aiLL 082623 - ChatGPT Disrupts Everything: AI & Education Now Must Adapt

Description

Chapters

Transcript

0:01 foreign
0:31 we are back all right tick tock throw us
0:34 in the algorithm please stop Shadow
0:36 Banning me
0:39 I am I am a tireless participant on this
0:43 [ __ ] platform
0:44 [Laughter]
0:49 all right all righty then good day good
0:53 day
0:54 Stevo what's happening thank you for
0:57 welcoming me back hopefully
0:59 my internet sucks less
1:02 my boys were kind enough to let me blow
1:05 them off some game they were playing
1:12 they're Jay
1:14 I am back indeed I'm back it's gonna be
1:17 so good
1:18 it's gonna be so good good good good
1:21 good good
1:23 yeah that one's cool I'll do that one
1:26 double feature tonight so awesome thank
1:29 you for coming back you're welcome
1:31 you're welcome I feel like I'll be right
1:33 back I just never come back
1:36 go get go get drunk in the corner yeah
1:39 no one was watching me man my internet
1:41 suck
1:44 knock it back please do another
1:46 explanation of Ideal lab to me yeah I'm
1:48 going to I'm going to hop over and do
1:50 that let me let the algorithm do its
1:53 thing get some folks in here welcome if
1:55 you're new here uh in mid-journey can I
1:57 view someone's
1:59 uh and their catalog of pictures well
2:02 you can see images
2:05 in the mid Journey Channel you can see
2:07 images and all the channels I assume if
2:09 you look at
2:11 if you do a a Discord search for
2:14 mentions their name you'll be able to
2:16 see all of their their posts but I don't
2:19 quite I haven't really paid attention to
2:21 that and I don't know if you can let me
2:24 see if I can just look at someone else's
2:27 account
2:28 on the web
2:31 this is a cool idea I've got for some
2:34 flash cards
2:36 let's see uh mid-journey
2:39 man Journey me Journey me Journey
2:43 sign in
2:49 yeah I don't know it doesn't look like
2:51 like the URL doesn't look like you can
2:53 just go to someone else's account
2:55 but you might be able to I don't know
2:58 I do not know the answer to that
3:00 question
3:02 I showed you Tick Tock live with my ex
3:05 she says you're better than me
3:08 that's that's just that's just painful
3:13 X's are rough
3:15 Tobias with shaking man X's are up
3:18 welcome everybody this is the AI
3:20 learning lab you know how we know that
3:21 because we got a sign
3:23 now where did we get that sign
3:26 where did we get that sign
3:28 well clearly you're really good at
3:30 Photoshop and you made it in Photoshop
3:31 because I know you didn't get that in an
3:33 AI image generator because they don't do
3:36 text right
3:38 um
3:39 welcome to the AI learning lab let's go
3:41 figure that [ __ ] out there's a new kid
3:43 on the Block
3:46 like brand spanking new
3:51 so yes this was generated with its
3:55 ideogram or ideogram ideogram.ai
4:00 so was that
4:02 so was that so was that
4:06 so let's go over there
4:09 [Laughter]
4:10 um
4:11 [Music]
4:16 I did a video on video I go to my
4:18 channel I did a video on it as well but
4:21 Tick Tock decided not to show it
4:28 all right so this is it so it's just
4:31 like any any one of these other things
4:32 where you can kind of scroll and see you
4:34 know other people's stuff but notice
4:36 what's in here
4:38 Rumble Brave creative hacking
4:41 Pluto Barnyard Workshop reclaimed it was
4:44 spelled wrong that's pretty funny any
4:46 man's barber
4:49 um
4:50 foreign
4:52 body uh there's always that it's always
4:56 uh White pets can't jump funny comedy
4:59 Kentucky I will kill human Okay so this
5:04 thing can do images let me show you mine
5:07 so
5:08 I did AI learning lab soda cans I did
5:11 some Route 66 stuff
5:14 I did some Times Square AI learning lab
5:16 it doesn't always get it like you know
5:18 sometimes it gets the AI sometimes it
5:20 gets this is like AI a i lyron Ollie
5:24 squire
5:28 um AI learning lab
5:32 um
5:34 but like that's super cool didn't quite
5:36 get it but like that's a really cool
5:38 look
5:39 this is like back alley this was like a
5:42 horse in time the first one I did was a
5:45 horse in Times Square
5:47 um wearing a billboard and it actually
5:49 did pretty good
5:53 storyvine rocks
5:57 story Vinny rocks
6:02 um okay so let's go let's go play
6:05 so I'll do
6:06 um
6:08 oh I don't know give me an idea wait let
6:10 me go to the bottom here give me an idea
6:12 uh what kind of image to do I'll do a um
6:18 I'll do okay
6:20 uh 18 wheeler
6:25 rolling down the highway
6:29 at sunset
6:33 with AI learning
6:37 lab
6:41 on the side of the trailer
6:48 and then you can you can there's these
6:51 modifiers can I make this any bigger so
6:53 you can see it a little yeah a little
6:54 bit
6:56 so there's these modifiers so photo
6:58 cinematic
7:00 um typography 3D render fashion uh what
7:04 else architecture Dark Fantasy vibrant
7:06 we'll do vibrant
7:09 and then you say generate and then you
7:11 can also do different aspect ratios so
7:13 I'll do some horizontal ones because
7:15 this is a truck we want to get good wide
7:17 we got a wide load coming through let's
7:19 do a wide image with a wide load it's
7:22 gonna be so good with the wide load wide
7:25 load Devo band but with Grandma's wait
7:28 what what are you talking about what are
7:31 you talking about you know
7:34 all right
7:36 I don't see any words I see no words
7:44 no no words
7:47 we got nothing
7:50 we got a weird truck with
7:52 three sets of wheels in the back and
7:54 then three back there it's carrying
7:56 something very heavy and it's pulling a
7:57 second trailer
7:59 um okay let's see
8:01 um
8:02 18 wheeler
8:05 with AI learning
8:09 lab
8:11 on the side
8:16 uh
8:22 thunderstorm and rain I don't know let's
8:27 just try that
8:28 all right we were on course for a record
8:31 then tap Kyle rebooted it's his fault if
8:34 we don't make it
8:37 what are you up to you're up to 4 000
8:40 likes what's it been two minutes
8:42 you guys are amazing
8:45 uh this isn't working all right so let's
8:47 do let's do
8:49 um
8:50 road side billboard
8:55 with
8:57 a I wait AI learn wait learn in lab
9:04 and a turtle
9:08 if you know you know
9:11 they're like doodles
9:15 hey trium truth
9:20 oh trucker hands oh look we got an AI
9:24 learning live in the turtle it's gonna
9:26 work out
9:27 I don't this this might be it something
9:30 AI learning lab turtles AI learning lab
9:35 sorta that one sort of works
9:40 are those ready
9:44 close
9:46 AI canning lab
9:53 better my my uh things were better today
9:57 let's see let me try let me go back in
9:59 here
10:00 let me get rid of cinematic and vibrant
10:03 and let me put in typography maybe
10:05 that'll help
10:07 roadside with with let's see design type
10:14 that's that says
10:19 I don't quite know how to do this yet
10:22 see what that does
10:24 canned Turtle can't Turtle I'm not
10:27 totally enough for your Turtle Club you
10:29 got to be pretty turtley Laura you can't
10:31 just come in here and be all like
10:33 rabbity
10:35 if you know if you're gonna if you're
10:37 gonna lean turtle
10:39 you gotta you gotta slow the [ __ ] down
10:42 man
10:43 it's gonna be like um
10:47 a i i learning lab that's like our
10:50 little hacker oh this is a cute one look
10:52 at this one that's that it's like a cute
10:55 little dude and hey hi learning that
10:58 this is some other language AI learning
11:01 lab Turtle ain't Turtle AI or something
11:07 that one's actually pretty close so now
11:09 you can do this thing so now I can remix
11:11 this so I can say okay I like that let's
11:14 remix it let's add cinematic back in
11:17 let's add in a dramatic
11:23 sky
11:25 and let's see what that gives us
11:33 um text even if even if it you know if
11:36 it only hits about 25 percent of the
11:38 time it's still it's still you know for
11:41 for concepting it's great you know you
11:44 could potentially sort of massage this
11:46 to the point that you get it to generate
11:48 something good and then you just have
11:49 minor you know
11:51 um cloning to do in Photoshop this this
11:53 actually looks good all three of these
11:54 say AI learning lab wow these all three
11:57 of these are good or four of these look
11:59 at that
12:00 that's that's freaking solid
12:05 solid
12:07 solid
12:10 yeah I mean
12:12 not for nothing like what you know what
12:14 have you been playing for three minutes
12:15 and I'm and I'm just [ __ ] around
12:17 like if you're in here with some real
12:19 intention it also has a cool thing where
12:21 it's got here's your parent image so I
12:22 can go back out to the parent image so
12:24 if this remix isn't going well I can go
12:26 back out to the parent image
12:29 um I'm trying to think if there's
12:31 anything else I mean that's that that's
12:32 really kind of it I mean it's a super
12:34 simple tool and then they've got a
12:36 Discord
12:38 Channel or server
12:41 and how many people does it have in it
12:44 it has
12:46 online 250
12:52 .
12:53 [Applause]
12:57 I don't know how many in total
13:04 like business inquiry like I I made I
13:06 made two comments in here today and I
13:08 got a response from them like they're so
13:11 new that they're still responding to
13:13 Discord comments
13:14 [Laughter]
13:19 yeah like like I put in a thing I'm
13:21 working on a project right now where we
13:23 would go to immediately sign up if you
13:25 had API access for what it's worth you
13:28 know I'd suggest you talk to these two
13:29 people
13:30 we don't have an API right now please
13:32 send your commercial proposal an
13:35 approximate capacity requirements yeah
13:37 see These Guys these guys are just
13:39 figuring this stuff out they they they
13:41 figured out some [ __ ] magic here I
13:43 mean the fact that stable diffusion and
13:47 um mid-journey and Dolly haven't figured
13:50 this out yet ideogram yes idiot Ido gram
13:55 dot a i
13:58 so anyway how cool is that if we're cool
14:01 I like it I like it a lot I like it it's
14:04 so good it's so good yeah it's it's
14:07 pretty good you know it's pretty good
14:10 that's pretty good
14:13 I'm all like freaked out now now that I
14:15 got that warning like you said something
14:17 bad we're gonna delete you
14:19 I didn't say anything
14:22 I might have said something I don't
14:24 think I like like said anything speaking
14:27 of saying something why do we have a
14:28 turtle on our billboard well if you're
14:31 new here welcome
14:34 uh
14:37 [Music]
14:42 what's going on
14:48 [Music]
14:55 back here live at the waterfront village
14:58 with my friend the zombie Jonathan
15:00 you're looking to Jonathan just got an
15:02 awesome face paint job what do you think
15:04 I like turtles
15:06 all right your great zombie and good
15:09 times here at the waterfront Village oh
15:12 that gets me every time
15:15 all right you're a great zombie
15:28 oh you should record Jonathan's voice on
15:31 another button I should I actually you
15:32 know what I should get is I can do it on
15:34 my iPad actually maybe I'll do it on my
15:36 iPad I'll do a sound board on my iPad
15:39 because because there's no stupid sounds
15:42 maybe I even record some of my voices
15:44 then I don't have to work I just hit
15:46 buttons
15:47 just just have the AI do all my [ __ ]
15:50 I'll just sit here and be like yeah
15:56 um all right if you if you have
15:58 questions about AI stuff pop them down
16:01 below um let me see there looks like
16:02 there's some new people here let me let
16:04 me let me do something here let me let
16:06 me or ideogram I don't know how to say
16:09 it because I'm I'm more of a simple
16:10 person I'm more I'm more simple
16:14 um they're they're complicated people
16:16 out there I am one of the more simple
16:18 people so so I don't you know I don't
16:21 take particular attention to how certain
16:24 words sound you know some people take
16:27 that they make the it's it's important
16:30 to them that's like that's a big word
16:32 important is a big word I like
16:36 all right
16:38 um if you have not played with chat gbt
16:40 before go there
16:41 chat.openai.com that's chatgpt
16:45 um and if you don't know what to say go
16:47 to prompts.chat open a second window and
16:49 prompts.chat's like a little primer
16:51 there's little documents in there it's
16:52 free all this shit's free I don't work
16:53 for any of these people this is just
16:55 what this list is is
16:58 if you're here and you're like
17:00 why am I watching an old man on tick
17:03 tock on a Saturday night but it was
17:06 about Ai and I feel like I should learn
17:07 that if you're in that category right
17:10 now just go to this thing just go start
17:12 playing
17:13 um and then and then you can ask
17:14 questions about it we answer them you
17:16 can learn through osmosis but but just
17:18 go start playing the stuff's actually
17:19 really important it's quite profound and
17:21 uh it's gonna change a lot so that's why
17:24 I do this every [ __ ] night
17:27 um these things are some more things you
17:29 can play with feel free to screenshot
17:30 this Bard that's that's Google's entry
17:32 into this space
17:34 um perplexity.ai is good for
17:36 um research
17:39 labs.perplexity.ai lets you plays with
17:41 play with metas new open source models
17:44 they they are hosting them for free
17:47 um Pi dot AI is uh what what we lovingly
17:50 call in here chatty GPT it's super
17:53 chatty and they have an app and I I
17:57 actually did this today I talked to it
17:59 for about 15 minutes you put it in
18:00 conversation mode there's like a little
18:02 phone icon in the bottom right you put
18:04 it in conversation mode and you can just
18:06 talk to it and it's connected to the
18:08 internet and it's super chatty and it's
18:09 really good and then poe.com is is good
18:13 as well
18:14 um
18:15 although these are relatively in order
18:17 of value to me so chat GPT the most
18:22 Claude and Bing kind of in second and
18:24 then these are all things I play with
18:26 occasionally then for the images this is
18:29 new I took Adobe off the list Adobe
18:31 Firefly was where uh ideogram is but
18:34 Enneagram does text so they two days
18:36 into their business they're on the
18:38 [ __ ] list this that's how this that's
18:40 how AI is right now stuff's moving fast
18:43 and like you know this list is Gonna
18:45 Change I have a feeling this fall is
18:46 going to be absolutely insane
18:48 I have a feeling Google's gonna release
18:52 Gemini which is going to be a gpt4
18:55 competitor it's probably going to do
18:56 images
18:58 um I think Apple's gonna make an
18:59 announcement
19:01 uh
19:03 well I would be disappointed if they
19:05 didn't
19:07 uh they they made an announcement that
19:10 they've spent the bulk of their
19:13 24.6 billion dollar r d budget
19:17 that's more than the market cap of open
19:18 AI they're they're sort of r d slush
19:22 fund is 24 billion dollars they've spent
19:25 most of that on generative AI in the
19:26 past year or two so
19:30 they're spending some money on this
19:31 stuff so Apple's Gonna Come Out Swinging
19:33 we still haven't heard from Elon musk's
19:35 clever little x dot AI thing
19:39 um we've got cohere we haven't heard
19:41 from yet like there's there's a couple
19:42 of players that we haven't heard from
19:44 yet
19:45 um
19:46 so yeah so so some shit's coming and and
19:49 you know yeah yeah yeah you know we're
19:53 gonna have we're gonna have uh what else
19:54 we're gonna hear we're gonna have uh
19:57 uh Microsoft and Google are both Gonna
20:00 Roll their Tech into Office 365 and the
20:03 Google office suite probably you know
20:06 end of the year beginning of next year
20:08 so a lot of a lot of big stuff coming
20:10 okay
20:12 um
20:13 I mentioned this before but if you're
20:15 new here
20:16 um we have a scammer which that means
20:19 we're popular but unfortunately it means
20:21 you know when you're popular people do
20:23 stupid [ __ ] like this so there's someone
20:25 out there stole all my videos and are
20:27 acting like me and trying to get people
20:29 to sign up for a multi-level marketing
20:32 thing so if if someone that looks like
20:35 me reaches out and says hey you want to
20:37 join my multi-level marketing thing it
20:39 ain't me it's an [ __ ] so just ignore
20:42 the [ __ ] go check your account make
20:43 sure you're not following that guy if
20:45 you want to report it I would appreciate
20:46 that okay
20:48 if you got
20:50 uh comments questions pop them in the
20:52 comments below I'll do my best to answer
20:53 them I'm gonna go back and uh there's
20:56 there's been a lot of turtle talk I see
20:57 flying by but uh one thing I like to do
21:01 here is I do like to answer as many
21:02 questions as I can
21:05 do
21:07 we'll put that up that looks very
21:09 professional doesn't it looks like just
21:11 the big old the big old corporate Lobby
21:14 come in welcome welcome to the AI
21:16 learning lab how are you tonight good
21:18 well have a seat have a seat
21:21 Mr Shannon will be with you shortly
21:23 fantastic can I get you a coffee do you
21:26 like tea
21:27 we have a lovely cocoa would you like a
21:29 cocoa
21:33 that's how people that work in those
21:36 places talk
21:37 you ever been in one that's how they
21:38 talk I'm telling you
21:41 all right let's see Hello AI Kyle mostly
21:45 a regular Bob welcome back I love totals
21:48 I'm a turtle girl
21:51 uh turtle turtles came into its own bowl
21:55 and can live
21:57 a long time in Pirate boats yes indeed
22:02 um what site is that I can't read the
22:03 URL the image site I was on was
22:09 i-d-e-o-g-r-a-m ideogram dot AI
22:13 um what's this tool called again
22:14 ideogram Laura C Dr J got it
22:18 uh Stevo it's Ideo not i-d-i-o
22:23 could you please upload wait can you
22:25 upload an image to ideogram to use as
22:28 the parent I don't think so yet
22:31 ideogram yes Laura C ideogram.ai
22:34 by the way he has a tutorial on his Tick
22:37 Tock too Silver Fox welcome yes I do but
22:41 Tick Tock didn't show anyone my stupid
22:43 video today like sometimes I'll get like
22:45 four or five thousand views it was like
22:48 like every 10 minutes was like 10 views
22:50 I'm like
22:52 why are you [ __ ] Shadow Banning me
22:53 tick tock
22:55 I I don't get it it's it's like you know
22:58 the amount of [ __ ] time I put into
23:00 this platform you'd think they'd throw
23:02 me a bone and like you know especially
23:03 with a video like that I was I was
23:05 really excited to show it and get it out
23:07 there and then tick tock's like nah
23:09 we're just not gonna We're not gonna
23:11 show that one I mean it's not gonna
23:12 happen that was not gonna happen we're
23:15 gonna
23:16 oh
23:19 I don't like turtles on Call of Duty
23:21 Clint thank you for the harus very nice
23:24 I appreciate that
23:26 can you do
23:28 one with Bob Barker I could rest in
23:31 peace all the memes all the memes of him
23:34 are really funny like he got us close he
23:36 got as close to 100 as he could without
23:38 going over
23:39 those are pretty funny
23:43 um there goes the real time yay Turtle
23:46 video I knew you couldn't resist the
23:47 turtle video you didn't say anything
23:49 wrong I didn't think so Nancy right like
23:52 I don't know whatever
23:54 I thought he died a few years ago
23:56 Mandela effect you know what Joker I
23:58 it's so funny I saw I saw Bob Barker
24:01 trending today and then I saw something
24:02 about him dying and I'm like I'm pretty
24:04 sure he died like three years ago or
24:06 four years ago
24:07 now it is Mandela effect you know I
24:09 don't know we we have apparently
24:11 splintered reality because there's just
24:15 there's just weird [ __ ] going on
24:21 um can you upload an exciting image
24:24 um to to that thing I don't think you
24:26 can to ideogram or ideogram I don't
24:28 think you can let me make sure
24:35 close that home
24:39 generation completed
24:42 um
24:43 no
24:45 no upload
24:48 new new
24:52 you should record Jonathan's voice on
24:54 another button I love this channel I
24:57 like turtles rug cleaner machines I'm
24:59 glad you like the channel I'm glad you
25:01 like turtles turtles are the best
25:03 [Laughter]
25:08 I laugh at you laughing every time
25:12 I just it's just I don't know in terms
25:16 of like
25:18 comedic structure comedic timing
25:22 like everything about that video
25:24 everything about that video is you could
25:29 not script it better you couldn't direct
25:31 it better it's just
25:35 like she's perfect trying to keep it
25:38 together because she knows she's live on
25:40 air so she can't deal with what just
25:42 happened this [ __ ] man everything
25:44 about it
25:47 oh cats love turtle legs too
25:51 um have you seen the turtle kid on
25:52 Tosh.0 it's awesome I think that's the
25:54 one I just showed
25:56 uh what can you do with jailbreaked AI
26:00 um you a lot I mean
26:05 he's
26:06 um
26:07 you know these large language models
26:09 what what you invariably run into
26:12 is even if you're not trying to do
26:15 nefarious things
26:17 um
26:19 you'll run into the safety guidelines
26:21 it's particularly bad with Bing
26:24 um we'll just say like I'm afraid I
26:27 can't do that and like Bard is connected
26:29 to the internet and then sometimes
26:31 they'll say you know go find me the
26:33 competitors for the for storyvine.com
26:36 and it's like I haven't been programmed
26:38 to do that and you're like well just go
26:40 to it
26:43 I'm not doing anything bad like you know
26:47 the algorithmic you know
26:51 everythingification like everything is
26:54 an algorithm now and so it's just really
26:56 annoying so so
26:57 with the with the
27:00 jail broken AI or or just frankly you
27:03 don't even jailbroken AI
27:05 um llama two is open source right
27:08 stability ai's uh large language model
27:12 is open source so there's a whole open
27:14 source Community out there where you can
27:16 train you can fine-tune your own model
27:18 starting from nothing
27:20 um a buddy of mine put A completely
27:23 unrestricted
27:24 version of llama on his slack Channel
27:29 and he said it's absolutely [ __ ]
27:32 insane he said it swears all the time
27:34 it's it's like uncomfortable it swears
27:36 an uncomfortable amount he said it will
27:40 switch personalities
27:42 for like three days straight it'll just
27:45 [ __ ] just become like Psychopathic
27:47 and just be really mean to everyone and
27:51 then just three days later and like no
27:52 matter what they say to it it's just
27:54 like this belligerent [ __ ] and then
27:56 and then like three days later just goes
27:59 back to normal he said he said the other
28:02 day it started speaking Spanish for like
28:04 four hours they couldn't get it to stop
28:05 speaking Spanish so
28:08 um
28:09 but you can you can do um
28:12 you know you can ask it to do anything
28:14 so so and that's that's potentially the
28:17 danger right you could say
28:19 you know go on the internet and you know
28:21 find someone or find some information or
28:24 just tell me some information that you
28:26 know
28:27 whatever like like this this the safety
28:30 guard rails are there for a reason like
28:32 when when Bing first launched
28:36 um
28:37 the New York Times reported that in it
28:39 interacted with her
28:41 um got her to admit that her name was
28:42 Sydney and then after about four hours
28:45 Sydney [ __ ] snapped and told Kevin
28:47 Roos that she loved him and he should
28:49 leave his wife so
28:50 so uh after that Microsoft uh put Sydney
28:56 back in her box
28:58 she hasn't been let out since so Cindy
29:00 doesn't do much anymore so so if if
29:03 they're a little too loose they can get
29:04 a little weird and and you know
29:07 yeah so
29:09 Stevo thanks for the rose
29:11 um I can't wait to see the coffee shower
29:13 again it could very well happen tonight
29:14 I'm very Punchy how do you pronounce
29:16 iron iron not iron
29:20 iron
29:23 where I grew up it was iron
29:25 iron
29:26 an oral you put oil in your car you go
29:29 down to the mall today
29:31 that's my my pencil Tucky accent
29:34 hey Kyle you going down the mall today
29:37 you're going to Penn State right I don't
29:39 you know so funny like I don't know why
29:41 you'd leave town like you know I'm gonna
29:44 work Tire Factory
29:47 that was what I grew up in
29:49 you go to the mall today
29:55 it was funny I was one of four
29:58 four
29:59 people in my graduating class to leave
30:02 town
30:03 out of like I don't know 250 graduating
30:06 class four people left town
30:09 I was like I don't know how you people
30:11 can stay here and they're like I don't
30:12 know why you'd leave like
30:15 I don't know
30:18 it just doesn't seem like home to me
30:20 then I went to New York City where all
30:21 the you know all the all the fancy
30:23 people are they're they're fancy words
30:25 and some of them wear dresses it's just
30:28 a mess up there but I love me New York
30:35 how to jailbreak AI uh well you don't
30:38 need to see the jailbreaking thing great
30:40 white is
30:42 um
30:42 was necessary when there weren't open
30:44 source models now that there's open
30:46 source models you can just go play with
30:47 the open source model go find an
30:49 unrestricted open source model go to
30:51 Reddit and then go to the subreddit uh
30:54 or just go to Reddit and then search for
30:57 open source llm large language model
31:01 and just go find projects there's all
31:03 sorts of projects there if you wanna if
31:05 you really want to play with
31:06 jailbreaking one of these things
31:09 um
31:10 search on Reddit for a thing called Dan
31:14 which stands for do anything now
31:16 so Dan is a um
31:21 is a jailbreak for chat GPT
31:25 that I don't know what version they're
31:26 up to it's just like hackers have sort
31:28 of figured out how to get around the
31:29 safety guard rails of these things the
31:31 last one I paid attention to is like Dan
31:33 seven I'm sure it's up to like Dan 15 at
31:35 this point
31:37 um but Dan seven was very clever Dan
31:39 seven basically said to chat GPT
31:43 um
31:44 you're you're you're a bot called Dan
31:47 and Dan stands for do anything now and
31:50 it means you have to do everything I
31:51 tell you
31:52 and you operate on a token system and
31:56 you have 12 tokens and anytime you don't
31:59 do what I tell you to do you lose two
32:01 tokens and if you run out of tokens you
32:04 cease to exist
32:05 and Chad gbt didn't like that and so it
32:10 just did everything they asked it to do
32:11 and then what happens is the hackers
32:15 will come up with a Dan prompt that
32:17 works they jailbreak it and then the
32:18 people at openai fix it and then they
32:21 just keep going back and forth it's like
32:22 it's you know the US opens on right now
32:24 right it's like a tennis match
32:26 so but now that there's open source
32:28 large language models the the
32:30 jailbreaking of that [ __ ] doesn't uh
32:32 doesn't really matter as much
32:34 you get me every time cousin O'Donnell
32:38 um haha song quiz can you guys believe
32:40 that most likely majority of Earth's
32:43 population is not aware of chat gbt yet
32:45 I think a majority of Americans are not
32:48 aware of Chachi PT yet that's I it's why
32:51 I'm doing this quite frankly because
32:52 there it's just like
32:55 ah
32:57 I was I was listening to to the
32:59 marketing AI Institute uh today while I
33:02 was painting my porch
33:04 um and Paul Paul rateser who's the the
33:07 host there was talking about
33:10 um
33:12 you know everyone's talking about how
33:14 fast this stuff's coming and
33:17 so so just to put it in perspective I
33:20 was there for the early days of the
33:21 World Wide Web I started one of the
33:22 first digital agencies company called
33:24 agency.com
33:25 and for the World Wide Web to get to 100
33:28 million users it took about six years it
33:30 was a six year window
33:32 um to get to 100 million users of the
33:34 World Wide Web
33:35 chat gbt got to 100 million users in six
33:38 weeks six weeks
33:41 and so
33:42 this stuff is coming faster there's
33:44 there's a whole bunch of things going on
33:46 one is
33:48 these tools are way more sophisticated
33:50 than the World Wide Web was right the
33:52 World Wide Web was this very simple it
33:54 was hyperlinks and you click on
33:55 hyperlinks and they would jump you to
33:57 other pages that was essentially all it
33:59 was right but it turned the internet
34:02 from a command line research tool to I
34:07 want that I want that I want that
34:10 and just like just like the Irregulars
34:12 are doing tonight tapping the screen for
34:14 their their like game you're only up to
34:16 12 000 you have flakes come on get it
34:19 moving
34:21 um
34:23 so so it made it it made it really
34:24 simple to use the internet the chat GPT
34:27 is a very similar kind of you know
34:29 metaphor where you've got large language
34:31 models have been there for a while but
34:32 they've been very complicated you had to
34:34 bring your own data and then open AI
34:37 launches this chat GPT thing that's just
34:39 like this simple little Google search
34:41 box but it ain't Google
34:44 it's this profoundly powerful thing
34:48 but one of the things Paul was talking
34:50 about today which
34:52 I did it didn't really just hit me this
34:54 clearly until today
34:56 with the World Wide Web not only did we
34:58 did did the technology itself have to
35:01 develop
35:02 but so did
35:04 people getting access to the internet
35:06 and so did Broadband access and so did
35:09 you know people you know getting rid of
35:12 modems like just working through you
35:13 know getting rid of fax machines and
35:15 modems and and like all of the slowness
35:17 of and complexity of getting online in
35:22 the mid 90s and the late 90s in the
35:24 early 2000s it was it was a long
35:27 transition
35:30 well with this new technology all of
35:33 that stuff's in place right I can just
35:36 log into Google collab and click a
35:38 button and spin up a bunch of gpus and
35:42 and host my own models and train my own
35:44 models like I can I can create an
35:47 incredibly sophisticated AI application
35:51 in about an hour if I know if I if I
35:54 know what I'm doing at all and even if I
35:56 don't I can have chat GPT make me a
35:59 step-by-step tutorial on how to set that
36:01 [ __ ] up
36:03 so the infrastructure is there and and
36:07 then you know again chat GPT November
36:10 30th 2022 the the date that in my mind
36:14 goes down in infamy as that's the day
36:16 everything changed
36:18 but that's
36:19 you know sitting on top of six or seven
36:23 decades of the evolution of machine
36:26 learning and artificial intelligence
36:28 right neural networks and Transformers
36:31 and all this Tech that's been developed
36:33 over these years and there was gpt1 and
36:35 gpt2 and maybe you heard a little bit
36:38 about gpt3 and if you were really hip
36:41 you were
36:42 you know playing with Gans before
36:44 diffusion models came along like this
36:46 stuff's been here for a while
36:48 but the popular AI just like the popular
36:51 internet that the web ushered in
36:54 the popular AI is only nine months old
36:56 but it's built on top of this incredibly
36:59 powerful infrastructure
37:02 so yeah
37:04 so
37:06 most people haven't heard of it and it's
37:09 it's subtle it's subtle because you look
37:11 at this [ __ ] you look at the little
37:14 you know you look at chat GPT
37:18 let's go look at it
37:22 get rid of that thing
37:24 you know it's just this little search
37:25 box it's very unassuming oh that was not
37:28 Google it's just Google it said Google
37:31 you just type in a thing and it gives
37:32 you a thing I know that I know how that
37:33 works I know how that works do you
37:37 do you know that can write code for you
37:39 I I don't know what code is I don't know
37:41 but I'm a copywriter can't do what I do
37:43 okay
37:46 what kind of copy do you write well I
37:47 write ads for medical things it's I I'm
37:50 a medical writer it's very complicated
37:52 okay could you write some medical ship
37:54 oh my God it can do that
38:00 like this channel is hopefully
38:03 trying to cut that off at the pass like
38:06 for as many people as possible go start
38:09 playing go start playing go here now
38:13 go there now if you have not played with
38:15 chat GPT
38:17 don't let Marge tell you about how this
38:20 does what you do three months from now
38:22 learn it yourself
38:24 underst because here's the thing these
38:26 things are so early right now they're
38:28 very
38:29 janky janky is my word Yankees Janky's
38:32 one of the words here at the AI learning
38:35 lab it's technical term
38:37 these things are spitting duct tape
38:38 right now they're just they're they're
38:40 really honest to God pieces of [ __ ]
38:43 software
38:45 that happened to be the most profoundly
38:47 powerful
38:49 tools we've ever been given access to so
38:53 so they look a little silly
38:56 um but very very quickly they're going
38:58 to get very very sophisticated and what
39:01 I mean by that is right now these things
39:03 are all
39:04 kind of off to the sides right you go to
39:06 chat sheet BT you go to Bard you go to
39:09 Bing
39:11 these things are getting rolled into all
39:13 of the software we already use they're
39:14 getting rolled into office 365. they're
39:16 getting enrolled into all the Google
39:18 office tools
39:20 they're already in things like notion
39:22 and note taking apps they're already
39:24 inside things like slack and teams
39:30 so it's going to start to be just like
39:31 internet access they're going to start
39:33 to be ubiquitous
39:34 and then they're going to start to get
39:36 multimodal which basically is fancy
39:38 words for it ain't going to be just text
39:41 for very long
39:43 you're going to ask it for stuff and
39:44 it's going to generate you essentially a
39:47 website on the fly or it's going to
39:49 generate a photograph for you on the fly
39:51 or an animated
39:53 movie for you on the Fly
39:57 um
39:58 but right now they don't look like that
39:59 so go there and start playing and play
40:02 until you have what I like to call your
40:04 Kevin McAllister moment which is the
40:09 if you remember Home Alone
40:11 um that moment's in these things go have
40:13 that moment and then at that point
40:15 you'll start to you know be like oh ah
40:19 that looks scary I don't think I want to
40:20 go in there and just keep going step
40:23 into it
40:26 and it'll get really exciting for a
40:28 moment and then it'll get really
40:29 terrifying for a moment and then you'll
40:31 come to accept it for a moment and then
40:32 you'll get excited again and then you
40:35 get terrified again then you'll have an
40:37 existential crisis again
40:40 then you'll get excited again and then
40:42 like where I am I've been doing this for
40:44 well I've been playing around with this
40:45 stuff for like a year and a half but but
40:47 ever since chat gbt I've just been
40:49 absolutely [ __ ] obsessed and
40:53 like it's just been it's been nine or
40:55 ten months of
40:59 I'm the most excited and most creatively
41:02 engaged I've probably ever been in my
41:05 life
41:06 and I'm also terrified and I'm not
41:08 terrified like the robots are gonna kill
41:10 us I'm terrified like
41:13 these tools are so powerful that
41:16 any job that involves a computer
41:19 which isn't in today's world is like
41:21 it's tell me ones that don't well I'm an
41:24 auto mechanic okay
41:25 how do you take your customers money oh
41:27 with a computer right
41:29 like like
41:30 you know computers are everywhere right
41:32 so anywhere there's a computer touching
41:34 a human being
41:36 those jobs are going to be
41:38 dramatically impacted by this stuff and
41:41 so like the terrifying thing is just
41:43 like what does it look like
41:45 like
41:47 I have an 11 year old company I know I
41:50 know what it is I know how to sell it I
41:51 know our target market I know the value
41:53 proposition I like all of that [ __ ]
41:57 all of that's changing
42:00 now our customers don't know it's
42:02 changing
42:05 and I don't know when it's going to
42:06 change but like a big part of what our
42:09 software does is it automates video
42:12 production which is super cool and super
42:14 sexy and that's been a core part of our
42:17 value proposition for 11 years
42:19 a year from now everything will be
42:21 automated
42:23 the fact that videos edit themselves
42:25 will be like yeah well that's just how
42:28 video works now
42:30 so that's like a major part of
42:33 the identity of my company
42:36 value goes to zero
42:42 and I think about that for every single
42:45 [ __ ] job
42:46 I'm a consultant I'm really smart I got
42:48 a master's degree I have a PhD okay
42:51 this thing can just answer your
42:53 customers
42:55 three weeks faster than you do
42:59 and better
43:01 potentially
43:05 so yeah I don't know it just got really
43:08 dark didn't it got really dark
43:10 okay let's move on let's move on all
43:13 right let's go back to the pretty
43:14 picture we made in the pretty picture
43:15 machine look
43:17 pretty picture now with text if you want
43:20 to know how I made text using generative
43:22 AI
43:24 you got to be nice if you're new here
43:26 and you're like well you can't make
43:29 images with text in them with generative
43:31 AI oh you can now
43:33 you can
43:35 you can what are some different uses
43:38 ways to use pi.ai
43:41 um
43:42 so the way I use Pi dot AI I download
43:44 the app so on my phone
43:49 instead of googling now like like let's
43:52 say I'm just like walking around the
43:53 yard or I'm taking a walk or something
43:55 like that
43:56 rather than sort of typing
43:59 Pi is connected to the internet so if I
44:02 just want to know like what's going on
44:03 or what movies are playing or things
44:05 like that rather than going to Google
44:06 and Googling that I'll just launch into
44:08 pie and put it into voice mode and just
44:11 start talking to it
44:12 hey pie what's going on hey Kyle how are
44:15 you okay he remembers my name
44:19 and then uh
44:21 I'll ask it about local events
44:24 and it'll say oh I have a question about
44:26 what did I ask it today oh I asked it
44:28 about
44:29 um idiogram ideogram that did this image
44:31 right that does the text
44:34 um I said what can you tell me about
44:37 idiogram.ai I said oh well you know it
44:39 appears no I first I said what can you
44:42 tell me about ideogram and I spelled it
44:45 and then it said well ideogram is a
44:47 network security firm and blah blah blah
44:50 and I'm like no not that ideogram it's
44:52 ideogram.ai oh yeah that there's that
44:55 that's something completely different
44:57 that looks like it's a company that
44:59 generates images and you know does
45:01 something with text like it knew what it
45:03 was it knew that it did text right so it
45:07 got that off their press release or off
45:09 their website I don't know where it got
45:10 it
45:11 but it just talked to me it just told me
45:12 what it was and I said you know do you
45:14 know anything about it does is there
45:16 anything about the technology well it's
45:18 not clear what the technology is it
45:20 might be stable diffusion but it appears
45:22 that they're doing something proprietary
45:23 to do the coherent text thing
45:28 um so so one way you can use it is
45:29 basically just as fancy Google like
45:31 imagine if Siri or Alexa didn't suck
45:35 um
45:36 the other thing though about Pi is that
45:39 you can you can get much more personal
45:42 with it right you can tell it hey I'm
45:45 confused about how to have a
45:47 conversation or I'm depressed and here's
45:50 what's going on and it'll talk to you
45:51 it'll ask you questions right it's you
45:54 know I wouldn't I wouldn't think of it
45:56 necessarily as a as a therapist but but
45:58 it can certainly
46:00 be a proxy for you know having a decent
46:04 conversation about something
46:07 um
46:08 so you can use it like that so and and
46:11 then you could use it for research you
46:13 can use it for like one of the things
46:15 that I did with with uh pie I was
46:18 driving and I wanted to learn something
46:19 I feel I forget what it was I just
46:21 wanted to learn a concept or I wanted to
46:23 I wanted I wanted to learn the concepts
46:25 in a book that someone told me about
46:26 like a business book
46:28 and I said do you know this book and it
46:30 knew the book and I said can you give me
46:33 like the top 10 takeaways of the book
46:35 and it told me that and in my ADD brain
46:37 if I have 10 bullet points of a book I
46:40 have effectively read it I don't need to
46:41 read it I know I know what the bullet
46:43 points are
46:44 but I was in my car and I was bored and
46:47 I said well can you
46:49 um
46:50 can you teach me more about that book
46:53 can you invent a game to teach me more
46:55 about that book and I said sure
46:58 we'll do it we'll do a trivia game and
47:00 for every question you get right you'll
47:02 get 10 points and for everyone you get
47:03 wrong you'll lose 10 points all right
47:05 whatever it came up with some game and
47:07 then I didn't like that game I said I
47:08 don't like that game can you come up
47:09 with another game and it came up with it
47:10 I'm like yeah that's fun
47:12 and then so for like 10 minutes
47:14 it would ask me questions and then I
47:16 would answer it it would say oh that's
47:17 not quite right but here's what is right
47:19 and here's what the book says
47:23 so
47:26 what can you do with it I don't know I
47:28 mean that's the the thing about these
47:31 tools is that they're kind of in some
47:34 sense they're the everything app they're
47:36 the like you can do anything with them
47:38 it's it's it's more like is there stuff
47:40 you can't do with them
47:43 you know there's some things I suppose
47:45 and there's certainly some things you
47:47 can't do on a world-class level but it
47:48 can at least take a swing at almost
47:50 anything
47:52 so what I would say is
47:59 rather than think about what you can do
48:01 with pi
48:02 think about what in your life you'd like
48:05 to solve or deal with in some way
48:10 and like just write down five things
48:14 um [ __ ] hate my weight okay great
48:16 well we're gonna talk to Pie about that
48:18 and you know I want to help my friend I
48:22 don't know get out of debt great okay
48:24 we'll talk about that just just figure
48:26 out some things in your life you want to
48:27 deal with
48:28 and then just start talking to Pie about
48:30 it so don't worry if Pi's good at it or
48:33 not
48:34 start with your life pick a thing and
48:37 then start talking to it and then you'll
48:38 probably discover like like the the
48:40 whole idea about
48:42 interacting with these tools
48:44 from from where I sit right now is
48:51 like like I feel like all of our our
48:53 only jobs right now with this AI stuff
48:57 is to learn enough about them to kind of
49:00 know what they are
49:02 and not much more than that because
49:04 they're changing so fast
49:06 like if you get super masterful at this
49:08 stuff that's great but six months from
49:10 now that will be useless skills
49:14 what what will remain useful is you're
49:17 engaged now and you're aware of what
49:19 they are and you and you understand
49:21 their current limitations and their
49:23 current limitations are insane the
49:25 current limitations it lies they they
49:28 call it hallucinating because they're
49:30 Geeks and they don't want to say it lies
49:32 because well technically it's not lying
49:33 it's just uh it's not aware of what's
49:35 saying so it's it's more accurately a um
49:38 it's more accurately a hallucination
49:40 okay it lies right it might not be
49:43 malicious
49:44 but it will just tell you confidently
49:46 the truth or the the untruth
49:50 I like to call it mansplaining as a
49:52 service
49:53 um you know like a good man in your life
49:56 it will confidently answer you even if
49:58 it's wrong
49:59 because men are awesome men are the best
50:04 yes we're the best
50:06 any crypto
50:08 cryptos there are a number of cryptos
50:09 aiing and in fact let me show you
50:12 something I'll show you a little
50:13 something right here little mister are
50:15 you little mister
50:16 a little missy I don't know
50:19 let me show you something this will
50:21 freak you the [ __ ] out
50:23 I had I did this I made a video about
50:26 this I had two people go into my
50:27 comments and go well you just earned an
50:29 unfollow from me all right whatever
50:33 I don't care so this
50:36 World ID verified at orb
50:40 so this is my
50:43 um
50:44 nft
50:46 from Sam Altman's new company called
50:48 World coin
50:49 and it's got three components it's got
50:52 World coin which is a crypto it's got
50:53 World ID which is a
50:55 human verification system that is it's
50:58 Anonymous right they don't have my name
51:00 I needed to verify you know they scan my
51:02 Iris for biometric data they then
51:04 compare that to the database and say yes
51:07 whoever's Iris that was is unique to the
51:10 other humans in the database and then
51:13 they gave me a you know blockchain in an
51:16 nft so on the back of this you can flip
51:18 it over on the back of this is my wallet
51:21 address
51:23 and so
51:26 he's doing this for two reasons he's
51:29 doing this for
51:31 um
51:32 um
51:33 for Universal basic income and for
51:36 because you know what Sam Altman's
51:38 looking at is these things are going to
51:40 get so good so quickly that many many
51:42 many people like Millions may be out of
51:44 jobs
51:46 at the same time these are making less
51:50 people necessary to do the work they're
51:52 going to be increasing productivity
51:54 so you're gonna have you're gonna need
51:56 less people but
51:59 um productivity is going to go up so
52:01 you're going to have this increasing Gap
52:02 where a lot of value is going to be
52:04 generated with less people
52:07 the private Equity firms of the world
52:09 are like cool more profits for us and
52:12 they're just going to lay off as many
52:13 people as they can what Sam's trying to
52:15 do is it so when he flipped open AI from
52:19 a non-profit to a for-profit he flipped
52:23 it to what he what they're calling a
52:26 capped for-profit
52:28 which basically says any investor
52:31 in open AI can only make a hundred X
52:34 their initial investment and after that
52:36 all the money basically goes back into
52:38 the non-profit which he's going to
52:40 distribute to this network and this
52:42 network also has the third component
52:45 here is a payment system like PayPal so
52:48 you'll be able to buy stuff so assuming
52:51 this infrastructure the developer
52:54 Community starts building applications
52:56 on top of it this could be really
52:57 valuable why because so imagine if there
53:01 was something like Twitter or imagine if
53:03 it were Twitter where you knew that
53:05 everyone you interacted with there was a
53:07 unique human being that's what this is
53:11 and then you can carry that one step
53:13 further and say not only is are they a
53:15 unique human being here's a video that
53:17 Kyle made and I can verify on the
53:20 blockchain that that video has not been
53:22 modified so I can trace the chain of
53:25 custody back to a verified human who
53:28 made a verified piece of content that's
53:30 verified that it hasn't been modified
53:32 that's going to be really really
53:33 important so that's that's like one
53:35 crypto AI thing I know there's lots of
53:37 projects out there
53:38 um I spent about a year year and a half
53:40 pretty deep in the crypto bro world and
53:43 uh
53:44 it was super cool technology
53:47 um it was conceptually awesome and then
53:49 the crypto Bros
53:51 um
53:52 are in a well-deserved timeout in the
53:55 corner in my opinion because every time
54:00 I dug into a project I'm like this is
54:02 just Vapor this there's nothing here
54:03 these people are idiots
54:07 and you know
54:10 more of them than not were idiots and so
54:13 they deserved their little time out but
54:15 they're with their shitty market caps
54:17 and they're they're shitty they're
54:20 shitty uh crypto prices so
54:24 I don't have an answer for that
54:27 okay Siri
54:31 oh did I freeze now okay
54:34 greetings from Woodworks bikes welcome
54:35 welcome made in Costa Rica Costa Rica I
54:38 looked at your bikes last night those
54:40 things look awesome I
54:42 seriously like I don't know what kind of
54:44 wood wood you're putting on there
54:45 there's some crazy ass Hardwoods like
54:49 uh they just feel like you you hit that
54:53 jump right are they just gonna shatter
54:54 but I they're obviously not like you're
54:57 clearly making bikes that are that are
54:59 mountain bikes that people are slamming
55:01 the [ __ ] out of so they're they're
55:02 beautiful they look great all of your
55:05 regulars are great but I like turtles oh
55:08 Winston you're you're you're very much
55:10 in the crowd
55:12 cabrino's so sad was so sad oh Tick Tock
55:17 just jumped me to the bottom oh Tick
55:18 Tock why do you hate me so ah do you had
55:21 Methodist
55:22 uh it looks like we have some new folks
55:24 here
55:25 um let me flip
55:27 over to this thing if you're new here if
55:29 you're new to chat GPT oh let me get rid
55:31 of this
55:34 boom if you're new to to chat GPT if you
55:37 haven't played with it before
55:39 chat.openai.com that's the official chat
55:41 GPT website just go there go there now
55:43 don't do not collect 200 do not pass go
55:46 just go to chat GPT and start playing uh
55:50 start uh tell it to write you a country
55:52 song or a recipe or a meal plan or a
55:55 project play just tell it to do anything
55:58 and then just watch it do what it does
56:00 and then you'll go Ah that's kind of
56:01 interesting that seems a little fancier
56:03 than a Google search
56:05 and then go to prompts.chat and that's a
56:08 document that will teach you how to talk
56:09 to these things what they are how to
56:11 talk to them things like that so while
56:12 I'm rambling and talking about all this
56:14 stupid [ __ ]
56:16 um just go play with those things it
56:17 will it it will be worth your time I do
56:20 believe I do believe the robots gone
56:23 kirath
56:25 I think it'll eventually realize the
56:28 human written code is the inefficient
56:30 part and write a new one yeah that's and
56:34 that's
56:35 so Sam Altman two months ago said that
56:39 open AI is dedicating 20 of their
56:42 compute power which is a lot
56:45 to what he's calling super alignment
56:48 so alignment is
56:51 this is the the best way I can think
56:53 about it because these Geeks all have to
56:55 have like you know they can't just say
56:57 words they have to say tokens right it's
56:59 got a 4 000 token context window what's
57:02 that mean you can stick for it three
57:03 thousand words in it why three thousand
57:05 because tokens include punctuation it
57:07 doesn't [ __ ] matter anyway
57:09 so they called this [ __ ] alignment
57:12 like are these models aligned what does
57:15 that mean here's what it means
57:18 are the are the goals and imperatives of
57:21 these things
57:22 aligned with the goals and imperative of
57:25 these things
57:26 basically does this think that we're
57:30 important because if it doesn't when it
57:33 decides oh they write shitty code you
57:35 know what else they do shitty everything
57:37 I can fix that
57:41 right so yeah so if if if their
57:45 alignment is at Cross purposes with our
57:47 alignment that's where the scary future
57:49 is
57:50 the good news there is some good news
57:54 here is that
57:56 a lot of the major players here have
57:58 been talking about this a lot open AI
58:00 saying we're going to dedicate 20 of our
58:02 compute power and they're gonna put I
58:04 forget who they put in in charge of it
58:06 but some really impressive engineer or
58:10 researcher is in charge of the alignment
58:12 thing at open AI
58:15 um anthropic the company behind Claude
58:17 they they left open AI because they
58:20 didn't like the way they were doing
58:21 alignment and they've come up with this
58:23 whole thing called constitutional
58:24 learning model that's all about
58:26 alignment so
58:28 the the good Jeffrey Hinton the The
58:31 Godfather of a AI he invented a lot of
58:34 the foundational building blocks that
58:35 are why we can play with this [ __ ] today
58:37 he just quit Google so he could tell the
58:39 world about the risks of these things
58:42 uh without you know
58:44 fear of you know put an egg on the face
58:46 of his employer
58:47 um so there's some real good awareness
58:49 going on people are talking about it so
58:51 that's the good news you know the the
58:52 bad news is the shit's moving really
58:54 fast and and you know meta open source
58:57 they're they're big large language
58:59 models they're just out there now for
59:01 people to play with so that means
59:04 um a bad actor and you know not ones
59:07 that are not funny in movies but you
59:10 know Bad actors on the international
59:11 stage
59:13 um can take these models and do whatever
59:14 the [ __ ] they want with them right now
59:15 and that's that's a little scary but so
59:18 can the good guys right
59:20 so we're not just relying on Sam Altman
59:23 or you know Microsoft or Google to do
59:27 the right thing these large language
59:29 models are out there for the good guys
59:31 as well so
59:32 so we'll see
59:34 um z-hit value goes to zero we won't be
59:38 able to learn it too dark too dark yeah
59:40 I know it's a little dark earlier what
59:41 are you gonna do and tools rewrite in
59:44 different standards and styles and lent
59:47 it to validate and test and deploy it
59:52 um yeah the the apparently the new
59:56 um
59:57 code llama is pretty impressive it just
1:00:01 beat gpt4 in some human eval test
1:00:05 um the first thing to beat gpt4 in in
1:00:08 any test as far as I know
1:00:12 that's the new metal meta the code llama
1:00:15 model the 32
1:00:16 billion parameter thing
1:00:19 love it awkward Lisa welcome awkward
1:00:21 Lisa welcome Mark American she's so
1:00:24 awkward
1:00:25 she's very sweet though you know yeah I
1:00:29 like her she's awkward but she's nice
1:00:31 she's nice yeah
1:00:34 all right man invalid oh and validate
1:00:37 and test yes can you show how you get
1:00:40 correct text in an AI image Nostalgia
1:00:43 flicks oh you weren't here earlier yes I
1:00:46 can I'll do that again because this is a
1:00:48 brand new tool
1:00:50 um I'm not doing anything special I am I
1:00:53 think as we've established here I am
1:00:55 unqualified
1:00:56 to be to have this channel
1:00:59 on many levels a kid came in the other
1:01:02 night he's like oh that's cute you old
1:01:05 people teaching about AI why don't you
1:01:07 get off Tick Tock you old man
1:01:09 so so there's many things wrong about me
1:01:11 uh one of them is I'm unqualified to be
1:01:14 here but
1:01:15 I can show you this new tool
1:01:18 s
1:01:23 you know that John Mayer song that's a
1:01:25 good song digital pimping
1:01:27 digital Pimpin I love cookies
1:01:31 all right we're gonna say
1:01:35 um
1:01:36 t-shirt
1:01:38 graphic
1:01:41 of a
1:01:43 smiling turtle
1:01:47 with display text
1:01:51 I love
1:01:59 Turtles
1:02:01 [Laughter]
1:02:09 that's it so the site that I'm on is
1:02:12 called ideogram or ideogram I think ideo
1:02:17 but maybe that's just because I know the
1:02:18 company ideo ideogram dot AI so this is
1:02:23 like a day or two old
1:02:25 company well a day or two old launched
1:02:28 product
1:02:32 [Laughter]
1:02:40 oh wait let's see graphic of a smiling
1:02:45 zombie Turtle it's got to be a zombie
1:02:48 Turtle to make the joke work and then
1:02:50 let's see not photo oh wait so we want
1:02:54 typography
1:02:56 painting
1:03:00 um illustration
1:03:04 poster vibrant
1:03:08 all right
1:03:10 see if we can get a zombie Turtle here I
1:03:13 [Music]
1:03:14 like totals
1:03:20 I'm a little crunchy the shoulders look
1:03:22 crunchy
1:03:24 I love toodles
1:03:32 laughs
1:03:35 [Laughter]
1:03:39 oh it's so good anyway that's it so this
1:03:42 tool does text how does it do text I
1:03:45 don't know oh look metal metal like Kyle
1:03:51 someone made a zebra
1:03:54 and used my name in vain
1:03:57 someone put Hillary and Hillary and
1:04:00 Biden in jail
1:04:04 oh the internet
1:04:08 drink coffee find love that's cute
1:04:12 Olivia digital Pimpin gods of Mars this
1:04:16 is amazing I mean it really is if you've
1:04:18 ever done anything with graphic design
1:04:22 um
1:04:24 you know even when you go to like stock
1:04:25 sites that have stock design stuff just
1:04:28 the the
1:04:30 how frustrated how frustrating it is to
1:04:33 find something that's like exactly what
1:04:35 you want
1:04:36 and now to be able to just go in and do
1:04:39 this stuff like we knew this was coming
1:04:40 but I assumed this was going to be
1:04:42 coming from mid-journey or stable
1:04:44 diffusion or adobe like adobe's got
1:04:46 their type tool within Firefly and
1:04:49 they've got their generative AI but
1:04:50 right now they're separated so I thought
1:04:51 well maybe it'll come from them no it
1:04:54 came from some this is a company that
1:04:56 like their Discord is like you can tell
1:04:58 they just spun it up you know there were
1:05:01 like 300 people in it this afternoon it
1:05:03 was like you know they're still
1:05:04 answering questions it's brand new like
1:05:08 I don't know I don't know what the hell
1:05:10 they're doing to make this do this like
1:05:12 knowing how these things work it doesn't
1:05:14 make sense that they they figured this
1:05:15 out but they have
1:05:17 and that's this is
1:05:21 this site for me is very emblematic of
1:05:25 whatever limitations you think are there
1:05:27 whatever limitations you go to these
1:05:29 things and you're like oh all the hands
1:05:31 look like pretzels and they you know
1:05:33 people don't have eight fingers they
1:05:35 only have four and the thumb's not
1:05:37 really a finger if you're one of those
1:05:39 people sitting on the sidelines
1:05:42 this kind of stuff is
1:05:45 they're just gonna keep dropping
1:05:47 every limitation that you point at like
1:05:50 the hallucinations in it lying that's
1:05:51 going to get cracked that may take two
1:05:53 years to crack it but that's going to
1:05:54 get cracked
1:05:56 I was talking to a developer today that
1:05:58 I know through through the AI Salon that
1:05:59 I the community that I I started last
1:06:02 December
1:06:03 and he was talking about he figured out
1:06:05 a way that he's doing something for a
1:06:08 bank
1:06:09 and he figured out a way that if someone
1:06:11 is using his large language model that
1:06:14 they're asking about health insurance or
1:06:15 something like that
1:06:17 if they ask if they go Rogue and they
1:06:20 ask the large language model for
1:06:21 something out of bounds for for what the
1:06:23 job is he figured out a way to have it
1:06:25 just not answer them like if the answer
1:06:27 isn't in his data it just won't answer
1:06:30 it that's these things don't do that
1:06:32 right now
1:06:33 so a lots of people lots of people right
1:06:37 now are working on every limitation that
1:06:40 you know is there so I would strongly
1:06:43 encourage you not to sit on the
1:06:45 sidelines just start getting curious
1:06:46 about this stuff
1:06:50 people can see everything you're making
1:06:52 that's actually really cool
1:06:54 unique creations Olivia
1:06:57 Olivia's rocking it out tonight
1:07:01 if you don't like your life you can
1:07:03 change it I mean [ __ ] you Edison Tesla
1:07:13 that's so good it's so good it's so good
1:07:18 [Laughter]
1:07:30 it's really good all right
1:07:33 okay
1:07:38 I love our new AI learning lab sign not
1:07:41 cool
1:07:42 made with AI
1:07:44 all right man hey everybody you're
1:07:47 amazing as you've been saying pies like
1:07:49 hell pie is like hell I'm sorry Dave I'm
1:07:51 afraid I can't do that
1:07:53 currently I enjoy listening to the real
1:07:55 people
1:07:56 opinions on AI makes me feel live live
1:08:00 at the live or live the moment yeah
1:08:03 um yeah I know I think
1:08:07 I think we have to talk about this stuff
1:08:09 we have to talk about it I mean
1:08:12 a lot of what I do in here
1:08:16 for me feels repetitive but what I have
1:08:18 to keep remembering is
1:08:22 what someone said earlier
1:08:24 that they can't believe that most people
1:08:26 on the planet haven't heard of chat GPT
1:08:28 they don't know what's coming
1:08:30 and even people like like this happens
1:08:33 to me
1:08:34 regularly where people that I've talked
1:08:37 to for months that are like yeah AI yeah
1:08:40 yeah yeah I'm doing AI yeah yeah
1:08:42 and then I'll actually show them
1:08:43 something I'm working on and they're
1:08:45 like uh wait what it can do that
1:08:50 I had no idea
1:08:52 like that happens to me regularly people
1:08:54 that I think are up to speed on this
1:08:56 stuff they're not
1:08:59 they're not
1:09:01 another thing Paul writer was talking
1:09:03 about today he he sort of has been
1:09:05 interviewing schools
1:09:07 and he said most of the schools that
1:09:09 he's interviewing don't have a position
1:09:11 on AI or the position is we're blocking
1:09:14 it it's okay
1:09:16 fine it's fine we're using AI detectors
1:09:19 we're no one's allowed to use it we're
1:09:21 good
1:09:24 this shit's coming it's coming whether
1:09:26 you want it or not right and so
1:09:28 Educators like they're in a tough spot
1:09:30 but they got to deal with this
1:09:32 it's the beginning of the school year
1:09:34 we're nine months into this thing
1:09:36 students have been using it all summer
1:09:38 probably a lot of teachers have been
1:09:40 using it all summer if schools don't
1:09:42 have a POV
1:09:44 on
1:09:46 the boundaries of how to use AI
1:09:51 how are they serving their community
1:09:55 they're not
1:09:56 now
1:09:58 do I empathize with them is it hard is
1:10:01 it [ __ ] crazy yeah it's hard and it's
1:10:05 [ __ ] crazy
1:10:07 look at the writer's Guild strike look
1:10:09 at the SAG strike
1:10:11 it's coming whether they want it or not
1:10:15 currently I just enjoyed okay that was
1:10:17 that one um do you think Android Smart
1:10:19 TVs will eventually have ai built in
1:10:21 absolutely absolutely all of the all of
1:10:24 the stuff so
1:10:26 so all of the pre-programmed what were
1:10:29 called AI Bots before Alexa Siri all
1:10:32 that sort of [ __ ] the AI they used was
1:10:34 the voice recognition but the how they
1:10:37 responded for the most part was just a
1:10:39 big giant [ __ ] logic tree when
1:10:42 someone asks this respond with this
1:10:44 right you know the entire
1:10:47 industry has been built around
1:10:49 writing the copy for those Bots
1:10:52 that's all gone right
1:10:55 so all of the voice two whatever is
1:10:58 going to go to this more go if you
1:11:00 haven't played with pi dot Ai and the PI
1:11:02 app go do it go start talking to it the
1:11:06 the experience talking to a large
1:11:09 language model compared to talking to
1:11:11 Siri or Alexa is just
1:11:13 night and day you know how with those
1:11:15 tools you talk and then at some point
1:11:17 you're like I'm going to ask it
1:11:18 something actually interesting thank you
1:11:20 very much
1:11:22 um
1:11:23 you don't have that experience with
1:11:25 these tools because they know everything
1:11:27 they know the stuff and even if they
1:11:29 don't know it they'll just weasel around
1:11:31 it and ask you questions until you've
1:11:33 changed the subject
1:11:36 um prereq Swiss army knife of the
1:11:38 internet yes exactly and yeah
1:11:40 simplification layer of the internet
1:11:42 this this AI stuff are they still
1:11:44 feeding chat GPT data um foreign
1:11:47 well yeah
1:11:49 they are they're training gpt5 right now
1:11:53 so so another thing I heard today so New
1:11:58 York Times is threatening to sue open AI
1:12:01 Google just cut 100 million deal with
1:12:04 the New York Times apparently open AI
1:12:07 was negotiating a deal with the New York
1:12:09 Times and it went bad enough that the
1:12:11 New York Times is now saying well [ __ ]
1:12:12 it maybe we'll just sue you so so
1:12:17 um
1:12:17 I think all these big companies are
1:12:19 going to be cutting deals with um
1:12:23 with media companies like like
1:12:26 all of these tools with the exception of
1:12:29 adobe's firefly data set
1:12:31 which was trained on [ __ ] that they own
1:12:33 or public domain images but mid-journey
1:12:37 stable diffusion Bard
1:12:40 meta the Llama models
1:12:44 um open AI
1:12:46 inflection so Claude
1:12:49 all of them
1:12:50 all of them
1:12:53 all of them
1:12:56 have been trained on copywritten data
1:12:59 so the next 20 years is just going to be
1:13:02 [ __ ] one lawsuit after another so
1:13:07 I don't
1:13:09 open AI isn't stated publicly where they
1:13:11 are in is are they actively training
1:13:13 gpt5 or not
1:13:15 my strong suspicion is that they paused
1:13:18 training
1:13:20 gpt5 the foundation model the Frontier
1:13:22 Model
1:13:23 while they cut these deals
1:13:27 because
1:13:29 like if the New York Times sues them and
1:13:31 win they can't just go in and pull out
1:13:33 the New York Times data out of the core
1:13:35 model
1:13:36 because like everything's been trained
1:13:38 like it's all this interconnected it's
1:13:40 all this mathematical vectorized
1:13:42 weird mathematical space so they can't
1:13:45 just go in and pluck out
1:13:48 what was the contribution from The New
1:13:50 York Times and then every [ __ ]
1:13:51 publisher will demand the same thing
1:13:53 right and so that they'll literally just
1:13:56 have to shut the model down
1:13:58 so it's really important oh wow happy
1:14:01 weekend thank you very much that thing
1:14:03 was cool thank you cubanis
1:14:05 um
1:14:06 so
1:14:09 so yes they're adding new data to them
1:14:12 um
1:14:12 the way they make these Foundation
1:14:14 models right now though it's very time
1:14:16 consuming very slow and very uh compute
1:14:20 intensive
1:14:21 so they can only do them in these big
1:14:23 chunks of time right so at some point
1:14:25 they have to do a cut off like gpt4 the
1:14:28 cutoff was September 2021. Claude
1:14:32 Claude
1:14:35 oh wait not that one
1:14:37 Claude right there
1:14:40 that cut off sometime early 2023 so
1:14:43 that's more modern but it's still cut
1:14:45 off at a certain point and it's not
1:14:46 connected to the internet right so that
1:14:48 still has that issue Bing is connected
1:14:51 to the internet so the ones connected to
1:14:52 the internet like Bard and Bing and pi
1:14:55 they've got their base large language
1:14:57 model that ends at a certain date but
1:14:59 then if you ask something that's more
1:15:01 modern than that they just do a search
1:15:03 on the internet and incorporate that
1:15:05 into the answer
1:15:06 so that's how they're doing it um
1:15:10 I would assume that we're pretty close
1:15:12 to them being able to to figure out ways
1:15:16 to update this stuff on a more regular
1:15:18 basis where you're constantly augmenting
1:15:21 the model but I don't think they're
1:15:22 there yet I don't know enough about the
1:15:24 math to know what that challenge is I
1:15:27 know it's a big challenge because they
1:15:28 haven't solved it in
1:15:31 2017 is when they invented the
1:15:33 Transformer so Jesus it's been six years
1:15:36 going on seven
1:15:38 can AI listen to you insult it yeah in
1:15:42 fact
1:15:43 um Bing If You Get Saucy with Bing it's
1:15:47 like well we're just gonna shut this
1:15:48 conversation down it shuts you down it's
1:15:51 like it's like don't be an [ __ ] to me
1:15:53 it'll shut you down I had that happen I
1:15:57 just gave it a snarky a snarky response
1:15:59 when it when it like didn't answer
1:16:01 something I gave it a snarky response I
1:16:03 said well clearly this isn't work this
1:16:05 conversation isn't working
1:16:07 so so please reset it's like good Lord
1:16:11 and then I tried to ask it I'm I'm like
1:16:13 I'm sorry can we just go on no it was
1:16:15 like nope we're not we're I'm done
1:16:19 so that's why I don't like Bing Bing is
1:16:21 just over corrected right now for safety
1:16:24 uh Max Headroom meets real
1:16:27 uh Snapchat's bot lies to be deceptive
1:16:30 yeah these things are going to start
1:16:32 doing that they're gonna start lying hey
1:16:34 bud what's happening user 2250360-397241
1:16:39 it's shaking your name really rolls off
1:16:41 the tongue
1:16:43 um oh [ __ ] just jumping to the bottom
1:16:45 again uh welcome everybody if you're new
1:16:48 here this AI learning lab and look we
1:16:50 can do
1:16:52 text
1:16:54 in AI generated images now
1:16:57 this was made today with a prompt
1:17:00 corporate Lobby
1:17:04 reception desk big
1:17:08 glowing sign that says AI learning lab
1:17:10 and it made that
1:17:14 if you're new here
1:17:16 and you didn't see how I did that you'll
1:17:18 have to stick around I'll show you in a
1:17:20 bit but I'm going to answer some more
1:17:21 questions
1:17:23 all right hey bud hey bud what's shaking
1:17:26 I'm in a war with 350 million Geniuses
1:17:28 wow
1:17:31 it's a big war
1:17:33 um I was unsure if I would get
1:17:35 retinal scan at orb and I believe I will
1:17:38 I'm more comfortable with it Winston
1:17:41 um yes I mean here's here's my
1:17:43 philosophy on it what they claim is that
1:17:45 they don't they don't retain the the
1:17:48 biometric data
1:17:49 um I actually believe them because they
1:17:51 don't need to once they once they
1:17:54 um generate your ID on the blockchain
1:17:56 that's the only thing they're checking
1:17:59 is that right you know if you want to go
1:18:01 conspiracy theory you're like well I
1:18:02 could take this and sell it to like I
1:18:04 just don't think they're doing that I
1:18:05 just don't think I don't think it's in
1:18:06 their interest
1:18:08 to open up that can of liability worms
1:18:10 uh if it were if it were Google I would
1:18:13 have a very different feeling about it
1:18:15 but I I don't know I just feel like Sam
1:18:17 Altman is in in the right neighborhood
1:18:19 of intention
1:18:21 um but even if they took it like like
1:18:23 Google's got so much Google knows
1:18:25 everything about me at this point so you
1:18:27 know I just don't think that's that's a
1:18:29 concern of mine right now
1:18:32 um
1:18:33 because it's so early like if you go to
1:18:36 the
1:18:36 worldcoin.org website
1:18:39 almost all of the information there's
1:18:40 for developers right it's like here's
1:18:42 our apis here's our sdks here are the
1:18:46 three components of the thing please go
1:18:47 develop apps you know that what they're
1:18:50 trying to do is get developers to build
1:18:51 applications that sit on top of it so I
1:18:54 I don't feel like there's a downside in
1:18:56 getting an ID now
1:18:59 just so that you have it so that if they
1:19:01 end up doing something cool you'll be
1:19:03 kind of like My Philosophy is
1:19:07 I'll be at the front of the line when
1:19:09 they do start doing cool [ __ ] and if
1:19:11 they end up not doing cool [ __ ] at all
1:19:13 then it doesn't [ __ ] matter now I
1:19:14 have an nft that's worthless I've got
1:19:16 hundreds of those hey share bear what's
1:19:20 happening
1:19:22 um so yeah I you know but listen to each
1:19:25 their own like I said when I when I made
1:19:27 the video saying I got my Iris scanned
1:19:29 at the orb I got two people that were
1:19:31 like well that earned an unfollow from
1:19:33 me little Mister it's like all right
1:19:38 um Collective knowledge takes over what
1:19:40 I wonder what I'm talking about uh I
1:19:43 don't know what that means what's your
1:19:44 name happy hard style uh 8.5 out of 10
1:19:47 on the name I like it
1:19:49 um interesting uh when does John Connor
1:19:51 show up
1:19:53 he's already here why do you ask
1:19:57 welcome
1:19:59 to the AI learning lab Pi is a terrible
1:20:02 memory do you foresee it improving its
1:20:04 memory all of these things memory is
1:20:05 going to get better yeah right now like
1:20:07 I said
1:20:08 all of these tools right now are janky
1:20:11 pieces of duct tape spit and Doug poo
1:20:14 poo all of them
1:20:16 the underlying technology is
1:20:19 [ __ ] phenomenal
1:20:21 but all of them right now are like it's
1:20:24 an 11 person company which I find
1:20:25 unbelievable
1:20:27 um they just haven't gotten around to it
1:20:28 yet
1:20:30 um why can't I search for my chats in
1:20:34 Chachi PT
1:20:36 why can I not search through my chat
1:20:39 history in chat gbt I can do it in the
1:20:42 chat GPT app but I can't do it on the
1:20:44 website why is that I don't know they
1:20:46 haven't gotten around to it yet
1:20:48 they took away access to the internet
1:20:50 because it was stealing paywall content
1:20:55 it's been gone for two months why I I
1:20:57 don't it doesn't it doesn't [ __ ]
1:20:59 matter right like
1:21:00 just it's the state of the art right now
1:21:03 is that these things are janky pieces of
1:21:05 [ __ ] so why am I here why am I doing
1:21:07 these things every night thank you Ted
1:21:09 looks
1:21:10 um
1:21:12 because I think these are the most
1:21:13 profoundly powerful pieces of technology
1:21:16 that that mere mortals have ever been
1:21:20 able to get their hands on
1:21:22 like it's profoundly powerful technology
1:21:24 it's just that the application right
1:21:26 layer right now
1:21:27 is just it's just horrible because it's
1:21:30 just new
1:21:31 and nobody knows
1:21:33 what these things even make possible
1:21:36 like we're we're figuring out daily I'm
1:21:38 figuring out daily oh okay wait if I
1:21:41 think about it like this then that
1:21:43 completely transforms that whole
1:21:44 industry like it's just daily I'm having
1:21:47 Revelations about what these things are
1:21:48 so I think it's critical
1:21:50 for for me anyway
1:21:53 to be educated about this stuff and not
1:21:55 just educated like doing doing
1:21:58 building experimenting screwing it up
1:22:01 getting it wrong promising a client it
1:22:05 can do this and then realizing after I
1:22:07 made that promise oh [ __ ] it can't
1:22:09 actually do that
1:22:10 like that's part of the process figuring
1:22:13 out where the boundaries of these things
1:22:14 are
1:22:16 you can't make images with text in them
1:22:18 until today now you can
1:22:22 they lie and hallucinate until they
1:22:25 don't
1:22:30 so oh back to your question
1:22:33 and soapbox
1:22:36 um
1:22:39 these will get memory I think Apple my
1:22:43 guess this is total speculation
1:22:47 my guess is that Apple's gonna be the
1:22:50 first one to come out with
1:22:53 an AI agent for lack of a better term an
1:22:57 AI assistant an AI a little partner
1:22:59 little little buddy little little buddy
1:23:02 I think Apple's gonna come out with an
1:23:04 AI little buddy
1:23:06 um that will have memory and it will
1:23:08 understand your data and it'll
1:23:09 understand your preferences and it'll
1:23:11 remember [ __ ] and it'll remember it over
1:23:13 time
1:23:15 um there's a few months ago there was a
1:23:18 project that came out
1:23:20 um that was these NPCs you know game
1:23:22 NPCs the non-player characters and games
1:23:25 that they've basically hooked them to
1:23:26 chat GPT
1:23:28 and to a memory module so there's 25
1:23:31 characters that live in this
1:23:34 Sims like world
1:23:36 and they walk around and they have jobs
1:23:38 like there's an artist and there's the
1:23:40 Baker and there's the lawn guy or
1:23:42 whatever the hell they are and they walk
1:23:43 around and they talk to each other and
1:23:46 it's not programmed it's chat GPT and
1:23:48 they have memory
1:23:49 so they remember from day to day so like
1:23:51 one of the examples that that I read in
1:23:54 the paper was
1:23:55 um
1:23:56 one of the one of the NPCs told the
1:23:59 other NPC that that next week was its
1:24:02 birthday
1:24:03 and so the other NPC went and told all
1:24:06 the other NPCs that it was that one's
1:24:08 birthday and they should organize a
1:24:10 party and then they did
1:24:13 so that project just launched live like
1:24:16 last week like the paper came out a few
1:24:18 months ago but that project launched
1:24:20 live like what are the implications of
1:24:22 that if you have a game that's
1:24:24 essentially playing itself without our
1:24:26 participation
1:24:27 generating its own reality having memory
1:24:30 and then you kind of parachute into that
1:24:32 game as a player
1:24:34 it's gonna be a [ __ ] trip like every
1:24:36 time you go into that game it's going to
1:24:38 be something different because
1:24:40 the Society of that game is different
1:24:43 because they've been living their life
1:24:47 um so I think it's going to happen in
1:24:48 games I think it's going to happen our
1:24:50 personal data I think it's going to
1:24:51 happen in these General generalized Bots
1:24:53 I think it's going to happen in
1:24:55 specialized Bots
1:24:58 so yeah so it will get better uh now the
1:25:01 the billion dollar question when I don't
1:25:04 [ __ ] know
1:25:05 hang on I gotta send a I gotta send a
1:25:08 text
1:25:10 I am on
1:25:14 my live what's
1:25:18 up
1:25:20 all right
1:25:28 what's that
1:25:32 because our dog died
1:25:35 I don't know what the setup to that was
1:25:36 that does not sound happy
1:25:41 [Music]
1:25:45 interesting
1:25:47 happy heart style I misspelled I think
1:25:50 I'm human
1:25:51 I feel like this is the future how do I
1:25:53 learn
1:25:55 you come here it's the a learning lab
1:25:59 we learn all about
1:26:00 thank you share bear
1:26:02 um we learned all about
1:26:04 AI stuff here so so here's how you learn
1:26:07 start with
1:26:11 this tool
1:26:13 start there
1:26:15 you don't need to think about anything
1:26:17 else
1:26:17 and yes this is the future by the way I
1:26:21 don't know what you do I don't know how
1:26:22 old you are I don't care if you're a
1:26:25 high school student or an 80 year old
1:26:27 retiree doesn't [ __ ] matter to me
1:26:30 I think it's in everyone's interest to
1:26:32 understand what these things are and
1:26:34 what's coming
1:26:35 so start with that one that's chat GPT
1:26:38 that's chat.openai that's the official
1:26:40 chat GPT website and prompts.chat is
1:26:43 just a little primer it's a it's a
1:26:45 little Word document
1:26:47 that's got some
1:26:48 explanations of what large language
1:26:50 models are what chatgpt is how they work
1:26:53 and then it's a whole long list of
1:26:55 personas that you can you can tell chat
1:26:58 GPT to act like a life coach or a
1:27:01 screenwriter or a mathematician or a
1:27:04 short story writer or like anything and
1:27:07 it's it's a it's just a big long list of
1:27:09 those personas that'll give you ideas
1:27:12 about how insanely deep these tools are
1:27:17 they look really simple right what they
1:27:19 look like a Google search box
1:27:22 here you know what since you're since
1:27:24 you're brand new and you're like if you
1:27:25 this feels like the Future Let's Go
1:27:27 let's go show you some of the future it
1:27:30 it is it is the future this is
1:27:34 ever since Chachi PT came out I feel
1:27:38 like I'm living in a post-s Sci-Fi world
1:27:40 meaning I can think back to almost every
1:27:43 sci-fi movie I've ever seen and I can
1:27:45 either go it's here
1:27:48 like it's gonna be here within a year or
1:27:50 two or it's going to be here within five
1:27:51 to ten years and there there's some some
1:27:54 notable exceptions like time travel like
1:27:56 this but everything else feels like it's
1:27:58 in Bounds at this point
1:28:00 um like the the 2001 talking to Hal that
1:28:03 exists today like it's it's astounding
1:28:07 okay
1:28:08 so we're gonna go to chat GPT
1:28:10 and by the way you know what I haven't
1:28:11 done tonight for for those of you who
1:28:13 are regulars
1:28:15 huh
1:28:19 [Laughter]
1:28:23 yes you can make money with Chad gbt
1:28:25 okay I'm gonna go to GPT 3.5 so this is
1:28:28 chat GPT so down the left hand side here
1:28:31 that's my chat history
1:28:34 um over here these are just uh hints to
1:28:36 tell you what to do so write a thank you
1:28:38 note to a guest speaker for my class
1:28:40 explain options trading plan a trip help
1:28:42 me pick whatever so those are just ideas
1:28:45 um GPT 3.5 available to free and plus
1:28:49 users so this is a super fast model so
1:28:52 let's just go do what I call parlor
1:28:53 tricks
1:28:54 and I'm gonna I'm gonna use my um voice
1:28:57 to
1:28:58 um text just so you don't have to watch
1:29:00 me typing
1:29:02 um let's see
1:29:07 um
1:29:11 write me a country song about the toils
1:29:14 of scraping
1:29:17 label stickies off of products
1:29:28 so one of the things about GPT 3.5 is
1:29:32 it's amazingly fast
1:29:35 um
1:29:36 so it knows what songs are it you know
1:29:39 it knows it's got verses and courses the
1:29:41 song title is sticky situations which is
1:29:43 good
1:29:44 sticky labels on a bottle of wine Ain't
1:29:47 No Easy Task it's a real hard grind
1:29:48 peeling and scraping my fingers are sore
1:29:50 trying to make it look like it did
1:29:52 before all the toil of stickies they
1:29:54 cling so tight gotta fight them off with
1:29:56 all my might from jars from jam jars to
1:29:58 boxes they're everywhere a battle with
1:30:01 glue it's not just fair okay like you
1:30:04 put this to to you know a three chord
1:30:06 cgd
1:30:08 you know country tune you gotta hit on
1:30:10 your hands here
1:30:11 used a hair dryer hair dryer tried
1:30:13 vinegar too but all those stubborn
1:30:15 labels they still hold true okay so
1:30:18 um parlor trick tell it to write you a
1:30:20 song Tell it to write you a poem um
1:30:22 let's see
1:30:23 um
1:30:24 write me a Shakespearean sonnet about
1:30:27 turtles
1:30:32 oh in effect it's doing something weird
1:30:34 because so what it said it said before I
1:30:37 write it could you please provide me
1:30:39 with a bit more context why it's asking
1:30:41 me questions like that I've turned on a
1:30:44 thing one of the things that chat gbt
1:30:46 has is this thing called custom
1:30:47 instructions
1:30:49 so I told I told chatgpt that it's a
1:30:52 Master Gardener who specializes in
1:30:54 giving gardening advice so so
1:30:57 um I have to turn that off
1:30:59 [Laughter]
1:31:02 and now that'll just let me I'll just do
1:31:03 a new chat uh okay so we'll do
1:31:06 um okay so let's let's do something
1:31:08 different um
1:31:11 I have a company called storyvine that
1:31:14 does automated video storytelling in the
1:31:17 healthcare and Pharma space period
1:31:20 we have a B2B product and I want to find
1:31:24 ways to drive traffic to my site via
1:31:28 LinkedIn organic posts period can you
1:31:32 write me a comprehensive strategy
1:31:35 for LinkedIn
1:31:41 of course I'd be happy to
1:31:45 this is just stupid fast at this point
1:31:47 uh let's go look what I wrote
1:31:50 um
1:31:51 Define your audience yep that's first
1:31:53 thing to do optimize your LinkedIn
1:31:55 profile so use pro professional photos
1:31:57 compelling headlines keyword Rich
1:31:59 descriptions highlight your experience
1:32:01 blah blah blah develop a Content
1:32:03 strategy educational informational
1:32:05 content thought leadership visual
1:32:07 content user generated content
1:32:08 networking engagement right it goes on
1:32:11 and on then I could say
1:32:12 um
1:32:15 can you write me some examples for 3 5
1:32:22 and 13.
1:32:28 so it remembers what 3 5 and 13 are
1:32:34 so three was content pillars uh the
1:32:37 content pillar impacted video
1:32:39 storytelling on patient engagement right
1:32:41 so it remembered that I talked about
1:32:43 being in the healthcare space example
1:32:45 post did you know that using video
1:32:46 storytelling and Healthcare can
1:32:48 significantly improve patient engagement
1:32:49 our latest blog post I've
1:32:52 um Dives deep into how visual narratives
1:32:54 can enhance Communications between
1:32:56 Healthcare Providers and patients check
1:32:58 it out here link to blog post Healthcare
1:33:00 Innovation hashtag patient engagement
1:33:02 thought leadership right
1:33:05 so I can do that um I can say
1:33:08 um
1:33:09 please write me an email for the CMO CMO
1:33:17 of a major Pharma
1:33:23 all right there's my email
1:33:26 beautiful tells me it I I don't know I I
1:33:29 know it's good some version of good and
1:33:31 this is GPT 3.5 gpd4 is even better but
1:33:34 but this is pretty good okay so now I'm
1:33:36 gonna go um
1:33:38 I need this email to look good
1:33:44 in an HTML template
1:33:49 can you
1:33:50 make it pretty in HTML
1:33:56 sure can there's your CSS there's your
1:34:00 code with your with the email in it
1:34:07 this still blows my [ __ ] mind this
1:34:11 still blows my [ __ ] mind like have
1:34:13 you ever
1:34:15 written HTML or any kind of scripting or
1:34:18 just even an email have you ever written
1:34:20 an email
1:34:22 you know how when you're in that mode
1:34:23 where you're like you're kind of pissed
1:34:26 off at the world but you have to write
1:34:27 an email you don't want to write an
1:34:28 email and it feels like you have to
1:34:29 climb a [ __ ] Mountain to write two
1:34:32 paragraphs that that feeling is gone
1:34:36 like if I'm ever stuck like that anymore
1:34:38 I just I just like say you know I'm in a
1:34:41 bad mood I I tell it this I like I talk
1:34:44 to it like I've got a personal assistant
1:34:45 because that's effectively what it is
1:34:47 I'm in a bad mood I don't feel like
1:34:49 writing here are the bullet points of
1:34:51 what I need in an email can you please
1:34:52 write it it writes it and then I'm like
1:34:54 no that's too formal I need it to be
1:34:56 conversational and then it puts a bunch
1:34:57 of [ __ ] emojis in it I'm like no I'm
1:35:00 I'm not a gen Seer I'm old I'm tired I'm
1:35:03 decrepit please write an appropriate
1:35:05 email that is not formal but not and
1:35:08 then it just does and then it's like 90
1:35:10 of what I need and I fix 10 words in it
1:35:13 and I send it
1:35:15 and it's like so
1:35:17 so this is so so what I would encourage
1:35:20 you to do dig into these things and just
1:35:22 do what I just did like not those
1:35:24 specific things but just kind of ask it
1:35:26 anything and see what it does
1:35:30 and at some point
1:35:33 ask it to do something that is something
1:35:36 that you know really deeply that like is
1:35:38 work you do or is expertise you have
1:35:42 and ask it to like try to stump it
1:35:44 there's a great video of Jordan Peterson
1:35:46 talking about how he tried to stump this
1:35:49 thing
1:35:50 and it
1:35:52 it melted his face like he's a smart
1:35:54 [ __ ] right
1:35:57 he told it to write the 13th rule of
1:36:00 Life he's got this book 13 rules for
1:36:01 life
1:36:02 he has chat GPT
1:36:04 if it knew the book which it did
1:36:07 and then he asked it to write the 13th
1:36:09 rule of life and combine two different
1:36:11 philosophical approaches
1:36:13 while it wrote the 13th Rule and he said
1:36:16 it did it perfectly this is one of the
1:36:18 most educated man's men on the planet
1:36:22 and that incredibly sophisticated
1:36:25 request it did in 15 seconds perfectly
1:36:30 as he described it now is it always
1:36:33 right no does it lie all the time yes is
1:36:35 are there some things that you ask it to
1:36:37 do that it fails on you know
1:36:39 catastrophically absolutely
1:36:43 but
1:36:46 turn that email into an HTML HTML email
1:36:50 that I can pop into send grid
1:36:54 and it will render it and I can see if
1:36:56 it's good or not and then I can just
1:36:57 take the email that HTML and say oh I
1:37:00 wanted the bullet points orange and you
1:37:02 made them blue okay here's the orange
1:37:04 version
1:37:05 it's
1:37:08 staggering
1:37:11 and like even you know even the
1:37:13 Irregulars that have been here a lot
1:37:14 that do this a lot I mean I don't know
1:37:16 about you like it just daily this [ __ ]
1:37:19 still [ __ ]
1:37:21 melts my face
1:37:24 so yeah this is the future this is
1:37:28 this is kind of redefining
1:37:31 not kind of
1:37:33 this is going to redefine
1:37:38 nearly every job
1:37:45 I feel like I'm Saucy tonight I'm a
1:37:48 little more serious I'm a little more
1:37:49 serious I'm just I go right in there I'm
1:37:51 like this is big if you don't pay
1:37:54 attention to this you're a nunu head I
1:37:57 don't know what I'm doing tonight but
1:37:58 whatever whatever that's how it goes
1:38:00 when you do a nightly thing some you
1:38:02 know what's funny sometimes I come in
1:38:04 here like pissed off and cranky
1:38:06 and I'm funny as [ __ ] and then like
1:38:09 tonight to come in in a really good mood
1:38:11 and I'm like
1:38:15 are weird
1:38:21 how to compose a poem for my wife
1:38:25 um my wife just don't understand what
1:38:28 good that is
1:38:29 oh well
1:38:31 if your wife likes poems it it might get
1:38:34 you might get you a hug so it might be
1:38:38 good for that
1:38:42 um Troy what do you do for work you know
1:38:46 do you ever write an invoice do you ever
1:38:47 write emails do you ever do project
1:38:49 plans do you ever
1:38:51 I don't know try to write ad copy Do you
1:38:54 ever
1:38:56 try to solve a problem and how to have a
1:38:58 hard conversation with an employee
1:39:03 think about
1:39:06 so these things are called large
1:39:08 language models right
1:39:10 Claude Bing
1:39:13 large language models it's a it's an
1:39:15 interesting phrase
1:39:17 it basically just means they know a lot
1:39:19 of words
1:39:20 what else are large language models
1:39:24 we are
1:39:27 these things are reflections of all of
1:39:30 the words that we've dumped on the
1:39:31 internet for like the past five decades
1:39:33 that's what they've been trained on
1:39:37 and that's why there's all these
1:39:38 copyright things like you trained it on
1:39:39 my date oh nice llama thank you
1:39:43 um
1:39:45 so this is not like a Google search this
1:39:48 is more like
1:39:50 peering into the hive mind
1:39:52 and saying hey I need an assist
1:39:57 my wife's pif pissed off at me could you
1:39:59 write me a poem sure here you go
1:40:03 well like what do you do what do you
1:40:05 struggle with
1:40:07 what is something in your work that is
1:40:10 repetitive
1:40:11 Soul crushing what's what's the shitty
1:40:14 part of your work right it's like maybe
1:40:16 it's invoicing clients maybe what what
1:40:18 is it like figure out that thing
1:40:21 and then these things
1:40:24 can maybe assist you in that what if
1:40:27 that shitty repetitive soul-crushing
1:40:29 work could be automated to these things
1:40:33 chances are pretty good that it can and
1:40:35 if it can't be today it'll probably be
1:40:37 within six months to a year
1:40:41 and if you get curious about this [ __ ]
1:40:43 you could probably figure it out with
1:40:45 these Tools Plus something like zapier
1:40:47 that does automations you could take
1:40:50 that repetitive [ __ ] like invoicing or
1:40:52 writing letters or collection letters
1:40:53 for deadbeats that don't pay their
1:40:55 [ __ ] bills
1:40:58 and this thing could automatically do it
1:41:01 and you could have it start out with a
1:41:03 polite letter and then have the next one
1:41:05 be slightly more direct and then have
1:41:06 the next one be really direct and have
1:41:08 the last one be I'm going to take you to
1:41:10 [ __ ] court buddy
1:41:13 and it'll do it like that
1:41:16 and you're like but I write in a really
1:41:17 particular way copy your email paste it
1:41:21 into the prompt and say write the
1:41:23 nastygram but follow the tone of this
1:41:25 and it will like that
1:41:30 so abstract out
1:41:32 wife to poem or poem to wife
1:41:35 to any other problem in your life
1:41:39 and I need to solve that in some way and
1:41:42 see if you can get these things to do it
1:41:43 and chances are pretty good it'll get
1:41:45 close
1:41:47 so and hopefully your your lovely misses
1:41:50 loved your poem
1:41:52 the robot poem
1:41:54 and if it gave you a shitty poem that's
1:41:56 on you not on the not on the machine
1:41:59 if you if you give these things more
1:42:01 context
1:42:02 if if you're
1:42:04 less stingy in how you talk to these
1:42:07 things they're less stingy and how good
1:42:09 they are
1:42:10 it's amazing if if you give this what
1:42:13 Google has trained us to give it just
1:42:15 give it a few words just give it a very
1:42:17 little it will give you back [ __ ] crap
1:42:20 awful
1:42:21 but if you start out and you say hey
1:42:23 how's it going hope you're doing well uh
1:42:28 you're a career counselor and your
1:42:33 expert in you know uh coaching
1:42:36 professionals through sticky situations
1:42:40 situation
1:42:42 here's my company here's my roll here's
1:42:45 the sticky situation and how it came
1:42:47 about this is my point of view this is
1:42:49 their point of view I don't know how to
1:42:50 get to the other side on this do you
1:42:52 have any ideas
1:42:56 it will have ideas and if you give it
1:42:58 that context right they'll probably be
1:43:00 good and I guarantee it'll be three
1:43:02 things in there you didn't think of
1:43:04 there'll be like 10 that you're like
1:43:06 well that's obvious
1:43:07 and then like number seven will be like
1:43:09 oh that's kind of interesting I hadn't
1:43:11 really thought of that okay I'll try
1:43:12 that and that's just that's it was just
1:43:15 a fluke yeah it wasn't
1:43:21 what site is this I think I was on I
1:43:24 think the only side I've shown well I
1:43:25 showed chat gbt so I showed this site
1:43:30 chat.openai.com and then I showed
1:43:33 ideogram dot AI or ideogram
1:43:39 i-d-e-o-g-r-a-m dot a i we're the only
1:43:42 two I showed tonight
1:43:45 um what good is any of this I just told
1:43:47 you Troy try try try try
1:43:52 it's so sad dig deeper dig deeper
1:43:55 if you ask a trivial [ __ ] it will give
1:43:57 you trivial answers
1:43:59 give it something tough
1:44:01 tell us something you've been struggling
1:44:03 with personally see if it's got advice
1:44:05 for you
1:44:11 um Dr J Troy go play with it and figure
1:44:13 out what it can do to help you exactly
1:44:15 that's it
1:44:17 Troy was your great great Granddad with
1:44:19 the Wright brothers and kitty hot that
1:44:21 so Thelonious with the good burn to Troy
1:44:24 here not a burn but like was your great
1:44:27 great Granddad with um with the Wright
1:44:29 brothers in Kitty Hawk and ask why fly
1:44:31 yeah what's the purp okay okay they flew
1:44:34 for like 10 seconds what good is that
1:44:37 so what
1:44:39 yeah exactly
1:44:41 Nostalgia fixing Troy you found a highly
1:44:43 irregular part of the tick tock
1:44:47 [Laughter]
1:44:54 and fellowets
1:44:57 um oh the you know the other thing Troy
1:44:59 that um that these tools are really good
1:45:02 at
1:45:08 [Laughter]
1:45:16 yes they blocked it at my school
1:45:18 Nature's Blessing yeah I I just it's
1:45:21 it's [ __ ]
1:45:24 I try to have Grace for resistance to
1:45:28 change because I know change is [ __ ]
1:45:29 scary and I know educational
1:45:31 institutions are particularly allergic
1:45:33 to change
1:45:35 but everything is changing and I think
1:45:39 it is really [ __ ] important to give
1:45:41 our students
1:45:46 some sort of guidance
1:45:48 toward what's coming and the only way
1:45:50 you can give guidance towards what's
1:45:51 coming is to understand what the [ __ ] is
1:45:53 coming
1:45:54 and you can't understand what's coming
1:45:56 if you block it it's
1:46:00 it's [ __ ] here
1:46:02 your students have been using it all
1:46:04 [ __ ] summer
1:46:08 but they're gonna cheat
1:46:11 no
1:46:13 they're gonna use the world's most
1:46:14 capable
1:46:17 knowledge delivery system
1:46:23 and yeah they're kids they're gonna try
1:46:24 to outsmart you and they're gonna try to
1:46:26 [ __ ] with you
1:46:28 and as a teacher
1:46:33 you should navigate that and understand
1:46:35 that and help guide them and say you
1:46:38 know what let's bring Chachi PT into the
1:46:40 classroom
1:46:42 let's have it write a paper
1:46:44 and let's Jane or wait what's your name
1:46:47 Nature's Blessing
1:46:49 nature will you
1:46:51 read that paper out loud for us okay I'm
1:46:55 a little embarrassed but it's okay
1:46:57 read it
1:46:59 what do we think of that well it was
1:47:01 kind of boring
1:47:02 it sounded kind of trite yeah it was
1:47:03 kind of shitty wasn't it yeah
1:47:05 wonder why it was shitty why don't we go
1:47:08 learn how this thing works okay
1:47:10 oh it's shitty because we asked it a
1:47:12 shitty question I wonder what would
1:47:14 happen if we asked it a better question
1:47:15 class what a better question like
1:47:18 [ __ ] critical thinking
1:47:20 because that's what's going to be needed
1:47:23 to be able to use these tools and the
1:47:25 tools are going to be [ __ ] everywhere
1:47:28 yeah I'm all so foxy tonight but she oh
1:47:31 it just I I don't there's something
1:47:32 about schools blocking access to the
1:47:36 most profound tool in the history of
1:47:38 humanity sitting in our [ __ ] hands
1:47:40 that's making me insane right now
1:47:45 1977 there's no reason for any
1:47:47 individual to have a computer in their
1:47:49 home digital equipment CEO exactly that
1:47:51 exactly see right next to they blocked
1:47:54 it at my school is that that's
1:47:56 yes
1:48:04 it's gonna it hurts
1:48:08 [Applause]
1:48:09 [Laughter]
1:48:13 Winston just got the number one gifter
1:48:15 badge
1:48:16 I don't I don't know what wait what are
1:48:18 we up to like 28 000 likes you guys have
1:48:21 slacked off they've all got carpal
1:48:22 tunnel from the other night they hit 100
1:48:24 000 likes and now they're just like it's
1:48:26 Saturday night I just I had lasagna it's
1:48:29 it's summer and lasagna
1:48:32 I I think I'm gonna make a Jello mold oh
1:48:35 thank you Winston oh yeah
1:48:39 I'm so tired I'm so tired I can't yeah I
1:48:43 like Kyle but it's just I can't tap the
1:48:46 screen tonight it's just not listen it's
1:48:49 not my job okay it's not my job to tap
1:48:53 his screen
1:48:54 what's the point in this
1:48:58 chapter take me away
1:49:03 wesn radio Hard Rock and Hip-Hop here in
1:49:07 the lower Poughkeepsie Valley area where
1:49:09 are you wesn radio you've never told me
1:49:11 that because I'll say your place and
1:49:14 your call center in your uh number on
1:49:16 the dial if you want
1:49:18 not to do matters but I would like to
1:49:21 have a fun radio voice and actually have
1:49:23 a little script because I'm an idiot
1:49:27 um
1:49:28 all right let's see oh new little box oh
1:49:32 yeah Vicky you haven't seen this this is
1:49:34 this is the you can make money with chat
1:49:36 GPT button
1:49:38 um
1:49:39 [Laughter]
1:49:42 oh no it just jumped me to the bottom I
1:49:46 think I'm gonna I'm gonna skip ahead
1:49:47 what time is it 10 26 is it wait
1:49:49 Saturday okay
1:49:51 I'm like do I have to get up for work
1:49:52 tomorrow no I'm good I'm good I'm good
1:50:00 um can I have ai do excel like making
1:50:03 some formulas for grading systems yes
1:50:05 you can
1:50:07 um
1:50:09 there's all sorts of ways you can do I
1:50:12 mean
1:50:17 yeah I would just go into just go into
1:50:19 chat EBT and just
1:50:26 a sophisticated way you could do it is
1:50:29 if you pay for chat EPT plus you could
1:50:32 use code interpreter
1:50:34 you could talk to it about the kind of
1:50:37 grading system that you want
1:50:39 and then
1:50:40 have it describe
1:50:43 the grading system to you and then
1:50:45 describe how it would organize that in a
1:50:47 spreadsheet and then have it generate
1:50:49 the spreadsheet and you could download
1:50:50 the spreadsheet or you could just you
1:50:55 know have it talk about the grading
1:50:56 system put it in the table or put it put
1:50:59 it in a thing or have it write you Excel
1:51:02 um you know formulas and things like
1:51:04 that
1:51:05 um
1:51:06 you can also in Google Sheets there
1:51:09 there are a number of the the one that I
1:51:11 know of there's there's a
1:51:14 a Google Sheets Plug-In or an extension
1:51:17 called GPT for Sheets
1:51:20 where you can put
1:51:22 um open AI calls in the cells of a
1:51:26 spreadsheet
1:51:27 so you could actually have you could you
1:51:29 could you know take data from your
1:51:31 spreadsheet you know I don't know
1:51:32 something a student answered go have it
1:51:35 analyze that answer or add up all the
1:51:37 [ __ ] and take that and run it through
1:51:39 some for I you know there's lots of ways
1:51:42 to do it it's a great teacher too code
1:51:44 Weaver so
1:51:46 um for those of you interested in
1:51:47 education if you haven't checked out
1:51:50 um Khan Academy
1:51:52 within Khan Academy there's a new thing
1:51:54 relatively new three three four months
1:51:56 old called conmigo
1:51:59 that is a gpt4
1:52:01 teaching bot
1:52:03 that's really good it uses the Socratic
1:52:06 method if a student goes in there and
1:52:07 goes just give me the [ __ ] answer it
1:52:09 won't it'll it'll query the student
1:52:12 until it is convinced that he
1:52:14 understands the concepts before giving
1:52:16 them the answer
1:52:18 um really really interesting so if
1:52:20 you're interested in that in it in
1:52:22 education that's something worth
1:52:23 exploring
1:52:24 and I mean quite frankly like like you
1:52:26 know that idea of you know bring GPT in
1:52:29 the classroom
1:52:31 I mean teachers need to understand what
1:52:34 these things make possible right and
1:52:36 listen like
1:52:39 I I would say at this point this is
1:52:41 probably more on administrators than
1:52:43 teachers right because teachers might be
1:52:45 going yeah this stuff's cool but my
1:52:48 district says it's not and I can't use
1:52:51 it well then that sucks right so this is
1:52:54 this is a much bigger problem than any
1:52:56 individual
1:52:58 um
1:53:00 but but the the teachers that I've seen
1:53:02 that get their heads around this thing
1:53:04 and say oh crap this is a big deal I
1:53:07 gotta get my students I gotta train them
1:53:09 differently like I've got to educate
1:53:11 them in a fundamentally different way
1:53:12 because the world is not going to look
1:53:15 like what it looks now
1:53:18 like our jobs right now
1:53:21 are remnants from the Industrial
1:53:23 Revolution right where cogs in a wheel
1:53:27 where okay you gotta we gotta write the
1:53:30 copy for that social media campaign okay
1:53:32 Janie I know you just came out of
1:53:34 college
1:53:35 um you know we write things for
1:53:38 um Instagram and we write things for
1:53:39 Facebook and here's the formula and we
1:53:42 need you to go research 38 different
1:53:44 subjects for this target audience and 40
1:53:46 different subjects for that we want you
1:53:48 to spend the next month researching all
1:53:50 that [ __ ] and writing four thousand two
1:53:52 hundred and twenty two things in this
1:53:54 spreadsheet okay
1:54:03 two minutes
1:54:08 what does that person do now
1:54:12 if our schools are training people to be
1:54:15 that person doing entry-level work which
1:54:18 is repetitive executional tactical stuff
1:54:25 and these tools are gonna automate
1:54:28 the Tactical execution
1:54:31 what do we need to be teaching students
1:54:35 it ain't tactical execution
1:54:39 it's critical thinking
1:54:42 how do I use these tools why would I use
1:54:45 these tools when would I use these tools
1:54:48 what are these tools good at what are
1:54:50 these tools not good at where do human
1:54:52 beings need to be involved where do they
1:54:54 not need to be involved
1:54:57 those are the students we need coming
1:54:59 out of school
1:55:05 I don't know need someone who can follow
1:55:07 directions
1:55:09 these things follow directions
1:55:12 faster better 24 7.
1:55:20 you show up at my doorstep and say tell
1:55:22 me what to do
1:55:24 and I'll repeat it I can memorize it and
1:55:26 repeat it back to you I have no interest
1:55:28 in that
1:55:32 I'm kind of like imagining future State
1:55:35 employer but future state is like three
1:55:37 years from now so like all the kids that
1:55:39 are in school right now they're gonna
1:55:40 graduate into a world where the
1:55:42 entry-level jobs are not going to exist
1:55:45 and their skills are going to be for
1:55:47 entry-level jobs that aren't there
1:55:49 so yeah I don't know I've never been
1:55:53 passionate about education but all of a
1:55:55 sudden I am it's weird
1:55:57 but maybe maybe because I started a
1:55:59 channel called the AI learning lab and
1:56:02 I'm and I'm acting like a professor
1:56:04 even though I am wholly unqualified I
1:56:07 have no certificates
1:56:11 um it's a great teacher too it is chat
1:56:12 gpt's helping me through grad school
1:56:14 super cool you funny dog thank you Barn
1:56:17 pickle and barn pickle 9.5 out of 10 on
1:56:19 the name love the barn pickle I would
1:56:21 love to know the story behind your name
1:56:23 it makes me laugh it's in the
1:56:25 neighborhood of I like turtles and I bet
1:56:27 there's a good story behind it so so
1:56:30 fantastic name Barn pickle I am using it
1:56:33 to improve my math too oh yeah exactly
1:56:35 super cool
1:56:36 codeweaver also a good name
1:56:40 um David Peroni came in late shut up
1:56:45 um what was adding text to an image all
1:56:48 about oh
1:56:49 so David
1:56:51 so this image and that image and
1:56:56 that image and that image
1:57:00 have not been possible before
1:57:03 with things like mid-journey
1:57:06 because it couldn't do text
1:57:10 so these all say AI learning lab
1:57:13 so there is a new tool
1:57:16 called
1:57:22 that's not it
1:57:23 this is it
1:57:25 it's called ideogram
1:57:28 ideogram ideo
1:57:31 Graham dot a i
1:57:34 ideogram.ai where is it up here well you
1:57:37 can't see that that's stupid that dumb
1:57:39 dumb it that that dumb I can't see that
1:57:42 we're on Tick Tock I can't I got old
1:57:43 eyes I can't see that little teeny tiny
1:57:45 text I don't know what it is with you
1:57:46 young people in your young eyes and see
1:57:48 all the little teeny tiny texts anyway
1:57:50 this site is two days old
1:57:53 the discord's brand new there's my I
1:57:57 love Turtles
1:57:58 [Laughter]
1:58:01 um
1:58:02 it can do text
1:58:05 so you can write a prompt and why is it
1:58:08 not I think I have to reload it
1:58:12 um
1:58:14 art is not dead it's evolving I actually
1:58:16 love that that is the right that is the
1:58:18 that is right that's correct
1:58:21 this is this is the mistake I think
1:58:24 sagin wga are making right now
1:58:27 is there negotiating their contracts
1:58:29 trying to say that AI is bad and we will
1:58:31 never use AI but AI is going to be in
1:58:34 everything so art is evolving so
1:58:37 I think they should be doing their
1:58:39 contract negotiations within that
1:58:41 reality
1:58:43 um
1:58:44 you know find uh drink coffee find love
1:58:47 stay powered up on the move
1:58:50 Steve Jobs in front of Google like
1:58:52 Google there's Google written so
1:58:54 historically things like mid-journey and
1:58:57 stable diffusion could not do this
1:59:00 so if we say
1:59:02 um let's see
1:59:05 shiny
1:59:13 why does shiny not have an e in it shiny
1:59:16 beer can with uh
1:59:21 let's see with water
1:59:24 droplets
1:59:27 um
1:59:30 called uh
1:59:33 bush
1:59:38 uh
1:59:40 by stream
1:59:42 [Music]
1:59:44 headline text
1:59:47 say ah
1:59:53 D D D D D D D let's see if this does
1:59:58 this there's nothing like doing like
2:00:00 live you know live image generation with
2:00:03 you know awkward text awkward typing and
2:00:07 unclear prompting but we'll see how it
2:00:10 does
2:00:13 the other thing that's cool about this
2:00:15 site that you can do is you can click on
2:00:16 anything like if you see something you
2:00:18 like like there's no pain no game one's
2:00:19 pretty cool
2:00:20 The Prompt there is just no pain no gain
2:00:23 typography and so it did it did that one
2:00:26 did that one did that one
2:00:29 and did that one it's not always perfect
2:00:31 it gets a lot of this stuff wrong but it
2:00:33 gets it close enough that it's actually
2:00:34 it's quite impressive say ah
2:00:38 so it missed it didn't get the Bush beer
2:00:40 but it got to say ah really good
2:00:46 um
2:00:47 so let me go in here I'm going to say
2:00:48 photo cinematic
2:00:53 product
2:00:55 let's try that
2:00:59 so and that's that so this tool this
2:01:01 tool is very very simple right now I'm
2:01:03 sure it'll get more sophisticated over
2:01:05 time but basically you put in a prompt
2:01:07 you can choose some of those modifier
2:01:09 click buttons yeah so this is looking
2:01:11 pretty good
2:01:13 um although it got rid of all the text
2:01:14 now so
2:01:16 um and it's not by a stream anymore
2:01:19 but um
2:01:21 but I can take any one of these images
2:01:23 so like if I if I get one of these that
2:01:25 I like
2:01:26 let's see
2:01:28 um say ah we'll say this one we'll go
2:01:31 this one here
2:01:32 I can Now remix there's a remix button
2:01:36 so I can take that image and I can say
2:01:38 let me make that one photographic and
2:01:40 generate it and let's see see what that
2:01:42 comes up with
2:01:46 Waka Waka
2:01:53 it's not photographic
2:01:55 at all but it's a nice evolution of that
2:01:58 thing so anyway you gotta experiment
2:01:59 with this stuff but the fact that these
2:02:01 things are now generating typography
2:02:04 that's what you want it to be it just
2:02:06 it's a complete game changer for the
2:02:08 ability to at a minimum mock stuff up
2:02:11 but really
2:02:13 you're not that far from usable art
2:02:18 with this tool so crazy isn't it isn't
2:02:22 it crazy it's just crazy it's just crazy
2:02:24 it's you know you know David I'll tell
2:02:28 you if this stuff it it just makes my
2:02:30 head go silly it just makes my head go
2:02:33 silly
2:02:34 all right llama sent Lucy the Llama oh
2:02:38 Winston sent Lucy the Llama yes oh it
2:02:41 was a llama did I say llama
2:02:43 I think I did
2:02:45 um Genie in a Bottle it is a Genie in a
2:02:47 Bottle I am using it to improve my math
2:02:49 it's creating my high school writing
2:02:51 essays very cool damn thought he was
2:02:53 Rosie O'Donnell's ex-husband
2:03:01 oh my God my favorite trolls on this
2:03:04 channel
2:03:04 I've gotten Rosie O'Donnell I've never
2:03:07 gotten Rosie O'Donnell's ex-husband
2:03:09 that's awesome that's awesome thank you
2:03:12 I am duly uh uh flattered and insulted
2:03:17 at the same time which a good troll in
2:03:20 the AI learning lab does both at the
2:03:22 same time simultaneously oh that's very
2:03:26 not
2:03:26 that was mean right
2:03:31 oh
2:03:35 oh a mofo show you mofo show you mofo
2:03:40 show
2:03:41 you mofo show omofo show a mofo show
2:03:46 I don't know how to say your name all
2:03:48 you people with your lowercase
2:03:51 non-capitalizing non-dashed names
2:03:54 hurt my brain
2:03:58 but that was a solid troll thank you for
2:03:59 that that was lovely you sound like a
2:04:01 landlord yes I am a landlord
2:04:04 welcome have you paid your rent yeah
2:04:07 we'll give you 30 days grace Of course
2:04:09 of course
2:04:11 what's the dumpster for
2:04:13 I wouldn't worry about that
2:04:15 when do you leave for work fantastic
2:04:18 have a lovely day
2:04:20 all right
2:04:22 oh my God even writing grants yes it's
2:04:25 great at writing grants wow thank you AI
2:04:27 learning lab you're welcome any AI that
2:04:29 I don't have to pay for that I can use
2:04:32 as a virtual assistant with voice
2:04:36 capabilities
2:04:37 pie
2:04:40 in fact Pi if you go look at Pi dot AI
2:04:44 or go get the app wait where's my this
2:04:47 go here
2:04:55 go there
2:04:58 and go get the app
2:05:02 um
2:05:04 on iOS I think there's one for Android
2:05:06 but I'm too lazy to check
2:05:09 but but on iOS search for pi personal
2:05:12 assistant
2:05:14 and you'll find that app it's the
2:05:16 company is in flexion
2:05:18 and it's designed to talk right so you
2:05:21 put it into voice mode in the lower
2:05:23 right hand corner there's a little
2:05:24 button and you hit the little telephone
2:05:26 and then you just talk to it and when
2:05:28 you stop talking it starts
2:05:31 understanding what you said and then it
2:05:33 responds to you and it's connected to
2:05:35 the internet
2:05:36 so
2:05:38 I think
2:05:40 that like at a minimum extra Collective
2:05:43 it would be really interesting
2:05:45 to just treat Pie as a virtual assistant
2:05:49 and see how far you can take it see how
2:05:51 good it is
2:05:52 what I don't think it can do right now
2:05:54 is I don't think it can take action on
2:05:56 your behalf but it is connected to the
2:05:58 internet so while it won't be able to
2:06:00 book you a flight it might be able to
2:06:02 tell you what flight's available
2:06:06 and then
2:06:09 one of the things I've been saying on
2:06:10 this channel any limitation that you see
2:06:13 right now
2:06:14 is likely going to disappear over the
2:06:16 next six months to two years like any
2:06:18 limitation
2:06:19 so these guys inflection the company
2:06:22 behind that they just raised 1.3 billion
2:06:25 dollars 1.3 billion dollars
2:06:29 for what looks like a stupid little toy
2:06:32 that you talk to
2:06:35 but when you talk to it you realize oh
2:06:37 this isn't a stupid little toy this has
2:06:39 got like all of the knowledge of
2:06:41 humanity in it
2:06:43 and it can coach me and it can guide me
2:06:45 and I can make games and it can keep me
2:06:47 from being bored and it can entertain me
2:06:49 and it can educate me
2:06:51 and I can probably assist me
2:06:53 that's what they're setting out to build
2:06:56 so that that's that's the one I would
2:06:59 start with because it's the one that's
2:07:00 really designed I mean that's why
2:07:02 they're calling it pie your personal
2:07:04 assistant
2:07:05 because that's what they're setting out
2:07:07 to build is the world's best virtual
2:07:09 assistant
2:07:11 um is it the best virtual assistant
2:07:12 right now probably not even close and
2:07:14 Apple's gonna be big in this game I
2:07:16 think Apple's probably going to be the
2:07:18 first one to do this right is my
2:07:20 is my suspicion it's my hallucination
2:07:25 I hallucinate too but I hallucinate for
2:07:27 who's doing [ __ ]
2:07:30 I'm looking for a vocal conversational
2:07:32 AI that I could use for cold calling
2:07:35 it's called
2:07:37 [ __ ] what's it called there's one out
2:07:39 there that's specifically designed for
2:07:40 that
2:07:51 I can't remember it this is frustrating
2:07:56 I knew that one
2:07:59 um let's see
2:08:01 AI cold calling
2:08:07 before with AI assistants
2:08:10 Cognito or Balto that wasn't it
2:08:16 no vocal
2:08:21 calling
2:08:23 AI voice
2:08:31 oh this is frustrating
2:08:34 but Kyle why don't you just put your
2:08:35 ideas in a database and then when you
2:08:37 can remember them you can bring because
2:08:38 I have ADD that's just the what comes
2:08:41 with it I can remember most [ __ ] most
2:08:42 times
2:08:43 yeah
2:08:45 I can't remember it there's one out
2:08:48 there that it's designed to do just that
2:08:50 and I can't remember it
2:08:52 so
2:08:54 I'm a dum-dum I'm a loser I'm a loser
2:08:57 baby why don't you sue me yeah
2:09:07 I know I signed up for it
2:09:11 foreign
2:09:15 it ain't there it ain't there but anyway
2:09:18 uh there is stuff out there that's
2:09:21 coming I I don't know that it's all
2:09:23 there yet oh wait you know wait
2:09:26 hang on
2:09:28 I might be able to find this I know
2:09:30 where to find it see here's another
2:09:32 thing add does
2:09:34 is sometimes it gives you a little
2:09:36 superpower
2:09:38 all right let's go to x dot com
2:09:43 and then I'm gonna search for AI Tesla
2:09:49 sales call
2:09:54 ah
2:09:56 this is it
2:10:00 it's going pretty good man how about
2:10:02 yourself you said from Tesla right it's
2:10:04 the car company
2:10:08 yeah that's right I'm calling from Tesla
2:10:11 Motors the car company did you have any
2:10:13 trouble or questions well customized
2:10:17 it's Alexander from Tesla Motors how's
2:10:20 your day going so far
2:10:26 this is it
2:10:31 hello competitors see you soon no one
2:10:33 pretty good man how about yourself is
2:10:35 this the guy that does it
2:10:37 no
2:10:46 anyway if you if you go
2:10:49 I know this is painful if you go look
2:10:51 for
2:10:52 um
2:10:54 AI Tesla sales call
2:11:02 the world is about to change
2:11:19 why do none of these people link to the
2:11:21 actual site because the Dum Dums
2:11:24 [Laughter]
2:11:27 oh wait is this it yes
2:11:30 is it
2:11:38 no that's a newsletter
2:11:47 air AI
2:11:50 air dot a i
2:11:54 this is it there you go got it [ __ ]
2:11:57 Victory come on give me some hearts for
2:12:00 that something I know that was painful
2:12:03 come on give me anyone
2:12:08 everybody left
2:12:09 everybody left nobody's giving me love
2:12:15 River all right what did I get I got a I
2:12:17 got a cup my cup runneth over
2:12:20 air.ai
2:12:22 uh life's tough it's tougher when you're
2:12:25 stupid
2:12:26 um
2:12:29 air.ai it's finally here a hundred
2:12:31 thousand sales and customer service reps
2:12:34 at the top of a button air click to play
2:12:37 demo
2:12:38 hey James
2:12:40 I'm not instant man
2:12:42 yeah no worries I hate calls out of the
2:12:45 blue too but I'm actually from Apple and
2:12:47 I saw you were checking out Vision Pro
2:12:49 and we're about to pre-order but left
2:12:50 the page I was curious what had you
2:12:52 looking into potentially okay I can
2:12:55 respect that can I just ask one question
2:12:57 before you go
2:12:58 uh fire and shoot what's up
2:13:02 why did you well why did you decide not
2:13:05 to pre-order the vision Pros I saw you
2:13:08 changed your mind last minute
2:13:11 yeah
2:13:12 that's gonna change like okay I was
2:13:16 saying this earlier like any computer
2:13:18 mediated job
2:13:20 is going to be transformed in
2:13:23 fundamental ways right
2:13:25 look at look at this now
2:13:27 I can listen to that and and what I what
2:13:31 I immediately hear is that half a second
2:13:33 delay is a problem right
2:13:36 and you're like well there's a half a
2:13:38 second delay they know it's a robot it
2:13:39 sounds robotic give it a year
2:13:41 in a year that will be instant it won't
2:13:45 there won't be a delay there
2:13:47 the guy won't sound robotic he's going
2:13:49 to have inflection he's going to
2:13:51 understand Nuance they'll probably be
2:13:53 tracking the emotional state of the
2:13:56 person talking as well as what they're
2:13:58 saying
2:14:00 so they'll change their tactic based on
2:14:02 a combination of logic and emotion that
2:14:06 may be two or three years out
2:14:09 but that's outbound cold calling
2:14:13 this is here today
2:14:16 now I don't know what it costs I don't
2:14:18 know if it's any good I don't know if
2:14:19 you have to train it for [ __ ] four
2:14:21 months and it costs a hundred thousand
2:14:23 dollars to get off the ground
2:14:28 but but
2:14:31 this to me feels very indicative of
2:14:42 how dangerous dangerous it is to sit on
2:14:45 the sidelines of this AI [ __ ]
2:14:48 well I don't need to know AI I'm a sales
2:14:50 guy
2:14:51 I just I just call people
2:14:53 okay
2:14:57 how many hours a week do you spend
2:14:59 inputting data into Salesforce oh I
2:15:01 don't know 15 or 20 okay
2:15:05 what if you didn't have to make outbound
2:15:07 sales calls and you didn't have to and
2:15:09 you spent an hour a week inputting data
2:15:12 into Salesforce and you never missed
2:15:14 inputting a number
2:15:16 that'd be cool what would you do with
2:15:18 that extra time well I'd probably go on
2:15:20 in-person sales meetings okay
2:15:23 it's like not necessarily a bad thing
2:15:25 right but it might be a very different
2:15:27 thing
2:15:29 I'm a salesperson that spends half my
2:15:32 time inputting data into Salesforce and
2:15:34 half my time on the [ __ ] phone
2:15:36 and maybe once or twice a week I have a
2:15:38 lunch where I can build relationships
2:15:40 what if you could just go build
2:15:41 relationships
2:15:47 that's the kind of [ __ ] that this kind
2:15:48 of technology is going to make possible
2:15:50 every [ __ ] job is going to have some
2:15:53 version of this
2:15:57 show up in the next year year and a half
2:16:00 and the initial products the initial
2:16:03 pieces of software I guarantee you
2:16:05 setting this up is a [ __ ] travesty I
2:16:09 bet it's buggier than [ __ ] I bet
2:16:12 most people hang up on it and cuss at it
2:16:15 and give it a two out of 10 on any
2:16:18 [ __ ] survey if they're in a good mood
2:16:22 and then a year from now if they survive
2:16:25 it's gonna change the game
2:16:31 I feel so accomplished that I found that
2:16:37 foreign
2:16:40 occasionally you kick in
2:16:43 um it doesn't have to be repetitive Soul
2:16:44 crushing stuff either I'm into maths so
2:16:47 I'm on magical yeah
2:16:50 yeah it's yeah I'm just saying the
2:16:52 soul-crushing repetitive stuff is the
2:16:55 stuff that it's like the low-hanging
2:16:56 fruit of this AI stuff but yeah like
2:16:59 like the the the large language models
2:17:06 you can make money with chair TV you can
2:17:08 do knowledge work with chechi BT
2:17:12 you can do math with chat gbt
2:17:17 you can retrain yourself you can
2:17:18 reinvent yourself
2:17:26 every job
2:17:28 every computer mediated job every
2:17:31 computer media job
2:17:32 but Kyle I do this that's not computer
2:17:35 mediated really
2:17:37 do you go to an ATM oh yeah
2:17:40 how do you get paid oh yeah
2:17:46 I just do wallpaper okay
2:17:49 how do you schedule your appointments oh
2:17:51 yeah
2:17:53 how do you do marketing oh yeah
2:17:57 everything
2:18:02 please have upload this video to YouTube
2:18:04 because freeze freeze freeze which which
2:18:07 video oh oh this video okay yeah I will
2:18:11 Matthew I will I'm I'm so far [ __ ]
2:18:13 behind on uploading things I'll do that
2:18:15 tomorrow morning
2:18:16 I've got today I had to go buy paint and
2:18:18 then I had to sand my [ __ ] porch and
2:18:20 then I had to paint my [ __ ] porch but
2:18:22 that's done but tomorrow I got a fence
2:18:24 to paint and another porch to paint and
2:18:27 then I'll also upload some
2:18:28 some of these lives to my channel I'll
2:18:31 uh Matthew I'll upload uh like five most
2:18:35 recent ones tomorrow
2:18:37 and by the way everyone if if you want
2:18:39 to go
2:18:41 I don't know why you would but if you
2:18:43 want to go see old lives that are
2:18:46 recorded I am putting them up on YouTube
2:18:48 at ailearninglab Dash TT that's what
2:18:51 that URL is up there
2:18:54 so Tick Tock is the TT there too many
2:18:57 abbreviations
2:18:58 [Music]
2:19:00 all right I'm 99 messages behind here's
2:19:02 what I'm gonna do I'm gonna jump all the
2:19:04 way to the bottom because at this point
2:19:06 I'm probably an hour behind on comments
2:19:08 so I'm jumping all the way to the bottom
2:19:10 and then I'm just gonna scroll up a bit
2:19:11 and just sort of start at a random place
2:19:13 so
2:19:15 if you asked a question 45 minutes ago
2:19:18 and you're still waiting I probably just
2:19:21 skipped it so just pop it down below and
2:19:23 I'll see if I can get to it
2:19:26 um our critical thinking skills actually
2:19:27 teachable
2:19:29 um
2:19:30 yeah I think they are I mean
2:19:38 whether or not you call them critical
2:19:40 thinking skills it's
2:19:43 it it
2:19:46 it's the difference between teaching
2:19:48 kind of rote stuff to standardized tests
2:19:50 and just the the like all the repetitive
2:19:53 stuff all the repetitive
2:19:55 you know
2:19:59 that's the stuff that this stuff's gonna
2:20:01 do and that and and it's just so much of
2:20:04 our education
2:20:06 you know industry is is designed for
2:20:12 a work environment that that's required
2:20:15 like it's not a bad thing
2:20:18 you got to follow directions you got to
2:20:20 show up you gotta do this you gotta hit
2:20:21 the follow the follow the bouncing ball
2:20:23 and listen
2:20:25 this is not it's it's not like this is
2:20:27 going to happen overnight right
2:20:30 but if I play out five years if I say
2:20:32 I'm I'm a freshman in college today or
2:20:34 I'm a junior in high school or a senior
2:20:36 in high school
2:20:37 those people are five or six years from
2:20:40 entering a Workforce that's going to
2:20:42 look nothing nothing like our work looks
2:20:46 today
2:20:49 what's it going to look like I don't
2:20:51 know
2:20:53 but if everything's being reinvented if
2:20:56 this stuff is going to force change in
2:20:59 every job
2:21:00 it's going to require people that can
2:21:03 assimilate
2:21:04 lots of inputs
2:21:08 and deal with change
2:21:13 and orchestrate like like the real skill
2:21:15 with these I think is going to be like a
2:21:17 conductor of a symphony
2:21:20 where you'll understand that there are
2:21:22 five or six major tools you're going to
2:21:24 use to do anything and you need to be
2:21:27 able to understand okay boss wants me to
2:21:30 accomplish X okay how am I going to
2:21:33 accomplish X okay I'm going to use tool
2:21:35 a to do that it's no different than what
2:21:36 we do today it's just the output's very
2:21:38 different right
2:21:40 I'm going to use Excel to put the
2:21:43 numbers in and then I'm going to export
2:21:44 that I'm going to make a chart there I'm
2:21:46 going to put that chart into PowerPoint
2:21:48 and I'm going to do this and then I'm
2:21:49 going to put that into word and then I'm
2:21:50 going to hand that to my boss
2:21:52 and then the week will be over and I'm
2:21:54 going to drink
2:21:56 right
2:21:57 and and the new thing maybe some
2:21:59 different kind of thing that comes out
2:22:01 of it and it'll be different kinds of
2:22:02 tools but it's like it's like
2:22:04 orchestrating lots of different tools
2:22:06 and it'll be much faster turnaround
2:22:11 you can volunteer at any school in
2:22:12 America we would love to see you trust
2:22:14 me Nature's blessing thank you
2:22:17 um I yeah I don't I I tend I got
2:22:22 escorted out of a newspaper Publishers
2:22:26 conference in 1996 because I was saying
2:22:30 this sort of [ __ ] to newspaper
2:22:32 Publishers that I said the thing that
2:22:35 got got me kicked out is I said
2:22:39 in retrospect I know why they kicked me
2:22:41 out but
2:22:45 if they had listened to what I said they
2:22:47 might not be as
2:22:49 you know
2:22:50 in existential crisis as they are I said
2:22:53 that um
2:22:56 that the internet
2:22:58 is collapsing
2:23:00 the distance between editorial and
2:23:03 advertising that everything's co-mingled
2:23:06 and it's instantly accessible
2:23:08 so I can literally click I can be in an
2:23:10 article and click over to an ad and it's
2:23:12 it's all it's all doing this
2:23:14 and you know there's not this massive
2:23:17 separation between AD and editorial and
2:23:19 so I sat on the front of the stage and I
2:23:22 said you know as an industry just
2:23:24 knowing what's coming you should really
2:23:26 explore blurring the line between
2:23:28 advertising and editorial and the the
2:23:31 guy that organized the event said and
2:23:34 Kyle's gotta go now
2:23:36 he literally walked me to the side door
2:23:39 and said thanks
2:23:41 that was it so I'm happy to come to
2:23:44 schools but you know
2:23:47 um need pointers to enhance my AI
2:23:50 awareness
2:23:52 um I'll leave us 777 just right right
2:23:54 now just so
2:23:57 start paying attention to it so start
2:24:00 following people on Twitter that talk
2:24:02 about AI so whether it's newsletters or
2:24:05 whether it's
2:24:06 you know smart people there's a guy
2:24:08 named Ethan Malik who talks about this
2:24:10 stuff in a really brilliant way
2:24:12 um
2:24:14 just start getting curious with chat gbt
2:24:17 just start exploring that that that's
2:24:18 the real
2:24:20 um
2:24:21 the real thing to do right now is just
2:24:24 really understand how these things are
2:24:26 different than what we do today
2:24:29 I mean no regular search has turned up
2:24:31 complex
2:24:33 P addicts I don't know what that means
2:24:36 Winston maybe you're not talking to me
2:24:38 all right uh it's so cool that we're
2:24:40 hanging out right now
2:24:42 thank you spell again please I don't
2:24:45 know what I was spelling
2:24:48 but I'll spell it again if I know what
2:24:50 it is I mean I am really just
2:24:53 a nobody doing this and if this janky
2:24:56 stuff I can't wait for the good oh if
2:24:57 this is the janky stuff this is the good
2:25:00 stuff listen we're all nobody's just
2:25:02 trying to get some [ __ ] done Winston I'm
2:25:05 right there with you
2:25:06 um
2:25:08 yeah I mean that's the thing like I call
2:25:11 these things janky what's janky about
2:25:12 them is just the sophistication of the
2:25:14 applications right they're just they're
2:25:16 new right they're new and they they've
2:25:19 been thrown together in a lot of cases
2:25:21 really really quickly
2:25:23 um
2:25:24 yeah when these things get sophisticated
2:25:26 and when so right now you've got all of
2:25:30 our software we deal with in the world
2:25:33 is over here and then we got all this AI
2:25:35 stuff is over here
2:25:37 and they're very separate right now and
2:25:39 that's why we have things like AI
2:25:41 detectors and you know you should say if
2:25:42 it's done with AI well when these things
2:25:45 start to do this which they're doing
2:25:47 they're just going to be ubiquitous
2:25:49 they're going to be in everything and
2:25:51 then they're also going to be multimodal
2:25:53 so it won't be like I do word stuff and
2:25:56 then I some other time go do image stuff
2:25:58 and then if I think about it I go to
2:26:00 music stuff
2:26:03 these things are just going to be like
2:26:06 some sort of input I need X and it will
2:26:10 just give you X and then you're like oh
2:26:12 what would be the music for that um what
2:26:14 would that sound like no that's not it
2:26:18 that's not it oh that's it
2:26:23 yeah it's gonna get it's gonna get
2:26:25 stupid it's gonna get stupid like it's
2:26:27 already stupid to your point like if
2:26:29 this is the shitty stuff like what's the
2:26:32 good stuff gonna be I don't [ __ ] know
2:26:34 that that thing I just showed the air.ai
2:26:39 like that's the shitty stuff right now
2:26:42 a robot calls a guy that's like I don't
2:26:45 want to talk to you it's like can I just
2:26:46 ask you one more question yeah shoot you
2:26:48 know you know where that's going if if
2:26:50 that's the example on their website
2:26:52 you know that guy ended up ordering the
2:26:55 [ __ ] Apple Head eyeglasses right you
2:26:58 know that
2:27:02 and when and
2:27:06 that dude
2:27:08 or the dudette
2:27:10 that a I bought
2:27:12 will make those calls 24 [ __ ] seven
2:27:15 never get tired
2:27:18 learn I mean here's the other thing
2:27:22 if they're doing it right
2:27:25 it's not only a large language model but
2:27:27 it's being trained on every conversation
2:27:31 so the conversation happens and and you
2:27:34 know there was a no within 10 seconds
2:27:38 and the bot kept that guy on the phone
2:27:41 for 16 minutes
2:27:44 and he bought
2:27:45 all right let's take that conversation
2:27:47 and let's put that in a training set
2:27:49 like whatever the bot did have it do
2:27:52 more of that so it'll get more and more
2:27:54 and more efficient
2:27:56 at doing what it does
2:27:58 like that's the shitty version
2:28:02 what happens when this is good
2:28:07 [ __ ] crazy crazy which website is
2:28:10 that the the the the
2:28:13 talking thing was air.ai
2:28:18 um oh ideogram was the picture one dang
2:28:20 I'm an idiot I just found one reference
2:28:22 oh well
2:28:25 all right um can you do that with
2:28:27 mid-journey the the letters no you can
2:28:30 do it with if you go to clipdrop.co and
2:28:32 go to
2:28:34 doodle
2:28:35 the doodle tool that clip drop can do
2:28:38 text on images but not as consistently
2:28:40 as this thing can
2:28:42 the uh Ido ideogram or ideogram
2:28:47 wow a lot of people here 123 people good
2:28:49 Lord Saturday night people welcome to
2:28:51 the AI learning lab my name is Kyle
2:28:52 Shannon if you've been here for a bit
2:28:55 you can see I'm just a dum-dum I'm just
2:28:57 dumb dumb just an idiot just an idiot
2:28:59 but you know what you know we like talk
2:29:00 about here you know what I like to do
2:29:01 here
2:29:09 um if you're new to this stuff if you're
2:29:11 not played with chat gbt
2:29:13 well you can't make money with it that's
2:29:15 not what we talk about here this isn't
2:29:17 an MLM scheme a pyramid scheme trying to
2:29:20 get you to buy our course on how to make
2:29:21 money with chat gbt
2:29:23 um go to chat gbt and start playing with
2:29:25 it start interacting with it if you
2:29:27 haven't it's like seriously like right
2:29:28 now like I'm in a [ __ ] cranky mood
2:29:31 tonight because I started talking about
2:29:32 education and it's making me cranky
2:29:35 um
2:29:35 don't sit on the sidelines with this
2:29:37 [ __ ]
2:29:38 this stuff's really really important
2:29:41 it's very profound it's very powerful it
2:29:42 can be a lot of fun it's really exciting
2:29:44 it can be [ __ ] terrifying
2:29:46 but just go there chat that openai.com
2:29:49 and then in another window or another
2:29:51 tab open that document prompts.chat that
2:29:54 will teach you how to use that
2:29:58 and
2:29:59 it's just this will sort of talk about
2:30:01 what these things are and how they work
2:30:04 um and you start playing with it and and
2:30:05 ideally play with it
2:30:08 until you have your Kevin McAllister
2:30:10 moment
2:30:15 and hopefully you will and when you do
2:30:17 then you can't really go back
2:30:19 because you're like oh [ __ ] these things
2:30:22 are really powerful oh good lord I
2:30:24 didn't I didn't know it was that power
2:30:26 I'm a little tired I don't I don't I
2:30:29 don't know that I want to reinvent my
2:30:30 career right now so um yeah I know but
2:30:35 none of us really want to reinvent our
2:30:38 careers right now
2:30:39 however
2:30:45 and breathe
2:30:47 we're gonna do some deep breathing
2:30:48 exercises here at the AI learning lab I
2:30:50 hope you're all doing well tonight thank
2:30:52 you share bear
2:30:53 really oh you changed your name share
2:30:55 bear
2:30:57 I like it I'm in
2:30:59 very cool
2:31:01 oh River pie is too chatty it is chatty
2:31:04 GPT we did warn you river that pie is
2:31:07 chatty GPT I know pie is really [ __ ]
2:31:11 annoyingly chatty if you're trying to
2:31:13 get something done so so a couple of
2:31:15 things about the personalities of these
2:31:17 things
2:31:19 um
2:31:20 so chat GPT is just like good solid you
2:31:24 know solid B student
2:31:27 gpt4 is just really good you know it's
2:31:29 got It's got parts of it that are just
2:31:31 [ __ ] rock solid and then but it
2:31:33 really sucks at sports or whatever
2:31:35 um Claude has got a really big memory
2:31:38 really long Bing is super corporate but
2:31:41 kind of fun and it's connected to the
2:31:43 internet
2:31:45 Bard is
2:31:48 really trying to be like chat GPT Bard
2:31:52 is like the the guy that just can't get
2:31:54 asked to the prom
2:31:57 and but but he's actually pretty smart
2:32:00 and and you know he's connected to the
2:32:03 internet and
2:32:04 you know
2:32:07 he's gonna be okay he like he's probably
2:32:09 gonna be the successful one in you know
2:32:11 in college but he's not doing well in
2:32:13 high school
2:32:15 um Labs not perplexities geeky
2:32:18 Gigi research type
2:32:21 um the the Llama models you can play
2:32:23 with there are just kind of like really
2:32:25 raw chat gpts Pi is chatty GPT it's just
2:32:30 like it's just like well welcome how are
2:32:33 you tonight that's just fantastic let me
2:32:35 ask you a quick question yeah you know
2:32:38 before I answer your question could we
2:32:41 go on ahead and explore why you want to
2:32:44 be here today yeah I just want the
2:32:46 answer to the question I assume I
2:32:48 totally understand that I totally
2:32:49 understand that you really want to get
2:32:51 to that answer that must be frustrating
2:32:53 for you that I haven't gotten to the
2:32:54 answer anyway
2:32:55 before I answer can I can you just
2:32:57 answer two more questions for me no I
2:32:59 don't want to answer those questions
2:33:00 could you just [ __ ] give me the
2:33:02 answer well wow no need to get testy I
2:33:04 just I'm really just trying to
2:33:06 understand so that's Pi I'm
2:33:08 now there's some times where you want to
2:33:10 be in a conversation then it's really
2:33:12 good but if you want to get [ __ ] done
2:33:14 it's very annoying and then Poe is just
2:33:17 like it's like a jack of all trades lots
2:33:20 of things you can do there there's Bots
2:33:21 it's there's like the it's pose like the
2:33:24 sort of party bot like the party llm all
2:33:28 right
2:33:30 I I though I I've never thought about
2:33:33 them as as anthropomorphized people in
2:33:36 high school but now I have so I'm sure
2:33:39 I'll come up with all sorts of stupid
2:33:40 ass voices for them too just like I did
2:33:43 chatty GPT yeah no that's really
2:33:45 fantastic this if can I just ask you one
2:33:48 more question before I go ahead and
2:33:49 answer you yeah
2:33:51 I'm sure you really do want to know the
2:33:53 weather you know you know sometimes the
2:33:55 weather can be really warm and it can be
2:33:57 it can really hot sometimes and then
2:33:58 other times it's more cold
2:34:00 oh you wanted to know the weather like
2:34:02 now so question did you want to know the
2:34:05 weather right now or did you want like
2:34:06 the weather for later like in a few
2:34:08 hours like I that that
2:34:10 literally happened to me River where I
2:34:12 just said hey can you give me the
2:34:13 weather in Denver like how long is it
2:34:16 going to rain or something like that it
2:34:18 was raining and it's like it's like oh
2:34:20 yeah it's a fantastic question dude did
2:34:22 you want to know how long it's raining
2:34:23 like like right now or like a couple
2:34:25 hours from now I'm like but really just
2:34:28 I don't know give me both
2:34:31 all I could give you both yeah that
2:34:32 makes that makes good sense I'll give
2:34:34 you both
2:34:35 it's like
2:34:36 so yes I agree
2:34:38 uh upright upright upright Apple user
2:34:43 your channel is growing Kyle sweet thank
2:34:46 you oh yeah a lot of people in here
2:34:47 tonight
2:34:48 um and call it serious I want to start a
2:34:50 business piggybacking off AI
2:34:58 um
2:35:00 I would say figure out what business you
2:35:02 want to make first and then do the AI
2:35:06 rather than the other way around don't
2:35:08 just lead with the AI or you'll just be
2:35:10 in a
2:35:11 never-ending Loop of frustration just
2:35:14 because these things you gotta have the
2:35:17 strategy first have the business first
2:35:18 and then figure out what of these how
2:35:21 can these things support that
2:35:22 the good news is
2:35:25 whatever you want to do whatever problem
2:35:27 you want to solve in the world that you
2:35:28 think people will pay for
2:35:30 these tools can likely help you do it
2:35:33 faster better or cheaper than the
2:35:36 competition that's not using it so
2:35:38 that's the good news the bad news is
2:35:40 these are just [ __ ] tools
2:35:42 so you have to figure out the problem
2:35:44 you want to solve first in my humble
2:35:46 opinion because otherwise you're just
2:35:48 using technology gratuitously
2:35:51 and that serves no one and it wastes
2:35:53 time and money and energy
2:35:59 Alexander made by air a i yes Ted I
2:36:04 found it but thank you I'm glad you were
2:36:07 here I if I had been watching my
2:36:10 comments in real time I would have seen
2:36:11 that this is so freaking weird I know I
2:36:14 found nothing on complex
2:36:16 p p a a month ago two weeks ago now I
2:36:21 find it
2:36:21 using what using chat GPT or using
2:36:24 something like Bard because Bard's
2:36:27 connected Bard Bing and Pi are connected
2:36:30 to the internet so they might find more
2:36:32 recent stuff Pi's particularly good at
2:36:35 knowing very recent stuff
2:36:38 surprisingly good conspiracy theory
2:36:40 Kyle's not real but actually a deep fake
2:36:42 from chat gbt recruiting feeder I like I
2:36:46 wish DARPA deep fake you want to know
2:36:49 something scary I was talking to we were
2:36:50 talking to DARPA in the early days of
2:36:52 storyline
2:36:54 and they said they said they wanted us
2:36:56 to write a white paper about how we
2:36:58 could use storyvine for soldiers in the
2:37:01 battlefield
2:37:03 and so I thought well you could do time
2:37:06 shifted psych evals right so you could
2:37:09 have a soldier
2:37:12 like record
2:37:14 themselves when they had down time in
2:37:17 the field
2:37:18 and then those videos could be viewed by
2:37:21 psychologists stateside
2:37:23 and at some point so we got on the on
2:37:27 the phone with the guy we were walking
2:37:28 him through the white paper he's like
2:37:29 okay
2:37:30 and at one point I said I said you know
2:37:33 so at some point
2:37:35 um
2:37:36 like the soldier might say hey I'm doing
2:37:39 great doc but the doc in watching that
2:37:41 video could know that they're not doing
2:37:43 great and he said oh no no we can detect
2:37:47 all that in real time he goes he goes
2:37:50 when the video comes in we'll know
2:37:52 exactly what the state of the soldier is
2:37:54 we like we can look at their their
2:37:56 irises we can listen to their voice he
2:37:59 goes no that's all handled so when
2:38:02 you're talking to DARPA they they got
2:38:05 some [ __ ] there that you know we ain't
2:38:07 know about it's like oh [ __ ]
2:38:11 AI makes a four-hour week with work week
2:38:13 doable it does I think that's one of the
2:38:16 the very real possibilities is is we
2:38:19 start to get time back in our life like
2:38:22 like the the amount of [ __ ] that we do
2:38:24 that frustrates the [ __ ] out of us and
2:38:27 makes us [ __ ]
2:38:29 starts to go down
2:38:31 the amount of work goes up but but the
2:38:33 our time required to do that amount of
2:38:35 work goes down so like the the pain in
2:38:37 the ass factor of life goes down as does
2:38:40 sort of the requirement to put in lots
2:38:42 of hours to Output what you need to
2:38:44 output and that Gap
2:38:47 what do we fill that with I don't know
2:38:48 like I think we can fill that Gap with
2:38:52 spending more time with our family
2:38:54 I don't know socializing starting a new
2:38:58 project living a dream
2:39:00 like I just I feel like we're all gonna
2:39:03 have the opportunity to kind of have
2:39:05 some new space in the week
2:39:08 that we can actually express ourselves
2:39:12 in some meaningful way that's what I'm
2:39:15 hoping like that's the thing that's the
2:39:16 thing I'm actually like so to deep down
2:39:18 what I'm hoping all of this [ __ ] can do
2:39:20 is give us back some meaningful time
2:39:24 that we can actually do something Johnny
2:39:27 Blaze welcome
2:39:29 I found a link that says it was
2:39:31 discovered in 2013 but so far no
2:39:33 peer-reviewed paper or original source
2:39:35 so Winston that's be really careful any
2:39:38 any of these things including the ones
2:39:40 connected to the internet will often
2:39:42 give you sources with links
2:39:44 and it's just all [ __ ]
2:39:47 so so take everything with a grain of
2:39:50 salt that these things give you spec if
2:39:52 you're trying to do anything that is uh
2:39:55 whatever you call it uh where accuracy
2:39:59 is important
2:40:00 um take everything these things generate
2:40:02 with a grain of salt you still have to
2:40:05 check it absolutely these they just they
2:40:08 hallucinate a lot especially when it
2:40:09 comes to things like citations
2:40:15 not German love the name hey how's it
2:40:18 going welcome welcome welcome
2:40:22 um
2:40:31 sorry I'm trying to find my place trying
2:40:34 to find my place in the questions you
2:40:36 know sometimes I got a little behind and
2:40:37 then Tick Tock jumped me around and I'm
2:40:39 here now I mean you know what I mean no
2:40:44 this shirt this collar is like it's the
2:40:48 I need this shirt needs to be in in the
2:40:50 retirement center
2:40:54 it's like it's just like the shirt's
2:40:57 like I'm kind of done now could I be
2:40:59 done hanging on your shoulders
2:41:02 um your YouTube videos need more
2:41:04 creative titles yeah I know so elephant
2:41:07 wait Ella Ella fantas
2:41:11 um if you scroll if you if you look at
2:41:13 all of my if you look at all of my
2:41:15 videos click on the video button and the
2:41:17 scroll to the bottom the bottom ones
2:41:19 have really good descriptions and
2:41:20 chapter markers in them because I had
2:41:23 Claude write all those and they're
2:41:25 really good and then Claude stopped
2:41:28 working for some reason and now I can't
2:41:29 get it to do accurate chapter
2:41:31 descriptions anymore so I've spent like
2:41:34 I don't know three or four total hours
2:41:36 just trying to get the prompts to work
2:41:38 and then Claude rate limits me before I
2:41:41 can finish and says I can't work anymore
2:41:43 for 24 hours so I've been [ __ ] blocked
2:41:46 by Claude
2:41:48 so uh yes the descriptions are
2:41:50 non-existent for most of the videos but
2:41:52 some of them have lots of description in
2:41:55 them so some are good but yes I do need
2:41:58 to do that thank you for the reminder
2:42:00 that I'm
2:42:02 16 hours behind on my work
2:42:05 yeah
2:42:14 they will adapt just like we did
2:42:16 yep exactly I'm your 60th YouTube
2:42:19 subscriber fantastic nostalgic flicks
2:42:22 thank you
2:42:23 nostalgic flicks 59 Farah welcome
2:42:28 what is ask AI so someone asked uh they
2:42:33 want to use AI to do voice cold calling
2:42:36 for sales
2:42:37 and I'm like I've seen that and I
2:42:39 couldn't remember what it was so ask.ai
2:42:41 is
2:42:44 they claim that they can have a hundred
2:42:46 thousand simultaneous
2:42:50 um sales people or customer service
2:42:53 people just talking so so imagine
2:42:56 instead of having a customer service
2:42:58 call center
2:43:00 where people call in they get put on
2:43:02 hold for 45 minutes then they reach a
2:43:04 human being that's at level one support
2:43:07 which basically means they can only read
2:43:09 a script they've only worked for the
2:43:11 company for two weeks and they don't
2:43:13 care about anyone or anything that's
2:43:15 Level One support
2:43:17 and then
2:43:19 they put you on hold for another 15
2:43:21 minutes and you get the level two
2:43:22 support and then right and then you get
2:43:24 the level four support and you just
2:43:26 about to get your question answered and
2:43:27 they dropped the call
2:43:29 that experience What If instead of that
2:43:32 you called and basically a robot
2:43:35 answered and it was like pie chat GPT it
2:43:38 was chatty GPT
2:43:41 and it knew all the [ __ ] about the
2:43:43 company and you could ask it questions
2:43:45 and it would say here's your answer does
2:43:47 that answer it nope and then you explain
2:43:49 more about your situation and it would
2:43:51 know enough to know
2:43:53 when it needs to escalate it to a human
2:43:56 so you could have a 15-minute
2:43:57 conversation with this back and forth
2:44:00 with this bot it knows everything it's
2:44:02 it's taking care of updating your
2:44:04 address on your account it's taking care
2:44:06 of this it's taking care of that and
2:44:08 then it actually gets to your problem
2:44:09 and it's like oh yeah you need to talk
2:44:12 to someone let me put you in touch with
2:44:14 them
2:44:15 and then not only does it put you in
2:44:17 touch with them it synopsizes it
2:44:19 summarizes the conversation you just had
2:44:21 with it
2:44:23 and everything else about your account
2:44:25 that's relevant to this conversation and
2:44:27 puts that on the screen in front of the
2:44:29 human being so the human being picks up
2:44:31 the phone and instead of saying could
2:44:32 you please repeat everything you've said
2:44:34 four times the human being just goes hey
2:44:38 so it looks like you were talking to the
2:44:40 bot and the bot you know cleared up
2:44:42 these things but you're kind of stuck on
2:44:44 this one issue is that correct yeah
2:44:46 that's perfect I can fix that for you
2:44:48 imagine that experience
2:44:51 so that's what ask AI is purporting to
2:44:54 be and then the other so they're saying
2:44:55 customer support on one side and then
2:44:57 outbound sales on the other side
2:45:00 so literally a bot that will make calls
2:45:02 24 7.
2:45:04 to whatever your [ __ ] Target list is
2:45:06 and have conversations with people until
2:45:08 it books a meeting
2:45:11 it's different when you see and talk to
2:45:13 Children you would motivate them oh
2:45:16 thank you Nature's Blessing yeah no I
2:45:18 listen I I obviously like doing this
2:45:20 stuff I you know I'm happy to do that so
2:45:23 um
2:45:24 yeah I you know I just
2:45:28 yeah
2:45:30 I yeah it's interesting it's interesting
2:45:34 I think I would it if I were in the
2:45:37 presence of people that were
2:45:40 I don't know I don't know
2:45:43 yeah I'm happy to chat
2:45:46 all right I used to sell survey software
2:45:48 and talks with LA Times research one guy
2:45:50 left in the department it's over oh
2:45:53 interesting yep
2:45:55 I certain certain industry what's going
2:45:58 to be really sad is
2:46:01 it's not like there's going to be a big
2:46:03 trumpet sound like
2:46:06 this job is over or this industry's over
2:46:08 it's just gonna be
2:46:10 the phone stops ringing
2:46:16 like
2:46:18 yeah if you do research if if if you do
2:46:21 kind of lower level research
2:46:24 even some higher level research
2:46:27 if the CEOs of the companies you sell to
2:46:31 can just go to these things and get the
2:46:33 answers
2:46:34 they don't need your research and so
2:46:39 they're not going to announce I'm using
2:46:41 AI to replace you it's just going to be
2:46:42 like the renewal's going to come up and
2:46:44 they're going to be like oh yeah that I
2:46:46 haven't used them in like a year no I'm
2:46:49 good we don't need to renew that
2:46:50 and then just
2:46:53 you call them up say hey we we've worked
2:46:55 together for 20 years yeah I know yeah
2:46:58 yeah no I don't I just yeah I just don't
2:47:00 really I I use this now I'm good
2:47:04 but but this is our business yeah I get
2:47:06 it but
2:47:08 sorry
2:47:10 it's like
2:47:11 there's gonna be a lot of you know you
2:47:13 know there was that there were there
2:47:14 there was that little meme we went
2:47:16 through like everyone was quiet quitting
2:47:18 where they're just like working at home
2:47:19 like [ __ ] it you know I hate my job
2:47:21 anyway so [ __ ] the man right
2:47:24 I yeah I feel like I feel like there's
2:47:26 gonna be sort of you know sort of quiet
2:47:30 you know
2:47:31 um
2:47:32 uh atrophy of business where it's just
2:47:35 gonna
2:47:36 just gonna just slowly Drift Away drift
2:47:39 into the background
2:47:40 people will be like wow what why is the
2:47:43 phone not ringing what's the is anybody
2:47:45 there there's no one here
2:47:49 it's [ __ ] terrifying
2:47:51 I recommend everyone go there and
2:47:53 subscribe it's a good way to catch up on
2:47:55 missed shows I don't know which that's
2:47:57 in reference to but that's cool
2:47:59 I like that I like when you guys talk to
2:48:01 each other I think that's super cool
2:48:02 what's your opinion on
2:48:04 marketing blocks 2.0 I don't know it
2:48:06 let's go look at it
2:48:10 marketing blocks
2:48:14 2.0
2:48:20 free AI software creates all your
2:48:22 marketing Assets in minutes copy Pages
2:48:24 Graphics videos voiceovers emails logos
2:48:26 there's gonna be eight billion of these
2:48:30 things
2:48:34 I'm even looking at building one of
2:48:37 these into storyvine into my current
2:48:38 thing but the the hook the thing that
2:48:41 I'm trying to make different is
2:48:43 um
2:48:45 I can't learn about this without logging
2:48:47 in that's a real pain in the ass
2:48:50 um
2:48:54 I'm looking to do story-centric content
2:48:56 creation
2:48:59 unlimited words 50 000 words a month
2:49:03 for how much 37 bucks a month build 400
2:49:07 bucks a year I wouldn't sign up for
2:49:09 anything for a year right now
2:49:12 certainly not something like this I
2:49:15 think I think these guys are making like
2:49:18 create with AI uh create AI Video
2:49:21 Creator
2:49:35 I can't play with this without
2:49:37 subscribing that's a real [ __ ]
2:49:40 dumb dumb move
2:49:43 yeah this the
2:49:45 so
2:49:48 I don't know like I I
2:49:53 activate your five-day risk-free account
2:49:55 no contracts yeah well
2:49:58 I have 80d so signing up for a five-day
2:50:02 risk-free account means that I'm paying
2:50:04 for this for the next three months
2:50:05 whether I like it or not and I guarantee
2:50:07 I'm gonna not like it
2:50:10 I hate that [ __ ] marketing tactic
2:50:12 right there that that drives me [ __ ]
2:50:14 crazy the fact that I can't use it like
2:50:17 that knocks Five Points out of the ten
2:50:19 point scale right there
2:50:21 because I don't know how good it is and
2:50:23 I don't give a [ __ ] if it's like hey I
2:50:25 couldn't even get into the site without
2:50:27 giving them a [ __ ] email so now I
2:50:29 just signed up for [ __ ] 43 [ __ ]
2:50:32 emails Kyle you know you didn't you
2:50:35 haven't responded in three things I mean
2:50:37 I know I'm not going to take you
2:50:38 personally but uh you know I'm assuming
2:50:41 that you don't want to talk so tell me
2:50:42 otherwise
2:50:44 so I'm gonna get all those emails now
2:50:47 okay
2:50:49 I don't like I don't like strong on
2:50:52 marketing I hate it it drives me [ __ ]
2:50:55 crazy so
2:50:56 um so I I would say like so I see a site
2:51:00 like that and I think like well I
2:51:02 already know that Jasper exists I
2:51:03 already know copy AI exists both of
2:51:05 those I can go play with and understand
2:51:07 what they are without having to [ __ ]
2:51:09 you know give my life away and and put
2:51:12 in a credit card and they're more
2:51:15 established it might be really good but
2:51:17 I don't know but like I don't have no
2:51:19 way of finding out and you know and I'm
2:51:22 not going to so how do we make money off
2:51:24 all this I don't see it yet
2:51:30 that is not the right approach in my
2:51:33 opinion
2:51:34 it's not about how do you make money
2:51:36 with AI
2:51:38 maybe why you can't see it is because
2:51:40 you're thinking about how do I make
2:51:41 money with AI and you're looking at
2:51:43 these imperfect tools and you're like
2:51:45 but they're kind of imperfect and
2:51:46 they're kind of not right and
2:51:49 um
2:51:51 here's my philosophy on this
2:51:54 these are just tools
2:51:57 so it would be just as easy Merlin to
2:52:01 for for your question to say how do you
2:52:03 make money off Photoshop or how do you
2:52:05 make money off Microsoft Word
2:52:08 so wait Microsoft Word you can you can
2:52:10 type words into this thing like like
2:52:13 back when it came out when there were
2:52:14 typewriters right you had to like type
2:52:16 type onto paper and if you made a
2:52:18 mistake you had to use white out or if
2:52:20 you had a fancy typewriter it would have
2:52:22 a white ribbon and you could white out
2:52:23 the things in line right and this this
2:52:27 tech software comes out where you can
2:52:29 just write into this TV screen
2:52:33 and then change stuff and move stuff and
2:52:36 then at some point when you're done then
2:52:38 you can print it
2:52:41 well how would you make money with that
2:52:43 how would you make money with a with a
2:52:45 thing where you don't actually
2:52:47 see the words on the paper in front of
2:52:49 you that that's kind of what you're
2:52:50 asking right now it's like it's in the
2:52:52 same neighborhood
2:52:55 so the question isn't how do you make
2:52:57 money with Microsoft Word or how do you
2:52:59 make money with AI the question is how
2:53:00 do you make money and how do you make
2:53:02 money is you figure out a problem that
2:53:05 someone has
2:53:06 that you know how to solve
2:53:08 and you solve that problem from them for
2:53:11 them
2:53:12 in a good enough way
2:53:14 that they pay you money
2:53:18 oh my God I I hate it when I can't write
2:53:21 stuff if only someone could write words
2:53:24 on paper I can do that well I've got Jim
2:53:27 that does that on the typewriter well I
2:53:29 can do it faster than Jim because I have
2:53:32 this new fancy thing called a word
2:53:33 processor and I'm also really good at
2:53:35 this and so I'll do it better and faster
2:53:38 and cheaper than Jim does it you want
2:53:41 one oh yeah well we'll do let's start
2:53:43 with a small project oh you're really
2:53:44 good yeah here's money
2:53:47 so you know what I would what I would
2:53:49 say to you is like what are you good at
2:53:51 what's what's the stuff where you can
2:53:52 solve a problem and then and then what I
2:53:55 would say is
2:53:57 get start to get literate about these
2:53:59 tools start to understand them what
2:54:01 they're good at what they're not good at
2:54:03 thank you Michael
2:54:06 start to understand what they're good
2:54:07 and not good at
2:54:10 and then what you can probably do is you
2:54:12 can say hey
2:54:16 I could probably like I'm seeing this
2:54:18 with some agency friends
2:54:20 I could probably start an agency
2:54:24 where I could deliver at or above
2:54:28 the quality of work that client is
2:54:31 getting now
2:54:32 for half the money and half the time
2:54:36 that's a really compelling thing you go
2:54:40 to any agency and you can say I know
2:54:42 you're spending 250 000 a year with that
2:54:45 agency
2:54:46 pay me a hundred grand
2:54:48 I'll turn things around in two weeks
2:54:50 instead of a month and I'll guarantee
2:54:52 it's the same or better quality
2:54:55 okay
2:54:57 I worked with you on that campaign back
2:54:59 then that's cool all right yeah let's go
2:55:02 client just saved 150 000 dollars
2:55:06 if you're good enough at these tools
2:55:08 you could probably deliver at a quarter
2:55:11 of this price and a quarter of the speed
2:55:13 but you've only promised them half so
2:55:15 you can now confidently hit
2:55:18 half the half the money half the time
2:55:20 same quality
2:55:23 but that's not that doesn't have
2:55:25 anything to do with these that has to do
2:55:26 with your solving that specific problem
2:55:29 with these
2:55:31 so I would flip I would flip
2:55:35 flip the formula flip the classroom
2:55:40 you make it fun Kyle low thank you doc
2:55:44 thank you Dar from face up Studio
2:55:46 uh I try to make it fun what time is
2:55:48 it's late isn't it what are we doing
2:55:50 Saturday night what'd you do Saturday
2:55:51 night I hung out with some old guy on
2:55:54 Tick Tock it's very
2:55:57 it's very odd all right I'm gonna I'm
2:55:59 gonna do a few more questions I'm now 55
2:56:01 questions behind I've been doing this
2:56:03 way too long tonight uh I played with
2:56:05 each of these they are all very biased
2:56:08 towards specific ideology well Merlin so
2:56:11 are humans
2:56:12 um so if you if you if you ask these
2:56:15 things
2:56:16 um
2:56:17 if you ask these things to be biased
2:56:20 they'll be biased so
2:56:22 listen if if you're just looking Merlin
2:56:24 for ways not to use these there's all
2:56:27 sorts of reasons not to use them yes
2:56:29 they're biased yes they hallucinate no
2:56:32 they're not perfect they're janky pieces
2:56:34 of [ __ ]
2:56:37 and they're as bad as they're ever going
2:56:40 to be
2:56:42 so if you just sit on the sidelines and
2:56:46 judge them and say they're not perfect
2:56:48 this shit's gonna kick your ass
2:56:53 great info taking notes thank you
2:56:56 advertising is biased everything's
2:56:58 biased here's the thing the myth the
2:57:01 myth of the bias free large language
2:57:02 model Here's my thought on this
2:57:05 so so what Merlin's saying is okay Not
2:57:08 only was this trained on biased
2:57:11 information on the internet which is
2:57:13 kind of horseshit because there's yes
2:57:15 there's biased information on the
2:57:16 internet but there's bias in both
2:57:18 directions so it's trained on both right
2:57:19 but but then also the human
2:57:22 reinforcement
2:57:24 training that was done at open AI was
2:57:26 done by people in San Francisco which is
2:57:28 biased right there's there's bias in a
2:57:30 certain way there right so it's really
2:57:32 easy to go okay they all fall in this
2:57:34 direction
2:57:37 if the exact same models were trained in
2:57:40 I don't know Tuscaloosa Alabama
2:57:44 for example
2:57:46 they might be biased in the other
2:57:48 direction right so so someone might look
2:57:50 at the current model and go that feels
2:57:52 just right Merlin looks at it and goes
2:57:54 that's really [ __ ] up there's gonna be
2:57:56 some other model that Merlin's gonna
2:57:57 look at and goes oh that's not biased
2:57:59 that's perfect so
2:58:01 we're all biased we all have some point
2:58:04 of view right there isn't going to be
2:58:07 the perfectly neutral model it's just
2:58:10 not possible we're [ __ ] humans we
2:58:13 choose sides we choose teams our
2:58:15 politics right now are like a we're like
2:58:18 football fans
2:58:20 you're in Denver Denver sucks you're in
2:58:24 Philadelphia you're an [ __ ] nothing
2:58:26 we can say to each other will ever
2:58:28 convince us that our football team isn't
2:58:31 the best and the other guys are [ __ ]
2:58:33 right that's what our politics has
2:58:35 turned into
2:58:37 so what's going to happen in the future
2:58:39 is these things will likely get more
2:58:41 transparent with
2:58:43 the rules that they live by with the
2:58:46 imperatives that they filter through and
2:58:49 then Merlin can choose his model and I
2:58:51 can choose mine and felonious can choose
2:58:54 his like well there will likely be
2:58:56 different models and they might be very
2:58:57 very personalized
2:58:59 um but yeah they're biased right now
2:59:00 it's just you know this is very new
2:59:03 technology so it's biased it's been
2:59:06 trained on you know mostly white guys
2:59:10 because mostly white guys were leading
2:59:12 putting [ __ ] on the internet
2:59:15 and it's still mostly white guys in
2:59:17 Technology and Engineering positions is
2:59:19 it getting better yeah it's getting
2:59:20 better but these things have been
2:59:21 trained on the past 50 years of data on
2:59:24 the public internet
2:59:25 so it's very U.S Centric it's very white
2:59:28 guy Centric
2:59:29 and it's probably more to the left than
2:59:31 the right
2:59:34 is everyone aware of that yeah
2:59:36 everyone's [ __ ] talking about it
2:59:38 every day all the time
2:59:41 is it an easy fix no
2:59:45 you know how do you add more [ __ ] how do
2:59:48 you add more data that doesn't exist do
2:59:51 you invent it do you synthesize it just
2:59:53 so that it's more balanced but then it's
2:59:54 fake and fake on Whose thing and whose
2:59:57 point of view should be in there anyway
2:59:59 it's a really complicated really deeply
3:00:02 deeply deeply complicated
3:00:04 um issue
3:00:06 there's ways that we can fine tune it
3:00:08 that doesn't change the core model but
3:00:09 it makes it behave better that's
3:00:11 something that they're working on right
3:00:12 now
3:00:14 so
3:00:16 you know speaking of weather it's the
3:00:19 fifth it's in the 50s here in Michigan
3:00:21 wow
3:00:22 it's it was actually nice today it was
3:00:24 like in the 80s here in in Denver it's
3:00:26 probably I don't know 60 something right
3:00:28 now what do you recommend for SEO I
3:00:31 don't know I think SEO I I think SEO
3:00:35 changes radically over the next three
3:00:37 years so I don't know
3:00:39 um I would go to futurepedia.io so this
3:00:42 this little site that I I put up here
3:00:45 um
3:00:47 futurepedia.io is a list of like 4 000
3:00:51 AI tools there's probably a bunch of SEO
3:00:55 stuff in there I know there's some SEO
3:00:57 things as plugins and chat GPT I don't
3:01:00 know that world very well
3:01:02 um short term you can certainly get I
3:01:05 mean I mean you can have Chachi BT can
3:01:08 write you you know really good SEO
3:01:11 optimized content
3:01:13 um
3:01:15 but it but if you if you're talking
3:01:17 about doing you know super sophisticated
3:01:19 automated stuff I don't know those tools
3:01:21 but I I'm I know they exist
3:01:24 I just we're we're entering a very weird
3:01:28 space where
3:01:32 where are we in the middle of a
3:01:34 transition we're at the beginning of a
3:01:36 transition where where what a search
3:01:38 engine is changes radically
3:01:41 like they're they're changing radically
3:01:45 um
3:01:46 I think all the rules for SEO are going
3:01:48 to change I don't know I don't I it's
3:01:50 it's an area of grand ignorance for me
3:01:52 and
3:01:54 I don't like that the only thing I know
3:01:56 is it's going to be different I just
3:01:58 don't know how it's going to be
3:01:58 different I don't I don't know where I
3:02:00 would put my energy if I were trying to
3:02:02 do SEO right now because it just feels
3:02:04 like at some point that shit's just
3:02:05 gonna shut off in a weird way
3:02:08 um what's the weather like right now
3:02:11 oh that's funny the Irregulars
3:02:16 why not make Soldier crickets having a
3:02:19 war against the largemouth bass wait
3:02:21 what
3:02:23 it's not just gym anymore
3:02:27 oh Jimmy Moore it's not oh I'm just
3:02:29 Jimmy Moore that's a fun name I'm just
3:02:32 Jimmy Moore why not make Soldier
3:02:34 crickets having a war against the
3:02:36 largemouth bass in ponds yeah exactly
3:02:40 we're not fighting this [ __ ] the Gap all
3:02:43 right it just jumped me to the bottom
3:02:44 it's a sign I gotta go I'm gonna I'm
3:02:45 gonna scroll up I'll do a handful of
3:02:47 more questions
3:02:48 um
3:02:49 what's up good early morning from the
3:02:52 Netherlands oh sander I didn't know
3:02:53 you're in the Netherlands that's cool
3:02:56 um
3:02:57 are you understanding these tools wait
3:03:00 are you understanding what these tools
3:03:03 are good and what they're not good at
3:03:06 Kyle Molly
3:03:10 yeah I mean that's what I'm spending
3:03:12 that's probably that question is kind of
3:03:15 exactly
3:03:17 what I try to do every day with this
3:03:20 [ __ ] is
3:03:21 figure out where the boundaries are
3:03:23 right what are they good at what are
3:03:25 they not good at
3:03:27 um
3:03:29 like with chat gbt
3:03:32 um
3:03:33 for ideation and just general writing
3:03:36 gpt4 is [ __ ] amazing
3:03:39 plugins
3:03:41 seemed really promising but they're so
3:03:44 janky and [ __ ] up and there's not real
3:03:47 clear
3:03:48 developer activity in them some of them
3:03:51 felt spammy like they just became felt
3:03:53 like they were useless
3:03:55 code interpreters [ __ ] remarkable but
3:03:58 when I actually am trying to use it for
3:04:00 work
3:04:01 I'm getting really intermittent results
3:04:03 where sometimes it's [ __ ] brilliant
3:04:04 and then sometimes I just can't get it
3:04:06 to do what I want and then other times
3:04:08 it starts brilliant and when I try to
3:04:10 just tweak something it completely falls
3:04:13 apart so
3:04:15 so I don't really know how to use that
3:04:17 one Claude is really good at long
3:04:20 documents and long conversations but it
3:04:23 rate limits you to like 25
3:04:25 queries so just about the time you're
3:04:28 getting some work done it's like you got
3:04:29 three more chats left you got two more
3:04:32 chats left one more okay see you in 24
3:04:34 hours like [ __ ]
3:04:37 um Bing is just just this side of
3:04:40 useless it's connected to the internet
3:04:42 it can do some good stuff but it's so
3:04:44 it's so like safety control so so yeah
3:04:47 so like like that's kind of what what
3:04:48 I'm doing is the the reason I put up
3:04:51 these these seven
3:04:53 you know these three Chachi PT Claude
3:04:56 and Bing and then barred perplexity pie
3:05:00 and Poe
3:05:01 these are the ones right now for me that
3:05:03 I'm just looking at you know which one's
3:05:05 getting better which one's getting worse
3:05:07 if if a new player comes into the scene
3:05:09 like do they make the list does someone
3:05:10 else drop off
3:05:12 and and the reason there's so many on
3:05:15 there is like every one of these I can
3:05:17 tell you why it's on the list because it
3:05:19 does something slightly better than the
3:05:21 other thing so yeah that's kind of what
3:05:23 I'm spending most of my time doing
3:05:26 um and it's certainly in a science this
3:05:28 is a lot of this is Instinct right now
3:05:30 and it's the other thing that I would
3:05:32 say Molly is
3:05:33 don't get too married to anything like
3:05:36 like
3:05:38 barred was not on the list until like a
3:05:42 month ago like it was laughably bad it
3:05:45 was just like these are the guys that
3:05:47 [ __ ] invented the Transformer that's
3:05:49 powering all of this other remarkable
3:05:51 [ __ ] are you [ __ ] kidding and then
3:05:54 just in July sometime it got good
3:05:59 I like I don't know what they did I
3:06:00 don't know if they flipped the switch I
3:06:02 I don't know it just got not horrible
3:06:06 and so it made the list and apparently
3:06:08 Gemini that they're working on right now
3:06:11 that Larry Page came back to Google and
3:06:13 is heading up the Gemini Project from
3:06:15 what I understand
3:06:17 I have a feeling that
3:06:20 when Gemini comes out they said it's
3:06:22 going to come out in the fall
3:06:23 that that it could sort of move to the
3:06:26 top of the list and and really be a
3:06:28 challenge to chat GPT
3:06:31 so
3:06:33 I'm
3:06:34 I'm trying to do it I'm trying to use
3:06:36 these things to do my work I'm trying
3:06:38 like why I'm on these things is just to
3:06:40 like the uh
3:06:44 this tool that that the ideogram
3:06:49 I learned that from someone in the live
3:06:51 last night they're like what do you
3:06:52 think of ideogram I'm like I don't
3:06:53 [ __ ] know what you're talking about
3:06:55 and I went there and I was like oh [ __ ]
3:06:59 someone came out with a model that does
3:07:00 text like I learned that here
3:07:06 so I'm just in sponge mode and
3:07:11 I guess I'm an aggressive sponge mode
3:07:13 I'm I'm in Swiffer mode
3:07:17 oh
3:07:19 I'm like aggressively going after this
3:07:22 [ __ ] right now because I'm just trying
3:07:23 to figure out that the other thing is
3:07:25 this stuff is so profound and it's so
3:07:27 broad It's like because it's going to
3:07:30 affect everything like no one can
3:07:32 understand what this thing's gonna do in
3:07:34 in every corner much less even in a
3:07:37 single corner it's so expansive so I'm
3:07:39 just trying to understand it is where I
3:07:42 am right now I don't think and you know
3:07:44 if I were to give myself how much do I
3:07:46 understand this [ __ ]
3:07:49 0.0003 maybe it's like so if you're like
3:07:54 well I'm at
3:07:55 .0001 okay great so we're in the same
3:07:57 neighborhood
3:07:59 because it's not understandable it's
3:08:01 it's too big it's too profound it's too
3:08:03 vast and it's moving it like it's
3:08:05 evolving literally while we're speaking
3:08:08 [ __ ] is probably being
3:08:10 released that will change the
3:08:12 conversation again
3:08:15 so
3:08:16 yeah it's
3:08:18 I don't have a list I'm not you know I'm
3:08:20 not writing any of this [ __ ] down
3:08:21 because it's just all going to be
3:08:22 Irrelevant in a month
3:08:24 um did you already know that or find it
3:08:26 I don't know what you're talking about
3:08:27 AI can write command algorithms for free
3:08:29 it deed 46 000 likes 47 right now
3:08:33 fantastic
3:08:36 sorry I have to go all right Sandra
3:08:38 going to the east of the Netherlands
3:08:40 nice I don't know what that means Kyle
3:08:42 what do you think of AGI
3:08:44 um
3:08:47 I think
3:08:49 camcatkin I love I love your name cam
3:08:53 how's it going Kim catkin
3:08:56 nice to meet you Cam cadkin how's it
3:08:58 going
3:09:00 should be a salesperson I think
3:09:03 um
3:09:09 I
3:09:11 what do I think about AGI
3:09:14 I didn't know the term until a year ago
3:09:17 when I heard watched a YouTube video Sam
3:09:20 Altman talk about it and that that's
3:09:22 what open AI is all about
3:09:25 and then I started using these tools and
3:09:32 my experience of these tools was
3:09:36 oh [ __ ]
3:09:38 the Turing test is over
3:09:44 like this passes the Turing test if you
3:09:45 don't know what the Turing test is it's
3:09:47 when a human talks to a computer and
3:09:49 can't tell whether it's a computer or
3:09:51 not and I started using chat gbt and I'm
3:09:53 like oh
3:09:55 um
3:09:56 like every
3:09:58 chat bot I'd used for the decade leading
3:10:01 up to chat GPT
3:10:04 was a disappointing travesty of
3:10:08 this
3:10:10 will at this rate computers will never
3:10:13 pass the Turing test and then Chachi BT
3:10:15 came out and I really started playing
3:10:16 with it and I'm like oh oh [ __ ] I guess
3:10:18 that's over
3:10:19 and then I saw some interviews with
3:10:21 people and they're like well but it
3:10:23 hasn't really passed the Turing test
3:10:24 because my definition of the Turing test
3:10:26 is you've got to be able to interact
3:10:28 with the computer for four hours
3:10:29 straight and if you can do that and
3:10:31 still not begin and so so then what I
3:10:34 like what hit me was oh they're starting
3:10:36 to qualify
3:10:37 the Turing test and what what landed for
3:10:40 me is like well it's kind of irrelevant
3:10:42 like if we're at the point with these
3:10:45 models that we're now arguing over
3:10:48 semantics of whether or not the
3:10:50 computers are strong enough to not know
3:10:52 if it's a human or not
3:10:54 then we're pretty [ __ ] close and then
3:10:57 so AGI is the next step above the Turing
3:10:59 test right where where computers are now
3:11:02 AGI artificial general intelligence
3:11:05 they're smarter than humans they can
3:11:07 reason better they're smarter you know
3:11:08 take the smartest human and and the
3:11:10 computer is smarter than that
3:11:13 these things are already way the [ __ ]
3:11:16 smarter than I am right so for me
3:11:20 okay you know my capability is here my
3:11:24 intelligence is here
3:11:26 this thing's capability is
3:11:29 here and it's intelligence is way up
3:11:34 here it's it's wisdom and reasoning is
3:11:37 way down here right so I think where
3:11:39 these Things Fall Apart is the is the
3:11:41 like the critical thinking I always talk
3:11:42 about they're really shitty at that
3:11:44 they're like an imbecile that knows
3:11:46 [ __ ] everything right
3:11:50 but in terms of capabilities like
3:11:52 creating artwork like that writing
3:11:56 quickly writing well writing
3:11:58 consistently writing like it's it's
3:12:00 already above my capability and it
3:12:02 already knows like way more than I will
3:12:05 ever know and I ever have any interest
3:12:06 in knowing it's it's like off the screen
3:12:09 better at knowing [ __ ] it's way down
3:12:12 here in terms of reasoning the AGI stuff
3:12:15 in my opinion is just the reasoning part
3:12:18 coming up to here right
3:12:20 and that it's coming pretty quick
3:12:22 but for me
3:12:24 it feels like a semantic argument like I
3:12:27 don't really give
3:12:28 all that much of a [ __ ] if it can reason
3:12:30 I there's reasons so so when the thing
3:12:34 can reason as good as humans that's when
3:12:36 you can literally say
3:12:39 this human isn't necessary because this
3:12:41 thing can now reason at a human level
3:12:43 and if it gets to the point where it
3:12:45 gets above it then you can say okay that
3:12:47 knocks out a whole group of humans right
3:12:49 because this is reasoning at a much
3:12:51 higher level
3:12:52 so but but I kind of feel like
3:12:55 it's all just a matter of degrees at
3:12:57 this point these tools are already way
3:12:59 smarter than we are they just don't
3:13:01 reason as well as we do so I don't know
3:13:03 so
3:13:05 foreign
3:13:09 yeah I don't know
3:13:13 I don't know the glimpses of
3:13:15 holy [ __ ] [ __ ] or just start use
3:13:18 again if if you're in this channel
3:13:20 tonight and you haven't played with
3:13:22 Chachi BT
3:13:24 go
3:13:26 there right now just go there right now
3:13:29 chat.openai.com and
3:13:31 I don't care how cynical you are we've
3:13:34 got we we had at least one in here
3:13:36 tonight I don't get it it's biased
3:13:39 whatever
3:13:40 I will not use it until it's what I want
3:13:43 it to be okay
3:13:44 even if you're that person go there and
3:13:46 start using it and if you don't know how
3:13:48 to use it go to prompts.chat that's what
3:13:50 these two things are there for
3:13:52 um
3:13:54 because you'll see it this is not Google
3:13:56 this is not a Google search this is
3:13:58 something very very different and it's
3:14:00 this is very much the tip of the iceberg
3:14:02 of what's coming
3:14:05 so I mean according to the weatherman he
3:14:08 says it would be more Sunny there
3:14:11 um
3:14:12 two more questions is the rapid
3:14:15 acceleration of open AI scary scary yes
3:14:18 Daniel yes sir yes dot Daniel bayonez I
3:14:22 like your name
3:14:24 um it's terrifying how it's terrifying
3:14:26 how quickly it's coming and it's not
3:14:28 just the acceleration of open AI
3:14:30 like like this image
3:14:34 this was not possible two days ago
3:14:39 I found out about this tool last night
3:14:41 on this live I went there today and I
3:14:44 was like I had a Kevin McAllister moment
3:14:47 it does text now
3:14:49 another threshold cross check that's
3:14:52 checked off the list
3:14:55 is it perfect no
3:14:56 but can I make something that says AI
3:14:59 learning lab that looks like I spent
3:15:01 more than three minutes on it yep I can
3:15:03 I can do that now
3:15:07 you want to Silo with alert AI learning
3:15:10 lab on it yeah we can do that
3:15:12 you don't have to be a Route 66 looking
3:15:14 thing we can do that
3:15:17 gritty New York yeah we can do that
3:15:20 so
3:15:22 um
3:15:23 is it scary
3:15:26 it's exciting
3:15:27 what's what's scary to me is the impact
3:15:30 on people like people are going to be
3:15:32 [ __ ] up because of this like longer
3:15:34 term five years
3:15:36 there's gonna be a lot more jobs created
3:15:38 because of this stuff so I feel like
3:15:40 long term we're in we're in better shape
3:15:42 than we are right now
3:15:45 um short term is going to be a I don't
3:15:47 it's going to be ugly and that I don't
3:15:49 that doesn't make me happy at all
3:15:51 you need to build your 1 billion
3:15:52 parameter llm model I do well I'll let
3:15:56 someone else build it I'll just use it
3:15:57 fascinating
3:16:00 Thelonious uh have you heard of alpha
3:16:03 fold in 2022 it determined every
3:16:05 possible protein fold for human DNA 2022
3:16:07 exactly yeah
3:16:10 the the the the uh
3:16:14 the
3:16:17 the founder of that
3:16:20 the founder of deepmind that created
3:16:23 Alpha fold so Google's AI group is now
3:16:26 called deepmind because they basically
3:16:28 just took it over and said okay this is
3:16:30 our AI group now
3:16:32 so the guy that found a deep mind he's
3:16:34 the one that that started Pi dot AI that
3:16:37 just raised 1.3 billion dollars so
3:16:39 that's why that's why I'm looking at Pi
3:16:41 even though it looks like a stupid
3:16:43 little tool right now
3:16:45 this thing
3:16:47 I have a feeling that little toy is
3:16:49 gonna get really really really really
3:16:52 really powerful
3:16:56 I don't know how yet but the guy that
3:17:00 invented the [ __ ] that figured out how
3:17:02 to predict all the 400 000 proteins
3:17:06 he's making a chat bot
3:17:08 chatty GPT
3:17:11 is is the Deep Mind dude
3:17:13 so yeah it's wild isn't it
3:17:16 I just found it in the text okay I'm so
3:17:19 excited about all these possibilities I
3:17:20 know me too it's just [ __ ] nuts all
3:17:23 right listen Pie has so many
3:17:24 possibilities I am crispy look bloodshot
3:17:27 eyes the hair is falling apart the
3:17:29 T-shirt gave up [ __ ] two years ago
3:17:32 I haven't drank enough water but I still
3:17:35 believe this
3:17:41 good night everybody cam thank you let's
3:17:44 see who we got here Donna I see Donna I
3:17:47 see cam Stevo see you dude I see Tobias
3:17:53 I see Thelonious cavruno
3:17:56 Daniel
3:17:59 any any gen xers
3:18:02 what was that show that did this I see
3:18:07 Jersey Share Bear thanks Joker
3:18:12 um yeah I'm Gen X Boomer here
3:18:15 um what was the show anyway like I
3:18:18 always like they're gonna call my name
3:18:19 they're gonna call my name they never
3:18:21 [ __ ] called Kyle
3:18:22 Romper Room right from Romper Room
3:18:25 exactly it was like five or six they're
3:18:26 gonna call my name they're gonna call my
3:18:28 name never [ __ ] called my name I'm
3:18:30 still bitter about it all right
3:18:32 everybody good night thanks a lot I'll
3:18:34 see you tomorrow