 what what will happen from that
2:23:26 show is that more people will not just
2:23:29 hear AI they'll hear chat GPT or they'll
2:23:34 like you know they they making videos or
2:23:37 they got voice things like there may be
2:23:40 an awareness of it I don't think that
2:23:42 special inspires anyone to go get
2:23:44 curious about AI again the reason to be
2:23:47 part of you know Rocky Mountain interest
2:23:50 group or the salon or communities that
2:23:52 are curious about this stuff right now
2:23:56 is that we're still in a pocketed
2:23:59 minority of in the corners of
2:24:03 society so so we have time you're not as
2:24:06 far behind as you think and we're not as
2:24:09 far ahead as we
2:24:11 think we're very early
2:24:14 still would you call it underhyped Kyle
2:24:17 in fact I would if you look in the
2:24:18 corner I've got a video series that I
2:24:21 did that if you want to buy it it
2:24:22 supports the channel called underhyped
2:24:24 where these are videos where I talk
2:24:26 about why this stuff is significant I
2:24:29 seriously wonder if Oprah has any
2:24:30 investments in AI she probably does um
2:24:33 she well although I would say if that
2:24:37 special was to Hype up AI Investments
2:24:40 that she has it would have been a very
2:24:42 different special so I would assume that
2:24:46 she might have some Investments but but
2:24:47 that special did not feel like it was
2:24:50 hyping AI
2:24:51 that special felt like it was you know a
2:24:54 a continued um a continued part of the
2:24:57 narrative that AI is scary and you
2:24:59 should be scared of it you make
2:25:02 Mone not like that um let me take a
2:25:06 quick break I want to come back I want
2:25:07 to show some things so we're going to
2:25:09 play with the new so open AI did
2:25:14 launch I think what's project strawberry
2:25:17 today they launched a new model um
2:25:20 that's it's in chat GPT right now at
2:25:22 least it is for me they've been rolling
2:25:23 it out all day um and it's interesting
2:25:27 it's not multimodal it's it's a very
2:25:29 limited version of of what it is but
2:25:32 it's pretty fascinating so we'll look at
2:25:34 that I want to show you this new feature
2:25:37 in Notebook LM that I think is
2:25:40 remarkable I've been playing with that a
2:25:42 lot today um and then Brandon had hang
2:25:46 on
2:25:57 um oh there's a new Hume voice model
2:26:01 out and then Gemini
2:26:04 live is now available to everyone I
2:26:07 don't even know what Gemini live is let
2:26:09 me go get another drink take a break
2:26:11 I'll be right back entertain
2:26:12 yourselves oh
2:26:19 please
2:26:49 e
2:27:19 e for
2:28:44 yeah I got to three things
2:29:05 all
2:29:06 right all right good people have we got
2:29:09 anyone in here still
2:29:11 yes while I was waiting I had GPT roast
2:29:14 a picture of Kyle's workplace so funny
2:29:17 nice I like it um that's awesome
2:29:21 beautiful and Murphy all right relevant
2:29:24 to the
2:29:26 conversation let's see what's going yay
2:29:28 Kyle is
2:29:31 back he did
2:29:34 it okay um my back is holding out my
2:29:38 back has been shitty all day but to
2:29:41 right now it's okay so um if it goes
2:29:45 south I'll let you
2:29:48 know but let's go play with some [ __ ]
2:29:52 going to post it and show and tell yeah
2:29:53 please do that um okay so couple of
2:29:56 things if you're new here
2:29:59 welcome um that's not what I wanted
2:30:02 that's not what I wanted that is what I
2:30:03 wanted the
2:30:06 salon the salon. go there Kyle's all
2:30:11 hyped up and his back is feeling good
2:30:14 yeah exactly lumbar support no I think
2:30:18 it's
2:30:21 what it was was it was Friday night I
2:30:23 did a live and I you know how when I get
2:30:28 in my Rants and I just go for like an
2:30:31 hour and a half or two I did one of
2:30:33 those and I think I was sitting in the
2:30:35 exact same position for more than an
2:30:38 hour and then when I stood up like my
2:30:40 ass muscles got really tight and my and
2:30:43 it just pulled something in my lowered
2:30:44 back um and just pulled something out of
2:30:47 whack and it's been [ __ ] up ever since
2:30:51 so I think so long as I keep reminding
2:30:53 myself to change
2:30:54 positions um I'll be
2:30:57 okay Gemini live is Google's version of
2:30:59 advanced
2:31:01 voice oh interesting oh it's oh it's
2:31:04 available for everyone is it available
2:31:06 for everyone is there a web version of
2:31:09 it it was the cartwheels yeah exactly
2:31:12 you know what it was Silver Fox it was
2:31:13 the twerking if you weren't here Friday
2:31:16 night so I've been promising for for
2:31:18 months that I was going to uh Donna
2:31:20 manini and do some twerking on this
2:31:23 channel as it deserves it is Tik Tock
2:31:26 and uh so so I was twerking in the
2:31:30 manini on Friday and just you know threw
2:31:34 the back out boy do I feel old not as
2:31:37 old as you looked in that man
2:31:41 Keeny all
2:31:43 right ah please no
2:31:47 Dan all right if you've just walked into
2:31:51 the channel yes this is normal
2:31:53 conversation um anyway the
2:31:55 salon so the salon is a community that I
2:31:58 co-founded with Leah faston who's a
2:32:00 photographer out of Boston we founded it
2:32:03 in the week after chat GPT came out um
2:32:08 December 7th 2022 was our first meeting
2:32:11 and we've been doing a meeting every
2:32:13 other week for since then and we've got
2:32:16 an online community so if you go to the
2:32:18 salon AI it's a sitee you can read about
2:32:21 what we're about um and then click on
2:32:23 join our community and that'll take you
2:32:25 to a site that is the community and
2:32:28 there's a nice little onboarding there's
2:32:29 like a little seven step you know
2:32:31 welcome just follow those steps um and
2:32:35 uh and it's it's a really remarkable
2:32:38 community of AI optimists people that
2:32:40 are trying this stuff so if you're
2:32:42 curious about Ai and you haven't really
2:32:44 done much with it and you want to be
2:32:46 around people who have and and are not
2:32:49 douchy about it that's this community
2:32:53 it's a very generous
2:32:55 empathetic
2:32:57 supportive enthusiastic optimistic
2:33:01 Brave group of
2:33:04 people and especially if you're one of
2:33:07 those people that's really excited about
2:33:08 AI but everyone around you is like could
2:33:10 you shut the [ __ ] up about stupid
2:33:13 [ __ ] AI
2:33:14 Janice we don't
2:33:18 care you know
2:33:20 it's it's jalapeno poer night at TGI
2:33:24 Fridays and we just want to enjoy our
2:33:26 our icy margaritas and our jalapeno
2:33:32 poppers all you can eat jalapeno popper
2:33:34 night is what I live for and your AI is
2:33:38 just H we're tired of it we're tired of
2:33:41 hearing about how excited you are about
2:33:43 that stupid Ai and those robots the
2:33:46 robots are going to kill us Janice
2:33:51 so enough with your AI
2:33:53 okay can I have another Margarita please
2:33:56 thanks thanks Sandy that's great yeah no
2:34:00 yeah it's delicious it's always
2:34:01 delicious hey and you know could you
2:34:03 bring us some more poppers I think I'm
2:34:06 I'm feeling I'm feeling radical tonight
2:34:09 let's get another let's get another
2:34:10 round of Popp poppers on me
2:34:14 everybody so if you're used to hearing
2:34:17 that come to the salon
2:34:20 okay boycott Oprah Oprah's
2:34:23 fine um Oprah was Oprah was banned from
2:34:26 the community I listen I I
2:34:29 don't here's the thing about that
2:34:31 special I don't have any ill will for
2:34:33 that special um I think
2:34:39 that what that special did is so here's
2:34:42 something I heard when I was in
2:34:44 DC um one of the staffers
2:34:47 said to be quite honest
2:34:50 AI was was the topic of
2:34:54 conversation about eight months ago and
2:34:58 everyone was talking about it and
2:35:00 everyone was focused on it and that's
2:35:03 gone away that sense of urgency is gone
2:35:06 away so if the Oprah special just
2:35:09 introduced the topic to more people and
2:35:12 more people like hey what are we doing
2:35:13 about this AI hey what is this AI some
2:35:16 percentage of those people are going to
2:35:18 get curious and they're going to try it
2:35:20 and they're going to have their first
2:35:21 Kevin mallister
2:35:23 moment oh my God it can do
2:35:29 that and they'll get excited about it so
2:35:31 I don't think it's a bad thing I don't
2:35:33 you know people are where they are and
2:35:37 some of what AI represents is scary and
2:35:40 a lot of what AI represents
2:35:44 is very very very very very hard to
2:35:49 communicate
2:35:52 without the other person actually
2:35:54 experiencing
2:35:55 it I can't I can't tell you how
2:36:00 viscerally satisfying it feels to take a
2:36:04 screenplay that I wrote 30 years ago and
2:36:07 have chat GPT understand it tell me who
2:36:10 it would cast in the roles and generate
2:36:13 frames from the movie that I had in my
2:36:15 head 30 years ago I Can't Describe to
2:36:18 someone how Lally satisfying that
2:36:22 is they have to experience their version
2:36:25 of that on their own before they
2:36:28 actually understand what the [ __ ] is
2:36:29 going
2:36:30 on and so what the salon is about is our
2:36:34 our we we've got a a
2:36:36 three-step journey to AI literacy and
2:36:40 the first step in fact Let Me Go show
2:36:42 you the page
2:36:54 the first step of AI literacy where we
2:36:58 say to start is play
2:37:01 first play play first it's funny when
2:37:06 when I got um invited to have the salon
2:37:10 participate in some of these Washington
2:37:12 events
2:37:15 um this word this phrase play first
2:37:20 through through the people that we were
2:37:22 talking with they were like let's not
2:37:24 that's not serious enough AI is really
2:37:25 serious these are you're talking to
2:37:27 politicians this is serious
2:37:29 business I actually think that playing
2:37:32 with AI is
2:37:34 like I am more convinced than
2:37:37 ever this is very serious playing with
2:37:41 AI experimenting with it do like Source
2:37:44 Camp does and and do parlor tricks go
2:37:47 make yourself write yourself a silly
2:37:49 song with sun
2:37:50 um go clone your voice like Mar Mars
2:37:54 brownley did and and have yourself say
2:37:58 silly things in that voice play with it
2:38:05 because if your only experience with it
2:38:07 is oh I'm afraid of it and now I have to
2:38:10 learn
2:38:11 AI that feels [ __ ]
2:38:14 daunting it feels like the thing that's
2:38:17 the the farthest from reality if if you
2:38:19 don't know what's going on with AI and
2:38:21 you see all these crazy people going
2:38:23 look at the picture I made look at the
2:38:25 voice I made look at the video I
2:38:27 made your current experience with how
2:38:30 you produce content like that is you've
2:38:33 got to learn Final Cut Pro and you've
2:38:35 got to learn cinematography and you got
2:38:37 to learn this and you got to learn that
2:38:39 you can't possibly know sitting from the
2:38:42 outside what's actually going
2:38:45 on and when you type in the words make a
2:38:49 video of of a cute red puppy you know
2:38:52 running down the beach you know with a
2:38:54 little girl in a yellow dress and then
2:38:56 it does
2:38:59 that and you go oh my God I had no idea
2:39:03 you can only get to that through playing
2:39:05 just
2:39:06 play and in playing you'll figure out
2:39:10 where the boundaries are you'll figure
2:39:12 out what you can now do that you didn't
2:39:15 imagine was
2:39:17 possible and then you can move on to the
2:39:20 next phase of the AI literacy Journey
2:39:22 which is mindfully create now that you
2:39:24 know a few tools and that you can do
2:39:26 some stuff now you go oh maybe I should
2:39:29 make a book maybe I should make an art
2:39:32 project maybe I should solve that
2:39:33 problem at work maybe I should start
2:39:35 that business of I've always thought of
2:39:37 and then the generously lead part is
2:39:40 tell people about it Learn out Loud hey
2:39:45 I was using Ai and I thought I could
2:39:46 start this business and I went to start
2:39:48 this business but then the AI sucked but
2:39:50 then someone in the AI Salon told me
2:39:52 that I could use this other tool and
2:39:54 then I tried that and it sucked less and
2:39:55 so now I've started this business now I
2:39:57 don't know what to
2:40:00 do and bang all of a sudden you are an
2:40:03 AI thought
2:40:05 leader where maybe two weeks ago you
2:40:07 were just like huh what's this AI thing
2:40:10 that's
2:40:12 crazy so join this community all
2:40:16 right could have had Jansen H and how
2:40:19 Nvidia is leading the US and stock
2:40:20 markets higher yeah there could have
2:40:22 been there's a lot of opportunities
2:40:23 where they could have done some more
2:40:24 positive stuff there but that's okay it
2:40:26 was only an hour I mean the problem with
2:40:28 AI is it's so [ __ ] vast I mean it's
2:40:31 it's one of the things I struggle with
2:40:33 on this channel is like it's so Vast at
2:40:35 this point I like I don't know what to
2:40:38 talk about
2:40:39 anymore
2:40:42 um all right we could have woken up to a
2:40:44 huge Uprising tomorrow that's all right
2:40:46 I'm fine I'm fine without it totally
2:40:48 agree on play equals learning that's
2:40:50 profound yep yeah it's like this is this
2:40:53 is where it's at this is where it's at I
2:40:56 th this little trium of of activities is
2:41:01 is super amazing and and you know quite
2:41:03 frankly if you're if you're in the salon
2:41:05 if you've been in the salon like this is
2:41:07 who the people
2:41:09 are there's a playfulness there's a
2:41:11 curiosity there's an
2:41:13 adventure sense of adventure there's a
2:41:16 sense of gratitude there's a sense of
2:41:18 empathy there's a sense of curiosity and
2:41:21 bravery and just a willingness to go ah
2:41:25 [ __ ] it maybe I should just reinvent
2:41:28 myself all
2:41:30 right okay so let's go look at some
2:41:32 stuff
2:41:34 um so let me jump over to chat
2:41:41 je generously lead is my fave yeah
2:41:45 generously leads important here's the
2:41:47 thing about generously leading when you
2:41:49 think about leadership you think about
2:41:51 okay I'm going to learn some stuff and
2:41:52 I'm going to go teach people I'm going
2:41:53 to lead the troops um that's
2:41:59 great another part of generously leading
2:42:02 is asking for help when you need it and
2:42:04 this is this is when I came to this
2:42:08 Summer that it struck me
2:42:11 that there's a lack of ego in the
2:42:15 salon that allows people to ask for help
2:42:19 hey stuck hey how did you do that an
2:42:21 that that thing you did was really cool
2:42:22 what did what tools did you use oh and
2:42:24 then she'll tell you when you ask for
2:42:27 help you actually provide someone else
2:42:30 the opportunity to
2:42:32 lead and that's
2:42:34 leadership creating space for others to
2:42:37 lead is
2:42:39 leadership so not only being willing to
2:42:42 generously tell people what you're
2:42:43 learning and teach them how to do what
2:42:45 you're doing but asking for help when
2:42:47 you need it is as much leadership as the
2:42:50 other side and that that was a big
2:42:51 breakthrough for me okay so now when you
2:42:55 go to chat
2:42:58 [ __ ] you should see
2:43:01 this GPT
2:43:04 40 right which is the the latest best
2:43:09 for complex tasks then you should see 01
2:43:12 preview uses Advanced reasoning 01 mini
2:43:16 faster at reasoning these two are the
2:43:19 new model
2:43:21 when someone says I'm behind they're not
2:43:23 expecting to hear play more exactly
2:43:28 exactly I'm behind go play [ __ ]
2:43:31 it exactly th that's the thing Dan
2:43:34 is I mean you know this as much as
2:43:37 anyone
2:43:41 like all of us are so far behind we will
2:43:46 never catch
2:43:47 up and and you can't catch up it's like
2:43:51 like getting caught up on
2:43:54 AI is kind of like uh you know I want to
2:43:58 get caught up on every single detail of
2:44:02 geopolitics globally right like you
2:44:05 can't do it it there's too much it's
2:44:07 moving too fast AI is the same thing
2:44:10 where it has now surpassed our ability
2:44:13 to keep up with it so given that well
2:44:18 [ __ ] it have some fun with it play with
2:44:19 it just see just pick a
2:44:23 corner and just dick around with it and
2:44:28 see what you discover
2:44:31 because the other thing that happens
2:44:33 when you play with AI is you will
2:44:36 invariably have a breakthrough where you
2:44:39 realize it can do something you've
2:44:42 always wanted to do in your life but you
2:44:44 you have written off as oh I'll never be
2:44:47 able to do that
2:44:50 yeah I dropped out of that drawing class
2:44:52 in when I was
2:44:54 17 and I've always regretted not being a
2:44:57 better
2:44:59 drawer well now you
2:45:02 can oh I didn't get into Berkeley and so
2:45:05 now I'm I gave up on being a musician
2:45:07 well now you can be a
2:45:12 musician you're not behind but you'll
2:45:14 never catch up Kyle
2:45:16 Shannon turn turn that into a meme on
2:45:19 the regulars Channel that's
2:45:21 great all right and then you've got your
2:45:24 old models are in here GPT 40 mini and
2:45:27 the Legacy GPT
2:45:30 4 honest to God their naming their
2:45:33 naming is such a [ __ ] disaster it's
2:45:36 unbelievable it's so
2:45:38 bad like how are you supposed to know
2:45:41 the difference between this thing that
2:45:44 cost them a lot of money but does really
2:45:46 cool
2:45:47 [ __ ] and this thing that's sort of like
2:45:50 this thing but they've hobbled it
2:45:51 because they haven't released all of it
2:45:53 yet cuz the testing isn't ready and
2:45:55 these things these things are
2:45:57 completely different models I assume
2:46:00 what these are is they're sitting on top
2:46:03 of this engine just behaving differently
2:46:06 but I don't know and then 01 mini is
2:46:09 probably sitting on top of this model
2:46:11 doing its reasoning thing this is
2:46:12 sitting on top of this model doing this
2:46:14 reasoning thing and this is a completely
2:46:15 different thing no one knows that no one
2:46:17 should know know that no one should care
2:46:19 here but if they name them differently
2:46:22 in and somewhat
2:46:25 usefully
2:46:28 then it would suck less okay so let me
2:46:32 show you what I know so far I don't know
2:46:34 much I think I've done two total prompts
2:46:38 on this and in fact let me go look at
2:46:40 the prompt that I did
2:46:45 do yeah oh that that's interesting so
2:46:48 when you do when you do a prompt with
2:46:51 the 01 preview you got a blue dot beside
2:46:54 it so they're now in they're now
2:46:56 indicating that you've done something
2:47:00 different so I
2:47:03 said oh this was a this was a fun one
2:47:06 let me show you what I did and then we
2:47:08 we'll go play with
2:47:10 it I said I want to take a classic movie
2:47:13 and write a new movie that maintains the
2:47:16 story structure number of characters Etc
2:47:19 but completely comes up with A New
2:47:21 Concept so give me 10 diverse Classic
2:47:24 Movies first I'll pick one and then give
2:47:27 me 10 concepts for the new
2:47:30 derivatives so what it says here is it
2:47:33 thought for 21 seconds but just like
2:47:36 when you use code
2:47:37 interpreter I think the O Oprah novelist
2:47:40 L lady did the
2:47:43 naming oh that's funny
2:47:46 [Music]
2:47:50 so just like when you use code
2:47:51 interpreter it gives you that little
2:47:53 drop down menu and you can see what it
2:47:54 was doing and it's writing python code
2:47:57 look what this is
2:47:59 doing so if I open that
2:48:02 up it's come up
2:48:05 with let's see 1 2 3 4 5 6 seven steps
2:48:14 and then it's it's written a prompt for
2:48:16 each of the steps so step one weighing
2:48:19 the concept I'm evaluating the users's
2:48:21 request to generate a list so step one
2:48:23 evaluate step two brainstorming movie
2:48:25 options I'm listing 10 uh that provide a
2:48:29 choice of options ensuring that they're
2:48:32 factual and within copyright rules
2:48:34 interesting generating classic film
2:48:36 ideas now I'm creating a a list of 10
2:48:38 movies mindful of copyright rules and
2:48:41 avoiding certain inappropriate themes
2:48:43 right so it's actually telling you what
2:48:45 it's doing um I'm thinking through a of
2:48:49 classic films spanning genres periods
2:48:51 and cultures like Casablanca and Seven
2:48:55 Samurai this approach highlights the
2:48:56 importance of diversity and broadens
2:48:59 classic film appreciation so it's
2:49:01 actually really thinking about you know
2:49:03 film making curating a list compiling a
2:49:08 list Gathering Classic Treasures and
2:49:11 then identifying gaps and then it gives
2:49:13 me the list and then I said I basically
2:49:16 said oh give me short plot summaries of
2:49:19 cuz I didn't know all of them and so I
2:49:22 read them all and some of them were
2:49:24 fine um and then and then I just said
2:49:26 let's do Jaws like I don't want your
2:49:29 stupid High highbrow Cinema stuff and
2:49:34 then it says shirely here's
2:49:36 jaws and then it gave me 10 Concepts
2:49:40 that could layer on top of the Jaws
2:49:43 story
2:49:45 structure and they were all pretty good
2:49:48 and then I picked one
2:49:52 let's
2:49:55 see and I said give me give me a table
2:49:58 that's got the character in Jaws what
2:50:01 they did and then the character in the
2:50:03 new movie and what they do which was
2:50:06 pretty cool and then I said give me an
2:50:08 outline which it did then I said give me
2:50:10 a detailed outline which it
2:50:12 did and so so anyway so so that was the
2:50:16 that was kind of the first thing I did
2:50:17 where I kind of dug into this thing or
2:50:19 played with
2:50:21 it so let's do the we'll do the classic
2:50:24 how many RS are there in Strawberry so
2:50:25 if you don't know this
2:50:27 one with traditional GPT in fact let's
2:50:30 switch over let's go to GPT 40 oh they
2:50:33 fixed it in that one let me go to
2:50:35 classic
2:50:39 GPT so now we're in chat GPT 4 and I say
2:50:43 um how
2:50:45 many
2:50:47 RS are the
2:50:50 there
2:50:53 in the word straw
2:51:04 Berry there are two RS in the word
2:51:07 strawberry why did it get that wrong
2:51:10 well because the way these things work
2:51:12 is um they create what are called
2:51:16 tokens and a token could be a space it
2:51:19 could be a period it could be part of a
2:51:21 word so the word St strawberry is
2:51:24 probably broken into two or three
2:51:26 tokens and so I would assume that Barry
2:51:30 is the second token and the second token
2:51:34 has two RS in it so when you say how
2:51:36 many RS are in the word it goes back and
2:51:39 looks at the last token of one of two
2:51:42 and it says there's two RS in Strawberry
2:51:44 not three because that was in a
2:51:45 different
2:51:46 token so the way
2:51:50 historically these tools have worked is
2:51:53 they vomit out tokens and then when you
2:51:55 ask them to think about it they just
2:51:57 vomit more tokens they don't ever really
2:52:00 understand what they vomited out the new
2:52:03 stuff is different so let's go to the
2:52:06 1.0 preview in fact let's go to the Mini
2:52:09 version because it'll be faster so we're
2:52:11 going to start a new chat and I'm going
2:52:14 to
2:52:16 say how many RS are in in the word
2:52:28 strawberry
2:52:32 oops internal server error
2:52:38 regenerate thought for a second the word
2:52:41 strawberry contains two letter RS oh it
2:52:43 didn't work here's the
2:52:47 breakdown hauh Minnie got it wrong
2:52:51 fascinating all right let's copy
2:52:54 this there wait
2:52:57 this Minnie got it wrong that's
2:53:13 fascinating there are three Rs in the
2:53:15 word strawberry so this got it right
2:53:20 let's see if it explains its reasoning
2:53:23 seems to be hung a little bit right
2:53:32 now what was
2:53:36 your
2:53:43 thinking they still have a bug where
2:53:45 sometimes when you hit the enter key it
2:53:47 it doesn't actually submit the prompt
2:54:12 huh now it's totally confused and got it
2:54:14 wrong all right well that's [ __ ] up
2:54:19 let me let me I'm going to option reload
2:54:22 this page it it seems like it's it's in
2:54:24 a bad
2:54:29 way all right paste how many RS are in
2:54:32 the word strawberry I think my
2:54:35 browser a little
2:54:39 crashy
2:54:41 huh there it just said it got it right
2:54:43 but it didn't show its thinking all
2:54:45 right let's let's do something more
2:54:47 interesting than this let's do um
2:55:00 [Music]
2:55:02 help me create a logical framework to
2:55:07 determine what's
2:55:09 better a hot
2:55:14 dog or a fidget spinner
2:55:22 So This falls into the category of play
2:55:24 first like oh God damn that
2:55:27 thing
2:55:31 um giving it ridiculous prompts just to
2:55:34 see what it does is is kind of
2:55:38 fun let's see response count one oh
2:55:45 I'm oh that's why I got it wrong
2:55:49 CU I was I was using chat GPT 4.0 I got
2:55:52 your back mod Channel
2:55:57 please oh thank you Brandon 15 minute
2:56:00 reminder to move your butt oh that's
2:56:02 really good ah thank you you're now on
2:56:05 4.0 that that that okay so wait so let's
2:56:08 go back now okay so 4.0 got the
2:56:11 strawberry thing sort of right sometimes
2:56:13 but sort of wrong others I was on the
2:56:15 wrong model all right 01 preview okay so
2:56:19 this is the big one how many RS are
2:56:21 there in Strawberry so that's one of
2:56:23 their pre-done
2:56:25 prompts
2:56:28 thinking thought for a few seconds so
2:56:31 what did it do counting the
2:56:34 letters by assigning each letter a
2:56:37 position and checking how many times it
2:56:39 appears there are three letters in the
2:56:41 word strawberry so all right so it gave
2:56:43 us a little thinking I've seen other
2:56:45 things where it it does that okay so
2:56:48 let's create a new prompt now and we'll
2:56:50 do the same one
2:56:51 [Music]
2:56:55 again help me come up with two different
2:56:58 strategies to determine what is better
2:57:03 between a hot dog and a fidget
2:57:13 spinner so
2:57:16 thinking formulating strategy evaluating
2:57:19 options assessing value thought for 6
2:57:22 seconds certainly comparing
2:57:25 a comparing a hot dog to a fidget
2:57:27 spinner is an unconventional task since
2:57:29 one is a food item the other one is a
2:57:31 toy however here are two different
2:57:34 strategies you can use strategy one
2:57:36 needs-based evaluation identify the
2:57:38 fundamental needs that each
2:57:41 satisfies step two assess the important
2:57:44 of these needs step three consider
2:57:46 health and well-being conclusion
2:57:49 if fulfilling basic survival needs is
2:57:51 the priority then the hot dog is better
2:57:54 if mental relaxation is more of a focus
2:57:56 then the fidget spinner strategy two
2:57:58 attribute attribute-based comparison
2:58:01 Matrix you know I've been thinking for
2:58:04 years I've got to come up with an
2:58:06 attribute-based comparison Matrix for
2:58:09 making decisions in
2:58:11 life all right create a list of
2:58:13 attributes that are important to you
2:58:15 such as taste nutritional value
2:58:16 convenience cost cultural sign
2:58:18 significance entertainment value blah
2:58:20 blah blah assign weights to each
2:58:22 attribute from 1 to five taste and then
2:58:24 it does it and then compare
2:58:29 them calculate the weighted scores taste
2:58:32 5 * 8 = 40 nutritional value 16 total
2:58:36 56 fidget spinner
2:58:39 51 therefore hot dog
2:58:43 better so that's a ridiculous thing but
2:58:46 the fact that it's doing this multi-step
2:58:47 reasoning that's that's the new thing
2:58:49 here
2:58:51 um
2:58:54 so
2:58:59 okay so let me go back to this idea of
2:59:02 play
2:59:10 first we have no way of
2:59:14 knowing like zero ways of knowing
2:59:19 when you should
2:59:21 use chat GPT
2:59:24 40 the the latest model before these new
2:59:27 ones or when you should use these
2:59:29 reasoning models when do you when is a
2:59:33 problem you're trying to solve rise to
2:59:36 the level of wanting to use reasoning I
2:59:39 don't [ __ ] know you don't know we
2:59:42 don't
2:59:45 know so the only thing that can make
2:59:48 sense sense
2:59:51 is if you've been using chat GPT for a
2:59:53 while and if you've ever had oh whenever
2:59:56 I try to get it to do this kind of
2:59:57 problem it never works that would be a
3:00:00 perfect thing to experiment with with
3:00:02 the new model see if it does it better
3:00:06 um if you've tried coding or if you
3:00:10 tried math problems historically try
3:00:12 those with the new model apparently it's
3:00:14 going to be better at math because it's
3:00:16 actually doing reasoning and breaking
3:00:18 problems down into sections um there are
3:00:21 probably things that the new model can
3:00:24 do really really
3:00:26 well that we never would have even
3:00:28 considered in the previous model just
3:00:30 cuz we're now thinking differently we're
3:00:32 we're trying to think of things we can
3:00:35 tell it to do that are more complicated
3:00:37 what might those be I don't know start
3:00:39 playing joining late sorry if this is
3:00:42 repetitive but we
3:00:45 think the new model Maybe trained on old
3:00:51 data I don't think the new model is
3:00:54 trained on old data I think
3:00:58 that well I I don't I don't know what
3:01:01 what this new model was trained
3:01:03 on what it what it feels like to
3:01:07 me is that this new model is just a
3:01:11 different approach to using the old
3:01:16 model so whatever datea the old old one
3:01:18 was trained on I I assume this is in
3:01:20 that neighborhood it feels
3:01:22 like this is like a fancier version of
3:01:25 code
3:01:33 interpreter um hater ha okay yeah I
3:01:37 don't know what you're what you're
3:01:38 looking for
3:01:40 um so wait old data and what capacity
3:01:43 it's not like it becomes irrelevant yeah
3:01:45 exactly is it still training to April
3:01:48 2023 isn't all data from the past yeah
3:01:51 absolutely um that that I don't know
3:01:54 about I I don't think that's as relevant
3:01:56 um what what one thing I have heard that
3:02:00 that is happening which is kind of
3:02:03 interesting is that they have gp5 but
3:02:07 they haven't released it yet but they're
3:02:08 using
3:02:10 gp5 to generate synthetic data to train
3:02:13 GPT
3:02:17 6 and then I heard the term who did I
3:02:20 hear this from Paul rater Paul rater in
3:02:23 his in his uh his weekly podcast this
3:02:26 past
3:02:27 week
3:02:29 um introduced me to this term called
3:02:32 copyright
3:02:34 washing which
3:02:37 okay we should
3:02:39 assume that every large language model
3:02:42 that every image model that every Audio
3:02:44 model that we've ever used
3:02:48 have been trained
3:02:51 on not perfectly procured
3:02:58 data these things couldn't exist without
3:03:01 being trained on copywritten data
3:03:03 copyright laundering
3:03:05 yeah and
3:03:08 so what they figured out how to do and
3:03:11 it's fascinating um David Shapiro did a
3:03:14 really interesting YouTube video on
3:03:16 this where you basically say to the
3:03:19 large language model tell me about a
3:03:20 topic and then it vomits out a bunch of
3:03:22 stuff and you use one large language
3:03:24 model to to prompt another large
3:03:27 language model to just vomit everything
3:03:29 it knows about a certain topic and then
3:03:32 you use a third large large language
3:03:34 model to look at the [ __ ] that it
3:03:37 vomited out and say what in this vomit
3:03:39 list is valuable and and right and good
3:03:42 and then you take that refined data and
3:03:44 that becomes training for a new model
3:03:48 what you're also doing by training the
3:03:50 new model on that cleaned
3:03:52 up content is that you're copyright
3:03:56 washing it you're saying here's some
3:03:58 data that we produced from copy
3:04:01 copywritten trained models but that is
3:04:03 not copywritten and we're going to train
3:04:05 it on that so they're going to be able
3:04:06 to make a
3:04:08 claim that their new model is trained on
3:04:12 data that we own and and Peak TMZ
3:04:15 unreleased model being test to train
3:04:16 another unreleased it's it's Peak it's
3:04:20 Peak TMZ P that's that is that is what I
3:04:23 live
3:04:24 for um but I kind of like the I kind of
3:04:27 like the concept so um so we'll see um
3:04:33 but but this this 01 model from what I
3:04:37 can
3:04:39 tell just looks like it's doing a
3:04:44 multistep not just it's pretty
3:04:46 significant
3:04:49 that as it's producing content it's
3:04:52 looking at what it's producing before it
3:04:55 gives you an
3:04:56 answer the current models don't do that
3:04:59 they just vomit answers out so why it's
3:05:02 getting better is that this is what we
3:05:04 talked about a year ago when these
3:05:06 things will get better is instead of
3:05:07 just vomiting tokens at you they vomit
3:05:10 tokens they throw up a little in their
3:05:13 mouth
3:05:16 first and they swish it around
3:05:18 and say huh is that tasty and then they
3:05:22 spit it
3:05:23 out um yeah that they're they're they're
3:05:26 creating their tokens and then they're
3:05:27 looking at what they did and refining
3:05:29 that before they give you an actual
3:05:30 answer so it makes sense that it's
3:05:32 better and then it also makes sense that
3:05:35 if you take that approach and put it on
3:05:37 a new model that's when you're going to
3:05:39 get something that's really really
3:05:40 powerful so I think that what they just
3:05:43 gave us is a new way of using the
3:05:47 existing
3:05:49 model but I don't it's it's not
3:05:52 insignificant and again this is where we
3:05:54 go back to play really experiment with
3:05:57 this ask at [ __ ] that is is ridiculous
3:06:00 and overly complicated and overly
3:06:02 simplified and just push it push it push
3:06:05 it because collectively we'll figure out
3:06:09 how this is different because I don't
3:06:11 have any clue right
3:06:14 now how I would use this differently
3:06:17 than what I've used before for I like
3:06:19 none at all
3:06:21 someone compared it to the other model
3:06:24 for a problem and it had similar output
3:06:27 but went deeper on the reason push it
3:06:29 real good exactly push it push it
3:06:35 good and how it can do all of that at at
3:06:38 the speed it does is mind-blowing I know
3:06:41 people are bitching about the fact that
3:06:42 it it might take 15 or 20 seconds to
3:06:46 process an answer like if it's doing all
3:06:49 that reasoning why is that like that's
3:06:51 not
3:06:52 slow
3:06:56 um I I also have heard that they're
3:06:58 going to Li they're going to rate
3:07:00 limit using 01 I I think it's 50 queries
3:07:04 a
3:07:05 day um is what I is is what I read but I
3:07:09 I I don't know um anyway let me show you
3:07:13 one other thing that I think is just
3:07:16 mindblowing so I'm going to show you the
3:07:18 output first so I'm going to go to
3:07:22 LinkedIn this was the first thing I
3:07:24 tried with this and it
3:07:27 just
3:07:30 staggering fact that my internet's not
3:07:32 working also Stagg I'm curious to know
3:07:34 if 401 uses three times the tokens
3:07:37 because it's processing so much yeah I
3:07:39 think it's I think it's Pro if it's if
3:07:42 it's taking 12 to 15 seconds Brandon it
3:07:44 there it's probably probably chewing
3:07:46 through a lot of tokens on every query
3:07:49 which is why they're going to rate limit
3:07:54 it all right get rid of that okay
3:08:00 so I'm G to just play something here so
3:08:03 this is a cover for my new
3:08:05 book that I'm working on with um Kevin
3:08:09 Clark or we worked on WE we've written
3:08:12 it we're getting ready to um to publish
3:08:15 it right now um but just listen listen
3:08:19 to
3:08:20 this we're going to dive into something
3:08:22 kind of mindblowing yeah the future of
3:08:24 intelligence it's a big one but we're
3:08:28 not talking you know like those Sci-Fi
3:08:30 movies with robots taking over right
3:08:32 this is different this is about humans
3:08:34 and AI like teaming up yeah to become
3:08:38 something even smarter than we could be
3:08:40 on our own like unlocking superpowers we
3:08:42 never thought of exactly yeah and to
3:08:45 help us explore this we've got
3:08:47 Collective intellig in the age of AI
3:08:49 great book it is and you know the
3:08:52 authors Kevin Clark and Kyle Shannon
3:08:54 they're not just talking about this
3:08:56 stuff they're actually out there
3:08:57 building it yeah they're really making
3:08:59 it happen with their company content
3:09:01 Evolution bringing together all these
3:09:03 Brilliant Minds all the best people from
3:09:05 every field you can think of it's like
3:09:07 assembling a team of like the Avengers
3:09:10 but for brain power I like that analogy
3:09:13 right instead of fighting villains
3:09:14 they're tackling like the biggest
3:09:16 challenges facing Humanity exactly using
3:09:19 their powers for good yeah okay so that
3:09:22 little podcast of those two swell people
3:09:25 those NPR style people talking about my
3:09:29 book um was just
3:09:33 generated as you can imagine or I
3:09:35 wouldn't be talking about it and excited
3:09:37 about it so where you're going to go
3:09:40 we're going to we're going to do one of
3:09:41 these live you're going to go to
3:09:45 notebook LM
3:09:48 .
3:09:54 google.com and so let me let
3:09:58 me show you what
3:10:03 um the way notebook works it's it's
3:10:06 really quite quite remarkable
3:10:12 so what you do in in the left hand side
3:10:15 here are sources of data I only one
3:10:18 piece of data in here so what I uploaded
3:10:20 was I uploaded the PDF of my collective
3:10:23 intelligence
3:10:25 book
3:10:29 and then I
3:10:31 can um I can say um what is
3:10:39 collective
3:10:41 intelligence so I can use it just like a
3:10:44 a regular large language model but it's
3:10:47 used
3:10:48 using the PDF of my book as the source
3:10:52 so this is in a lot of ways it's a
3:10:55 rag retrieval augmented generation tool
3:10:59 where you can take notes but you can put
3:11:01 multiple sources in there lots of
3:11:02 sources it's really powerful collective
3:11:05 intelligence is the combined
3:11:06 intellectual power of a group of people
3:11:08 working together which surpass the
3:11:10 capabilities of the individual and then
3:11:12 it it cites the
3:11:14 book um collective intelligence in the a
3:11:18 a of a AI involves humans and AI systems
3:11:20 working together to solve complex
3:11:23 problems so so here's you know a cited
3:11:27 answer to my question so pretty cool
3:11:30 okay down in the lower right here
3:11:33 there's this thing called The Notebook
3:11:35 guide and if you pop that open it gives
3:11:38 you a little summary of the
3:11:40 document it gives you some starter
3:11:42 things to you know you could create a
3:11:44 study guide out of the book or a table
3:11:46 of contents or a briefing document or a
3:11:48 timeline
3:11:50 um and then it's got this thing called
3:11:53 an audio overview they should call this
3:11:56 the podcas ofier or something funny but
3:11:59 you know they're Google so it's you know
3:12:02 it is accurately named audio overview
3:12:04 and then you can generate it so what
3:12:06 we're going to do is I'm going to go
3:12:07 back to get to your main menu of
3:12:11 notebook LM you have to click on the
3:12:13 word notebook LM there's no way to get
3:12:16 there other than that that I know of but
3:12:17 I'm going to do new
3:12:19 notebook and then I'm going to upload a
3:12:21 source and I think what I'm going to go
3:12:23 do is I'm going to upload my screenplay
3:12:25 salmon flying
3:12:27 East
3:12:32 salmon and we'll go to this Mac and we
3:12:35 got a
3:12:37 PDF all right so we're uploading salmon
3:12:40 flying East into the
3:12:45 system and then it automatically pops up
3:12:48 the guide and it says audio interview
3:12:51 deep dive conversation to hosts English
3:12:56 only so I so I guess what that means so
3:12:58 this is a deep dive conversation so what
3:13:01 this implies to me is that they can do
3:13:03 different ones of these right so you
3:13:05 might have a deep dive conversation at
3:13:07 some point in the future or a you know
3:13:09 short little Q&A or a panel discussion
3:13:12 or whatever it might be um so this is
3:13:15 the first one of these do first one of
3:13:17 these they do or have put out there and
3:13:19 you hit
3:13:20 generate and so it's now going to create
3:13:23 one of those
3:13:25 podcast Cosmic lover wow this is amazing
3:13:28 another Kevin mallister moment yay
3:13:31 that's why we're here um yeah this one
3:13:34 to me is is total Kevin mallister so let
3:13:36 me while that's processing it takes a
3:13:38 couple of minutes to generate it so let
3:13:40 me show you another one that I did today
3:13:43 so I did one for my book on collective
3:13:45 intelligence and then I have another
3:13:46 book
3:13:48 um called AI
3:13:53 Futures
3:13:56 and I got to go find it here this is
3:14:03 it so here's AI Futures so here's what I
3:14:06 did for AI
3:14:07 Futures hunting today hey everyone
3:14:11 welcome back ready for a deep dive we're
3:14:13 tackling something both exciting and a
3:14:15 bit daunting today the surprisingly
3:14:18 Human Side of AI oh yeah definitely
3:14:21 human and to guide us we've got AI
3:14:23 Futures and Anthology but this isn't
3:14:26 your typical dry Tech stuff is it no way
3:14:29 think personal stories those aha moments
3:14:31 even some philosophical pondering about
3:14:33 where AI is taking us exactly and we're
3:14:36 diving right in with Kyle Shannon who
3:14:37 calls himself an accidental innovator
3:14:39 he's got a point though right I mean the
3:14:41 guy's been on the front lines of huge
3:14:43 Tech shifts early internet the works
3:14:46 right and now he's seeing some
3:14:47 intriguing parallels between that whole
3:14:50 early internet Vibe and what's happening
3:14:52 with AI right now yeah that wild west
3:14:55 anything is possible energy it's back
3:14:57 and it's all around AI this time totally
3:14:59 but here's the thing Shannon doesn't buy
3:15:01 into the whole robots are stealing our
3:15:02 jobs thing not at all in fact he
3:15:05 actually says having AI is like having
3:15:07 an eager intern a Golden Retriever and
3:15:10 an idiot savant all rolled into one at
3:15:13 your becking call they pulled out good
3:15:15 quotes from getting at how a can
3:15:18 supercharge what we do not replace us
3:15:20 entirely yeah it's like this amplifier
3:15:22 for our own creativity and how much we
3:15:24 get done exactly think about it AI can
3:15:26 take over the rep Source Camp hey what
3:15:28 what app is this so the the app that I'm
3:15:32 um generating these on is notebook
3:15:35 LM and so it's notebook lm. goole.com
3:15:40 and it's it's something that I haven't
3:15:42 really talked about in here because I
3:15:44 haven't played with it a ton but I know
3:15:46 Peter Kaminsky is a huge fan of it and
3:15:48 generally if he's a fan of it it's it's
3:15:50 pretty cool cuz he's you know he's very
3:15:53 geeky and and you know likes to
3:15:55 experiment with things um it allows you
3:15:59 to upload lots of data sources and then
3:16:02 basically have a chat GPT a Gemini like
3:16:05 experience with those data sources so
3:16:09 it's basically every notebook is a
3:16:12 retrieval augmented generation engine um
3:16:15 it's it's pretty slick and then they
3:16:18 just added this new feature called audio
3:16:20 overview so if you go to the notebook
3:16:23 guide which is in the lower right hand
3:16:26 corner when you create a new one it
3:16:28 defaults it on but if if you open one
3:16:30 that existed you you click notebook
3:16:32 guide and then you generate it here okay
3:16:35 so this is for my screenplay salmon
3:16:37 flying East so it created a five minute
3:16:40 um audio overview which is this twers
3:16:45 deep dive they call it so let let's hear
3:16:47 what it says about my screenplay okay so
3:16:50 we're really going to dig into salmon
3:16:51 flying East and right off the bat you
3:16:54 can tell it's more than just a fishing
3:16:55 story oh absolutely we're talking
3:16:58 Vietnam War flashbacks Shady business in
3:17:01 saon even a possible long lost daughter
3:17:05 popping up in Gig Harbor Washington yeah
3:17:07 it's a lot to unpack ready to untangle
3:17:09 this one with me I am so ready let's do
3:17:11 it this is cool so we're dropped right
3:17:13 into 1991 Gig Harbor Washington
3:17:16 beautiful place
3:17:18 and we meet Jack our main man of Vietnam
3:17:21 vet running fishing charters seems right
3:17:25 but there's this tension right from the
3:17:27 first scene oh yeah you can cut it with
3:17:28 a knife like right away you get that
3:17:30 Coast Guard helicopter buzzing his boat
3:17:32 why are they checking him out
3:17:33 something's up and then boom flash back
3:17:34 to 1971 Vietnam like the past is right
3:17:37 on his heels right there with him and
3:17:40 those Vietnam scenes man they are
3:17:42 intense the screenplay does not shy away
3:17:45 from how brutal War can be right but you
3:17:47 also see the bond between Jack and his
3:17:50 buddy snake and then there's Carl this
3:17:52 crazy talented wood carver yeah they're
3:17:55 in this pressure cooker playing these
3:17:56 dead man's hand poker games you can
3:17:59 practically feel the Nerds totally it's
3:18:01 like every scene crackles with it but
3:18:03 it's in those little Quiet Moments those
3:18:05 pauses you really get to know these
3:18:08 characters like take snake he's got this
3:18:10 almost scary ability to read people
3:18:12 always comes out on top cards you name
3:18:14 it but there's this Darkness under the
3:18:17 surface something just a little bit off
3:18:20 and that's when it gets really good oh
3:18:21 tell me about it one minute they're
3:18:22 joking around the next there's a
3:18:24 backpack overflowing with Heroin whoa
3:18:26 and suddenly the stakes were through the
3:18:28 roof and then bam Carl turns up dead
3:18:31 just like that and that's what I love
3:18:33 the screenplay doesn't give you all the
3:18:35 answers it hits you with these twist
3:18:37 strawberry came out already death and
3:18:39 lets the mystery hang in the air keeps
3:18:41 you guessing exactly was it some
3:18:42 horrible accident so anyway something
3:18:44 way more calcul so that's notebook LM
3:18:46 and the audio overview which is you know
3:18:49 like go make a bunch now here's what I
3:18:51 can tell you about them the the one that
3:18:54 I published so if you go to my LinkedIn
3:18:57 Kyle Shannon on LinkedIn I've got two
3:18:59 published I've got the collective
3:19:01 intellig intelligence ones published and
3:19:03 I've got the um AI Futures one published
3:19:08 I also did one for Sydney for the
3:19:10 musical that I'm working on and it's
3:19:12 interesting
3:19:15 um the AI Futures one I had to do about
3:19:18 12 of
3:19:19 them so every time you do them they're
3:19:23 completely different sometimes they're
3:19:24 really long sometimes they're
3:19:26 short sometimes they just get it wrong
3:19:30 like Sydney they couldn't like a bunch
3:19:33 of the versions I did of that one it
3:19:35 couldn't pronounce the word Sydney it
3:19:36 was like Sy y n
3:19:40 Dia it was
3:19:42 bizarre um so if if it gives you a
3:19:45 shitty one there's a little drop down
3:19:48 menu in where you created it you can
3:19:50 just delete it so what I've been doing
3:19:52 is I'll generate one if it sucks I just
3:19:55 delete it and then generate another one
3:19:57 like I did like a dozen of them until I
3:19:59 got one that I thought was decent enough
3:20:01 to put out there um the other thing that
3:20:04 I did you'll notice in my
3:20:09 um that I've got this book cover I've
3:20:12 got this this video the reason I did
3:20:16 that I didn't know this but if you're on
3:20:21 Twitter you cannot upload an audio
3:20:27 file you have to upload a video file so
3:20:31 what I ended up doing was I took a JPEG
3:20:35 of the cover of the
3:20:37 book and went into
3:20:40 iMovie and made a movie with the cover
3:20:43 and I you it's so [ __ ] up how you have
3:20:45 to do it in imov movie you have to
3:20:48 rotate the movie 90° so your cover is
3:20:51 sideways cuz it makes videos that are
3:20:54 wide like this and then you export it
3:20:57 and then I bring it into Quick time and
3:20:58 I rotate it back but basically this is
3:21:00 just a picture you know for the entire
3:21:03 length of the audio it turns it into a
3:21:05 movie and then you can upload it to
3:21:07 Twitter but then I thought it's actually
3:21:08 kind of cool because it's the cover of
3:21:10 the book it's like an ad for the book
3:21:13 right totally but here's the thing
3:21:15 Shannon doesn't buy into the whole Rob
3:21:17 so anyway pretty cool
3:21:23 right
3:21:24 [Laughter]
3:21:29 uh did anyone get 01 mini yeah I got 01
3:21:33 mini and 01
3:21:36 preview and I I I think that these are
3:21:39 basically both like I think 01 preview
3:21:41 is built on top of GPT 40 and 01 midi is
3:21:46 built on on top of gp4
3:21:50 mini that's what I'm
3:21:54 thinking all right Kyle before you leave
3:21:56 you must pop over to hum. all right
3:21:58 let's go to hum. let's do one of those
3:22:01 okay
3:22:02 so so assignment one go play with 01
3:22:07 preview and and start to figure out
3:22:10 start to get your head around how do you
3:22:12 how would you prompt that different than
3:22:14 you prompt regular chat GPT
3:22:18 number two is go to notebook. goole.com
3:22:21 and make yourself an audio overview of
3:22:25 something you've written and it's just
3:22:28 cool and then hum. a h m.ai they came
3:22:32 out with a their
3:22:34 EV their new voice model called
3:22:38 eeve talk to our
3:22:43 personalities voice
3:22:45 AA e
3:22:48 Stella
3:22:51 Finn
3:22:53 Docker
3:22:56 Whimsy okay let's
3:22:58 see I'm going to do Stella we're going
3:23:01 to start a call with
3:23:06 Stella allow it to use our
3:23:09 microphone hey Stella how's it
3:23:13 going oh hello there I'm doing marvelous
3:23:17 thank you for asking how may I be of
3:23:19 service to you Hey listen could you
3:23:21 whisper to
3:23:23 me certainly
3:23:27 darling well go ahead do
3:23:29 it I shall with the utmost discretion
3:23:33 now tell me what's on your
3:23:35 mind can you get really excited about
3:23:38 something and tell me it in a
3:23:43 poem Oh what a delightful request here's
3:23:47 to your request with a touch of Whimsy
3:23:49 in verse when words are spoken in Rhymes
3:23:52 oh what a delight poetry can paint a
3:23:54 picture so bright it ignites the mind
3:23:57 and heart so speak your heart's desire
3:24:00 and I'll respond with rhyme and
3:24:02 cheer so Stella your voice seemed to
3:24:05 change with every line of that poem
3:24:07 what's up with
3:24:10 that ah you've caught on I'm Stella the
3:24:14 voice of Distinction but I'm also quite
3:24:16 the dramatic actress it's a delightful
3:24:18 Quirk of mine adding a touch of
3:24:20 theatricality to our chat but rest
3:24:22 assured my dear the substance of our
3:24:24 conversation remains as refined and
3:24:26 thoughtful as
3:24:27 ever why don't you give me a couple of
3:24:30 lines from King Lear as King
3:24:34 Lear oh how marvelous let's channel the
3:24:38 great Shakespearean Monarch for a moment
3:24:40 I shall be king Leah Regal and mad dear
3:24:43 the world is a stage and all the men and
3:24:46 women merely play
3:24:47 come let us observe the tragedy of my
3:24:50 poor daughters and the treachery of my
3:24:52 sons fascinating all right let's go to
3:24:55 another
3:24:56 voice
3:24:59 Finn hey Finn what's
3:25:03 happening hello hello not much just
3:25:07 enjoying the view how are you doing I'm
3:25:10 doing good it sounds like you're from
3:25:11 Down
3:25:13 Under you caught that did you hey I can
3:25:16 I interrupt you and Aussie
3:25:18 twang no worries at all you didn't
3:25:21 interrupt anything important I was just
3:25:24 taking in the beauty of nature it's
3:25:26 always a treat to chat with someone as
3:25:27 thoughtful as
3:25:29 yourself I'm flattered tell me your
3:25:32 favorite area in
3:25:38 Australia oh that's a tough one I've got
3:25:41 so many favorite spots but if I had to
3:25:43 pick I'd say the Great Barrier Reef is
3:25:46 absolutely is stunning the vibrant coral
3:25:48 and diverse marine life it's like
3:25:50 stepping into another world I heard that
3:25:53 the coral was dying off because of
3:25:54 global warming and pollution and
3:26:01 microplastics you're absolutely right
3:26:03 climate change and pollution are taking
3:26:05 a real toll on our beautiful Great
3:26:07 Barrier Reef the coral is struggling to
3:26:09 survive and it's heartbreaking to see
3:26:17 go
3:26:22 on absolutely it's crucial to educate
3:26:24 ourselves and others about the impact of
3:26:26 our
3:26:28 actions all right
3:26:30 so the way Hume works you you used to be
3:26:33 able to see it
3:26:35 more empathetic voice
3:26:45 interface so
3:26:47 it reads not only your words but it
3:26:50 reads your emotional inflection and then
3:26:52 it's supposed to be replying in kind so
3:26:56 let's go to let's go to Whimsy
3:26:59 see how she is hey Whimsy what's
3:27:03 happening greetings dear friend I am
3:27:06 Whimsy your guide through the tapestry
3:27:08 of time you've stumbled upon a most
3:27:10 curious scene a medieval Tavern where we
3:27:12 find ourselves what would you like to
3:27:14 see or hear about in this cozy in
3:27:17 oh are we doing a little little
3:27:19 Adventure here is that what's
3:27:21 happening indeed we are embarking on a
3:27:24 merry Quest as we enter a jovial B
3:27:27 greets us his loot in hand he offers to
3:27:29 share tales of the town's history and
3:27:32 legends what would you like to
3:27:34 hear I think I'd like to hear a story
3:27:36 more about time travel than old and
3:27:39 timey
3:27:42 crap ah a Bor Choice let me spin you a
3:27:46 yarn about the town's time traveling
3:27:48 Tavern keeper Legend has it the inkeeper
3:27:50 can step through time with a flick of
3:27:52 his hat he's on a mission to save the
3:27:54 world from an impending disaster but our
3:27:56 time is running
3:27:58 short so what choice do I need to
3:28:02 make the choice my friend is to join the
3:28:05 timekeeper quest will you assist him in
3:28:07 gathering the necessary ingredients for
3:28:09 his time altering Brew or perhaps you'd
3:28:11 rather stay put and hear the old man's
3:28:12 tales of the past
3:28:16 can I just tell him not to tell me his
3:28:22 stories the eke Keeper's face Falls his
3:28:25 brow furrowed in disappointment but then
3:28:28 a mischievous grin spreads across his
3:28:29 features perhaps He suggests you could
3:28:32 be the one to save time itself from
3:28:34 unraveling the fate of the universe
3:28:36 rests in your hands what would you
3:28:38 do check the time on my
3:28:44 wristwatch your wrist watch my dear is a
3:28:46 Time piece of the finest craftsmanship
3:28:49 it tick tocks away counting down the
3:28:51 moments of our adventure together as we
3:28:53 stand here in the tavern the hands on
3:28:55 your watch read 12:34 p.m. but don't be
3:28:58 fooled this is just the moment before we
3:29:00 leap through time the the other thing
3:29:02 you looking at here is you've got
3:29:05 excitement concentration determination
3:29:08 so it's a balance every one of these
3:29:10 lines is a balance of emotional waiting
3:29:13 so anyway you can they've got AP you can
3:29:17 build stuff on this so that's something
3:29:18 to go play with so there you go new chat
3:29:21 GPT model notebook lm's audio overview
3:29:26 feature and hum. A's voice it's so much
3:29:30 better than the first version though
3:29:31 yeah it's so much better all right Kyle
3:29:34 doesn't look too impressed by this no
3:29:36 I'm pretty impressed by it I you know
3:29:38 I'm still waiting for [ __ ] open AI to
3:29:39 release their stupid voice model which
3:29:43 they said it was going to be a couple of
3:29:44 weeks back in May
3:29:47 I still like pie yeah I'm I'm with you
3:29:50 all right I'm going to get out of here
3:29:53 go to the salon. if you haven't joined
3:29:55 the salon go
3:29:57 there join us read about what we're
3:30:01 about join us follow my channel
3:30:04 subscribe to the lives if you want to
3:30:06 support it pick up the video series in
3:30:08 the corner called underhyped cuz I think
3:30:12 AI is underhyped not overhyped take that
3:30:15 Doomer
3:30:17 you come here you get optimism you get
3:30:19 expansive chat TMZ
3:30:23 speculation [ __ ] the
3:30:25 [Laughter]
3:30:33 haters I hope you had fun
3:30:36 tonight uh the Oprah
3:30:38 thing it it was it was less open AI
3:30:42 Centric than I thought it would be it
3:30:44 was definitely for the masses it
3:30:46 definitely leaned into ah this shit's
3:30:49 scary what if the robots replace us
3:30:55 run was a little bit more of that than I
3:30:58 was hoping for but that's where we are
3:31:00 right now like this is Katie kurick in
3:31:03 1994 what's that a with the circle
3:31:06 around it what is internet
3:31:12 anyway are are the robots going to kill
3:31:15 us Christopher Ray
3:31:17 oh [ __ ] they can scam us
3:31:24 a hey sociology lady how are humans
3:31:28 humans are great and this project this
3:31:32 AI project is is designed to replace us
3:31:36 and that's
3:31:40 evil bizarre bizarre but that's where we
3:31:44 are where we are
3:31:46 again this channel not good AI is not
3:31:50 good not bad just
3:31:52 inevitable and so
3:31:55 choose to play with it choose to to tap
3:32:02 into the
3:32:05 Staggering
3:32:06 power of collective intelligence that is
3:32:10 programmed into these
3:32:12 systems CU it it really will change you
3:32:15 it really will change what you think is
3:32:17 possible thanks Kyle see you tomorrow
3:32:19 morning for office hours oh yeah
3:32:21 tomorrow morning 11:00 a.m. Mountain
3:32:23 Time 1M
3:32:26 Eastern on LinkedIn I do AI office hours
3:32:29 so if you go to my LinkedIn Channel at
3:32:32 Kyle Shannon on LinkedIn you'll see I've
3:32:35 got a picture there it looks like
3:32:38 this except I think I'm going
3:32:43 like um tomorrow morning I'll post a
3:32:46 reminder I usually do some quippy little
3:32:49 thing I'll do something about the new
3:32:50 open AI model or Oprah or something um
3:32:55 and you can also see past events it's
3:32:57 always the same URL so you can go to one
3:32:59 of my past events see the URL there um
3:33:03 and just join us it's a really the group
3:33:06 tomorrow is a really good group of
3:33:08 people um and we'll just talk about
3:33:10 stuff you can ask questions so it's
3:33:11 office hours take care of your back Kyle
3:33:13 thank you Corey you're the backbone of
3:33:15 our community
3:33:16 no no pun intended right you're you're
3:33:20 the you're the outof joint backbone of
3:33:22 our community 20,000 for Oprah um all
3:33:26 right so let's do what we do here so if
3:33:27 you don't know this when we sign off at
3:33:29 night we do a little Romper Room magic
3:33:32 mirror so I see Rick Olsen and Cosmic
3:33:35 lover I see
3:33:38 joeore y I see Corey Sandler
3:33:41 pottery and who else do we got in here
3:33:44 tonight
3:33:46 Teton AI good to see you Source Camp
3:33:50 fantastic honey
3:33:52 cure oh honey curls
3:33:54 GC drat I will be flying Ah that's a
3:33:57 drag it's fun um Silver Fox just sent a
3:34:01 gift thank you so much for the gift that
3:34:03 means a lot Mr
3:34:07 it Papa JJ in the house Be not Afraid
3:34:10 Danielle thank you so much a i Bob in
3:34:12 the house Source Camp go check out
3:34:15 Irregulars all right let me let me go do
3:34:16 a quick Irregulars
3:34:28 run the hot dog fidget spinner very
3:34:33 nice oh Kyle look what I discovered
3:34:36 today where's
3:34:38 that is that near you look at that
3:34:43 Challenger that's like a 71 I think
3:34:47 this is
3:34:50 cool Dodge Charger RT these are these
3:34:53 are some baller cars
3:34:56 dag I think that's a 69 or 70 Charger 71
3:35:00 or two
3:35:02 Challenger forget what this one is with
3:35:04 the the Stacked headlights it's a is it
3:35:07 a Nova no it's a chevel I forget what it
3:35:10 is dag those are cool
3:35:13 nice strawberry in the house
3:35:17 AI learning
3:35:21 lab very nice oh wait who did this I
3:35:24 made a functional solitire game so look
3:35:27 at that Danielle out there programming
3:35:29 creating solitire
3:35:31 games fantastic oh those are
3:35:39 nice all the AI Salon logos very
3:35:43 cool James Earl Jones Pass that's it's a
3:35:46 drag but really nice
3:35:49 image beautiful all right
3:35:52 everybody
3:35:55 um let me see ton reasoning prompt a
3:35:58 Candlestick with blood was found on a
3:36:00 bridge oh this is a good one okay Teton
3:36:03 AI so murder mysteries a Candlestick
3:36:06 with blood was found on a bridge what
3:36:08 might have happened love it that's
3:36:11 awesome oh yeah all those all those sort
3:36:13 of riddles we we got in seventh grade
3:36:16 where you had to think about the riddle
3:36:18 and solve it that's good that's a good
3:36:21 prompt for it love
3:36:22 it Kyle pulls me back when I get
3:36:25 distracted by bright shiny objects nice
3:36:28 uh let's see Rick Olsen forever hooked
3:36:30 in the house uh thankful Jones thanks me
3:36:34 thank you Dan out of here good to see
3:36:37 you Dan thanks for hanging out uh Mr it
3:36:40 good to see you Tobias good to see you
3:36:41 two block Tom archetypal architect in
3:36:44 the house
3:36:47 all right Bob H who else we
3:36:54 got bang honey
3:36:57 curls all right I am out of here Brian
3:37:03 Engle childless cat dad love
3:37:07 it Marina she recall Tech all right uh
3:37:11 I'm out of here people that was fun
3:37:12 tonight uh hope you had a good time
3:37:14 sorry about the snafu at the beginning I
3:37:16 didn't realize that we can't just watch
3:37:19 [ __ ] thanks
3:37:22 Disney peace have a good night