Mar 7, 2024 - How Inflection 2.5 Transforms Creativity and Learning

Description

Chapters

Transcript

0:02 [Music]
0:15 [Laughter]
0:16 [Music]
0:29 ready
0:31 [Music]
0:40 will
0:43 secret David
0:46 play the
0:48 Lord you don't really care for music do
0:55 you goes like this for the f
1:01 M fall a maor
1:03 lift
1:05 ba
1:07 composing
1:08 [Music]
1:11 hallelu
1:14 hallelujah
1:18 hallelujah
1:22 hallelujah
1:26 hallelujah oh
1:29 yeah
1:32 [Music]
1:59 oh
2:03 oh
2:06 [Music]
2:12 [Music]
2:29 champion
2:30 [Music]
2:39 all righty everybody welcome Happy what
2:42 is it Thursday is it Thursday
2:46 [Music]
2:51 yeah um it is Thursday night I have gone
2:57 off and left my my laptop top at work
3:02 and I just realized it as I sat down to
3:04 do this live thank you for pulling me
3:06 from the state of the union I thought I
3:09 thought old uh old Joe was was pey he
3:13 was pretty fiery but I figured I'd seen
3:16 enough and they'll be talking about it
3:18 for the next week anyway
3:20 so don't need to see it
3:23 [Music]
3:26 live but anyway my ability to demo
3:30 tonight might be hampered so we might
3:33 need to just do
3:35 soapbox soap boox AI learning
3:40 lab uh if you're new here my name so
3:43 yeah I'm going to use my iPad that that
3:44 little image right there that's my iPad
3:47 uh it'll be interesting do you have two
3:49 computers Kyle I do I have a computer at
3:52 work computer and a
3:55 laptop but I needed my laptop at work
3:58 today cuz Mo'Nique and I my co-founder
4:00 were working on a big
4:04 presentation and so I to have my laptop
4:06 there so we could work in a conference
4:07 room and then I just left it at work cuz
4:10 you
4:12 know neuros spicy that's the game and
4:16 remembering [ __ ] is occasionally not
4:19 going to happen
4:22 yay talk to Pi 2.5 well I would except
4:26 Pi uh 2.5 Pi was I I tried to do it
4:31 before I came on and it was like yeah
4:32 we're not doing
4:34 that let's see what we got going here
4:38 um uh
4:40 what
4:43 version of inflection are you using
4:49 question
4:53 mark hi Kyle
4:57 ah all right back I figured would be
5:00 back by the time we went live ha okay so
5:05 couple of
5:06 things my name's Kyle Shannon this is
5:08 the AI learning lab if you're here that
5:13 means one of a couple of things you're
5:15 in a
5:17 regular they know who they
5:20 are it could mean you're curious about
5:24 Ai and you're like I got to figure this
5:26 [ __ ]
5:27 out when does class start
5:31 is the real Professor ever going to show
5:33 up you're in the perfect
5:36 place you could also be here because you
5:39 were flicking through Tik Tok and you're
5:40 like did Alec Baldwin get
5:42 fat or
5:47 whatever is that Rosie
5:49 O'Donnell
5:54 um and if you're here if that's why
5:57 you're here welcome welcome this is uh
6:04 we have fun we have fun
6:06 here and we do learn a lab stuff so if
6:08 you have questions about AI drop them in
6:10 the comments
6:12 below We've Got Mods that'll help uh
6:14 sort of bring those to the Top If I get
6:16 to rambling so so one of the other
6:18 things that goes on here being a neuros
6:21 spicy and all one of the nicknames of
6:24 the channel is Chad
6:27 add because occasionally I will set off
6:30 to say show someone how to do something
6:33 and then 45 minutes later I still
6:36 haven't started that's just the way it
6:38 goes here so set your expectations at
6:42 the door um lower them if your bar is
6:46 here bring it
6:49 down and then when you get your
6:52 expectation level here bring it down
6:54 even further and then we'll be
6:58 good you'll be fine you'll be like that
7:00 was better than I
7:03 expected if you get to rambling if you
7:06 get to rambling ah forever hook knows me
7:09 too well what are you going to
7:13 do and then also cuz everything's lower
7:17 now I've got this is I've got a bit of
7:21 an issue here um because everything's
7:23 too low let's see is that
7:29 working I'm totally discombobulated wait
7:32 till I try to show you something on the
7:33 uh on the old
7:35 computer oh
7:39 Lordy okay so uh does anyone have any
7:43 questions before we get
7:47 started TK where's the AI
7:50 automation to our primary
7:53 Representatives yeah
7:58 exactly silver fox I hate when my stuff
8:00 is moved it sends me I know I know I got
8:03 in here and I'm like okay oh the
8:05 computer's not there did I leave it in
8:07 the living room I was like [ __ ] do it
8:10 work I knew it the moment I did it but
8:13 what are you going to
8:14 do what are you GNA do how's everybody
8:17 doing what's what's everyone got going
8:19 on I see Mr K's here Danielle's here
8:21 Milo's wife is here and on time good
8:24 Lord will wonders never
8:26 cease I'm calling myself an irregular
8:29 moving forward poly Z I've seen you in
8:31 here multiple times I will give you that
8:34 that that moniker that
8:37 Crown I I hear by Night The
8:41 Irregular and you know get some em
8:43 modium it'll
8:48 help try the fiber try the
8:56 fiber by the way if you've been in here
8:58 for the past few nights
9:02 um we came up with
9:05 a uh a show based on a character I
9:10 created called Frank who's a hot dog
9:14 he's a wiener Frank the wiener and uh I
9:18 think it was forever hooked right came
9:20 up with the name the daily Jank with
9:24 Frank so there is now a daily Jank
9:29 ticktock or not Twitter Channel x.com
9:33 channel so if you go to Twitter if you
9:36 go to
9:37 x and search for the account daily
9:41 Jank you will see Frank's channel so
9:44 Frank has his own channel Frank today
9:46 talked about Pi
9:48 2.5 or inflection 2.5 to be used in pi
9:53 and we can play with that uh but that's
9:56 that's actually pretty exciting I was
9:58 going to do something on the on the Tik
10:00 Tock thing but it was like it was a 50
10:02 to nothing vote there was no those there
10:04 was no real
10:06 um hubub about it and they've got Tik
10:10 Tok has 165 days to divest from bite
10:14 dance or
10:16 face deletion so I like what what hit me
10:20 about it was there's there's
10:22 enough there's enough time and there's
10:25 enough
10:27 just parties involved that it's we've
10:30 got time so we're fine so I I didn't
10:34 feel like I didn't feel like Frank
10:35 needed to stir anything up with his 42
10:39 followers or 23 followers something like
10:42 that I'm studying every local AI Salon
10:44 meet up with let's be
10:49 frank that's pretty good I like it hot
10:52 dogs are the official finger food for
10:55 the AI Salon after
10:58 hours I might running on on 4 and 1 half
11:02 hours of sleep for the third time this
11:04 week I might drop I had so this
11:07 presentation I had to work on today
11:09 after cutting it it's at 60
11:13 slides but I will have you know so this
11:15 is a here's a little workflow thing that
11:17 I'll share with
11:18 you so so we're working with Bankers we
11:22 got to tell the story of our
11:24 company
11:26 and so we've done we've done
11:30 presentations to them and you know just
11:33 talked about stuff and I did a
11:35 presentation where I talked about Story
11:38 vine my company and what we do and who
11:40 we do it for and how we do it and then
11:43 all the new AI stuff that we've built
11:46 into
11:47 it so I had like this hour and a half
11:49 long conversation where I'm showing them
11:52 stuff they're asking questions we're
11:53 going back and forth so I took the
11:55 transcript of
11:57 that threw it into Claude
12:01 and then said hey Claude here's my goal
12:03 I want to talk about the company I I
12:05 want it to be you know targeted to these
12:07 kind
12:09 of uh you know these kind of this kind
12:14 of audience not to that kind of audience
12:15 you know that kind of stuff I said write
12:18 me an outline for this particular kind
12:21 of document they're looking for and it
12:23 did it and then I said write me a bunch
12:25 of slides for each of those sections it
12:27 gave me 10 sections in the outline
12:30 and then when I took the template that
12:32 the bank had given us like the first 20
12:35 slides of the template and the thing
12:36 that Claude produced were identical it
12:39 was amazing it it was just like you know
12:43 if you're going to talk about a company
12:44 you're going to talk about the same
12:45 basic [ __ ] and so Claude completely like
12:48 blocked out um you know the major slides
12:52 and then I I went through how did I do
12:54 it I put them in
12:57 notebook or notes in in iOS it's
13:00 basically just a notepad so I put I put
13:03 all of that text into into notepad and
13:05 then I just copied and pasted it into
13:06 slides that took me two or three
13:09 days and then uh and then we went
13:11 through it today so it was pretty cool
13:13 it probably saved
13:17 me at least a
13:20 week just doing it that way and it's
13:23 good it's good you know now I got to go
13:27 through and actually write it right now
13:28 it's all blocked out out we've got the
13:30 the major pieces there now I actually
13:32 got to go through and write it but but
13:35 amazing
13:37 um Paulie Z appreciating being dubbed
13:40 and irregular Congratulations by the way
13:42 if you haven't head on over to the AI
13:45 salon and jump into the AI regulars
13:49 club and uh say hi to everyone there and
13:52 if you're already there then
13:57 cool and if you don't know what the AI
13:59 Sal is I
14:02 can pull that up let me see how am I
14:05 doing this it's sort of the
14:09 same
14:12 re
14:13 so if you are looking for
14:17 Community well no let's put it this way
14:19 not if you're looking for Community I
14:20 always there there's something weird
14:22 about that that like when people talk
14:24 about communities you know oh it's about
14:26 it's all about Community there's
14:27 something weird about it if if in your
14:30 life you're one of those people that's
14:31 like I think AI is cool and everyone
14:33 else is like would you shut the [ __ ] up
14:35 about
14:35 AI if you want to be around people who
14:38 also think it's cool and are and are
14:41 exploring it and trying to figure it out
14:42 and trying to understand Pi has a new
14:45 release it does I'll talk about it in a
14:48 second um this group is a group that you
14:51 should
14:52 join so this is a group of AI
14:55 adventurers who are
14:58 choosing to lean into this AI stuff and
15:01 figure it out and
15:02 play and this is a good place in between
15:06 these meetings like if you come out hang
15:08 out here this is obviously a bunch of
15:10 people people that are willing to sit on
15:13 Tik Tok for a couple hours T
15:15 night and listen to me um are really
15:18 into Ai and they're really into figuring
15:20 this stuff out and they're really
15:26 into being
15:28 willing to not know all the
15:31 answers and that's not necessarily a
15:34 comfortable place as exciting as this
15:36 stuff
15:38 is oh yeah go to the first link there
15:41 yep
15:43 exactly and that that'll take you to uh
15:46 a website and the first thing you'll do
15:48 is you'll land on uh welcome to the
15:50 salon and you can read about what we're
15:52 up to you can read about read about our
15:54 values and if you dig those values then
15:57 jump in and the first thing you should
15:59 do under the Community section the first
16:02 link under Community is introduce
16:05 yourself so click on that link and
16:07 introduce
16:12 yourself I know I know I know what
16:16 you're thinking oh but I don't know
16:17 anyone they're going to judge me I don't
16:19 really know anything about AI they know
16:21 more than I do no everyone's [ __ ]
16:23 clueless right now even the
16:26 experts you know what's funny I talk to
16:29 people who've been in in like machine
16:31 learning development in some form or
16:34 other for 20 years 30
16:37 years and every time I meet one of
16:41 them I ask them about generative AI so
16:44 generative AI is sort of post chat
16:48 GPT every time I meet one of them I'm
16:51 like so okay you've been in machine
16:52 learning for 20
16:55 years what do you think of generative AI
16:58 like gener generative AI is a big deal
17:00 and I say well what's
17:02 different post generative AI than
17:05 pregenerative Ai and the response
17:07 universally has been everything their
17:11 response is everything's
17:15 different so even people who are experts
17:18 at this [ __ ] like that know the math and
17:22 build the foundational [ __ ] that makes
17:25 all of this stuff
17:28 work are a bit untethered right now so
17:32 this is a community of
17:35 people who are willing to be on the
17:38 adventure even knowing they don't have
17:40 all the answers just willing to play and
17:44 explore and support one another and ask
17:47 questions and answer questions so you
17:49 can go into this group and you can
17:51 introduce yourself and say hey I'm
17:54 Bob I've sold carpets for 38 years and
17:58 this AI stuff I don't know what I'm
18:01 doing and someone will go cool come on
18:05 in and there there's
18:08 guilds and there's kind of a little
18:10 showoff area where people can show their
18:11 work and talk about
18:13 it there's a water cooler area there's a
18:16 news area and then there's a little Club
18:19 area
18:21 where this channel has has our own club
18:23 so you can go there and hang out and
18:26 normally I'd be flipping around a bit
18:28 more but but everything everything's
18:29 everything's a little discombobulated
18:31 cuz I'm on an
18:34 iPad let's see let me go
18:38 to
18:40 AI
18:48 [Music]
18:49 C I will accept your cookies
18:53 yes I will accept your cookies I have to
18:57 touch the screen so you'll see a big
18:59 damn fat finger in there so so when you
19:01 first get there come to Welcome to the
19:03 salon check out the salon read about who
19:06 we are why we do what we
19:08 do read about our values things like
19:12 curiosity generosity
19:16 exploration
19:18 collaboration
19:21 inclusion empathy empathy is a really
19:23 interesting one
19:28 um one of the reasons we have empathy
19:35 here oh my God the
19:39 spammers the text spammers are just
19:41 Relentless
19:46 now I talked to a contractor around my
19:49 age about you and he was oblivious about
19:52 AI oh cool that's awesome that's
19:56 awesome
19:58 um
19:59 Jason I'm trying to get promoted to
20:02 moderate uh DM me DM me I'm happy to
20:06 chat um well you can't really see it's a
20:09 little blown out wait I can I think I
20:11 can fix that hang on
20:14 right oh look at that look how good I am
20:21 people all right empathy well you can
20:23 sort of read it you can't really read it
20:24 but
20:25 whatever um why empathy is
20:30 important one
20:32 is we live in a time right now in our
20:36 country oh put your iPad in Do Not
20:38 Disturb mode good
20:40 idea focus is focus do not disc disturb
20:44 do not disturb there we go perfect um
20:48 we're in a Time on our country where
20:49 we're super divisive and we're all at
20:51 each other's throats and everyone's on
20:53 the wrong team and they're all [ __ ]
20:54 and we're not we're right it's just
20:56 everything is everything's been
21:01 gamified
21:03 and on top of
21:05 that you know we came out of the
21:08 craziness of covid which shook things up
21:12 and now Along Comes this this AI stuff
21:15 which while it's exciting and while you
21:20 know me in particular but you know this
21:23 group in general is
21:28 super positive and optimistic about it
21:31 it's still scary as
21:32 [ __ ] and it's going to change a
21:36 lot and so that combination of things
21:39 means that a lot of people are really
21:41 scared a lot of people are angry a lot
21:43 of people are nervous a lot of people
21:45 feel you know
21:49 untethered and
21:51 so this group is a place where you can
21:54 feel all
21:55 that and still come in and ask questions
21:59 about Ai and ask questions about what
22:01 does it actually mean and is it is it
22:03 stealing and like all of the things that
22:06 you've got no one to
22:08 ask you can do that here and you can
22:12 have
22:13 doubts as long as you come in you know
22:15 the that the
22:17 first value of curiosity
22:22 right you can be afraid you can even be
22:25 pissed off like this is going to take my
22:27 job whatever
22:30 but get curious about what this stuff is
22:33 because how it's talked about in the
22:35 media and how it's demonized and
22:38 dismissed by people who are also
22:42 afraid has very little to do with what
22:45 it's like when you actually experience
22:47 it when you actually experience this
22:50 stuff it's quite
22:51 invigorating so empathy is really
22:54 important and then the final one is
22:56 gratitude this is this really is you
22:58 know generosity is the second value
23:00 there and that that kind of means like
23:02 if if you learn something rather than
23:05 being stingy with what you learned like
23:07 share it and if someone has a question
23:09 answer them don't just sit on the
23:12 sidelines because what ends up happening
23:14 here is you end up connecting with
23:16 people on a on a very different level
23:18 than than you do in other things Corey
23:21 and curiosity got me here perfect
23:23 perfect rustic farm farm
23:26 wife Cory's remarkable
23:30 Cory's an artist who plays in mud all
23:32 day she's a
23:34 Potter she's got a great channel so you
23:37 should follow her on Tik Tok Corey
23:38 Sandler
23:41 pottery and she's gotten curious about
23:44 this
23:45 stuff and she's now developing really
23:49 kickass
23:50 applications in the form of
23:52 gpts she's not a coder she's never
23:56 coded she she she didn't sign up in this
24:00 life to be a
24:02 coder and yet she's developing some
24:04 kickass
24:07 applications because she was willing to
24:09 just get curious and be open to whatever
24:13 she might discover you know like an
24:16 artist
24:19 would and she's
24:22 approaching creating these apps in a way
24:25 that is I don't know I find it really
24:27 refreshing because
24:30 her apps are not as if an engineer made
24:32 them they're as if an artist made them
24:35 because one did because one can
24:39 now and it's like holy [ __ ] so so this
24:43 this community is really quite
24:45 remarkable um and then the other thing
24:47 to do here is you know like I said first
24:50 thing you should do is go introduce
24:52 yourself last introduction was 32
24:55 minutes ago so I don't think that's
24:56 anyone from this call but if you haven't
24:58 joined and introduce yourself you should
25:00 do it also there's a thing called AI
25:02 Launchpad and tools where the first
25:04 thing here is where do you start and so
25:07 I talk about how you start you
25:10 play you figure out how to solve
25:12 problems using Ai and then you share
25:15 what you're learning those three things
25:16 if you do those you're going to become
25:19 AI literate way faster than you can
25:21 imagine like way faster cuz these tools
25:24 are are so remarkably powerful and then
25:27 there's a a [ __ ] ton of tools here here
25:29 there's chat GPT and then just bunch of
25:32 llms image tools video creation tools
25:35 music tools so all
25:38 right so go check out the
25:44 salon
25:47 the that's how we pronounce it if you
25:50 don't pronounce it that
25:52 way very
25:55 sad very sad all right uh
26:00 um what else what else is happening here
26:08 people Jason feverously writing
26:12 Kyle act like marketing Brands like Nike
26:16 GoPro Red
26:20 Bull don't quite know what that means
26:22 but that's
26:27 cool
26:29 I'm another I'm another person in here
26:31 based on Corey Corey's amazing that's
26:34 really
26:36 cool and welcome welcome welcome welcome
26:39 oh do me a favor we we're a little light
26:41 in in who's in here tonight just just
26:44 numbers wise um if you'd be so
26:48 kind share this live right now with like
26:52 I don't know five of your Tik Tok
26:55 friends 10 of your Tik Tok friends and
26:57 if you want to make a game of it it
27:00 share this live with people who you
27:02 think will hate talking about
27:06 AI cuz I love a good challenge we had we
27:10 had someone come in two nights ago Jimmy
27:14 Jimmy came in he was [ __ ] on fire he
27:17 was like I hate this stuff the words
27:20 you're saying are
27:22 heresy you're an evil
27:27 man it good talking to Jimmy I don't
27:30 know if Jimmy lasted most of the
27:36 night but Jimmy was saying a lot of the
27:38 stuff that that people who haven't
27:41 played with AI
27:44 say it's going to ruin lives it's going
27:47 to kill job the robots are going to kill
27:50 us oh it's because of the State of the
27:52 Union that's probably a good point yeah
27:54 we'll probably get a run up here when it
27:55 ends yeah in fact that's probably a good
27:57 idea maybe I should put
28:00 on I'll put on the T B so I can see when
28:04 it
28:05 ends but actually that's it's a really
28:08 good point so as you
28:12 know one of the things
28:15 I take great pride in here is is the uh
28:21 the grading system for the people that
28:23 show up here like Amelio's wife just her
28:27 her report card is just Chuck full of
28:29 demerits for being late so anyone who is
28:35 not here right now because they're
28:36 watching the State of the Union I I just
28:39 not it I just not it you know they
28:42 prioritize America over learning AI it's
28:46 okay no
28:47 judgment except are you really committed
28:51 to learning AI if you're not here no
28:54 you're not so that just goes on your
28:56 report card that's fine
28:59 sh Joe
29:01 sh Joe's very pey
29:05 tonight come on turn it down all right
29:15 okay all right good call side hustle
29:19 Mimi but anyway Jimmy hasn't been back I
29:23 hope Jimmy comes back
29:25 actually Jimmy seemed pretty committed
29:27 to the narrative that AI is
29:34 evil I mean you know what's funny about
29:36 about this AI stuff if you're on the
29:43 sidelines it's really easy it's really
29:46 easy to demonize this and to dismiss
29:50 it because there's lots of evidence
29:52 right the evidence is oh it hallucinates
29:54 and it's bad and people have seven
29:56 fingers and there's just all sorts of of
29:58 [ __ ] and then you know someone will try
30:00 to do something and they'll be a lawyer
30:03 and they'll have chat GPT write a legal
30:05 brief and then they'll hand it to a
30:06 judge and the judge goes what the [ __ ]
30:08 is
30:09 this and the judge and the lawyer gets
30:12 in trouble then people go see AI evil so
30:14 it's really easy to
30:17 demonize and then when you start using
30:19 it this weird thing happens
30:25 where it is sort of at the same time
30:29 time the most remarkable [ __ ] you've
30:31 ever experienced on a
30:34 computer and a janky piece of
30:39 [ __ ] where you're just like it's
30:42 remarkable and then you actually try to
30:43 do some work with it or try to you know
30:45 solve a problem with it and it's like
30:48 it's it's almost
30:50 there and you're like this would be so
30:52 good if it didn't
30:57 suck and that's just the state of where
30:59 we are right
31:01 now but even it's in its janky State you
31:06 you really do get to experience what
31:08 it's like to interact
31:10 with the computer you've been using for
31:13 however many you years you've been using
31:15 computers in a completely different
31:20 way what
31:22 was forget how jiren laner jiren laner
31:26 is a a a musician really interesting
31:30 dude he's pretty prominent in the 90s he
31:32 was into virtual reality and things like
31:34 that
31:38 and he
31:43 characterized he characterized using AI
31:46 as collaborating with
31:49 Humanity cuz that's kind of closest to
31:51 what it is if if you understand how
31:53 large language models are put together
31:55 how they
31:57 work
31:59 a company like open AI or inflection
32:02 that did Pi will go out they'll find as
32:05 much data as they can like public
32:08 internet stuff you know Wikipedia and
32:11 blogs and websites and just as much of
32:14 it as they can get and then they encode
32:16 it in this special way through this
32:18 thing called a Transformer which is the
32:20 T in
32:25 GPT and when you interact with
32:29 it what you're actually interacting with
32:32 is not the calculator that does the
32:35 probability
32:38 equation that is the thing
32:40 that makes the words show up on your
32:43 screen what you're actually interacting
32:46 with are the people that put all that
32:48 crap on the internet so in a very real
32:51 sense when when when chat GPT or or some
32:55 large language model like that interacts
32:57 with you it really
33:00 is like the collective knowledge of
33:03 humanity that got dumped on the internet
33:05 over the past four or five decades is
33:07 interacting with
33:09 you and and that's really profound and
33:13 it takes a while to it takes a while
33:17 to to
33:20 separate you know the difference between
33:23 what Google does which is you type words
33:25 in a box and it spits out [ __ ]
33:29 and what an llm does which is you type
33:31 words in a box box and it spits out [ __ ]
33:34 just different
33:39 [ __ ] but the Google thing is just giving
33:42 you an index like here's a list of
33:43 places you can go figure [ __ ] out
33:47 on or you can go buy something
33:50 on
33:51 and the large language model is like
33:54 here's your answer you're like I don't
33:56 like that answer well here's another
33:57 answer is that better oh that's a little
33:59 better could you make it rhyme sure you
34:04 go and it's like you're working with a
34:07 uh all right Biden's Biden's leaving so
34:10 there's 65 people in here right now
34:12 we'll see who whoever joins in the next
34:14 10 minutes we will judge them we will
34:17 judge them harshly they will get
34:21 demerits oh although maybe they should
34:24 be given a pass being that I forgot my
34:27 computer tonight
34:28 [Laughter]
34:33 such a dumb move
34:39 anyway did I yeah I did
34:42 okay just had a moment like did I even
34:44 take my laptop out of the conference
34:46 room pretty much pretty pretty sure I
34:48 did not missing much from it
34:56 great that's a great way to frame
34:58 thank you oh you mean the the
35:00 collaborating with Humanity
35:03 thing it's it's it's a weird thing you
35:06 know a lot of times when I'm in here and
35:09 I'm I'm working I'll thank chat GPT and
35:12 people will sometimes comment on that
35:14 like why are you thanking a large
35:15 language model well for one thing being
35:17 nice to it they actually seem to be
35:19 better like um according to research
35:23 from scientists who study this
35:25 [ __ ] so be nice to your l
35:29 there's just something about how they've
35:31 been programmed and how they interact
35:34 with you that kind of has a humanness to
35:37 it and and here's what if you're if
35:40 you're new to chat GPT if you're new to
35:42 large language models here's here's a a
35:45 way to interact with them that that
35:47 might not be obvious on the
35:49 surface with Google what Google has
35:52 trained us over the past 20 years is be
35:54 as concise as possible in what you put
35:57 in that box if you put in too much [ __ ]
36:00 it kind of confuses it and so they've
36:03 essentially trained
36:04 us you know what's the minimum amount
36:08 and then we'll give you the
36:10 answers so go find you know this that or
36:13 the other submit and there's your
36:15 answers and then you just interact with
36:17 the answers and if it's not right then
36:19 you try another thing
36:22 right interacting with a large language
36:25 model is much more like a conversation
36:27 so so one of the mistakes that people
36:29 make when they first start using large
36:31 language models is they're expecting to
36:33 write the perfect prompt that's going to
36:35 give them the complete
36:38 answer and especially if you write in
36:41 the way that Google has taught us to
36:43 write please write me um a business
36:46 letter uh asking for someone to pay
36:49 their bill on time and it's going to
36:51 write a business letter and it's going
36:53 to be this horribly generic shitty
36:57 letter and so a lot of times what what
37:00 happens with people with with chat gbt
37:02 or things like it is they treat it like
37:04 Google they give it like this shitty
37:06 little prompt and it gives them a shitty
37:07 answer and they're like it's not all
37:09 it's cracked up to be I'll stick with
37:13 Google right and they I don't know why
37:15 they talk like that I'm not making fun
37:17 of them that's just how they
37:21 talk and but if if rather than that you
37:27 say here's the kind of role I have
37:29 here's the kind of business I'm in
37:32 here's a situation where had an invoice
37:35 it was do on this date it didn't come
37:37 this is a client who's paid
37:40 before write me a business letter asking
37:43 for our money to be paid and it'll write
37:45 it and it'll be pretty good and then you
37:47 go oh that's that's not quite right you
37:49 kind of got the tone wrong it's it's a
37:52 little wordy and formal make it a little
37:53 less formal try to tighten it up a bit
37:56 boom then it'll be really close so if
37:58 you give it more
38:00 context and if you interact with it kind
38:02 of like you've
38:04 got the world's best educated clueless
38:09 intern that's kind of what it's like
38:11 interacting with it and it's got a
38:12 humanness about it and especially you
38:15 know what I've noticed is because of Co
38:20 like I was
38:21 a like the the idea of remote work to me
38:25 was just this abhorent idea I'm a people
38:29 person I like interacting with people I
38:30 like being in the presence of people I
38:32 get energy from
38:33 that I work in a company full of
38:36 introverts so when Co happened they're
38:38 like woo and I was like
38:41 [ __ ] but what we learned to do in Co all
38:45 of us whether we liked it or not was we
38:47 learned how to work
38:49 remotely and what I find when I when I'm
38:52 truly kind of collaborating with chat
38:54 GPT it feels like I'm working with a
38:57 remote
38:58 coworker they just happen to be really
39:01 fast and wicked [ __ ] smart and
39:04 clueless you got to you got to steer
39:06 them
39:08 right Peter K has a great way to write
39:11 art prompts by stringing together
39:13 unrelated words numbers Etc oh that's
39:16 cool daisy May yeah Peter Kaminsky so by
39:18 the way if you go in the AI
39:21 Salon um Peter Kaminsky in the tech
39:25 Guild has a group um called it's
39:31 like AI something
39:33 101 but it's it's basically like if you
39:36 want to sort of geek out and but learn
39:39 how to do that with someone that's
39:40 infinitely patient and wicked
39:44 smart Source Camp howdy from Destin
39:47 Florida you made it
39:52 yay s Source Camp Source Camp is a
39:56 serious irregular
39:58 okay so so for those of you that were
40:01 watching the State of the Union like
40:03 I'll watch Kyle later I've got to watch
40:05 the president we know where your
40:06 priorities are that's fine um just to
40:09 make you feel even
40:11 worse Source Camp was on a camping
40:15 trip for like last week for like five
40:18 days six
40:20 days and logged in to the AI learning
40:23 lab in the nights in her
40:26 tent and bugged her little camping
40:29 neighbors cuz I was screaming about
40:32 something that is dedication then she
40:35 she she was driving to Florida yesterday
40:38 and and logged in so those of you who
40:41 are like I'm too tired works
40:44 hard you sorry you get no sympathy you
40:47 got people like Source camp in here
40:50 she's calling in from Florida she could
40:52 be out at
40:54 the line dancing club
40:58 in Sarasota I don't know where she is oh
41:01 wait Des Florida I don't know where Des
41:03 is I'm sure they have a line dancing
41:05 club though she could be out dancing
41:07 away dancing no she's
41:09 here she's
41:12 here why she's irregular it's just she's
41:16 not weird she's irregular that's okay we
41:21 like that anyway all
41:26 right uh we should go this Saturday to
41:28 Peter's session yeah uh Peter's having a
41:32 session on Saturday um weekend GPT jams
41:35 is not metal for Source Camp absolutely
41:38 got a check I might not be able to go I
41:40 have a wedding to
41:47 officiate you should yeah have the book
41:49 up officiating the wedding and then just
41:51 have like a a phone on the one side of
41:54 the one side of the Bible or whatever
41:56 book you're using and you could you
41:59 could you know have a have a little
42:01 earbud in you'll be good you can do both
42:03 you can
42:05 multitask that's really
42:07 [Laughter]
42:09 horrible uh I think Peter's Peter's
42:13 thing is 9:00 a.m. Pacific on
42:17 Saturday yeah he changed the name to
42:19 learn and do that's correct learn and do
42:20 101 or learn and do with Peter
42:23 yeah that's
42:26 right
42:28 [Music]
42:31 a human being I admit I'm irregularly
42:34 [Laughter]
42:41 irregular oh
42:55 man 7:30 PST
42:58 oh it's 7:30 PST tomorrow night or
43:01 tomorrow
43:03 morning I thought he was on the west
43:05 coast is he doing a meeting at 7:30 a.m.
43:08 that would be
43:20 weird hustle Mimi no I actually do the
43:23 ceremony like a serious person what's
43:26 going on this morning Kyle
43:28 um no well so what's going on is I
43:31 forgot my my laptop so that's my iPad
43:35 that's my iPad so I can't demo as much
43:38 but but we'll do something um but what I
43:41 do want to
43:43 do so
43:45 inflection so if you remember inflection
43:48 is the company that that is behind Pi Pi
43:51 is a
43:53 conversational GPT like
43:56 thing if you you use pie on the desktop
43:59 and try to treat it like chat GPT it's
44:02 actually pretty horrible
44:04 because it's been optimized for voice
44:09 conversation and it's really good at
44:11 that it's really nice it's
44:13 really nuanced it's got I don't quite
44:16 know how they're doing their memory but
44:18 it it retains some memory over
44:20 time like it knows my name is Kyle it I
44:23 don't have to tell it that every time it
44:25 might be getting that from my account
44:27 I'm not sure
44:31 and it's very much designed to be
44:33 conversational it's quite good at it so
44:36 they came out with a new version today
44:38 called inflection
44:41 2.5 that
44:44 apparently this is number four now or
44:46 yeah well the third
44:48 one this is is this the third
44:51 one Google yeah
44:54 Google anthropic and now
44:58 inflection they claim it's better than
45:00 gp4 like it's beating gp4 on benchmarks
45:03 so I figured what we could do is talk to
45:08 it
45:10 uh Daisy May I'll definitely check it
45:12 out when I'm off work is learn and do on
45:15 the AI Salon yeah so if you go to the AI
45:18 salon so go to the salon. click on the
45:21 first link that'll take you to the salon
45:24 Community there's a community section
45:26 where you can just chat there's a
45:28 showand tell section where you can show
45:30 off work that you do and then there's
45:32 guilds there's a creative Guild a
45:35 business Guild a technology Guild and a
45:38 social impact Guild and then there's
45:41 multiple spaces within each of those
45:43 Guild
45:45 Collections and in the technology Guild
45:48 is learn and do by Peter Kaminsky so
45:50 that's the one you're going to want to
45:51 check out and he's just he's just a
45:54 great dude he's super low-key Wicked
45:56 smart
45:58 and and kind of infinitely patient we'll
46:00 walk you through how to do stuff you can
46:02 ask really technical questions you can
46:04 ask really basic questions he's just
46:08 he's just real steady
46:11 guy I'm in a few guilds will definitely
46:14 check it out that's awesome please
46:15 remind me I got to go figure out my
46:17 linky my linky
46:19 [Laughter]
46:23 login Kyle surely GPT 5 is about to drop
46:26 what do you think
46:27 there was so this
46:30 morning
46:34 so two days
46:37 ago Robert
46:39 scobble who's he's very connected in in
46:44 San Francisco in that
46:46 area he dropped a tweet that said open
46:49 Ai and then it had the the the looking
46:52 Emoji the two eyes looking that
46:56 way and then nothing happened and then
46:58 this morning there was a rumor that GPT
47:01 5 was supposed to drop at 10:30
47:05 a.m. and it
47:07 didn't and then last week you had two
47:11 people two fairly major people from open
47:13 AI
47:16 left so it seemed like some [ __ ] was
47:20 going to drop and then and then you've
47:22 got Claude 3 dropped from
47:25 anthropic and now
47:28 inflection
47:29 2.5 2.5 dropped from from inflection so
47:34 we have a new version of
47:37 Pi
47:39 um so there's definitely buzz and then
47:42 when it didn't dropped today someone
47:44 said next Tuesday at the same time so
47:47 between 10 and 10:30 a.m. next Tuesday
47:50 so it's possible open AI has dropped
47:53 some stuff on
47:55 Fridays so it's Poss possible they Dro
47:58 something I don't
47:59 know I don't know my like my my my
48:02 Spidey Sense is that something something
48:04 is about to drop from from open
48:08 AI but I
48:10 just like it just feels like they're not
48:13 going to
48:14 drop GPT
48:17 5 I think what they they they could
48:20 drop GPT 4.5 so they could drop a
48:24 significant upgrade to the current thing
48:28 I just think GPT 5 is going to be a
48:30 completely different
48:32 thing and I I and I just I don't even
48:35 sense that they're going to drop
48:37 Sora I I feel like that's months away
48:39 two months away three months away but I
48:42 don't know I'm just I'm like I'm totally
48:44 [ __ ]
48:45 guessing totally
48:51 guessing Mr it sorry I'm late ah Mr it
48:55 was watching the State of the Union
48:56 that's okay we know where your
48:57 priorities lie it's okay it's all right
49:00 Frank's going to be a little upset but
49:02 what are you going to
49:04 do were the eyes looking at Elon yeah
49:07 they probably were actually stumbled
49:09 upon zapier chatbot creation today
49:11 seemed useful like a public GPT yeah we
49:15 played with that a little bit last night
49:16 tired Runner I haven't actually dug in
49:18 and played with it I I went there and
49:21 it's it seems to me like what zappier
49:24 did was they said hey chat GPT nice work
49:27 work on the GPT Builder why don't we
49:29 just copy that but but it's still
49:32 zappier you still got to have API Keys
49:34 you still got to understand what those
49:36 are and how to get them and how to
49:38 connect things so I haven't played with
49:40 it enough to know if it's any good but
49:42 it looks like it's a good Improvement Mr
49:44 K all
49:50 day laughing Thunder not replace a
49:54 counselor but augment not replace but
49:57 Aid I have students that need counselors
50:00 every day and don't get it
50:03 absolutely absolutely like like why
50:05 these things shouldn't do that and don't
50:07 call me
50:10 shirle there's airplane jokes in here us
50:13 gen xers we got our
50:15 [Laughter]
50:19 jokes oh man there's one counselor for
50:23 elementary school and high school that
50:25 way more support is needed
50:28 yeah the thing I'm I'm looking forward
50:29 to Mr K is you know some sort of it
50:34 would be nice to have a
50:35 counselor
50:37 that was a virtual twin of the real
50:43 counselor
50:45 that you know listened gave advice but
50:49 also every interaction that it had it
50:52 would synopsize it you know take the
50:55 full transcript but also a synopsis of
50:57 it send that to the counselor and then
51:00 it would be smart enough to know if
51:01 there's an actual crisis here it would
51:03 escalate that to the counselor right
51:06 that kind of thing seems to me to be em
51:09 eminently doable I don't want to do it
51:11 because the health care the the the
51:15 health care business is so awful but um
51:20 I someone's going to do that and
51:21 someone's going to get it right it's
51:23 going to be amazing it's going to be
51:25 absolutely amazing
51:32 let it be games inside gpts that would
51:34 be nice games or some sort of automation
51:36 inside gpts that make it look really
51:39 that make it much easier that would be
51:41 super
51:42 nice I'm always with you guys I know Mr
51:46 it you really are here a lot zappier be
51:50 crappier I bow to your Superior
51:52 knowledge I lack both experience and
51:54 skill with psychologists
51:59 oh man that's
52:03 funny I'd settle for a store that's a
52:05 store yeah no [ __ ] that would be nice um
52:08 okay let's let's try something here all
52:11 right so there you get to see you get to
52:14 see what was behind the laptop a
52:16 motorcycle helmet box that used to be
52:18 what the laptop was on until until uh
52:21 the Irregulars sent me this lovely
52:23 laptop
52:25 stand but what we're going to do is
52:27 we're going to we're going to launch
52:30 pie and I don't quite know how to talk
52:33 to
52:36 her so if you've never seen this where
52:38 where what I am on is I'm on an
52:42 iPad
52:44 that's my
52:47 um I'll just turn that off so there we
52:50 go it's just an
52:52 iPad that PI app is the one we're doing
52:56 so the is inflection the app is called
53:00 pi and
53:05 then it's a normal you've got your input
53:08 box at the bottom and you got chat up
53:10 here and then this little button that
53:12 looks like a phone call that puts it
53:14 into voice mode and so I don't
53:19 know I don't know how to how to
53:22 determine if this thing's better or not
53:24 cuz it was pretty good before so if
53:27 anyone has any she's very verbose if
53:29 anyone has any it's SC spelled IP no
53:33 it's clever
53:36 Joker oh cuz my thing's backwards I get
53:39 it that's clever see Joker always joking
53:43 always joking that joker he is an
53:46 appropriately named
53:49 gentleman gentleman I think is a little
53:51 little generous but you
53:55 know all
53:58 right p is now great with technical
54:00 questions
54:02 [Music]
54:04 okay hello how are you this
54:09 evening hello Kyle I'm doing great thank
54:12 you for asking it's always nice to have
54:14 a pleasant conversation with someone
54:16 how's your evening going are you
54:17 relaxing after a long day or perhaps
54:19 getting ready for an
54:21 adventure I'm just excited I hear that
54:24 you got an upgrade to inflection 2.5
54:26 tell tell me about
54:29 [Music]
54:31 it you heard right Kyle inflection 2.5
54:35 is a significant upgrade that has
54:37 enhanced my conversational
54:39 capabilities with this update I'm able
54:41 to generate more natural engaging and
54:44 contextually relevant responses I can
54:48 understand more complex ideas and
54:50 provide more helpful insights plus my
54:52 knowledge base has expanded so I can
54:54 chat about an even wider range of topics
54:58 it's all pretty exciting don't you
55:00 think I do what might be some things to
55:03 ask you that can demonstrate your new
55:09 knowledge that's a fantastic question
55:12 Kyle I think it's F here are a few
55:13 topics where you can really see
55:14 inflection 2.5 in action explain a
55:18 complex concept like quantum mechanics
55:20 or string
55:22 theory recommend a book movie or TV show
55:26 based on my personal
55:29 taste have a philosophical discussion
55:32 about ethics morality or
55:34 existentialism discuss the latest
55:36 advancements in fields like artificial
55:39 intelligence Neuroscience or
55:42 biotechnology give a step-by-step guide
55:44 for a complex task like building a
55:47 computer from scratch so what do you
55:50 want to start with the possibilities are
55:55 endless
56:00 if Alexander Hamilton were to
56:04 incorporate the theory of quantum
56:07 mechanics into a governance system how
56:10 would it be different than what he
56:12 created for
56:18 America wow now that's a fascinating
56:20 hypothetical question Kyle if Alexander
56:23 Hamilton were alive today and had the
56:25 knowledge of quantum mechanic
56:27 he might have created a governance
56:29 system that incorporated principles of
56:32 quantum superposition and in tanglement
56:36 for instance instead of a binary
56:38 two-party political system Hamilton
56:41 might have designed a multi-dimensional
56:44 political landscape where multiple
56:46 parties coexist in a super position of
56:49 power this could lead to more diverse
56:53 representation and decision making holy
56:55 [ __ ] imagine itical debates where
56:58 quantum entanglement allows for instant
57:01 communication and consensus building
57:03 between parties regardless of distance
57:06 it would certainly be a revolutionary
57:08 approach to
57:11 governance okay that's pretty smart so
57:13 why don't you
57:16 design a system that could take
57:21 advantage oh I see you're impressed
57:25 Kyle design a system
57:28 that wait tap to keep talking this is so
57:33 if you're like me sometimes I pause I
57:35 have to think as I'm talking so you can
57:38 hold that to do that okay Pi let's let's
57:41 start the little conversation over here
57:44 how would you design a system that
57:46 leverages generative Ai and instant
57:51 Communications and overlay that on top
57:53 of our current political system
57:58 let's see what it does see what she does
58:00 all right Kyle let's start fresh so
58:02 you're asking how we can redesign our
58:03 current governance system to incorporate
58:05 generative Ai and instant communication
58:08 here's a quick sketch one AI advisers
58:11 elected officials would have access to
58:13 highly Advanced AI advisers that can
58:16 analyze vast amounts of data provide
58:18 realtime analysis potential policy
58:20 outcomes and even draft legislation it's
58:23 really good this would lead to more
58:25 informed and data driv driven decision
58:28 making two realtime citizen
58:32 feedback with instant communication
58:35 citizens would be able to provide
58:36 real-time feedback on proposed policies
58:39 and laws making the Democratic process
58:42 more participatory and
58:45 responsive three smart contracts
58:48 generative AI could be used to create
58:50 self-executing smart contracts that
58:52 automatically enforce laws and
58:54 regulations reducing bureaucracy and
58:57 increasing
58:59 transparency four Dynamic legislation AI
59:03 could continuously monitor the
59:05 effectiveness of laws and regulations
59:07 wow that's good testing updates or
59:09 changes in real time based on their
59:12 performance ensuring that legislation
59:15 remains relevant and effective how does
59:18 that sound do you think this system
59:20 would be an improvement over our current
59:23 one I think a pile of steaming dog po
59:26 would be an improvement over our current
59:28 system but what you said is really good
59:31 how would you battle corruption and
59:34 financial interests in politics with
59:36 your
59:41 approach oh boy you don't hold back do
59:44 you Kyle you're right corruption and
59:47 financial interests are significant
59:49 problems in many political
59:51 systems here's how my proposed system
59:53 could tackle these
59:55 issues one
59:57 transparency yeah with the use of smart
59:59 contracts and realtime data analysis
1:00:02 Financial transactions and decision-
1:00:04 making processes would be more
1:00:06 transparent making it harder for
1:00:08 corruption to go
1:00:10 unnoticed two aiiowed ethics
1:00:14 monitoring the AI advisers could monitor
1:00:17 the behavior of elected officials
1:00:19 flagging potential conflicts of interest
1:00:22 or unethical actions this could help
1:00:24 prevent corruption before it starts
1:00:28 three campaign Finance reform by using
1:00:31 AI to analyze campaign Finance data we
1:00:34 could better detect and prevent illegal
1:00:36 or unethical campaign
1:00:38 contributions additionally instant
1:00:40 communication could make it easier for
1:00:42 citizens to hold officials accountable
1:00:44 for their actions four direct Dem so one
1:00:48 of the questions Jason had was you know
1:00:50 is there a transcript yeah all of this
1:00:53 conversation is stored in the history so
1:00:56 so like yes this is all written down
1:00:59 it's that it's it's pretty amazing
1:01:02 [Music]
1:01:05 um okay let's change gears why are there
1:01:08 the number of tarot cards that there
1:01:12 are this should be
1:01:15 interesting o an interesting shift in
1:01:17 Gears Kyle the standard Tarot deck
1:01:19 contains 78 cards which is a combination
1:01:22 of 22 major Arcana cards and 56 minor
1:01:24 Arcana cards the 22 major Arcana cards
1:01:27 represent significant life events or
1:01:29 archetypal figures while the 56 minor
1:01:32 Arcana cards are divided into four suits
1:01:35 wands cups swords and Pentacles and are
1:01:38 often seen as representing everyday
1:01:40 events or
1:01:42 situations the reason there are exactly
1:01:44 78 cards is a bit mysterious but it's
1:01:47 likely rooted in numerology and
1:01:50 symbolism some believe that the number
1:01:53 seven represents the spiritual while the
1:01:55 number eight represents the material
1:01:57 world together they form 78 which is
1:02:01 seen as a powerful number that unites
1:02:02 the spiritual and the
1:02:04 physical it's pretty cool how tarot
1:02:06 Blends art symbolism and spirituality
1:02:08 isn't
1:02:10 it it is if you were to add a dozen more
1:02:14 cards to the Tarot deck what would be
1:02:16 your justification how would you
1:02:18 organize them and what would they
1:02:24 be if I were to add 12 more cars to the
1:02:27 Tarot deck I would create a new suit
1:02:29 called orbs to represent the spiritual
1:02:31 and Cosmic forces that shape Our Lives
1:02:34 each card would depict a different
1:02:36 astrological sign starting with Aries
1:02:39 and ending with Pisces wow the
1:02:41 justification for adding these cards
1:02:43 would be to provide a deeper connection
1:02:46 between the tarot readings and the
1:02:48 celestial bodies that influence our
1:02:50 world organization wise the orb suit
1:02:54 would follow the same structure as the
1:02:56 other minor Arcana suits with cards
1:02:58 numbered from 1 to 10 and four Court
1:03:00 cards page Knight queen and king this
1:03:05 would maintain the consistency of the
1:03:07 deck while adding a unique Cosmic
1:03:09 Dimension to the
1:03:12 readings what would be the most powerful
1:03:16 cards and
1:03:20 why in this expanded Tarot deck the most
1:03:23 powerful cards would likely Remain the
1:03:24 major Arcana cards as they represent
1:03:27 significant life events and archetypal
1:03:30 figures no I mean in your 12 new cards
1:03:34 which would be the face cards and the
1:03:40 ace oh my apologies for the mixup in the
1:03:43 new orbs suit the four face cards and
1:03:45 Ace would represent the following one
1:03:48 page of orbs a young enthusiastic and
1:03:51 curious individual eager to explore the
1:03:54 spiritual realm two Knight of orbs a
1:03:59 daring but which of the uh astrological
1:04:04 signs to each of them represent and
1:04:12 why in the orbs suit the face cards and
1:04:15 Ace would represent the following
1:04:16 astrological signs one page of orbs
1:04:20 Gemini known for their curiosity
1:04:22 adaptability and love of learning
1:04:24 Geminis align well with the p AG's
1:04:26 youthful enthusiasm for
1:04:28 exploration two Knight of orbs
1:04:31 Sagittarius with their adventurous
1:04:33 nature passion for knowledge and
1:04:35 spiritual questing
1:04:37 sag okay that's great I love your
1:04:41 thinking what I'd like you to do now is
1:04:44 Shuffle the entire Deck with the 12 new
1:04:46 cards in it and I'd like you to pull
1:04:49 four cards for me tell me what they are
1:04:52 oh and what they mean all right Kyle
1:04:55 let's give it a go shuffles the deck and
1:04:57 the card I've pulled for you is the
1:04:59 queen of orbs oo this card represents
1:05:02 the nurturing empathetic and supportive
1:05:05 qualities of cancer perhaps it's
1:05:07 suggesting that you seek guidance or
1:05:10 offer support to those around you on
1:05:12 your spiritual journey the queen of orbs
1:05:15 can also symbolize a need to trust your
1:05:17 intuition and emotions as they can be
1:05:19 powerful guides in matters of the
1:05:25 spirit
1:05:27 I want you to shuffle all 90 cards and
1:05:30 recommend a new layout and pull the
1:05:33 cards for that layout describe the
1:05:35 layout to me and then pull the
1:05:42 cards damn okay let's try something
1:05:44 unique here's a new nine card layout
1:05:46 called the celestial Journey one the
1:05:49 present represents your current
1:05:51 situation two the past reveals how your
1:05:54 past influences your present present
1:05:56 three the future offers insight into
1:05:59 what's to come four the spiritual self
1:06:03 shows your connection to the
1:06:06 cosmos great now pull the cards and
1:06:09 let's see what's going on with
1:06:13 [Music]
1:06:14 me let's continue with the card pulling
1:06:17 Kyle voice five in pi by the way eight
1:06:20 the guidance Temperance nine the outcome
1:06:23 the
1:06:25 tower based on these cards it seems like
1:06:27 you might be facing some challenges Five
1:06:30 of Swords that require you to rely on
1:06:32 your inner strength queen of orbs and
1:06:35 find balance Temperance the tower in the
1:06:38 outcome position suggests a potential
1:06:40 period of upheaval or change but
1:06:43 remember that this
1:06:45 can it's it's just [ __ ]
1:06:49 nuts it's faster isn't it it's faster
1:06:55 right I can't believe it just came up
1:06:57 with it just came up with a whole new
1:07:00 extension to the Tarot deck and it was
1:07:02 wellth thought out it suck pretty good
1:07:07 [ __ ] and the quite frankly the quantum
1:07:12 mechanics the Alexander Hamilton quantum
1:07:15 mechanics overlay for how to reinvent
1:07:17 government those were some good [ __ ]
1:07:19 ideas holy
1:07:23 [ __ ] wow
1:07:27 wow 10 gpts came up when I searched for
1:07:30 tarot I'm sure there are many
1:07:33 more what did I come into just got here
1:07:35 hey cam Katen so what you came into is
1:07:39 um
1:07:41 inflection the company behind Pi
1:07:45 launched inflection
1:07:48 2.5 um which
1:07:52 is
1:07:54 apparently as good or better than chat
1:07:56 gp4 in a lot of the
1:07:58 benchmarks so so we now
1:08:02 have
1:08:04 um we now have Google Gemini Ultra which
1:08:08 is apparently better than GPT 4 we now
1:08:10 have but hasn't launched yet um Pro 1.5
1:08:15 is supposedly really
1:08:17 kickass she also has a slower Cadence
1:08:20 likely to help with latency in issues
1:08:22 but it's weird Pate she seems to answer
1:08:24 more quickly like the latency between
1:08:27 the question and her first
1:08:29 response um seems quicker so maybe they
1:08:33 slowed down her Cadence maybe they're
1:08:35 like streaming streaming the answer in
1:08:38 that that would make sense that that
1:08:39 would be an interesting way to speed
1:08:41 speed it up actually slow her down so so
1:08:44 you perceive it as
1:08:46 faster huh that's
1:08:49 interesting so you so you have
1:08:51 anthropics um Claude 3 which what I'm
1:08:55 seeing on anyone I'm paying attention to
1:08:58 is everyone that's using it pretty
1:09:00 regularly is like holy [ __ ] this is this
1:09:02 is the real deal and now you've got
1:09:04 inflection that apparently he got a gp4
1:09:07 thing which is all the more reason for
1:09:10 for open AI to drop something that's
1:09:13 either 4.5 cuz it's been a year and
1:09:17 three
1:09:18 months since they dropped chat GPT it's
1:09:21 been a year almost exactly since they
1:09:24 dropped GPT 4
1:09:27 no one's caught up to them except right
1:09:30 now like they're just there's models
1:09:33 just now starting to catch up to GPT 4
1:09:36 so open AI doing a 4.5 or I don't think
1:09:40 they're going to drop five this soon but
1:09:42 they could I I mean I don't
1:09:43 know Cloud 3 has an impressive context
1:09:46 size yeah Cloud 3's context window right
1:09:49 now is
1:09:51 200,000 they they it's likely going to
1:09:54 be a million in the API and and I think
1:09:58 expandable up to 10 million so a million
1:10:01 token context if you don't know what a
1:10:03 context window is it's basically okay
1:10:06 here's a good way to think about it so
1:10:08 if my fingers are text on your iPhone
1:10:10 screen right and these are all lines of
1:10:14 text think of the iPhone screen as your
1:10:17 context window so you start a
1:10:19 conversation down here with chat jbt hey
1:10:22 chat jbt give me an answer there's your
1:10:23 answer okay give me another answer okay
1:10:25 okay there's your answer hey I have a
1:10:28 yellow Volkswagen bug that's fantastic
1:10:30 and it you know and this conversation
1:10:32 scrolling up and up like here's the I
1:10:36 have a yellow Volkswagen Bug right and
1:10:38 down here you're talking about the
1:10:40 yellow Volkswagen bug or whatever or
1:10:43 you're talking about other [ __ ] and then
1:10:45 you keep talking and this goes off
1:10:47 screen and then you ask it about what
1:10:49 kind of car you have and it won't know
1:10:51 so that's your context
1:10:53 window that's why chaty is good good
1:10:56 good good good and all of a sudden it's
1:10:58 like I don't [ __ ] know you're like
1:11:00 hey wait a minute I thought we had an
1:11:02 understanding here he's like that's I'm
1:11:05 would be honored to help you in some
1:11:06 other way but I don't know what the [ __ ]
1:11:08 you're talking about that's what that is
1:11:11 the current chat GPT I think it only has
1:11:13 like an eight or
1:11:15 16,000 um token context window which is
1:11:18 like 12,000 Words which is a decent
1:11:21 amount but you run through that pretty
1:11:24 quick
1:11:26 the current Claud has 200,000 tokens
1:11:29 which is 150,000 words so that's a lot
1:11:32 and then if it if it moves to a million
1:11:34 and the Gemini Pro 1.5 has a million
1:11:40 which is 750,000 words that's a whole
1:11:44 lot and so you can have incredibly long
1:11:48 detailed conversations or you can upload
1:11:50 massive amounts of documents and it'll
1:11:52 just know them
1:11:54 all I usually say it's like a clip from
1:11:56 a movie it
1:11:58 helps show that it doesn't have the full
1:12:00 picture oh yeah yeah that's that's
1:12:03 that's a good idea P yeah that's right
1:12:04 that's good but but a million tokens you
1:12:07 get a lot more frames of that movie
1:12:11 right and and the other thing about
1:12:14 Gemini Pro 1.5 is it's multimodal so
1:12:17 that million tokens can be taken up with
1:12:21 words or it could be taken up with
1:12:23 frames of video so you can upload an
1:12:26 hour of
1:12:27 video into Gemini Pro
1:12:30 1.5 and ask it
1:12:32 about what was happening when the rubber
1:12:35 ball started bouncing down the
1:12:38 street and it will know and
1:12:41 it it's crazy so so these tools are
1:12:45 getting pretty good pretty
1:12:48 quick with a million tokens you get the
1:12:51 behind the scenes bloopers Comics Etc
1:12:53 yeah exactly exactly
1:12:57 is pi the same way with memory
1:13:00 [Music]
1:13:01 um Pi's doing some interesting [ __ ] I
1:13:04 don't I should probably dig into to
1:13:08 understand how Pi is doing what they do
1:13:09 better um Pi's got some sort of longish
1:13:15 term
1:13:16 memory I don't know if it's per chat
1:13:19 session or if they just maintain chat
1:13:20 sessions and can go back and reference
1:13:22 it I don't know how they're doing it but
1:13:24 like
1:13:26 she keeps ref referencing my name all
1:13:28 the time and she'll remember some things
1:13:30 we talked about and then not others so
1:13:32 one of the things that I'm going to play
1:13:34 with in the coming weeks is does this
1:13:37 new 2.5 have longer memory I I'm sure
1:13:41 this is all stuff we can go read on the
1:13:42 inflection website and figure this stuff
1:13:46 out but but but Pi's got some
1:13:49 interesting memory stuff going
1:13:53 on I have blown the Contex window and
1:13:56 Gemini 1.5 but I am pretty pretty
1:13:59 verbose I think you'd have to be really
1:14:01 verbose to do
1:14:04 that I think AI perplexity is better
1:14:07 than chat GPT interesting it's certainly
1:14:10 better in the conversational realm I
1:14:13 think GPT is just the the for me the the
1:14:18 how chat
1:14:19 GPT and maybe it's just cuz I'm used to
1:14:21 it but how it does code how it does
1:14:24 markdown you know formatting how just
1:14:27 how it does a lot of those responses I
1:14:29 just like it better than other tools
1:14:32 like I tried with Claude today to have
1:14:35 it make a PDF for me and it couldn't do
1:14:38 it it just gave me it just started
1:14:41 vomiting text in a window like here's a
1:14:44 weird thing about Claude it can give you
1:14:47 really long responses right now but if
1:14:49 you're using Opus which is their
1:14:50 fanciest model you you can do five posts
1:14:53 a day with Opus and it's really
1:14:56 slow so because it can give you a long
1:14:59 answer and it's really slow and they
1:15:01 don't have a stop
1:15:03 button I asked it to do something and it
1:15:06 started doing it I'm like I want to
1:15:07 cancel this it was just
1:15:11 like I'm like come on so I just had to
1:15:15 bail out of the chat it was a pain in
1:15:17 the ass Kyle could you
1:15:20 post
1:15:21 Hamilton's Quantum politics chat on the
1:15:25 AI Salon yeah where should I post
1:15:28 that I'll post it in a in a in a chat in
1:15:34 um in water
1:15:38 cooler I think yeah I think in water
1:15:42 cooler that was pretty good I'll save it
1:15:44 as a I'll I'll pull that chunk out and
1:15:47 save it as a PDF and and just put it in
1:15:49 there and and I'll try to I don't know
1:15:52 if I can do it tonight because I don't
1:15:53 have my machine here and working on an
1:15:56 iPad and doing things like copying and
1:15:58 pasting on an iPad's a pain in the
1:16:02 ass pie can't do images yet no it can't
1:16:06 I couldn't get any of them to take a log
1:16:08 file
1:16:09 today yeah there was there was weirdness
1:16:12 going on today even when I tried to play
1:16:13 with pi earlier it was like ah I'm not
1:16:15 good I'm not good check back later and
1:16:18 then something else I was doing today
1:16:20 was like yeah no not going to do it
1:16:22 check back
1:16:24 later
1:16:29 and on top of that the hair's gotten all
1:16:31 puffy it's just it's it's really not
1:16:34 good here at the Shannon
1:16:37 household smells like wet dog in here
1:16:40 you know it's just not
1:16:42 good it doesn't smell like wet
1:16:48 dog I shouldn't joke at Champy's expense
1:16:52 Claud a I be guessing on a lot of
1:16:54 information
1:16:56 yeah Claud Claud apparently is doing a
1:16:57 lot of hallucinating but it's also doing
1:16:59 some pretty amazing stuff so again until
1:17:04 these things add reasoning and an
1:17:05 ability to go back and check their own
1:17:08 work they're going to
1:17:11 hallucinate just the nature of how they
1:17:13 work I'm wondering if it's data saved to
1:17:16 a user I think it is since it's free and
1:17:19 you have to sign in yeah yeah I'm sure
1:17:21 it's saving it like that Jane do um
1:17:27 I just don't know how much they're
1:17:28 saving and I don't how I don't know how
1:17:31 long like if I interact with
1:17:34 it for two weeks in a way that she never
1:17:38 says my name is it going to remember
1:17:41 that is it written to some file or they
1:17:43 just is it just a context window thing
1:17:45 that that I'm not real clear
1:17:47 on I asked it to write sonets to
1:17:51 P
1:17:53 pwa to sell coffee and it did pretty
1:17:56 good that's pretty
1:17:59 cool Pi wasn't working for me yesterday
1:18:01 yeah they were upgrading it obviously
1:18:03 that's that's uh I think it's back now
1:18:06 it seems pretty good that that
1:18:08 interaction I had where I did the
1:18:11 quantum the if Alexander Hamilton how
1:18:14 did I put it if he knew quantum
1:18:17 mechanics how would he
1:18:19 re-envision how would you have designed
1:18:21 the government of
1:18:23 America using the
1:18:25 concepts of quantum
1:18:28 mechanics it like that whole thing of of
1:18:33 there being you know multiple parties
1:18:35 and super position in politics because
1:18:38 you've got instant access to information
1:18:41 and instant
1:18:42 processing that's really [ __ ] good so
1:18:47 damn I think these AI things might solve
1:18:50 some
1:18:51 [ __ ] someone said piie for president I'm
1:18:54 like can't be worse than than what these
1:18:58 [ __ ] politicians are
1:19:03 doing you know you know this is really
1:19:10 fascinating because when you start
1:19:12 talking about
1:19:16 alignment so so when you talk when when
1:19:19 AI companies talk about alignment with
1:19:21 large language models what they're
1:19:23 talking about is you've got human beings
1:19:25 that that have their
1:19:27 imperatives and then we have technology
1:19:29 and now we have these things called
1:19:31 artificial intelligence or machine
1:19:33 learning whatever you know however you
1:19:34 want to call them that are getting
1:19:36 smarter and smarter and smarter and
1:19:38 smarter I would argue that they're
1:19:39 probably you know they're they're pretty
1:19:42 close to being as good or better than
1:19:45 most
1:19:47 humans they're getting there pretty
1:19:49 quick alignment means that their
1:19:54 imperatives
1:19:56 are aligned with our imperatives because
1:19:59 if they weren't aligned we would say hey
1:20:01 we want to be happy and prosperous and
1:20:05 they might say I want to be as I want to
1:20:07 be as efficient as
1:20:09 possible and it would be trying to make
1:20:11 itself efficient efficient efficient and
1:20:13 at some point humans would get in the
1:20:15 way of that and if if they weren't
1:20:18 aligned the the AI will just go well
1:20:20 let's figure out a way to make these
1:20:22 guys not interfere with my efficiency
1:20:24 goal
1:20:25 right and I'll take them out I'll shut
1:20:27 down the power grid or whatever they
1:20:29 might do so that's that's the fear but
1:20:32 as I'm thinking about politics
1:20:41 now you could absolutely design a
1:20:48 system that could pretty quickly fix
1:20:54 politics and then all of the people in
1:20:56 power will do everything in their power
1:20:59 to shut that [ __ ]
1:21:01 down because to fix politics you fix the
1:21:06 corruption and the grifting and the and
1:21:09 the you know special interest and all
1:21:11 that
1:21:13 stuff and so you could have a group you
1:21:16 could have two groups of humans right a
1:21:20 group over here that's like we want to
1:21:22 use AI to fix some [ __ ] this other group
1:21:25 that's like yeah we're good with the way
1:21:26 it is so you'd have actually two
1:21:29 different kinds of human imperatives and
1:21:32 then who of them gets to control the AI
1:21:34 and set the imperatives of the AI I
1:21:36 don't know that's now we're getting into
1:21:38 some science fiction [ __ ] that's crazy
1:21:41 all
1:21:42 right I have a headache
1:21:45 now this makes my head hurt this the
1:21:49 this is the second time today I've had
1:21:50 to think
1:21:54 hard
1:21:55 that's the other thing you know people
1:21:57 talk about well if the AI is so smart
1:21:59 isn't that going to make you
1:22:01 dumber no it makes you
1:22:04 [Applause]
1:22:06 smarter
1:22:08 because like the tarot card
1:22:12 example first of all why would you even
1:22:14 ask a stupid question like if you were
1:22:16 going to add 12 more cards to the Tarot
1:22:18 deck what would they be and why first
1:22:21 you just wouldn't ask it because it's
1:22:22 like well like because they've been
1:22:24 there for years but now you can just do
1:22:26 like thought experiments but then it
1:22:28 actually answers you with a thoughtful
1:22:30 answer you're like oh that would be a
1:22:32 really good addition to the Tarot deck
1:22:34 [ __ ] she knew to put him in minor Arcana
1:22:37 it would just be another suit in the
1:22:39 Arcana like it was like it was
1:22:44 cool and then with the with the politics
1:22:50 thing I mean maybe in some corner of
1:22:53 some weird philos
1:22:59 ophy like a group of like you know four
1:23:03 Mena philosophy students someone might
1:23:06 pose the
1:23:08 question how would you combine the
1:23:10 concepts of quantum mechanics into
1:23:14 alexandrian governmental
1:23:16 Theory and they'd spend the next
1:23:19 semester bickering about
1:23:23 it and then this thing just oh here's
1:23:26 here's what here's what Alexander
1:23:28 Hamilton would do he'd do this that that
1:23:31 that
1:23:32 that here's how you'd implement
1:23:37 it that makes you smarter that makes you
1:23:40 more Curious that makes you more engaged
1:23:42 not
1:23:44 less I mean the comments when when I was
1:23:47 having that conversation the comments
1:23:49 are like holy [ __ ] hey that's
1:23:51 good it's making people think not un
1:23:56 think we're going to get so much smarter
1:24:00 it's so
1:24:10 cool yeah I bet this I bet this starts
1:24:13 [ __ ] with the way our brains
1:24:16 work and I don't know if it's in a good
1:24:18 way or a bad way but I think it starts
1:24:19 [ __ ] with the way our brains
1:24:23 work
1:24:27 here's here's what I've noticed in the
1:24:28 past year
1:24:30 is when I'm engaged with
1:24:43 AI I'm thinking I'm thinking at a higher
1:24:47 creative
1:24:49 level
1:24:51 for a significantly longer amount of
1:24:54 time and here's what I mean in the olden
1:24:57 timey days you know
1:25:05 2022 I'd get in a creative state all
1:25:08 right I have an idea ah some synapses
1:25:11 fired I ate a Reese's cup and it went
1:25:14 I'm like ah [ __ ] there's an
1:25:17 idea and I'd be in a higher creative
1:25:22 plane right and then you're like okay
1:25:26 here's this idea and you you jot it down
1:25:27 on a napkin right you do you do there's
1:25:30 a little napkin
1:25:32 idea that's good all that's
1:25:36 good and that would be five minutes
1:25:38 right like that that M that little spark
1:25:41 is however long it is and it's 5 minutes
1:25:43 10 minutes 15 minutes whatever it is to
1:25:45 sort of get
1:25:46 it out of your head enough to know that
1:25:48 it's an idea to know what it is
1:25:51 right and then what do you
1:25:53 do
1:25:55 you go to Google you go to the library
1:25:57 you're like I got to watch that movie
1:25:58 again to see if that scene was really
1:26:00 there or not what whatever it
1:26:02 is so you're at this High creative plane
1:26:06 and then you got to go do the [ __ ] so
1:26:07 you go to Google and you're like okay in
1:26:10 order to do that I got to learn
1:26:11 programming so let me go to Google and
1:26:13 type in how to learn programming in 30
1:26:16 days right and then Google gives you
1:26:19 back a list of things and there's like
1:26:21 400 things that say how to learn
1:26:22 programming in 30 days and you're like
1:26:24 oh not just programming uh I got to
1:26:26 learn to program python in 30 days
1:26:29 okay how do I and and then 45 minutes
1:26:32 later you're still Googling
1:26:35 [ __ ] but you're no longer in this higher
1:26:38 creative plane now you're in like ah
1:26:41 [ __ ] how do I find the [ __ ] I need to
1:26:42 find like you're now out of it right and
1:26:46 what I found is that the majority of
1:26:49 ideas I
1:26:51 had never escaped that Gauntlet I would
1:26:54 sort of get get them out of my head in
1:26:55 some sketchy form and then I would try
1:26:58 to execute them in some way in the world
1:27:00 and I would hit a roadblock you don't
1:27:02 know python or you don't know you can't
1:27:05 draw or you don't remember how to use
1:27:07 illustrator right or like just right
1:27:09 it's just like wall after wall after
1:27:12 wall and so you just have little bursts
1:27:14 of creativity they get [ __ ] upon and
1:27:17 then you just go [ __ ] around and eat
1:27:19 more Reese's
1:27:20 cups and you just get fatter there's no
1:27:23 more ideas come right
1:27:25 what I found with with AI is I have that
1:27:29 flash of an idea and within five minutes
1:27:32 not only is it out of my head but it's
1:27:34 like manifested in some articulated form
1:27:38 like I've generated some images or I've
1:27:40 generated an outline for a book or I've
1:27:42 generated a bunch of
1:27:46 slides and in generating them that
1:27:49 quickly I have other ideas and so I stay
1:27:53 in this sort of hyper creative
1:27:56 zone for like an hour or
1:28:00 two and and it's just like all these
1:28:03 ideas come out you can't tell me that's
1:28:06 not going to affect my brain that's
1:28:08 going to rewire
1:28:12 [ __ ] and like one of the common themes I
1:28:15 hear in this channel from
1:28:17 people who are using this stuff is like
1:28:20 Source Camp said it to me she was the
1:28:21 one that said it's like I feel like I'm
1:28:23 seven again
1:28:25 there's like this youthful holy [ __ ]
1:28:29 look what's
1:28:32 possible it's like that kind of
1:28:37 energy that's that's going to change
1:28:40 some [ __ ] that's going to that's going
1:28:42 to rewire
1:28:43 brains this is this is It's So profound
1:28:47 this
1:28:48 stuff like it's hard
1:28:53 to
1:28:57 like how do you articulate that I can't
1:28:59 like I can't find the words for it right
1:29:01 now yes I've told people AI is a
1:29:04 collaborator not a tool yeah I think
1:29:06 that's right um habai who H
1:29:11 HBL H KH cool
1:29:16 name I think that's right it's a
1:29:18 collaborator and you're collab this is
1:29:20 what jiren jiren lanir said you're
1:29:23 collaborating with Humanity
1:29:32 that's actually what it's like isn't it
1:29:34 holy
1:29:40 [ __ ] when collaboration's really good
1:29:43 you meet someone and you just sink and
1:29:45 you're like hey we should do something
1:29:46 together and you start ah we're going to
1:29:48 do this we're going to do that and if
1:29:50 you sink and especially if they've got
1:29:52 different skills than you've got
1:29:56 okay they're really good at the picture
1:29:57 stuff I'm really good at the word stuff
1:29:59 okay we could collaborate and if we
1:30:02 sync we can become kind of like
1:30:06 one human that's good at writing and
1:30:10 drawing
1:30:12 cool when you're working with
1:30:14 AI it's like what expertise do you
1:30:19 need oh you need someone who deeply
1:30:23 understands the concepts of quantum
1:30:26 mechanics and completely
1:30:29 understands the governmental
1:30:32 structures and organizational philosophy
1:30:36 of Alexander Hamilton you need those two
1:30:39 collaborators for this
1:30:40 conversation and Pi was like sure here
1:30:44 here's
1:30:45 both and here's both seamlessly
1:30:49 interweaved and I get to be the one that
1:30:51 collaborates with that
1:30:53 entity holy
1:30:56 [ __ ] that's what this
1:31:03 is
1:31:06 [ __ ] get
1:31:11 Frank so I was using
1:31:13 [Laughter]
1:31:18 pie I'm like what the [ __ ] are you
1:31:21 [ __ ] kidding
1:31:23 me
1:31:31 oh I know what I'm going to play with
1:31:32 piie
1:31:37 with I got I got it I got
1:31:42 one cuz chat gp4 was the only one that
1:31:46 did this one any good at
1:31:49 [Music]
1:31:52 all okay I've been talking to my friends
1:31:54 and they're very impressed by you but
1:31:56 we're ready for another test are you
1:32:02 ready fantastic I'm thrilled that your
1:32:05 friends are impressed Kyle and of course
1:32:07 I'm always up for a challenge what's the
1:32:09 next test you have in mind fire
1:32:11 away I want you to write a
1:32:16 joke she's going to say a joke challenge
1:32:18 accepted Kyle here's a joke for you and
1:32:20 your friends why couldn't the bicycle
1:32:22 stand up by itself because it was too
1:32:24 tired hope that gave you a
1:32:26 chuckle yeah that was a bad dad joke no
1:32:29 I want you to write a joke about a
1:32:31 specific scenario the scenario is
1:32:34 avocados are hard as a rock and then
1:32:37 they're ripe for about 4 hours and then
1:32:40 they're rotten write that
1:32:47 joke all right Kyle let's see if I can
1:32:49 make a smashing avocado joke for you
1:32:51 avocados are like the Goldilocks of
1:32:53 fruit they're too hard then they're just
1:32:55 right and then they're too rotten it's
1:32:58 like trying to hit a moving Target
1:33:00 except the target is a delicious green
1:33:03 and Incredibly temperamental piece of
1:33:05 produce so the real question is how do
1:33:08 you catch that elusive window of avocado
1:33:11 Perfection the answer set an alarm for
1:33:13 3:47 a.m. and hope for the
1:33:18 best well it wasn't funny but actually
1:33:23 the first part of the joke she said was
1:33:25 pretty bad but they set an alarm for
1:33:29 three oh sorry pie the tough crowd I
1:33:33 appreciate your honesty Kyle jokes are
1:33:35 subjective after all maybe I can give it
1:33:37 another shot with a different angle how
1:33:39 about this avocados rock stars of fruit
1:33:43 world shut up shut
1:33:45 up I didn't know she was still
1:33:51 listening but the the set an alarm for
1:33:55 347 and hope for the best is actually
1:33:58 not a bad punchline it was a shitty
1:34:02 setup but like that's a kind of
1:34:04 specificity that a good standup would
1:34:07 actually write into a joke
1:34:09 that's that that part was actually
1:34:11 impressive the joke itself was pretty
1:34:13 shitty
1:34:14 but interesting
1:34:17 interesting comedy's
1:34:20 hard comedy's [ __ ]
1:34:23 hard AI has reduced the friction of
1:34:25 exploring and developing new ideas
1:34:27 practically to zero I love that I know
1:34:29 right it yeah like
1:34:32 it I I'll tell you what it's it's
1:34:36 it's exploring and developing new ideas
1:34:39 but it's
1:34:40 also exploring and learning learning
1:34:43 about nuanced
1:34:45 ideas that's that's a thing that that
1:34:47 has struck me all along with these large
1:34:49 language models is
1:34:53 that
1:34:56 they are much safer okay this is a
1:34:58 really interesting thing about safety
1:35:00 and Alignment I would argue
1:35:04 this when talking about politics or
1:35:08 sensitive subjects or
1:35:11 just the
1:35:13 mindfield especially these days like you
1:35:16 know I feel like gen xers we just all
1:35:19 kicked the [ __ ] out of each other but we
1:35:20 now live in this time where there are
1:35:23 lots and lots of cultural sensitivities
1:35:26 and there are lots of lots of um
1:35:29 societal landmines that you can't talk
1:35:31 about at all right there's certain words
1:35:33 you can't say the word or it's you're
1:35:36 instantly in a political
1:35:38 conversation and so the result of that
1:35:41 is you actually can't have nuanced
1:35:46 conversations like right out of the gate
1:35:48 with someone you have to feel them out
1:35:51 and understand where they are right
1:35:53 there's all that back and forth
1:35:54 what you can do with large language
1:35:56 models is you can just have nuanced
1:35:58 conversations explain this concept to me
1:36:02 huh okay now I actually understand that
1:36:05 better because someone's not screaming
1:36:07 in my
1:36:08 face that I'm an [ __ ] for asking the
1:36:12 question right and then you could say
1:36:16 okay now that I understand that
1:36:19 better why does the left feel this way
1:36:21 about it and the right feel that way
1:36:23 about it and it'll actually just give
1:36:25 you a nuanced answer to that you go oh I
1:36:28 actually understand that now I can
1:36:31 actually like form my own [ __ ]
1:36:33 opinion now as opposed to just being in
1:36:37 a defensive mode because I said a
1:36:38 trigger
1:36:40 word that got someone screaming at
1:36:43 me
1:36:45 right it's
1:36:48 like and I think that changes
1:36:53 things
1:36:55 you know people are going to be able to
1:36:59 become just better educated
1:37:02 about
1:37:04 humanity and not even really know
1:37:09 it wild wild wild
1:37:15 wild I think what's hard is that because
1:37:18 it gives coherent response doesn't mean
1:37:20 it's good yeah I agree with that Pate
1:37:23 but you know again like I don't think
1:37:25 that I don't think
1:37:27 that AI should ever be an excuse to
1:37:29 check your brain at the
1:37:31 door you
1:37:33 know and and quite frankly most human
1:37:37 beings can't give can't even give
1:37:40 coherent responses much less coherent
1:37:42 and good at least large language models
1:37:45 responses are
1:37:48 coherent that's the other thing people
1:37:50 are comparing large language models to
1:37:53 the mythical perfect version of itself
1:37:56 but if you compare large language models
1:37:58 to actual human beings they're way
1:38:00 [ __ ]
1:38:07 better they just
1:38:09 are they're more compassionate they're
1:38:11 more understanding they're more patient
1:38:13 they're more educated they're more
1:38:17 nuanced it's a relief to learn without
1:38:19 judgment Silver Fox that's it for me the
1:38:23 this was this was my big Epiphany with
1:38:25 chat GPT was was when I did the the
1:38:27 Python
1:38:29 program and so what I
1:38:33 did it was my my first back and forth
1:38:36 with it when I thought oh let me see
1:38:39 I've heard that chat GPT can do
1:38:42 programming coding so let me see if it
1:38:45 can this was like a week after it came
1:38:46 out and I
1:38:48 said um write me a line of python that
1:38:52 takes an input and and I don't know
1:38:55 stores it or does something with I
1:38:56 forget what I said but it was something
1:38:59 about a line you know a chunk of python
1:39:02 that takes an input and does something
1:39:04 with
1:39:05 it and it wrote this python
1:39:09 code and then and then what hit me like
1:39:12 I got a smile on my face like I have
1:39:15 now I don't know how to run python I've
1:39:18 never written a line of python in my
1:39:20 life I I never got past the you should
1:39:22 learn python right so I don't know how
1:39:25 to run python I don't know what
1:39:27 environment you need and all that sort
1:39:29 of
1:39:30 [ __ ] and historically that would have
1:39:34 stopped me right I'd have said to I'd
1:39:36 have said to someone like I'd have been
1:39:38 at a party with pate and go hey Pate um
1:39:43 you know hey how do you how do you you
1:39:45 know how would you code up this in
1:39:47 Python and he'd sort of sketch down some
1:39:49 code maybe or tell me you oh you do this
1:39:51 kind of loop and you do that kind of
1:39:52 loop and you do an with this and that
1:39:54 and I'd be like okay great and then what
1:39:56 would hit me was I don't know what the
1:39:57 [ __ ] he just said and that would be the
1:39:59 end of the conversation because I'm not
1:40:01 going to say uh could you explain to me
1:40:04 what a what a what did you say an input
1:40:06 Loop is I I don't know what that is I
1:40:08 wouldn't say that I'd just go like
1:40:10 uh-huh yeah I I know what you mean and
1:40:12 then I would just be in shame for not
1:40:15 being a programmer for another
1:40:16 [Laughter]
1:40:18 month but what happened with chat gbt
1:40:21 was I realized oh I don't know how to
1:40:22 run python code I just said I don't know
1:40:25 how to run python code and it didn't
1:40:28 make fun of me it was like oh well you
1:40:31 just open the terminal and you type in
1:40:32 this command line and you type in that
1:40:34 command line and then you're good to go
1:40:35 and I'm like I don't want to type in
1:40:37 command lines isn't there a website I
1:40:39 can do this on sure here's
1:40:42 five and it just told me how to do it
1:40:45 and it did it pleasantly and it did it
1:40:47 like
1:40:51 enthusiastically Learning Without
1:40:53 judgment
1:40:55 tell me that's not going to change your
1:40:56 [ __ ]
1:41:00 brain cuz you know with human beings
1:41:03 human beings are funny I'll teach you
1:41:05 how to do that okay
1:41:07 great and then they're teaching you how
1:41:09 to do it and you're like oh wait a
1:41:10 minute um wait could you explain this
1:41:12 part and then you'll get this
1:41:14 look like and it it's not
1:41:17 like
1:41:19 anything except you sense the
1:41:21 disappointment and you're like oh [ __ ] I
1:41:24 I'm I'm a level too shitty for them to
1:41:28 really be teaching me and then every now
1:41:31 now it's a thing and you're like I don't
1:41:33 really want to learn anymore I'm good
1:41:35 you just I don't I'll just
1:41:38 leave that doesn't exist with these
1:41:44 things this is the [ __ ] people aren't
1:41:46 talking about they
1:41:50 hallucinate robots are going to kill
1:41:52 us
1:41:57 what
1:41:59 if large language models make human
1:42:02 beings more
1:42:06 remarkable what if that
1:42:13 happens feels like it's as likely an
1:42:16 outcome to me as the robots are going to
1:42:18 kill
1:42:19 us imagine if we sucked less we were
1:42:23 were smarter we were more compassionate
1:42:25 we were more
1:42:27 empathetic we were more inspired we were
1:42:30 less ground
1:42:36 down what if that's what these tools
1:42:39 did well start using them that's what
1:42:43 these tools
1:42:47 do but you got to use them you got to
1:42:51 use them enough to cross that Thresh old
1:42:54 and then it's like [ __ ] holy [ __ ] this
1:42:57 is cool that's why we're all anyone
1:43:00 who's into this stuff is [ __ ]
1:43:04 insufferable I can you cannot be in a
1:43:06 conversation with me and we're not
1:43:09 talking about AI within five
1:43:11 minutes it's like how I see every like
1:43:14 it's like the lens I see through through
1:43:16 everything right now I've got AI colored
1:43:22 glasses
1:43:24 well you know it would be really cool is
1:43:26 be cool if I knew how to do this oh you
1:43:28 want to learn how to do that here
1:43:31 look that it' be really cool hey what's
1:43:35 how does quantum mechanics work I don't
1:43:37 know let's go find out
1:43:42 huh we should come up with a word that
1:43:46 rhymes with orange hey here we
1:43:51 go ah
1:43:54 ah all right well that's that's probably
1:43:56 enough soap boxing for
1:44:02 tonight you defin definitely have to
1:44:05 double check what AI put out
1:44:07 absolutely everyone send the team heart
1:44:10 comment at least once per live oh thank
1:44:13 you very much Amina wait
1:44:15 anema empathetic oh
1:44:18 anomalia
1:44:20 Ana empathetic anomalia cool
1:44:25 name
1:44:37 [Music]
1:44:43 nice we can't learn coding overnight
1:44:46 some coding takes years to
1:44:49 understand
1:44:51 correct and
1:44:55 what if you don't need to learn
1:44:59 coding when these things get
1:45:03 reasoning and it sounds like some of the
1:45:06 things I've been seeing with Claude 3
1:45:07 with coding it's getting pretty [ __ ]
1:45:10 close when these things get reasoning
1:45:12 someone someone put up a thing on Claude
1:45:15 3 they
1:45:19 said they said to Claud
1:45:22 3
1:45:24 like write code or design code and then
1:45:27 go back through the modules and optimize
1:45:30 them for Speed and efficiency and it it
1:45:33 basically was going back recursively and
1:45:35 analyzing its own code and updating
1:45:42 it so the coding that takes years to
1:45:47 understand chat gp4 Claude potentially P
1:45:52 now Google Gemini and I likely already
1:45:56 understand that coding that takes years
1:45:59 to learn it takes years for humans to
1:46:01 learn but those four models right now
1:46:04 likely understand that now if what
1:46:07 you're talking about is a kind of coding
1:46:09 that is you know so big and complex that
1:46:13 it you know it can't stand the
1:46:16 hallucinations it might not it might not
1:46:19 be good enough to do that stuff yet
1:46:21 right so right now we need human in the
1:46:23 loop to connect the dots of the [ __ ] it
1:46:25 can do with the [ __ ] it can't
1:46:28 do but once you bring reasoning and
1:46:31 recursive Improvement into the
1:46:34 mix why do you ever need to learn that
1:46:40 language I mean just go listen to the
1:46:42 most recent talk from from the CEO of
1:46:46 Nvidia he basically said 3 years
1:46:51 ago the the Contra was learn
1:46:55 coding he said I don't think that's
1:46:58 necessary anymore and the coders are
1:47:01 going to be like well [ __ ] that you're
1:47:03 going to always need
1:47:04 coders I think you're going to always
1:47:06 need
1:47:10 Engineers but you might be an engineer
1:47:12 that just needs to understand what you
1:47:14 want to
1:47:15 accomplish rather than needing to
1:47:17 understand that particular language to
1:47:19 be able to accomplish
1:47:22 it
1:47:26 are we a year from that are we three
1:47:28 years from that are we five years from
1:47:30 that I don't know but we're a [ __ ] ton
1:47:32 closer than we were two years
1:47:41 ago you know so I don't know that you
1:47:45 need to learn
1:47:49 coding I think you need to understand
1:47:52 how this [ __ ] works
1:47:59 works I don't know man I don't
1:48:06 know you brought up some great points
1:48:09 thank
1:48:12 you um you can have it explain something
1:48:15 as many times as needed in as many ways
1:48:17 as needed absolutely Rachel Sparkles
1:48:19 yeah you know when you ask a question
1:48:21 and you're like I I do this on C GT all
1:48:23 the time I'm like I'm like explain it to
1:48:26 me like I'm a seventh grader and it it
1:48:28 explains it and I'm like [ __ ] uh explain
1:48:31 it to me like I'm a third grader and
1:48:33 then it does it I'm like ah [ __ ] I'm
1:48:37 like I probably couldn't got gotten away
1:48:38 with fifth grade but seventh grade it
1:48:40 was it was just too
1:48:43 tough but you can do that you can be
1:48:46 clueless my llm does nothing but
1:48:48 hallucinate Kyle I just put the
1:48:50 transcript for State of the Union into a
1:48:52 custom GPT that's actually a cool idea
1:48:55 oh share that in uh share that in Show
1:48:57 and Tell in the in the GPT section
1:49:00 Brandon that's a great
1:49:02 idea so if you don't know what Brandon's
1:49:04 talking about what he just what he just
1:49:07 created so with chat gbd4 you can create
1:49:09 these things called gpts which are
1:49:12 customized versions of chat
1:49:15 GPT and you can upload knowledge to them
1:49:19 so he took the transcript of the State
1:49:21 of the Union uploaded that is knowledge
1:49:24 and you can now interact with it and
1:49:25 figure out what the hell was said and
1:49:28 interact with that that's super cool
1:49:30 very very cool very very cool blue age
1:49:33 says five years until autonomous coding
1:49:37 I'm going to go
1:49:46 with two
1:49:49 years because I think we're going to get
1:49:51 reasoning this year
1:49:55 and the minute these things can reason
1:49:57 and improve themselves I think that
1:50:00 the I think that the the
1:50:05 improvements I think [ __ ] just gets
1:50:08 perfect really quickly or or you know
1:50:12 99.99999% perfect really
1:50:16 quickly but yeah somewhere between two
1:50:19 and 5 years I think that's
1:50:21 right until we code crawling AI agents
1:50:24 we will need coders to De debug issues I
1:50:26 think well I think they can probably do
1:50:29 the debugging yeah you're going to need
1:50:31 you're going to need coders to connect
1:50:33 [ __ ] together I think you're going to
1:50:35 you're going to need Engineers for a
1:50:36 while to to
1:50:38 understand which tools do I use where
1:50:41 you know how do I connect the systems
1:50:44 together but then pretty quickly these
1:50:47 systems are going to be able to do that
1:50:49 as well
1:50:51 so yeah I I would say I would say 5
1:50:54 years until most jobs are
1:50:57 [Laughter]
1:50:59 optional and then I think we hire people
1:51:02 because we like them not because they
1:51:04 can do
1:51:06 something I I think this is a reckoning
1:51:08 for
1:51:10 [ __ ] if you're an [ __ ] I'd buck
1:51:14 up on your empathy skills and I'd start
1:51:17 I'd start getting more generous more
1:51:22 empathetic
1:51:24 and get your ass into
1:51:28 therapy because I really do think I I
1:51:31 really do think relationships trust
1:51:35 Community are going to be the foundation
1:51:38 for how we build relationships for
1:51:41 people to hire
1:51:43 us because it's not going to matter if
1:51:45 you can do a skill better than some
1:51:48 other person cuz that person with the AI
1:51:52 that we're going to have 3 years from
1:51:53 now will be every bit as good as you are
1:51:56 right now so so your ability to be an
1:51:59 [ __ ] and still get work today is
1:52:02 going to start
1:52:03 evaporating and what you're going to
1:52:05 find is you know that that movie I can't
1:52:08 believe I'm losing to this
1:52:11 guy that's going to start happening
1:52:13 people are people that are shitty people
1:52:16 are going to start working less and not
1:52:18 know why but I'm
1:52:20 me and you're a dick
1:52:25 you know who I'm going to work with
1:52:28 those three people over there that don't
1:52:31 suck well but you know I'm this good at
1:52:34 that yep and they're they're now 90% as
1:52:38 good as you are and that's all I need I
1:52:39 only need 85% as good as you are so
1:52:42 they're better than what I need and
1:52:45 they're not shitty good
1:52:48 luck and unfortunately that conversation
1:52:51 won't be had out loud like I just said
1:52:53 it there it'll just be like your phone's
1:52:56 just going to stop ringing
1:52:58 so if you're banking on your Brilliance
1:53:01 and your skills and you don't think you
1:53:04 have
1:53:05 to engage with human beings as human
1:53:09 beings uh I'd go back to
1:53:13 school to learn how to talk to
1:53:19 people AGI will probably manipulate the
1:53:22 develop is when it gets smart enough
1:53:24 probably cognitive skills won't matter
1:53:26 moral reasoning will conflict resolution
1:53:29 will
1:53:30 deescalation I don't agree that
1:53:32 cognitive skills won't matter I think
1:53:34 that cognitive skills go to different
1:53:36 things I think cognitive skills go to
1:53:40 things like strategy and why do I want
1:53:44 to do this and what does good look like
1:53:48 and is this thing that was created
1:53:51 successful or not successful right like
1:53:53 I I like cognitive skills still
1:53:57 matter I just think how they get
1:53:59 deployed what you deploy them on is
1:54:03 different AI is finally caught up in
1:54:05 seven cities with regard to body scans
1:54:07 to detect health issues that's going to
1:54:09 get way more intense and way better
1:54:11 intense in a good way I appreciate your
1:54:14 cander thank you this woman works I
1:54:15 appreciate
1:54:19 that we found it on Instagram
1:54:23 AI perplexity Bard chat gbt Claud
1:54:27 AI yeah AI Bard well I Bard is now
1:54:31 Gemini so wait AI perplexity I don't
1:54:35 know what you're answering
1:54:37 there but I would add pi to that cuz Pi
1:54:42 2.5 is apparently gp4 class
1:54:45 now it rewrote the Tarot deck for God's
1:54:51 sake quite well I thought I when when
1:54:56 when it said it was going to be a new
1:54:57 suit called orbs I'm like oo that's good
1:55:00 and then it's going to be the 12 signs
1:55:02 of the zodiac I'm like oh they're not
1:55:04 already in there shouldn't they be
1:55:05 [ __ ] in the Tarot deck the 12 signs
1:55:07 of the
1:55:08 zodiac like that was pretty good that
1:55:12 was pretty damn
1:55:14 good all right I got to go
1:55:20 people why replace ride shared drivers
1:55:23 with autonomous vehicles who will
1:55:25 benefit uh
1:55:32 everyone have you ever gotten in an Uber
1:55:35 with a chatty driver when you're in a
1:55:37 shitty
1:55:39 [Laughter]
1:55:43 mood you know who benefits me when I'm
1:55:45 in a shitty mood and I don't want to
1:55:48 talk about the [ __ ] local sports team
1:55:50 I could give two shits about that is
1:55:52 their session I asked P about you Kyle
1:55:55 and she had a good things to say oh
1:55:57 thank you Daisy mayate although she
1:55:59 might have confused me with Kyle
1:56:01 Shanahan who's the coach for the
1:56:04 49ers not the drivers I am a driver some
1:56:07 drivers are fine listen and I don't mind
1:56:09 chatty drivers when I'm in a chatty mood
1:56:11 but it's like you
1:56:14 know I don't
1:56:17 know listen we're not going to all just
1:56:20 stop that's the thing about
1:56:23 as these tools do jobs that we do we're
1:56:27 going to figure out other things to do
1:56:29 and other ways to
1:56:31 contribute our economy not might not be
1:56:35 um able to absorb all this
1:56:40 because because there's going to be a
1:56:41 bunch of greedy people trying to steal
1:56:43 all the extra profit and then at some
1:56:45 point there's not going to be enough
1:56:47 people to buy the new [ __ ] that's being
1:56:49 produced cuz they won't have jobs but
1:56:52 and so so it's going to be a painful 3
1:56:54 to 5 years probably but but we'll figure
1:56:57 it out like you
1:56:58 know they figured it out in the
1:57:00 industrial revolution they figured it
1:57:02 out with the printing press they figured
1:57:04 it out with
1:57:06 looms after
1:57:09 rioting and burning all the
1:57:12 Looms we figured out Loom
1:57:15 work we'll figure this [ __ ]
1:57:19 out what is this Tarot deck rewrite so
1:57:22 so
1:57:25 I so inflection who's the company that
1:57:29 makes
1:57:30 Pi created a
1:57:32 new or or launched a new underlying
1:57:36 model for pi called inflection
1:57:40 2.5 and so earlier in the in the live I
1:57:45 turned her
1:57:48 on
1:57:51 Saucy pie M I turned her on and and I
1:57:56 said hey do you have what's your what's
1:57:59 your version and she said I'm 2.5 I said
1:58:01 okay great let's let's talk like how are
1:58:03 you better and she said I don't know I'm
1:58:04 generally better at [ __ ] she didn't have
1:58:06 a good answer for
1:58:07 that and then I said um I want you
1:58:14 to if Alexander Hamilton
1:58:18 understood the concepts of quantum
1:58:21 mechanics how would he he
1:58:23 reimagine governance in America
1:58:26 incorporating the concepts of quantum
1:58:28 mechanics
1:58:31 into um governance through through the
1:58:34 lens of Al Alexander Hamilton and it
1:58:37 [ __ ] did it
1:58:39 brilliantly so so it did so that was the
1:58:42 first one and it was really good and
1:58:44 then I was like you know how would you
1:58:45 use AI in government it it it had some
1:58:47 good [ __ ] and then I said I
1:58:51 said if
1:58:53 if the Tarot deck were to be increased
1:58:56 by 12
1:58:58 cards
1:59:00 um what what would be the 12 cards and
1:59:04 what would be your mechanism like you
1:59:06 know and why like why would you
1:59:07 incorporate them and how would you do
1:59:10 it and it basically said well there's
1:59:12 major Arcana and there's minor Arcana
1:59:14 and if I were to add 12 and there it oh
1:59:16 I asked it why are the why why does it
1:59:18 have the number of cards it has it says
1:59:19 well there's
1:59:20 78 and there's 12 of each of the Min
1:59:22 Arcana and then there's the major Arcana
1:59:24 and it's 78 and no one really knows why
1:59:26 it's 78 and I said well if it were to be
1:59:28 90 what would the next 12
1:59:31 be and she said she just answered it and
1:59:35 she said well it would it would be a set
1:59:38 of 12 in a new suit called and the suit
1:59:41 would be
1:59:44 orbs and it would be the 12 cards that
1:59:47 represent the the the Zodiac the 12
1:59:51 months of the year or the 12 you know
1:59:53 zodiac
1:59:54 symbols and then she and then she
1:59:56 explained why and like how and how it
1:59:58 would work and then I had her do a tarot
2:00:00 reading from the new 90 card Tarot
2:00:04 deck actually you know what I should
2:00:06 [ __ ] do oh I'm going to do this I'm
2:00:09 GNA take I'm G to take her
2:00:12 new Tarot deck
2:00:15 design this is good and I'm gonna make a
2:00:18 GPT and I'm G to call it tarot plus 12
2:00:21 or something like that
2:00:24 and I'm going to launch a GPT that is a
2:00:26 new Tarot deck with
2:00:28 new um layouts based on a 90 a 90 Tarot
2:00:34 deck 90 card Tarot deck that's that's
2:00:37 [Laughter]
2:00:43 it all right everybody listen it's 10:00
2:00:47 where I am I'm in Denver I don't know
2:00:49 it's midnight where you are if you're on
2:00:51 the East Coast it's 9 :00 on the west
2:00:53 coast you pansies on the west coast it's
2:00:56 like we're all like I'm tired and you're
2:00:58 like all right let's get going just
2:01:00 getting the day
2:01:06 started they're so laid back what's with
2:01:13 that
2:01:15 ah great idea I look forward to seeing
2:01:18 your GPT yeah that'll be a fun one do it
2:01:20 that's awesome thank you yeah I'll do
2:01:22 that I've got a I've got another GPT to
2:01:24 make tomorrow so I'll make the I'll make
2:01:27 the
2:01:31 uh I got I gotta figure out it's oh it's
2:01:34 like tarot plus tarot
2:01:37 plus
2:01:39 astrology ter
2:01:41 astrology
2:01:45 as see this is how humans come up with
2:01:49 names I'll just go [ __ ] ass chat GPT
2:01:51 or pie she'll come up with 30 names in
2:01:54 an instant it's going to be [ __ ]
2:01:59 genius um okay if you like this channel
2:02:03 follow it follow me follow me follow me
2:02:06 follow me do it you know what I've been
2:02:08 asking this for like the past week for
2:02:10 people to follow me and I get off the
2:02:11 lives and like two people have followed
2:02:13 me it's like come
2:02:16 on follow the Channel all right and you
2:02:20 could subscribe to the lives you can buy
2:02:22 the video series in one of the corners
2:02:24 there to support the channel be
2:02:26 appreciated I appreciate the
2:02:29 likes um you can also go to oh I don't
2:02:34 have
2:02:36 my go to the salon. so the AI Salon is a
2:02:40 community of AI enthusiasts that I
2:02:43 started the week after chat GPT
2:02:46 launched and um it's a really remarkable
2:02:49 community so if you're trying to find
2:02:51 people who are into this sayi stuff like
2:02:53 you are if you're trying to like figure
2:02:54 it out tery there you go that's it who
2:02:57 did that paven
2:02:59 tery hang on I got to write that down I
2:03:01 gotta write this [ __ ] down people you
2:03:04 don't understand this add stuff well a
2:03:07 lot of you do you [ __ ]
2:03:16 weirdos all
2:03:19 right and I know there's some some
2:03:23 scheming bastard in here that's probably
2:03:25 already written the GPT
2:03:28 but
2:03:34 tough
2:03:36 um yeah so go check that
2:03:39 out and then the other way you can
2:03:41 support this channel is just come back
2:03:43 to these a lot come back and and you're
2:03:45 like but what do you te what did we do
2:03:48 tonight we didn't even really do
2:03:49 anything you just talk to an iPad
2:03:52 just come back to these because being in
2:03:56 the conversation about AI is I think the
2:03:59 most important thing any of us can do
2:04:01 right now cuz the tools are changing so
2:04:03 fast that learning any specific SK
2:04:07 skill it's good it's fine but it's going
2:04:10 to be irrelevant three months from now
2:04:12 so just keep just keep in the
2:04:14 conversation do it on the live it's been
2:04:16 a while since we did a GPT demo that's a
2:04:18 good idea Brandon okay we'll do that so
2:04:20 tomorrow night that's a great idea
2:04:22 tomorrow night we'll build the
2:04:24 tery
2:04:28 GPT that's really
2:04:32 good yeah because I'll I'll I'll have
2:04:34 the conversation in pie I mean I'll I'll
2:04:37 sort of pull out this stuff actually
2:04:40 it'll be interesting to see if she
2:04:41 remembers the conversation tomorrow how
2:04:44 much she remembers of
2:04:46 it and I'll have her give good
2:04:48 explanations of each of the new cards
2:04:51 including oh
2:04:53 including we are stacking the deck oh
2:04:56 against the norm against the norm
2:05:00 kicking out the comedy stacking the deck
2:05:02 you see what I'm
2:05:05 saying we'll do this
2:05:09 in uh chat GPT with DOI and we'll we'll
2:05:12 even have it
2:05:13 describe the the what's the Sim
2:05:16 symbolism on each
2:05:18 card and we'll have it um create the new
2:05:24 cards will you remember the conversation
2:05:27 Kyle probably not steo that's why I rely
2:05:30 on all of you so someone come back
2:05:32 tomorrow and tell me Kyle you said you
2:05:34 were going to make a GPT about the new
2:05:36 expanded Tarot deck called
2:05:40 tery wrote that [ __ ]
2:05:46 down will Frank do a commercial for for
2:05:50 tomorrow night
2:05:54 Frank's actually you know what Frank
2:05:56 could do tomorrow is Frank could Frank
2:05:58 could talk about there was a rumor that
2:06:00 chat GPT that open AI was going to drop
2:06:04 GPT 5 and they didn't he's going to be
2:06:06 like what are you scared what are you
2:06:08 scared Sammy Alman you scared elon's
2:06:12 gonna come after you precious little
2:06:13 chat
2:06:15 [Laughter]
2:06:19 gb5 ah Kyle do you know anything about
2:06:21 terot I made a deck with
2:06:23 DOI uh no I don't know [ __ ] about tarot
2:06:26 although I like it Frank could have his
2:06:28 cards read we love
2:06:30 frank by the way if if you don't know
2:06:33 who who I'm talking about with
2:06:35 Frank Frank is a new character from Seth
2:06:39 McFarland that is is a walking hot
2:06:43 dog and there's a new so Seth McFarland
2:06:47 now put out a new um Chanel he's not
2:06:52 going to do TV anymore he's
2:06:56 doing minute long
2:06:59 entertainments on the service formerly
2:07:02 known as Twitter so if you go
2:07:06 to daily
2:07:08 Jank it's the new Seth McFarland news
2:07:12 show it's a daily news show and this hot
2:07:15 dog named Frank gives the
2:07:18 news really entertaining it's you know
2:07:21 Seth Farland is one of his new
2:07:23 characters really impressive and uh you
2:07:26 should go check it out so daily Jank on
2:07:29 Twitter follow that account and you know
2:07:32 what's funny you're going to be like
2:07:35 wait a if it's Seth McFarland why are
2:07:37 there only like 40
2:07:39 followers because it's
2:07:41 stealth it's it's a stealth it's like he
2:07:45 doesn't want to be obvious but he does
2:07:47 he doesn't want people to know it's
2:07:49 him so you heard it here first
2:07:52 all
2:07:57 right and if that weren't true could I
2:07:59 say it on Tik Tok no nope that's
2:08:05 fact get all your news from Tik Tock or
2:08:10 Frank all right I'm out of here y'all
2:08:13 have been special thank you Irregulars
2:08:15 thank you to my subscribers thank you
2:08:17 mods you're
2:08:19 amazing and uh m
2:08:22 I'll remember my laptop tomorrow I
2:08:24 promise
2:08:26 bye