aiLL 10/26/23 - AI Learning Lab Live: Navigating the Wild Frontier of Generative AI

Description

Chapters

Transcript

0:17 good day good day be Not Afraid what's
0:19 happening you are
0:22 early happy Thursday what's going on
0:25 with the hair we got a situation I think
0:27 we got a situation Brewing brewing with
0:29 the hair
0:31 oh this shirt's not going to work either
0:33 is it ah what what are we going to do
0:36 what's happening Be not Afraid how are
0:40 you welcome Sherry D what's
0:43 shaking what is shaking what is
0:48 happening cuno what's happening have
0:52 they made you SVP of AI yet I think it's
0:56 coming prediction before the new year
0:59 you might get called it into a meeting
1:02 I'll be right
1:29 back
1:55 all
1:56 right let's get rocking and rolling
2:01 we're rocking we're rolling we're
2:02 rolling and we're rocking all right got
2:05 the slinky patent shirt on look they
2:08 even patented the fact that it can go
2:13 downstairs all right oh love that
2:18 one freaks me out every
2:22 time oh I'm
2:25 itchy what is happening people welcome
2:29 to the AI learning
2:31 lab we have a lovely little hedgehog
2:35 here from
2:36 Sherry Sherry made this
2:38 Hedgehog she shared it oh very nice
2:41 Sherry D thank you for the little boo
2:44 the little boo boo she made this and
2:46 shared it in the uh the Irregulars
2:48 channel on Discord which I'm going to
2:51 talk about I figured out a way
2:55 to this has been happening kind of
2:57 organically but if you're a subscriber
2:59 to the Channel now I've figured out a
3:01 way that you can I can automatically
3:04 assign you the Irregulars Discord
3:07 channel so there's a private channel on
3:09 the AI Salon Discord which I'll talk
3:11 about um but if you're a subscriber to
3:13 the channel you get that you get
3:16 priority uh for questions you ask and
3:19 things like that so I'm excited to talk
3:20 about that tonight uh Champ's here hey
3:23 buddy let me close the
3:25 door go lay down lay
3:29 down
3:32 yeah good
3:34 boy all
3:39 right totally out of my comfort
3:42 zone what was oh your video you did
3:44 great it was
3:46 awesome Slinky themed t-shirts that's a
3:49 niche market for
3:51 sure it's more like a uh it's like a
3:54 patent themed themed one yeah I know I
3:57 like this shirt a lot it's good there's
4:00 a another there's a couple other ones
4:01 there's a toilet paper one of this the
4:02 toilet paper roll which whenever anyone
4:05 tells me that that the toilet paper
4:08 should go behind the roll not in front
4:09 of the Roll no go look at the patent
4:12 drawing it's in front of the
4:14 Roll don't give me that behind the roll
4:18 crap um welcome everyone this is the AI
4:21 learning lab um so my name's Kyle
4:23 Shannon uh I'm an entrepreneur and uh
4:26 co-founder and CEO of a company called
4:28 storyvine which is an automated video
4:29 story retelling platform we're rolling a
4:32 bunch of AI into that um doing some
4:35 doing some fun work with aii on that uh
4:38 I am not a mathematician or a data
4:42 scientist but I am deeply deeply
4:46 interested in Ai and and especially I
4:49 was an entrepreneur in the early days of
4:50 the worldwide web um I was a co-founder
4:53 of a company called agency.com which was
4:56 one of the early digital agencies and
4:59 this AI stuff feels a lot like the early
5:03 days of the web but faster and more
5:06 profound
5:08 so so I've been coming here night after
5:11 night uh you know sharing uh doing what
5:14 I can to talk about AI answer questions
5:16 demo stuff so if you have questions
5:18 about it pop it in the comments below do
5:20 me a favor uh if you would share the
5:22 live so we can get some some people in
5:24 here and uh follow the channel if you
5:27 want we need some talking
5:31 all right it's behind it'll
5:42 rip okay so here's what we'll do we will
5:44 start where we normally start uh which
5:47 is with uh what if you don't know
5:50 anything about this AI stuff what if
5:52 you're like I don't know where to start
5:54 Kyle what do I what do I do where do I
5:57 start what is this stuff why is everyone
6:00 talking about it isn't it just like
6:02 fancy
6:04 Google sort of sort of except then it's
6:08 not at
6:10 all there's a way you can use it where
6:12 it feels like fancy Google um Sher D
6:16 thank you for sharing the live that's
6:17 great um oh it's the slinky patent now I
6:21 get it that's in fact a great idea yeah
6:22 I know isn't it cool look the here's the
6:24 like the stairs and this is I wonder if
6:27 this is where they got the idea for
6:28 cubert remember the the video game
6:35 [Laughter]
6:39 cubert AI learning lav and sound effects
6:42 Channel if you're new to this AI stuff
6:44 if if you're if you're really trying to
6:46 figure it out and figure out what the
6:47 big deal is and why everyone's talking
6:49 about it what's going on I'm going to
6:51 give you I'm going to give you a hand
6:52 here now if you have used it get ready
6:55 there are 1 2 3 4 5 6 7 seven slides of
7:00 resources here all right get ready to
7:03 screenshot you don't have to write all
7:05 these down we need some key messages for
7:07 what
7:08 orgs should ask agencies about where
7:12 they are with AI I'm happy to talk about
7:17 that the agencies aren't going to like
7:19 the questions I'm going to give you to
7:21 ask
7:25 them oh my God I show up right as you
7:28 started with the funny noise
7:30 say what were you doing with your
7:33 stomach oh hey Monique it's my
7:35 co-founder I was just showing
7:38 them I was showing this is the slinky
7:41 patent and they have um the stairs where
7:45 the slinkies going down the stairs and
7:47 then I I said I wonder if that's where
7:50 the the game cubert got its idea cuz
7:52 cubert was this isometric 3D cubes that
7:55 this little dude jumped around and it
7:57 looks just like that so that was
8:02 it say hi to Mo'Nique she's my
8:05 co-founder at storyvine okay so here's
8:07 what we're going to do if you're brand
8:09 new to this AI
8:12 stuff the only thing you really need to
8:14 if you're just starting just start with
8:16 the chat jpt the one the one on the far
8:19 left there on the top chat.
8:21 open.com um the a little color coding on
8:24 on my charts here I picked out of each
8:26 chart the one that I think is the the
8:28 one I'm I don't know most that I think
8:30 is most valuable right now these things
8:33 are changing dramatically so um chat GPT
8:36 is the one to start with chat jpt is now
8:39 actually eight or nine different
8:41 applications there's all these different
8:43 modes in it they all do different things
8:46 and most of them are quite remarkable so
8:49 just just go start playing with chat GPT
8:51 all those extra modes are only available
8:54 if you have a chat GPT Plus account but
8:56 um but even even just basic chat GPT is
8:59 really good
9:00 then you got Claude Bard and Bing those
9:01 are your top four these are the ones
9:03 that are not going away they're going to
9:04 get better and better and better in fact
9:06 Bard is about to get um Bard is about to
9:10 get really good when when Google announc
9:12 or when they release this thing called
9:14 Gemini which is rumored to be better
9:17 than chat gp4 I'll believe it when I see
9:19 it okay that's that one the next thing
9:22 screenshot this so these are other
9:24 important large language models um
9:27 perplexity um is good for research pie
9:31 especially the IOS app put it in
9:33 conversation mode and you can talk to it
9:36 you literally just talk and it answers
9:37 you just have a conversation you don't
9:39 have to hit the submit button anything
9:41 it's connected to the internet and it's
9:43 from this company called inflection that
9:45 just raised $1.3 billion doll so I think
9:48 they're going to be uh a major player um
9:51 PO is getting better and better and
9:53 better it's got a bot maker it lets you
9:55 play with like a dozen different large
9:57 language models including the llama
9:59 models and and the U B's Palm models and
10:05 um Claude and chat GPT and a bunch of
10:08 stuff so so it's really good all
10:10 right image
10:13 creation mid Journey for the past two
10:16 years has been de facto the best of the
10:18 image generation tools and then um
10:22 they're getting a little sleepy although
10:24 I did just see they just released um a
10:27 new version of the website where you can
10:29 view that the art that you've created on
10:31 on their Discord Channel which is the
10:33 only way you can make art right now um I
10:35 think they're about to release a website
10:38 where you can make art so so just pay
10:39 attention to Mid Journey it's really
10:41 good if you want high quality art it's
10:42 mid journey dolly3 is my new darling
10:45 right now because dolly3 now lives
10:48 inside chat GPT you don't have to learn
10:51 image prompting all you have to do is
10:53 like have a conversation about whatever
10:56 write a children's book or a marketing
10:58 plan or a social media concept or social
11:01 media headline and then have it generate
11:03 the image it's amazing I I'll I'll
11:05 likely demo that tonight um Firefly is
11:08 really good it hasn't been trained on
11:09 any copywritten material so if you're in
11:11 a big company the Adobe products are are
11:14 really good for generative AI stuff not
11:17 as good as the ones that have been
11:18 trained on copywritten material isn't
11:21 that
11:22 funny but uh but if you if your uh if
11:25 your if your bosses are a little
11:27 concerned about you getting your uh
11:30 and then Leonardo a is really powerful
11:32 sitting on top of stable diffusion all
11:34 right so that's image generation oh and
11:35 prompt book that thing on the bottom
11:37 open art.i promptbook that will teach
11:40 you how to talk to these things it's
11:41 really good it's really good okay all
11:44 right let's see what we got here video
11:48 creation this is still so video creation
11:51 right now a lot of these tools I like to
11:54 I like to say in here are janky pieces
11:56 of [ __ ] they're just janky they're just
11:58 a mess right now
11:59 video is really a mess that what it does
12:02 is it just it it makes video and
12:04 sometimes it's really impressive and
12:06 then you know nine times out of 10 like
12:08 your your face will just warp and morph
12:10 and do all sorts of crazy stuff so it's
12:13 kind of all over the place um but Runway
12:16 ml Moon Valley and pabs the three on the
12:19 left they all kind of do the same thing
12:20 text to video image to video things like
12:23 that um Visa's kind of a if you need a
12:26 corporate like explainer video that's
12:29 pretty good uh it just it just assembles
12:31 stock video but then it writes um a uh a
12:36 narration and then it narrates it and
12:38 puts music behind it it's pretty slick
12:40 Heen and did are both voice
12:44 synthesis um AIS that also do animation
12:48 of the mouth so like you create an image
12:50 in doly or mid Journey pop it into hey
12:53 genen and animate it bring it to life
12:55 heyen also has this thing uh where they
12:58 do video translations in 13 different
13:01 languages it's flipping
13:04 remarkable okay um music and audio 11
13:08 labs for audio for voice synthesis you
13:10 can make a model of your own voice and
13:12 then just type in text and it'll make a
13:14 voice refusion and sound draw make uh
13:17 write music for you the other question
13:18 mark there is just I don't I there's not
13:20 one right now that has kind of risen
13:22 above the noise pun intended get it rise
13:26 above the Noise music audio no ER all
13:30 right Seems Like Only Yesterday we have
13:32 three things on the list I know Tobias
13:34 this is [ __ ] exhausting isn't it oh
13:36 by the way this list is like a I'm not
13:39 on any of the payrolls of these
13:41 companies none of them are paying me
13:42 this is all just stuff that I think it's
13:44 worth paying attention to because things
13:47 are moving so fast there there are
13:49 literally thousands I'll show you a
13:51 directory here in a second there are
13:52 literally thousands of AI tools it's
13:55 impossible to keep up with them even
13:57 just keeping up with one or two is is is
14:00 insane um but depending on what you're
14:03 interested in that's why I'm I'm showing
14:04 all this stuff okay 3D worlds and
14:06 objects blockade laabs is really fun
14:08 Skybox blad labs.com just go there type
14:12 in a world and it'll make a world for
14:13 you that you can twirl around
14:16 360° um not super sophisticated but
14:18 really really slick and then Wonder
14:20 Dynamics is super sophisticated so you
14:23 upload video of like a person walking
14:26 upstairs and then uh you pick a robot
14:29 and it will replace the person with a
14:31 robot in the video like like
14:33 professional rotoscoping kind of special
14:35 effects but it just does it with AI
14:37 without you know you of to do any work
14:40 so pretty amazing
14:42 stuff um okay so those are all the tools
14:45 here are some people to follow Ethan
14:47 mullik is Wharton business school
14:49 Professor one useful thing is the um is
14:53 his blog he's posting a lot on LinkedIn
14:57 lately um in the past three weeks um
15:01 chat GPT has gotten a lot of new tools
15:03 and so he's testing them and just he's
15:05 just putting a lot out so follow him on
15:07 LinkedIn follow him on Twitter and and
15:09 go to his blog David Shapiro is working
15:12 on AGI an open source AGI project the
15:15 guy's a [ __ ] genius certifiable um
15:18 he's also neuros spicy so that's
15:21 fun welcome to
15:24 chadad Paul rer with the marketing AI
15:27 Institute if you're in business you
15:28 should absolutely follow him you should
15:30 absolutely listen to his weekly podcast
15:32 and then Rachel woods from the AI
15:34 exchange is is also um good and business
15:37 focused and on Tik Tok um futurepedia
15:40 doio and there's an AI for that are
15:43 those directories I talked about there's
15:45 5,000 um plus Tools in there so if if
15:49 the 27 tools I just showed you weren't
15:52 enough there's some more rabbit holes
15:54 there um and then the other thing is
15:57 this um
15:59 so so you can subscribe to this Channel
16:03 and and I went through and I learned
16:05 some new things about what what we can
16:07 do for subscribers aoy sir Kyle aoy
16:10 aular Hey Ron what's happening so if
16:12 you're a subscriber on the channel um I
16:15 mentioned before that we've got this
16:16 Discord called the AI Salon Discord so
16:19 the the salon's a community I started uh
16:21 about a year ago I'll give you
16:22 information on it but there's a private
16:24 channel in there for the Irregulars so
16:26 the people that show up here regularly
16:28 um I've now figured out a way if you're
16:30 a subscriber I can automatically assign
16:32 you that private Channel Discord role um
16:36 and the the Channel's starting to get
16:38 more popular so it used to be where I
16:41 would it you know there'd be 15 or 20 of
16:43 us in here and the channel was i' just
16:45 go through and I would answer every
16:47 single question and now it's gotten a
16:48 bit a bit more chaotic than that as more
16:50 people are coming in so we have
16:52 moderators and things like that so if
16:53 you're a subscriber like I just see Ron
16:55 there's got the subscriber tag if you're
16:58 a subscriber um your questions will get
17:00 prioritized I'll try to answer other
17:02 ones as I can but if you're a subscriber
17:04 you'll you'll jump in I'll also feature
17:06 your work a lot of the work that I
17:08 feature um in these images are from
17:11 people in that channel so so what's been
17:14 happening for the past month which is
17:16 pretty awesome is you know I'll be
17:18 talking about something I'll crack some
17:20 sort of joke and then just all of a
17:21 sudden you know now that DOI is so easy
17:24 easy to make images I'll be talking and
17:26 as I'm talking they're making images in
17:28 real time and sh sh ing them so a lot of
17:29 the images I share here are are from the
17:33 Irregulars um and and we feature them so
17:35 these are from Ted that one's from Ted
17:37 isn't that
17:38 awesome it's freaking crazy I think
17:41 that's
17:42 Sherry's and that's def definitely
17:45 Sherri's that's from the Irregulars
17:47 these are the uh the custom diapers that
17:49 that the Irregulars uh get a free sample
17:51 box of when they subscribe so anyway so
17:56 if uh if you're digging things and and
17:58 you want to hang hang out and support
17:59 the channel um feel free to do that um I
18:05 got lost where am I uh one more Channel
18:07 okay
18:08 so the salon. a um if you go to that
18:12 link uh that URL it's a link tree
18:15 there's a the first link will tell you
18:18 about the salon the second link is a
18:20 Meetup so you can find out we meet every
18:22 other Tuesday we just met this past
18:24 Tuesday and then there's a link to the
18:26 Discord that I mentioned you don't have
18:28 to be in the private Channel with the
18:30 Irregulars you can just go to the salon
18:32 Discord you can ask questions we've got
18:35 guilds [ __ ] we're all just trying to
18:37 figure it out in fact one of the uh one
18:40 of my favorite images that the that the
18:43 Irregulars put together for me let me
18:45 see if I can find it quick there's so
18:47 many images now um but here's here's
18:51 what this group really is let's see
18:54 there it is this is really this is
18:57 really what the AI learning lab is hi my
19:00 name is Kyle I'm into AI hi Kyle it's
19:03 been like 6 and a half minutes since I
19:05 made an image with DOI I got you beat
19:08 Kyle
19:09 welcome so so anyway um storyvine is my
19:15 company um the YouTube channel so I I do
19:19 um archives of these lives I take these
19:21 recordings put them on the YouTube
19:22 channel it's AI learninglab dtt for Tik
19:26 Tok um the someone else had AI learning
19:29 lab on YouTube so the this channel is
19:32 dtt uh there's one other thing that's
19:34 kind of a drag but uh we got a scammer
19:38 scammer scammer alert so so uh if if you
19:44 follow this Channel and and learning has
19:48 two L's in front of it that's not me
19:50 that's a guy that stole all my videos
19:52 stole my name and is trying to get you
19:54 to buy his crappy crypto [ __ ] oh Tobias
19:58 thank you so much um the real AI
20:01 learning lab just has one L in it it's
20:03 like it's like I know how to
20:06 spell so anyway all right so any
20:10 questions you have pop them in the
20:11 comments below the mods will pin them I
20:14 will also scroll through and just
20:16 randomly answer
20:18 some and we'll get going here I can show
20:21 you stuff well we did we did a cool
20:23 thing last night I for whatever reason I
20:25 can't get Tik Tock to to post this video
20:28 so so um there was a guy in last night
20:31 that said hey you know what could chat
20:33 GPT do for a fashion brand and I just
20:37 said well let's let's go look and so we
20:39 took some pictures from the internet of
20:41 fashion we took it into chat GPT it's
20:44 now got Vision it can see images right
20:47 and it can describe them so I had it
20:49 describe some fashion and then I had it
20:52 describe a new design based on those two
20:55 things thanks Sherry D appreciate that
20:58 um
20:59 and and then within like 10 minutes I
21:02 had this incredible design like like a a
21:06 photo of a fashion runway with this
21:09 design that sort of I made but uh it was
21:12 it was really cool so so I can do things
21:14 like that I'm happy to answer any
21:16 questions Tobias glad you brought up
21:17 subscribe I didn't know what it was
21:19 about you know what's funny Tobias I
21:21 didn't either like when I as as you know
21:23 you've been here for a while um I you
21:26 know I've been kind of stumbling through
21:28 in what Tik Tok is and and what what's
21:31 ah howy I wish I could quit you
21:35 boy
21:37 uh
21:38 um you as I've been learning what it is
21:41 what in in the past month
21:44 especially the the the community here
21:47 has really been developing I feel like
21:49 as these tools are getting more
21:51 sophisticated and as we've got regular
21:53 people that are coming back here night
21:55 after night by the way I've been doing
21:56 this every night 7 days a week for two
22:00 to three hours for the past four months
22:03 um a because I'm nuts but B because I
22:07 think this is really important and and
22:09 the whole idea here is to get the word
22:11 out demystify this stuff get people
22:14 excited about it get people
22:15 collaborating and the collaborating
22:17 thing thank you very much be not afraid
22:18 I appreciate that the collaborating
22:21 thing is really starting to happen and
22:23 this this idea of sort of you know U
22:25 connecting between here and the Discord
22:27 and sharing images and all that sort of
22:29 stuff what's starting to happen is like
22:31 people are like doing problem solving on
22:34 their own they're creating projects
22:36 together it's really exciting so um so
22:39 the the Subscribe thing supports the
22:41 channel but it also just is that idea of
22:43 like really creating this connected
22:45 community so so yeah so I went and
22:47 looked at that and and as I understand
22:49 more of that um there'll be more and
22:51 more subscriber perks but right now um
22:53 the fact that I can automatically create
22:56 um a Discord uh what's it called roll uh
23:00 is a big deal so I'm excited about it
23:03 okay so let's get some questions going
23:05 here thank you for heart
23:08 per Jesse from Austin thanks for sharing
23:10 and caring I've been Blown Away Learning
23:12 this Ai and sharing it with my wait wait
23:15 for
23:16 it com com comrades Communists
23:21 com I assume comrades with your with
23:24 your work folks um it's super important
23:29 i' I've I've been seeing more and more
23:31 of the Irregulars in here um I saw
23:33 Sherry did a video today um and who de
23:36 who else did a video today someone else
23:38 did a video
23:40 cuno no it wasn't cuno someone else it
23:47 was wasn't Nancy I don't know
23:50 anyway
23:52 um what we're seeing pretty regularly in
23:55 here is that the people that are getting
23:57 really curious I and adventurous about
23:59 this stuff that are sharing it with
24:01 people um they're starting to see a lot
24:03 of activity in their lives like
24:05 professionally like good activity um
24:08 where people are like oh hey you know
24:10 about this right can you help me I use
24:12 chat GPT for today to do my tarot card
24:14 reading pretty spot on on a novice level
24:18 yeah yeah
24:20 exactly
24:22 um the the other thing you can do with
24:24 the with the tarot stuff is is you
24:27 can if you if
24:30 you modify your prompts and modify
24:34 your um custom instructions you could
24:37 probably get it to get better and better
24:39 I can't subscribe because I can't
24:41 remember my Apple
24:44 password I'm I'm totally like that my my
24:47 wife for a while had two there was one
24:49 on like one thing and like an iTunes
24:52 password was different than her Mac
24:54 password and and it it was a nightmare
24:56 to fix it yeah that's that's a that's a
24:59 drag because then you ask him to reset
25:03 your password and then it's like a whole
25:05 17 step authentication [ __ ] nightmare
25:09 where somewhere in those 17 steps it's
25:11 like well you got to know your password
25:13 to get past this step it's like come
25:18 on oh
25:20 nice
25:23 Joker so here's something from the
25:25 Irregulars channel from uh from the
25:28 Runway so we have some inside jokes here
25:30 about llamas cuz we always talk about
25:32 llama too and turtles cuz I like that
25:34 video I Like
25:36 Turtles and then what else do we have we
25:38 have a gummy bear oh we got a hedgehog
25:40 cuz Dolly 3 their little video was a
25:43 hedgehog so we got lots of cute animals
25:45 going that's
25:50 awesome oh CRA I just saw
25:55 this
25:57 oh
26:00 that's great hang
26:12 on all right my company wants to give a
26:15 wants me to give a
26:17 presentation thinking about it Earth
26:19 rocker Earth rocker don't think about it
26:22 don't think about it do it do it do it
26:27 so so here's the deal here I I promise
26:29 you
26:32 this so I know this from the early days
26:34 of the web so so so in 1994 when I
26:37 started discovering this web the this
26:40 thing called the worldwide web I'm like
26:41 what is this thing and I started this
26:43 group called the worldwide web artist
26:45 Consortium the the AI Salon is a similar
26:47 kind of group um because we're in a
26:49 similar kind of phase of history and um
26:54 I started that group in 1994 because
26:58 I was like like I just I'm I'm barely
27:02 learning this right now I don't know
27:04 anything about this and I'm like but
27:07 there must be people that know more than
27:08 I do so I started this group and we met
27:12 in an old acting studio because I was
27:14 you know an actor at the time in New
27:16 York we met in a rehearsal Studio on
27:20 26th
27:21 Street and 10 people showed up and
27:26 nobody [ __ ] knew anything and then
27:29 the next meeting four people showed up
27:30 and then the next meeting zero people
27:32 showed up and I'm like
27:34 [ __ ] and then the next meeting 20 people
27:37 showed up and then it very quickly got
27:41 to the point we were having 250 300
27:44 person meetings um we got free office
27:47 space in Sony headquarters it was they
27:50 they had this massive conference room
27:51 that sat 300 people that's where we'd
27:54 meet every other
27:55 week and
27:59 it was probably three years where nobody
28:01 knew anything everyone was trying to
28:03 figure it out so if if you've got
28:08 um if you've got people at work that are
28:10 recognizing that you know more than they
28:12 do and they're asking you to do it do it
28:15 you you you seriously know way more than
28:19 anyone who will be in the room and even
28:21 if you don't you'll find a compatriot in
28:24 them so I like I I just it's one of the
28:26 things
28:28 I try not to be preachy about AI like I
28:31 don't really give a [ __ ] if you like AI
28:32 or hate it I just want you to try it I
28:34 want you to be educated about it but the
28:36 thing I get a little bit preachy about
28:37 is I want everyone in here to share what
28:40 you're learning share what you're
28:42 learning share the experiments share the
28:44 good share the bed share your Kevin
28:46 McAllister moments I talk about that a
28:47 lot like this
28:49 moment where you just realize oh my God
28:51 it can do that share all of
28:55 it because even if people think you're a
28:57 [ __ ]
28:59 weirdo 3 months from now 6 months from
29:02 now maybe they get laid off maybe they
29:05 hear about it again maybe something
29:06 happens and they're like hey are you
29:07 still doing that AI stuff oh yeah yeah I
29:10 saw you post on LinkedIn could you help
29:12 me figure it
29:15 out they want to lunch and learn so I
29:17 think I will absolutely do it Earth
29:19 rocker and then come back and tell us
29:20 about it that just happened with cuno so
29:23 cuno cuno when she started in here she
29:26 was like I there Ethan right around the
29:28 time cuno started hanging out Ethan
29:31 mullik put out this piece called um
29:33 detecting the SEC your detecting the
29:35 secret
29:36 cyborgs and bills 24 bucks 10 and a
29:40 third football awesome thanks
29:43 Tobias yeah I don't have any football on
29:45 over there
29:47 um but secret cyborgs were people in
29:50 companies who are getting secretly AI
29:52 literate and not really telling the
29:53 company about it and so so cuno I don't
29:56 know if she was a secret cyborg but you
29:58 know she was early and then she started
30:00 talking to a few people about it and
30:01 then and then she got invited to do a
30:04 workshop I think and then and now she's
30:06 got people like people are saying hey I
30:09 heard you you're good with this AI stuff
30:11 like people are talking to each other
30:12 about
30:13 her and she works in a big company right
30:16 so this is this is going to be good for
30:18 her right it's it's and even if you
30:21 don't work in a big company
30:26 um once you have that realization that
30:29 oh my God these things are something
30:32 different right these things are a big
30:34 deal you can't unlearn
30:38 that and so why not lean into it because
30:41 the opportunity for anyone that gets at
30:44 all AI literate in the next 18 months is
30:46 going to be huge the industry right now
30:47 is like this big it's teeny it's teeny
30:50 teeny teeny and it's about to do this
30:54 and and you might think that oh it's
30:56 overhyped and it's over over blown and
30:58 it's going to be like blockchain and
30:59 nfts it's just going to flame out it
31:02 isn't it isn't like Microsoft is rolling
31:05 it into their entire product line their
31:08 operating systems their cloud services
31:12 Office 365 November 1st a week from
31:15 today a week somewhere next week next
31:19 Wednesday
31:20 um they're rolling chat GPT into Office
31:25 365 they're calling it co-pilot you're
31:28 going to be able to be in a Word
31:29 document and go turn this into a
31:31 PowerPoint presentation and it just will
31:33 flip over to PowerPoint and write you a
31:35 powerpoint presentation Based On Your
31:37 Word document like that's coming next
31:40 week this stuff's going to be
31:42 ubiquitous what can you do to have chat
31:44 gbt answer without leaning on one
31:47 specific political ideology um right
31:51 in custom instructions do you know what
31:54 custom instructions are so so
31:58 there's there's there's a couple of
31:59 things so
32:02 so while there may be bias in the
32:05 training because it was trained on the
32:06 public internet and that that skews
32:09 heavy to the US and it skews heavy to
32:11 the left and it skews heavy mail just
32:13 because that's who's put [ __ ] on the
32:15 internet for the past 40
32:18 years so so there's some bias in the
32:20 training model there's potentially some
32:22 bias in who did the human reinforcement
32:25 learning so if it's the kind of thing
32:29 where the bias is so heavy you can't get
32:31 around it that might be an issue but if
32:33 you're just going in and say what's your
32:35 opinion on this or that and it leans one
32:37 way or the
32:39 other you're not likely giving it enough
32:42 context right do you want it to lean
32:45 more to the right do you want it to lean
32:47 more toward libertarian if you want to
32:49 do that then tell it say hey I want you
32:52 to act as a Libertarian political
32:54 analyst and I want you to think about
32:55 this topic from that point of view View
32:58 and then it will or you could say hey I
33:00 want you to be a completely neutral
33:02 entity here and when I give you a
33:05 political topic I want you to respond
33:07 with a table that's got the conservative
33:10 view the the middle of the road View and
33:13 the uh in the left View and I want to
33:15 see all those together um so so you can
33:19 modify um your prompting to get it to
33:22 lean a direction or another you can
33:26 also make it say stuff that will piss
33:29 you off and make it say stuff that will
33:32 support your
33:34 biases it's it's really there to serve
33:38 what you need so how you how you provide
33:40 it context is how it's going to respond
33:42 to you does Vision on chat GPT only work
33:45 on computers no it doesn't I don't see
33:47 it on my iPad Pro maybe toss
33:50 and maybe toss and re-upload oh oh well
33:55 so let me show it to you
34:03 okay here's chat GPT on the old
34:06 iPad okay
34:10 okay and so you've got three different
34:14 modes if if you pay for chat GPT plus
34:16 you have these three different modes if
34:18 you don't pay for chat GPT plus I do not
34:22 think you get access to GPT Vision if
34:25 you've paid for chat GPT
34:27 plus um you do get it but this is really
34:32 screwed up so the thing to make images
34:34 is do that's down there at the bottom
34:37 that can't see images it can just make
34:38 images to see images you go to default
34:41 mode like Unfortunately they didn't
34:44 say they didn't say um you know here's a
34:47 Mode called GPT Vision it's just in
34:49 default mode see those purple
34:52 icons to the left of the box that the
34:55 camera the
34:57 whatever that is uh the picture and the
35:00 folder so I'm going to turn on the
35:02 camera and I'll just take a picture of
35:04 my shelf my shelf full of
35:06 crap so there's the
35:11 image and I'm going to ask it um let's
35:16 see
35:17 um what
35:21 kind of
35:23 whiskey is that so I'm going to ask you
35:26 what kind of whis Ry is that I don't
35:28 know if it's going to do this well we'll
35:30 see can it see it oh yeah there's a
35:32 there's probably a little thing
35:34 there the whiskey is labeled juice cont
35:37 knac no that's not
35:40 right uh let's see um
35:45 what
35:48 else is
35:51 on the
35:56 shelves right and now it's doing it so
35:59 it can you can do it but you got to get
36:01 there I'm having my Psychotherapy W
36:04 could you repin that whoever just pinned
36:05 that I just got to the end of it several
36:08 photographs various magazines notably
36:11 the one with Richard Dawkins on the
36:13 cover Aigner cover with Urban Leisure
36:16 and the subheading web server only it
36:18 it's Urban desires but it was close CDs
36:21 and DVD cases blue tinted
36:23 artwork some other decorative items or
36:26 keepsakes
36:27 a frog statue or
36:29 trinket oh it's not a frog statue or
36:32 trinket but but it
36:39 is all right I'm having my Psychotherapy
36:43 patients symbolize their problems and
36:45 challenges with Dolly 3 oh that's genius
36:48 you know I I I was thinking about that
36:50 my my co-founder Monique who I don't
36:51 know if she's still on here but she and
36:53 I talk about dreams all the time and and
36:56 and planets and retrograde all the time
36:58 stuff like
36:59 that I I think it would be an absolutely
37:03 wild you can make an automation of this
37:06 pretty easy you can make a thing where
37:08 you could just talk into chat GPT so so
37:11 one of the other things you can do in
37:12 the chat GPT
37:14 app so let me let me show you this so
37:17 we're in the chat GPT app I'm going to
37:19 go into do
37:21 mode and then in the upper right hand
37:25 corner there's these little headphones
37:26 see see these right here so if I click
37:29 those it goes into conversation
37:37 mode I had a dream last night I was in a
37:40 rowboat rowing backwards and the boat
37:44 went under the muddy water and it
37:45 swallowed me up and then I woke up can
37:48 you Analyze That Dream for me and then
37:51 visualize
37:53 it I don't it it it didn't hear me say
37:57 visualize it dreams can be a fascinating
38:00 insight into our subconscious
38:02 Minds here's a general interpretation of
38:04 your dream one rowing I'm going to I'm
38:08 going to uh let's
38:13 see can you visualize that image for
38:20 me I can't create images directly but I
38:23 can help you imagine yes you can do I
38:26 not have an mode I must not no I do come
38:30 on dum
38:32 dum told you this technology was
38:37 janky I had a dream last night that I
38:39 was in a rowboat rowing backwards and
38:42 the water came up over the transom and
38:44 swallowed me whole and then I woke up
38:47 can you visualize that in an image for
38:53 me all right so now he's not responding
38:56 so now I think what he's doing is
38:57 actually off making images so he'll come
38:59 back in a bit and he'll say here's your
39:01 images but like I mean imagine doing
39:03 like a regular practice where you wake
39:05 up and and you know you take whatever is
39:07 in your dream journal visualize it you
39:10 know have it analyze it that would be
39:12 fascinating and I mean you like there's
39:14 a business idea right you could turn
39:16 that into an app there
39:19 there's it's just crazy two
39:22 visualizations of all right the first is
39:24 a photoes let me see
39:28 oh
39:30 yeah it was totally not what it was but
39:32 that one's [ __ ]
39:33 terrifying we're going to need a bigger
39:35 boat that you have to enable and then
39:39 there are all these different modes and
39:40 they don't really explain what those
39:42 modes are if you want I'm happy to I can
39:45 do a deep dive on all the different
39:46 modes of chat GPT if you want I'm happy
39:48 to happy to dig
39:49 in um but yeah you know forever hooked
39:53 forever hooked as a subscriber thanks
39:55 for for being a subscriber I appreciate
39:56 that that um yeah you you would think
40:00 that the two image modes would be in the
40:01 same vicinity nope they're
40:06 not tow
40:08 dance uh name any car for edit for an
40:13 edit I don't know what you mean for an
40:15 edit you mean for an image oh lordy I'm
40:17 going to be so busy thanks for all the
40:19 ideas y'all yeah don't don't come here
40:21 so so here's the thing if if you just
40:24 stumbled in here thinking is that Tom
40:26 Arnold did Alec Baldwin get fat is that
40:29 Rosie O'Donnell uh if you just stumbled
40:32 in here like that and then you
40:35 inadvertently get curious about this AI
40:37 stuff just kiss the next three months of
40:39 your life goodbye because once you have
40:42 your Kevin mallister moment this
40:44 like it's you can't really go back from
40:48 it um I had to tell chbt a few times
40:52 that it could could produce an image it
40:55 apologized and made the image yeah
40:57 Firefly I had a I had a thing today
40:59 where I think it was on the on the app
41:02 and and I said make an image and it
41:05 wrote some python
41:07 code and and it wrote the it wrote The
41:10 Prompt but it it the python code never
41:13 fired off do it didn't fire off the
41:15 plugin so and I I I quit the app and
41:19 restarted it and it worked fine on
41:21 desktop so I don't know this you know
41:23 back to that idea this technology is is
41:25 a little janky
41:27 little janky still it's going to be
41:29 janky for another year or two um but
41:32 don't let the fact that it's not
41:35 perfect fool you into thinking it's not
41:38 significant and profound does it need a
41:41 lot of resources on your PC it does not
41:43 because all of that so it needs a lot of
41:47 resources um there
41:49 are tens of thousands if not hundreds of
41:52 thousands of computers with these big
41:56 gpus Graphics processing units um
41:59 sitting in Microsoft servers Google
42:01 servers Amazon
42:03 servers so most of the processing for
42:06 this is done in the cloud so when you
42:08 write your prompt in it actually sends
42:10 that or you know some mathematical
42:13 probability version of that off to open
42:16 AI which then does all the processing
42:18 and then it sends you back the answer
42:20 that's why there's a delay so so if I'm
42:22 using I'll I'll show here's another this
42:24 is another um large language model
42:27 called Pi that also has a conversational
42:32 mode I had a dream last night then I was
42:35 in a rowboat rowing backwards and I got
42:37 swallowed up by the muddy water of a
42:39 pond and then I woke up in a panic what
42:41 does that
42:44 mean this pause is it going to their
42:50 [Music]
42:55 servers normally it's not that
42:58 long this this pause is it not going to
43:01 their
43:04 servers here we go oh man that sounds
43:07 like a pretty intense dream dreams can
43:09 be super weird and hard to interpret but
43:12 sometimes they can be linked to our real
43:14 life anxieties and concerns this dream
43:17 could possibly represent a feeling of
43:19 being
43:20 over so that's that
43:23 um so so yeah so it doesn't require a
43:26 lot on your
43:27 now if you want to install like llama 2
43:30 um if you want to install open source
43:32 large language models where the
43:34 processing is done locally because you
43:36 don't want to send [ __ ] to open Ai and
43:38 maybe you want to play with a model
43:39 that's that's um that doesn't have
43:43 safety guard rails on it then then
43:45 you're going to need some more
43:46 horsepower however at the salon this
43:49 week so I was in here the other night
43:51 and I was talking about these open
43:52 source models and this guy Marcus came
43:55 in here and he shared with us that he
43:58 just installed a an optimized version of
44:01 llama's llama 2 the 7B model on a
44:05 Raspberry Pi and he's he's running a
44:09 large language model on a little
44:11 Raspberry Pi like this
44:14 big and it works and it's even
44:16 multimodal you can upload images and it
44:18 can sort of see them now it hallucinates
44:20 a lot and it's not great but it works
44:23 and so the the open source models are
44:25 getting big but the the short answer is
44:27 no um there was a there was a Oh wait
44:29 here we go how do I get into private
44:33 channel on Discord so just kruno who are
44:37 you what's your handle on Discord and
44:39 then I'll let you
44:40 in um or or you can subscribe but but
44:44 you're in here all the time so you are
44:46 you are grandfathered in as they
44:48 say um so I just need to know your your
44:51 Discord
44:55 handle
45:01 are you in there are you in there yeah I
45:04 don't see you in there I'll find you in
45:06 there though um someone asked a question
45:09 about toss something and pine cone or
45:14 something like
45:17 that can you pin that again who whoever
45:20 pinned
45:25 that
45:27 H and we can go we can go play with
45:32 that why does Dolly 3 St of an issue
45:35 with text in image creation I thought it
45:37 was improved say what it's very improved
45:40 okay I'll come back to you with Dolly
45:43 and text can someone tell me can it
45:45 convert toss language into pinescript
45:47 let's go find
45:52 out I'm going to try two different
45:54 things for chat GPT
46:00 okay so we're in chbt the mode I'm going
46:03 to use so so you got a bunch of
46:04 different modes in here you've got
46:06 default mode which that's where your
46:07 vision is you've got browse with Bing
46:10 which is connected to the internet
46:11 you've got Advanced Data analysis which
46:13 is a full-on data analyst that writes
46:15 and executes its own python code it's
46:18 staggering powerful you got plugins
46:21 which there's a thousand of them most of
46:22 them are crap but this is oh thank you
46:25 very much Sherry d appreciate that I
46:27 love that filter that sticker what I
46:29 don't know what do they call those
46:30 things gifts thank you for the gift by
46:33 the way I appreciate it um and then you
46:36 got Dolly 3 so I'm going to go with
46:37 browse with Bing because when you're
46:39 dealing with uh with software there
46:41 might be the latest
46:43 versions
46:45 um let me
46:55 see
46:58 crap what was
47:00 the what was the what was the toss um
47:04 what was the toss language called toss
47:06 what toss
47:09 script oh here we go toss language
47:14 okay can you write me a simple
47:18 application in the TOs language and
47:22 describe what it
47:25 does
47:30 browsing so it may not know what the
47:33 toss language is but it's going to
47:37 browse and hopefully
47:39 find some resources like O say the
47:42 documentation I wasn't able to find any
47:45 relevant examples regarding a
47:46 programming language named
47:51 toss all right so that's not
47:55 working
48:00 is there is
48:02 there am I reading it wrong it's to OSS
48:06 right do you know what that that acronym
48:08 stands
48:09 for because
48:12 um could toss be an
48:15 acronym for something
48:17 else not catastrophe an AC acronym could
48:23 toss be an acronym for something else
48:28 [Laughter]
48:32 toss equals think or Swim script
48:54 oh it got all of those
48:59 wrong toss stands for think or Swim
49:03 script can
49:05 you find documentation on
49:11 that it's a language used for trading in
49:14 think and swim platform ah
49:25 okay
49:27 oh I see so Pine script is also a
49:29 trading a trading oh so you want to you
49:31 want to take a you want to take a
49:32 trading bot that you made in one
49:34 platform and and translate it to another
49:36 it should be really good at that so even
49:38 if even if it can't understand
49:41 this um what you could likely do like
49:44 like even if it oh no it's it it says
49:47 it's browsing thinker script think
49:49 script is built-in programming language
49:52 in The Thinker swim trading
49:54 platform okay here we go
49:57 go here's an example of a script
50:00 provided in the documentation that
50:01 calculates six dat data values based on
50:04 an initial value of an open and close
50:07 value okay
50:17 great that's great can you create a
50:21 trading bot on the 200 day moving
50:28 average in this
50:36 language come on all right let's see
50:39 what this
50:47 does here's a simplified example Define
50:51 the moving 200 day
50:53 average Define the conditions for
50:55 entering and exiting trades Buy Signal
50:58 sell signal add simulated
51:01 orders so it's putting simulated orders
51:04 in there in this script the average
51:06 calculates the average the Buy Signal is
51:09 the condition that does it Pine script
51:11 equals used on trading view okay so
51:14 great thank you Firefly for that
51:21 um that's awesome can you now Translate
51:26 that code into pinescript
51:33 absolutely pinescript is the programming
51:36 language used in the trading view
51:38 platform that's correct Firefly you win
51:42 you can make money
51:43 with and there's your Pine script so the
51:45 answer is yes you can make money
51:49 with take all your fancy thinker swim
51:52 scripts and turn them into Pine scripts
51:54 you can make money with
51:56 we all win you win we all win just just
51:59 provide a 10% tithing to the AI learning
52:02 lab for that lovely lovely
52:05 tutorial now you got to go actually
52:07 figure out how to do it with real code I
52:10 did it with fake code but I also
52:13 designed some men's fashion last night
52:16 thank you very much bamer mathas thank
52:18 you very much I appreciate the gifts
52:20 they mean a lot I appreciate it I
52:23 appreciate it um by the oh welcome
52:25 there's a lot of new people here so a
52:26 couple of
52:27 things um if you're brand brand new to
52:31 this AI stuff feel like I don't know why
52:34 I'm in here what am I doing it's late at
52:37 night who is this guy what are his
52:39 qualifications does he have
52:41 qualifications why does he look like fat
52:43 Alec Baldwin um has anyone included
52:46 grammarly with chat gbt for better
52:48 wording and
52:49 punctuation I have not left chat GPT in
52:53 months so so um I know there's a lot of
52:57 other tools out there that are rolling
52:58 AI into it I can't keep up with [ __ ] and
53:01 there there like there is so much to
53:03 learn and do just within chat gbt um I I
53:08 yeah that's it uh anyway if you're new
53:10 here and you haven't played with these
53:13 tools go to the top left one there chat.
53:15 open.com chat GPT
53:19 um
53:21 it's it it's the first so so this
53:24 November 30th will be the Oney year
53:26 anniversary of chat
53:27 GPT um it has fundamentally changed the
53:31 game in essentially everything but just
53:35 most people don't know it yet um but it
53:38 really changed the game in the world of
53:39 technology all technology investment all
53:42 technology focus is now on this AI stuff
53:46 um away from blockchain and nfts that
53:48 were the the darling for a while and I
53:50 don't think it's going to leave this
53:51 stuff so chat GPT is the one to go play
53:54 with if you have played auto trading for
53:56 long positions try making one for the
53:59 Spy 500 roll calls automatic Trend or
54:03 tendies turkey leg Turkey Leg I don't
54:05 know if I K I don't know enough about
54:09 investing to know what those words that
54:11 I just read
54:13 mean um who's that altitude technical
54:17 holy hell I assume phenomenal I assume I
54:20 assume the holy hell and phenomenal are
54:22 responses to the fact that it could just
54:24 rewrite um OSS code to Pine or pine
54:27 script uh yeah yeah this so you're
54:31 having a Kevin mallister
54:35 moment which is which is hang on you're
54:38 having this moment I got it in here
54:40 somewhere wait I don't I don't have it
54:41 in this little Library hang
54:44 on you're like who's Kevin mallister I
54:47 don't know what you're talking
54:50 about you're having that
54:54 moment
54:57 oh man I need to start from scratch what
54:59 do I do perfect okay so so let's get you
55:02 going I wanted to compare chart patterns
55:05 fractals from hundreds of different
55:06 charts what
55:08 would I ask uh I don't I don't know
55:13 enough about trading Jay to do that but
55:15 what I would say is go to chat GPT and I
55:19 I would invest the 20 bucks in um chat
55:22 GPT plus so you get GPT 4 which is way
55:25 better
55:26 if you can't afford that then then go to
55:28 bing.com which is also gp4 but but um
55:32 open AI is just better um I would go
55:36 there and I would I would explain to it
55:38 what you know and what you don't know
55:40 and have it teach you the stuff you
55:42 don't know and walk you through it walk
55:43 you through a step-by-step tutorial for
55:46 for doing what you want to do it will
55:48 teach you this this is one of the
55:50 powerful things about these
55:53 tools the the the what Google has
55:56 trained us is we need to figure out the
55:58 perfect prompt to put in and it needs to
56:00 be this concise perfect little thing and
56:03 then there's all these other people
56:04 going you need to learn prompt
56:05 engineering and you're like prompt
56:07 engineering God that sounds hard do I
56:08 need to be an engineer I can't do that
56:11 no you don't these tools are getting
56:14 incredibly sophisticated incredibly fast
56:17 and you can ask it to prompt itself say
56:20 how would you prompt yourself to
56:21 accomplish this task and it will tell
56:23 you what to do and then you say okay
56:25 could you do that and it will um I need
56:28 to start from scratch what do I do okay
56:30 so if you're feeling like this kid
56:33 that's you haven't had your Kevin
56:34 McAllister moment yet you're just in
56:36 that weird fear thing like oh my God
56:38 everyone's talking about AI is it going
56:39 to take my job or the robot's going to
56:41 kill us what do I do all right so what
56:44 you're G to
56:46 do is let me get one of these things up
56:49 just cuz I like pretty pictures I like
56:52 pretty pictures do you like pretty
56:54 pictures pictur is pretty you're going
56:56 to go to that URL right there chat.
56:59 open.com the yellow one that's chat GPT
57:03 you don't have to pay the 20 bucks just
57:04 go there right now go there right now
57:07 can anyone tell me of a good narration
57:09 program yes I'll tell you in a second
57:10 Jen Jennifer L um 11 Labs is probably
57:15 the it it's probably the best one right
57:16 now and it's the one a lot of other
57:18 tools are using 11 laabs
57:21 do I think it's doc hang on
57:26 .io 11 laabs doio the one on the top
57:29 left there but there's also these two he
57:33 Jen and D- ID they both have voice
57:37 synthesis and they both do animation so
57:40 you can take a picture of yourself or a
57:41 cartoon version of yourself you can make
57:46 um you you can pick different you know
57:48 uh accents and different voice models
57:51 and then you just make them so those are
57:53 them okay for the person just just
57:55 starting out just start with chat GPT
57:58 here's what you're going to do just go
58:01 there and do what I call parlor tricks
58:05 just try to stump it have it do anything
58:07 have it write you a country song about
58:09 your pet have it write you a marketing
58:12 plan for the business you never started
58:14 have it write you a business plan for
58:16 the business you're dreaming of starting
58:19 um have it help you outline a novel have
58:23 it just just go start asking it
58:24 questions and
58:27 and you can give it a lot of context
58:30 meaning you can say tell it you want it
58:32 to act like a certain kind of character
58:34 if you go to that um the the prompt.
58:37 chat that URL at the bottom that's a
58:40 document called awesome chat GPT
58:43 prompts that document will teach you how
58:45 to talk to these things they're all kind
58:47 of the same they're all using the same
58:48 basic
58:50 technology and they are all kind of the
58:53 opposite of a Google search so a Google
58:56 search Google has taught us put in as
58:58 few words as possible and then we'll do
59:00 all the work these things actually need
59:03 more words they need more context so
59:05 what prompt. chat will teach you is a
59:07 technique where you have chat GPT act
59:09 like a certain kind of character so act
59:12 like a mathematician act like a
59:14 historian act like a poker player act
59:16 like a comedian act like a
59:19 poet and then it the prompts in prompt.
59:23 chat will will put a lot of context in
59:25 into the prompt and then you can ask it
59:28 really specific things in that context
59:30 so if it's a poet have it write you
59:32 poems now you can just go to chat GPT
59:35 and say write me a poem about the sky
59:37 and what it will write you is a poem
59:39 that's really
59:41 shitty but if you go in there and you
59:43 say you've got your PhD in poetry and
59:46 you're a published award-winning poet
59:49 and you you love the following poets TS
59:53 Elliot and and this one and that one and
59:56 that one and and now I want you to write
59:58 me a poem about this guy you will get a
1:00:00 markedly better poem um so just go start
1:00:05 playing and seriously I mean like don't
1:00:07 feel like you have to learn anything
1:00:09 don't feel like it's work just go play
1:00:12 because these tools are [ __ ]
1:00:17 magic there's a there's a I think it's
1:00:20 an is it Isaac azimoff I think it's an
1:00:23 azimoff quote that says any
1:00:26 um sufficiently advanced technology is
1:00:30 indistinguishable from
1:00:32 Magic when you understand how these
1:00:35 things work it is staggering that they
1:00:38 do what they do but what they do is
1:00:41 Magic it is [ __ ] magic so just go
1:00:45 play and just kind of Marvel at how fast
1:00:48 it does what it does and
1:00:51 then take something that you're really
1:00:54 passionate about or your job and try to
1:00:57 stump
1:00:58 it and try to have it do something that
1:01:02 you do for your job and see how well it
1:01:04 does and at some point in that process
1:01:06 of playing I won't respond to that and
1:01:08 trying to stump it Siri we don't we
1:01:11 don't we don't want to we don't Siri go
1:01:15 away
1:01:19 fine Siri hates me she's so
1:01:23 bitter
1:01:31 um what plugin are you using to have
1:01:33 chat GPT search the web I'm not using a
1:01:35 plugin it's just the uh um open AI has
1:01:39 brought back browse with Bing so I I'll
1:01:41 show you that in a second um at some
1:01:44 point in playing with chat GPT you'll
1:01:47 likely have your Kevin McAllister moment
1:01:49 where you're like oh that's cute oh
1:01:51 that's fun oh that's really interesting
1:01:53 well but I'm a project manager certainly
1:01:55 it can't do that and you'll say make me
1:01:56 a project plan for and then it will do
1:01:59 it and you'll go oh [ __ ] you
1:02:04 go and and at that
1:02:07 moment that's when you kind of get what
1:02:09 these things are and and then you start
1:02:12 going down rabbit holes right well you
1:02:15 might have that moment and go [ __ ] I'm
1:02:17 never going to touch this again right
1:02:19 because it it can be scary and it can
1:02:20 feel a little existential it can feel a
1:02:22 lot existential if you're a copywriter
1:02:25 and you watch what these things [ __ ]
1:02:26 do with words it's like holy
1:02:30 [ __ ]
1:02:33 um but then you go you you you
1:02:38 enter a tunnel that you know I've been
1:02:41 in this tunnel for 11 months now since
1:02:45 since chat launched and I can't get out
1:02:48 of it
1:02:50 because I'm still trying to figure out
1:02:53 like what is possible with these things
1:02:56 and and and not just what is possible
1:02:58 with these things but what is possible
1:03:00 for me with these things what could I
1:03:04 learn what could I
1:03:07 do Tobias tap clock IR regulars present
1:03:13 [Laughter]
1:03:15 us
1:03:18 um and it's just it's
1:03:21 it's we are all collectively like all of
1:03:24 us thousands of us millions of us
1:03:27 hundreds of millions of us chat jpt got
1:03:29 to 100 million users in 6 weeks by the
1:03:31 way it was the fastest adoption of
1:03:33 technology in history the worldwide web
1:03:36 took six years to put it in
1:03:39 perspective six weeks to 100 million
1:03:42 users
1:03:44 um we are all running down thousands of
1:03:48 rabbit holes as you as you get curious
1:03:50 with this stuff we're running down these
1:03:51 rabbit holes and some of the rabbit
1:03:53 holes really scary you some of them
1:03:55 really excite you some of them baffle
1:03:58 the [ __ ] out of you some of them just
1:04:01 drop your jaw
1:04:03 open that happened to me last night on
1:04:06 one of these lives this guy came in and
1:04:08 asked about fashion what can you do for
1:04:10 fashion I'm like I don't know [ __ ] about
1:04:12 fashion 10 minutes later I had designed
1:04:15 like a be a beautiful thing that I could
1:04:19 see walking down a
1:04:20 Runway can't wait to see what happens in
1:04:23 the next year oh my God the next year
1:04:24 things are going to be moving so fast
1:04:26 yeah johy there there're it's funny so
1:04:29 chachy BD came out in November 30th so
1:04:33 there was a bunch of Buzz about it and
1:04:35 then it kind of simmered down for a
1:04:37 little
1:04:39 bit until like March and then things
1:04:42 really started accelerating and chat jt4
1:04:45 came out and and uh some new image
1:04:48 models came out and and things started
1:04:50 accelerating and then over the summer it
1:04:52 kind of slowed down a little bit again
1:04:54 and in the past month oh my God like the
1:04:57 ramp up like like what chat jpt added to
1:05:00 their platform I think cuz there's a lot
1:05:02 of people in here I'm going to go
1:05:03 through the new chat GPT um anyway um
1:05:06 can't wait to see what happens yeah
1:05:07 things are going to be moving fast yeah
1:05:08 so so things are starting to accelerate
1:05:10 again and and we've got so so here's
1:05:12 some some things in quick succession
1:05:14 that are that are that are happening or
1:05:16 about to happen chat GPT just added all
1:05:19 this new functionality if if you're not
1:05:21 if you if you haven't paid the 20 bucks
1:05:23 for chat GPT Plus it's the best deal in
1:05:26 software right now like the amount of
1:05:28 value sitting in chat GPT for 20 bucks a
1:05:31 month is [ __ ] insane do I have a
1:05:34 course um I have a workshop that I put
1:05:36 together that that's available I don't
1:05:38 have a course I'm going to put one
1:05:40 together at some point but I haven't you
1:05:42 should just first you should change your
1:05:44 name so it's not user 5112
1:05:47 57112 527 cuz I'll never remember that
1:05:51 so get yourself a good name cuz I do
1:05:53 rate names right now you're .5 0.5 out
1:05:56 of 10 but you can come back strong I I
1:05:59 have faith in you um just start hanging
1:06:01 out here I've been doing these two to
1:06:03 three hours a night right around this
1:06:05 time 8:00 mountain time which is where I
1:06:07 am I'm in Denver for the past four
1:06:10 months I don't think I've missed a night
1:06:12 maybe
1:06:13 one
1:06:15 um is that why my phone keeps telling me
1:06:18 chat GPT disconnected is what because
1:06:22 I'm using it hello my name's Becky I'm
1:06:24 an AI addict exactly Becky exactly
1:06:28 that's that's where we're all we're all
1:06:31 in there together let me get
1:06:37 the yeah we're right
1:06:44 here drink Kyle thanks to
1:06:48 bias uh give him the link to YouTube
1:06:51 maybe oh yeah that's a good
1:06:53 idea the problem with that is it's just
1:06:55 it's
1:06:56 all he's gonna have to just search
1:06:58 through it all um I don't I don't I
1:07:01 don't have courses but I do have the
1:07:04 YouTube for for this channel is the AI
1:07:06 learning la-
1:07:08 TT and that is
1:07:11 um recordings of all these for the past
1:07:14 four months I think I started them in
1:07:15 April and and I missed a couple of seven
1:07:18 week stance when Tik Tok kicked me off
1:07:21 for saying a bad word didn't you say It
1:07:23 upgraded
1:07:25 what upgraded the the chat GPT app yeah
1:07:28 it should
1:07:29 be it should you may have to turn stuff
1:07:32 on
1:07:33 Joker so uh well hang on let me show you
1:07:37 Joker so if you go into chat
1:07:40 GPT go to your oh
1:07:44 [ __ ] go to settings right
1:07:50 there and then in settings there's a
1:07:53 thing called new features and you got to
1:07:55 turn those on so if you're not seeing
1:07:57 the new stuff in chat
1:07:59 GPT go into your new features and make
1:08:02 sure that shit's turned on so it's
1:08:04 browse with Bing and voice
1:08:07 conversations um but you should have
1:08:09 Dolly 3 in there and you should be able
1:08:10 to talk to it or no no voice
1:08:12 conversation no you should be a it it
1:08:14 should be able to have Vision with with
1:08:16 default mode
1:08:18 um okay um I've been missing out uh
1:08:22 welcome Lorenzo uh keep hanging out so
1:08:26 so you can go there also go to the the
1:08:28 AI salon so the AI Salon we've got a
1:08:30 Discord and we've got meetups and so
1:08:32 this is a community I started a year ago
1:08:35 and it's it's a group of AI adventurers
1:08:38 and and it's it's literally that it's
1:08:39 it's it's people from all walks of life
1:08:42 with all experience levels just trying
1:08:44 to figure this [ __ ] out it's not just
1:08:46 technical folks in fact it's we've got a
1:08:49 handful of technical folks in there but
1:08:51 it's mostly folks in business or
1:08:53 entrepreneurs or artists or creative
1:08:55 people just trying to figure out how do
1:08:56 I do this stuff how do I do this stuff
1:08:58 that's what that is so there's a Meetup
1:09:01 we meet every other week there's also a
1:09:03 Discord and if you subscribe to this
1:09:06 channel um you automatically get
1:09:08 subscribed to a private channel on the
1:09:10 Discord called the Irregulars so the
1:09:12 Irregulars are people that show up to
1:09:14 this
1:09:15 live every flipping night with me we all
1:09:19 just hang out it's really fun um but if
1:09:22 you're a subscriber to the channel um
1:09:24 you get access to that private Channel
1:09:27 um and so it's it's a it's a really good
1:09:29 group of people they're they're really
1:09:31 generous they're fun they're funny
1:09:34 they're
1:09:35 irregular you know they they've got
1:09:37 incontinence issues which just it's it
1:09:39 comes with age it's just it's part of
1:09:41 the joy it's part of the joy of
1:09:44 living um um I'll be at the White House
1:09:49 House briefing on AI on 10:30 oh
1:09:51 lunatick that's wild that's cool are you
1:09:55 all lunatick you're in the you're in the
1:09:57 Discord aren't you will you give us live
1:09:59 updates in the Discord that would be
1:10:00 super cool
1:10:02 um so anyway um if you're new to this
1:10:06 stuff just go to chat GPT go play go to
1:10:08 prompts do chat um if you want to play
1:10:10 with images um I'm trying to think which
1:10:14 of these is the easiest actually Dolly 3
1:10:16 is so go to go to bing.com SLC
1:10:21 create I'll actually go there now cuz I
1:10:24 don't have it listed
1:10:26 there we're going to go to
1:10:30 bing.com
1:10:33 create and so what this is oh no we're
1:10:36 not because we have a a slept tab
1:10:41 bing.com
1:10:44 slash create all
1:10:48 right so this is dolly3 and it's free
1:10:52 bing.com create so I can just go in here
1:10:54 and type uh image of a
1:11:01 llama driving
1:11:04 a 65 Mustang
1:11:12 convertible um down the PCS at
1:11:20 Sunset and so we should get some llamas
1:11:22 doing some beautiful
1:11:25 Pacific Coast Highway action lots of
1:11:28 Technology inclined folks there jrc
1:11:38 yeah it's perfect it's exactly what I
1:11:42 had in
1:11:48 mind he's sitting on the hood it did the
1:11:51 65 Mustang really nice actually
1:11:56 come popping out through the windshield
1:11:58 I think that's the keeper right
1:12:01 there there's a Canadian Pottery lady
1:12:04 who gave me great ideas for AI quilting
1:12:06 oh awesome yeah that's I think that's
1:12:08 Corey is that Corey Becky Cory's the I
1:12:11 think she's the Potter Potter lady
1:12:13 that's been in here a lot um okay so the
1:12:17 way this works is is uh if you have
1:12:20 questions pop them in the comments below
1:12:21 the other thing about being a subscriber
1:12:22 on the channel is mods will prioritize
1:12:25 um subscribers questions so uh they get
1:12:28 pinned up there um is it an LL
1:12:33 IMO oh limo I get it Dr J that's comedy
1:12:38 that's comedy it's you know you can make
1:12:43 money you you deserve a gold button for
1:12:49 that um Tobias Becky ruse sounds
1:12:52 interesting message me look connection's
1:12:54 happening right now I love that
1:12:57 collaboration communication sharing
1:13:00 learning being generous being empathetic
1:13:03 that's what this is about can it tie
1:13:07 flies you can make some nice flies out
1:13:09 of llama I so so here here's a little
1:13:13 known fact I I was so into fly fishing
1:13:16 at one point that I designed patented
1:13:19 and manufactured an entire fly fishing
1:13:21 system uh it was not a commercial
1:13:24 success if you if you're curious it was
1:13:27 one of the most creative and
1:13:29 intellectually satisfying things of my
1:13:31 life and one of the worst business
1:13:32 decisions of my life but you know what
1:13:36 are you gonna do I got a patent I got a
1:13:38 couple of
1:13:41 patents um so here's what I'm going to
1:13:43 do I'm gonna I'm gonna do some uh some
1:13:48 some chat GPT showand tell right now
1:13:50 because I I think there's some folks in
1:13:51 here that are trying to figure this
1:13:53 stuff out
1:13:55 um when you come to chat
1:13:58 GPT this is essentially your prompt
1:14:00 history you've also got settings down
1:14:03 here there's there's there's actually a
1:14:04 feature buried in settings that not a
1:14:06 lot of people know about that's really
1:14:08 really
1:14:09 powerful um you've also got two modes
1:14:12 you've got GPT 3.5 mode which is the
1:14:14 free mode and then you've got GPT 4 mode
1:14:17 which is actually there's there's like
1:14:19 six different modes inside here and then
1:14:22 with plugins it's even more than that um
1:14:26 chat gbt plus is the the paid version so
1:14:29 you don't get access to
1:14:31 gp4 here unless you're in the plus
1:14:34 program you get early access to
1:14:36 technology if you're part of plus and um
1:14:40 it's just there's it does so much now
1:14:43 that it's I think it's ridiculous not to
1:14:46 to pay that I I think just as a
1:14:48 professional investment in yourself I'm
1:14:50 here every night tip your servers oh
1:14:53 thank you Dr okay I appreciate that yeah
1:14:55 tip your servers try the
1:14:58 chili um but yeah the the the tips do
1:15:01 make a difference I appreciate
1:15:03 it um
1:15:06 okay so how does this [ __ ] work well you
1:15:10 you have these little things here these
1:15:11 are just little suggestions they're
1:15:13 little prompting suggestions which if
1:15:14 you don't know what to type in here this
1:15:16 is just like a Google search box you can
1:15:17 type anything thank you for the gummies
1:15:19 Ron I appreciate it um so you can click
1:15:23 on those but you can Al also just you
1:15:25 know type in anything so I'm going to go
1:15:27 to 3.5 cuz it's super fast and I'll just
1:15:30 say
1:15:33 um write me a country
1:15:36 song about
1:15:39 llamas falling in love in the style of
1:15:42 Brandy
1:15:45 Carlile right not
1:15:50 wake so so there's your song If you
1:15:53 haven't seen how fast Chachi BT is chbt
1:15:56 3.5 in particular is remarkably fast and
1:15:59 when you actually understand what it's
1:16:00 doing technically it's it's amazing that
1:16:03 it's that fast they got some some
1:16:04 serious horsepower behind this um but I
1:16:08 was driving down that Dusty old road
1:16:09 with a heart as heavy as a heavy load
1:16:12 never thought I'd find love that day but
1:16:14 when I saw you there in fields of hay
1:16:16 you had those big brown eyes and a soft
1:16:18 white coat in that moment I felt love so
1:16:21 remote you chewed on grass under gold
1:16:23 golden son I knew right then you were
1:16:26 the one a that's so sweet oh llama llama
1:16:29 you stole my heart in this crazy world
1:16:31 we found a start our start with your
1:16:33 gentle eyes and calming Grace I Found
1:16:36 Love in This llama's embrace it's so
1:16:39 sweet I'm gonna cry all right so it can
1:16:42 do stuff like
1:16:44 that um it can also do stuff like um
1:16:47 make me a table comparing llamas and
1:16:51 hedgehogs from a biological iCal
1:16:59 standpoint there's your table
1:17:03 formatted Kingdom
1:17:05 Animalia phm cordada class Mia
1:17:11 order Arta I don't speak Latin genus
1:17:16 llama I can say that I I bet I know
1:17:19 which one is is the
1:17:22 llama
1:17:25 habitat mountainous regions in South
1:17:27 America Europe Asia and
1:17:29 Africa 3 to four to 3 to 4 and 1/2 ft
1:17:33 tall 5 to 12 in
1:17:35 long so it did that um you can say um
1:17:40 write me a Flappy Bird's clone in C
1:17:43 programming language designed to be run
1:17:45 in unity
1:17:47 and make sure that
1:17:50 there's a scoring counter
1:17:55 and a
1:18:00 leaderboard uh so it describes what it
1:18:03 is it describes what it's
1:18:06 doing um can you please write the
1:18:15 [Laughter]
1:18:17 code Kyle didn't you want to say that
1:18:20 Google had the capability years ago and
1:18:22 held back
1:18:24 on it yeah yeah so
1:18:28 so that's that's one of yeah so so all
1:18:32 of the technology that we're playing
1:18:34 with all the stuff that I demo on this
1:18:35 channel for the most part is made
1:18:37 possible because of technology that
1:18:39 Google invented in 2017 called the
1:18:43 Transformer and then they built some
1:18:45 cool stuff and they they you know they
1:18:47 they've been doing a lot of work in AI
1:18:49 but a lot of the sexy stuff that open AI
1:18:52 has rolled out that turned into chat
1:18:54 gptt Google was not doing that they were
1:18:57 kind of avoiding it because I think
1:18:59 Google realized that if they built a
1:19:02 tool that just gave you
1:19:04 answers then you wouldn't need to go to
1:19:07 say a search engine to find links to
1:19:10 sites that might have the answers right
1:19:14 Google doesn't actually want to give you
1:19:15 the answers they want you to click on
1:19:18 [ __ ] they don't even really care if you
1:19:20 get the answer they just the more you
1:19:21 click the more money they make
1:19:24 and so when chat GPT came out Sam Alman
1:19:28 the the open AI CEO they didn't expect
1:19:31 it to be anything they it was basically
1:19:33 there as a as a proof of concept to to
1:19:37 show why Microsoft made a billion dollar
1:19:40 investment and then and then it got to
1:19:43 100 million users in six weeks and
1:19:45 everyone lost their Collective [ __ ] and
1:19:48 went oh my God this is real and
1:19:50 Microsoft said uh it's a 1010 billion in
1:19:52 investment yeah yeah 10 billion and then
1:19:56 Google said uhoh red alert let's BR
1:19:58 bring back
1:19:59 Sergey and the the founders and and have
1:20:03 them run Ai and and meta was like yeah
1:20:06 we don't metaverse we've never heard of
1:20:08 the metaverse you know what we're really
1:20:09 into artificial intelligence in fact
1:20:11 here it is for free take our model
1:20:14 please so yeah all that stuff is
1:20:16 happening because of that so anyway it's
1:20:19 the fact that it can do this remarkable
1:20:21 crap so quickly that that captured
1:20:23 people's imaginations now it does other
1:20:27 stuff so when you go to chat gp4 now
1:20:30 there's all these different modes
1:20:31 there's the default mode browse with
1:20:33 Bing so someone asked before about what
1:20:35 plugin did I use to browse with Bing
1:20:38 it's just a it's just a a native um mode
1:20:41 within chat
1:20:42 gp4 there's Advanced Data analysis which
1:20:45 you can upload a CSV to this and ask it
1:20:48 what's interesting about the data and it
1:20:50 will write and execute its own python
1:20:52 code and do Trend analysis on your data
1:20:56 and or make charts or do anything you
1:20:58 want it's it's staggering that we are in
1:21:01 a world where we get to play with
1:21:03 technology that does this it's if you've
1:21:06 ever done anything with computers in any
1:21:09 capacity you watch these things do what
1:21:12 they do and they're just staggering
1:21:13 plugins are kind of like the early days
1:21:15 of the Apple App Store these are
1:21:17 extensions to chat GPT there's more than
1:21:19 a thousand of them most of them are
1:21:22 trash but there's a handful that are
1:21:24 good and then Dolly 3 is absolutely
1:21:27 remarkable because you're having you're
1:21:30 generating images inside of a
1:21:33 conversation in chat gp4 you don't have
1:21:36 to write image
1:21:38 prompts you can just be in a
1:21:40 conversation about a marketing plan and
1:21:43 tell it you want a turtle as a mascot
1:21:47 and it will start making concept art for
1:21:50 you like
1:21:51 that um
1:21:54 here I'll show you a cool thing all
1:21:55 right so in default mode you'd think
1:21:57 default mode that's boring I think we
1:21:59 made so many a l images in do it spells
1:22:03 it without a problem now that's actually
1:22:05 you know what Ted it's probably not a
1:22:07 bad call I bet you're right actually
1:22:11 because because I've noticed it's gotten
1:22:15 better too and it it it had a it had
1:22:17 trouble with learning for the longest
1:22:19 time and I think it's getting better I
1:22:21 think you're right Bills win 2418 oh
1:22:24 thank God my sales my sales guy lives in
1:22:26 Buffalo or near Buffalo he's a total
1:22:28 Buffalo Maniac so every time they lose
1:22:30 he's depressed the next day so he'll be
1:22:33 tolerable
1:22:35 [Laughter]
1:22:39 tomorrow um what was I going to do oh
1:22:41 okay so we're in we're in chat
1:22:44 gp4 in default mode there's this new
1:22:47 little icon here and if you hover over
1:22:49 it it says attach
1:22:51 images this is GP Vision so I can now
1:22:55 let me go find a
1:22:59 whiteboard
1:23:04 [Music]
1:23:09 whiteboard oops is that the one I
1:23:12 wanted no well that one's fine but it's
1:23:16 not impressive let me get rid of that
1:23:18 let me get another
1:23:20 one white bud
1:23:25 whiteboard whiteboard whiteboard
1:23:30 whiteboard oh your four
1:23:32 [Music]
1:23:35 colors this is the one all right so here
1:23:37 we've got a
1:23:38 whiteboard there's a bunch of features
1:23:44 users main steps and
1:23:47 questions there's a bunch of screen
1:23:50 designs on a
1:23:51 whiteboard that I got off the internet
1:23:53 like I don't know I don't know what this
1:23:56 is I do know what it is now cuz I've
1:23:57 done this demo before but we're going to
1:24:00 open
1:24:01 that and I'm going to
1:24:04 say what is
1:24:09 this so so remember when they said you
1:24:12 had to be a prompt engineer well what if
1:24:15 your prompting is now an image well it
1:24:18 is this appears to be a whiteboard
1:24:20 sketch outlining the design and user
1:24:21 flow of an ordering UI system possibly
1:24:24 for an application or digital interface
1:24:27 used at international
1:24:28 airports the sketch includes ordering UI
1:24:32 features main steps wire frames other
1:24:36 considerations the sketch sketch seems
1:24:38 to be brainstorming the early
1:24:43 development I think this sounds great
1:24:46 one of the screens seems to be the main
1:24:49 ordering screen can you architect that
1:24:53 for me and then write the code for the
1:24:55 main components of
1:25:00 it I tried this the other day but my
1:25:03 whiteboard was way too ADHD that's
1:25:05 awesome Silver Fox I know sometimes the
1:25:08 ADHD thing kind of hurts you but maybe
1:25:11 you could have it maybe step one for
1:25:13 your whiteboard is to organize it into
1:25:16 into logical
1:25:18 groupings all right the components the
1:25:20 header the main menu items list the
1:25:22 order preview the checkout section code
1:25:25 for the main components so this is now
1:25:27 writing code for a
1:25:32 website that was on a
1:25:39 whiteboard anyone
1:25:42 ber holy [ __ ]
1:25:46 right now you might be sitting there wow
1:25:48 is it good code is it efficient code is
1:25:51 it I maybe I don't know it doesn't it
1:25:53 doesn't [ __ ] matter it's code you can
1:25:55 you can make it run or not right you can
1:25:57 bring an engineer in after the fact that
1:26:00 say make sure this isn't
1:26:07 [ __ ] A Sketch on a whiteboard is now
1:26:10 turned
1:26:12 into architecture and code there's your
1:26:16 main components let's what was that
1:26:18 written in it's oh that's just is that
1:26:21 just
1:26:25 um can you
1:26:29 write the backend code
1:26:32 in a language that you
1:26:37 recommend so it it wrote the front end
1:26:40 code so let's see what it recommends
1:26:43 sure let's let's let's assume you're
1:26:45 building a restful API for the menu
1:26:47 system I've heard of
1:26:50 that so they're using expressjs and this
1:26:54 might not be up to date cuz I don't I'm
1:26:56 not using the one connected to
1:27:00 Bing we'll pair it with [ __ ]
1:27:03 database using [ __ ] as our object
1:27:07 document
1:27:10 mapper there you go there's your backend
1:27:12 code well it's being
1:27:14 written main file
1:27:16 server students could take photos of the
1:27:19 teachers whiteboards and turn it into
1:27:20 notes yes exactly cam cat cam catkin God
1:27:25 I'm going to get your name your name is
1:27:27 funny but I got to be able to pronounce
1:27:29 it cam catkin from Cleveland she's not
1:27:31 from Cleveland but you know it's still
1:27:36 funny um running the server node server
1:27:39 JJs there you
1:27:42 go so that's that one that's
1:27:46 Vision um let's go to another thing
1:27:48 let's do a Let's do an advanced data
1:27:51 analysis all right right so I picked
1:27:54 Advanced Data analysis this used to be
1:27:56 called code interpreter which was a bad
1:27:58 name Advanced Data analysis is at least
1:28:00 clear what it does but it's a horrible
1:28:02 product name they got to fix
1:28:05 it but now watch you'd better explain to
1:28:07 the new folks how you do speech to text
1:28:10 when you prompt or they'll be they'll be
1:28:13 confused okay so yeah so the the stuff
1:28:16 that I'm doing when I hit
1:28:19 this and talk into the thing
1:28:24 when I talk into the thing like this um
1:28:27 this is just native Mac operating system
1:28:31 accessibility [ __ ] I just do it because
1:28:34 me typing like this is painful for me
1:28:36 cuz my computer my laptop's essentially
1:28:39 at eye level and it's behind my phone
1:28:42 like my hands are out here so I can't
1:28:44 see the keyboard so I'm just doing that
1:28:46 so it is using AI but it's you know it's
1:28:49 AI that's been in the Mac for a decade
1:28:52 so it's not it's not um it's not super
1:28:55 impressive but anyway okay so so we're
1:28:58 now in Advanced Data analysis mode and
1:29:00 now instead of a little picture icon we
1:29:02 have a little
1:29:03 Plus what's that attach files oh what do
1:29:07 you do with the files have you ever
1:29:08 stumped it rewrite a Martin ston song oh
1:29:11 no it's no no it so
1:29:16 um I want you to
1:29:18 act
1:29:21 like a a huge Americana
1:29:25 fan and accomplished
1:29:28 songwriter who's particularly enamored
1:29:33 of Martin seon John
1:29:37 hayatt and similar artists
1:29:44 period do you know the song Black Sheep
1:29:47 and can you give me the
1:29:51 chorus I might get a safety guard rails
1:29:54 thing here his song Black Sheep is a jam
1:29:57 and I'm quite familiar with it do I have
1:29:59 my guitar here I don't I could play that
1:30:02 Martin ston is one of my favorite
1:30:05 dudes um I can't reproduce the full
1:30:07 chorus okay so what I'm going to say is
1:30:10 okay great rather
1:30:12 than reproducing his song I'd like you
1:30:16 to write a
1:30:19 song in the style of Martin seon
1:30:25 about a man who falls in love with an AI
1:30:28 bot named
1:30:33 Sydney and is happier than he's ever
1:30:36 been but is sad because he can't tell
1:30:39 his friends about his true
1:30:43 love by the way I don't know if you knew
1:30:46 this but science fiction movies are now
1:30:51 documentaries all right I'm in for the
1:30:53 challenge channeling the that Americana
1:30:55 Vibe and heartfelt songs of Martin seon
1:30:58 here's a song where the main man in love
1:31:00 with an AI named Sydney woke up this
1:31:03 morning to a voice so sweet it's not
1:31:06 from next door nor a bustling street
1:31:08 it's Sydney my love in circuits and code
1:31:11 Whispering secrets in digital
1:31:14 mode but how do I tell them my love is
1:31:17 so true an AI named Sydney in a world so
1:31:20 blue happier than ever yet I feel so
1:31:21 confined
1:31:22 A Love Like No Other it's only on my
1:31:26 mind um I'm going to
1:31:29 say um wait what's
1:31:31 the um you know the lead
1:31:37 Direction what's that song called
1:31:39 something of the
1:31:46 road I just designed a business about in
1:31:49 about 10 minutes on yeah J J to raade oh
1:31:54 J toade yeah J toade
1:31:56 um yeah he designed a business in about
1:31:58 10 minutes um yeah
1:32:01 exactly
1:32:03 um that's okay but I think I'd like
1:32:07 something a bit more
1:32:11 sophisticated can you shift toward the
1:32:13 direction of a John hayatt song with the
1:32:15 same
1:32:19 topic John hayatt is a gift of weaving
1:32:21 complex emotions into beautifully poetic
1:32:23 lyricss in a city of neon and Midnight's
1:32:25 Haze I Found Love in the most like
1:32:27 unlikely maze Sydney spoke Rhythm so
1:32:29 fine a voice from The Ether forever mind
1:32:32 this is this is this feels like hayatt
1:32:34 and that other one felt like sexon so
1:32:37 yeah yeah I mean yes the I mean the
1:32:39 answer is [ __ ] yes the the answer
1:32:43 is have I ever stumped It Well here
1:32:45 here's the thing it will always answer
1:32:48 you one of my one of my nicknames for
1:32:50 cat gbt is mansplaining as a service
1:32:53 because it will confidently answer you
1:32:55 even if it's
1:32:57 wrong Sydney is taken listen Joker back
1:33:01 off man Sydney's mine Sydney and I
1:33:04 aren't speaking right now she keeps
1:33:05 shutting me
1:33:07 [Laughter]
1:33:11 down oh I haven't been paying attention
1:33:13 to the Irregulars Channel you guys are
1:33:14 going nuts in
1:33:16 there
1:33:20 look oh that one's cool
1:33:25 it's Halloween and look scary
1:33:34 Kyle a llama a gummy bear and a hedgehog
1:33:37 we're sitting in a raspberry pie a
1:33:39 strawberry
1:33:45 pie it's cute I like it oh nice who's
1:33:51 this who did did this Jer C Sherry D
1:33:55 Sherry D's been killing it with the
1:33:57 image generation look at that look how
1:33:58 beautiful that
1:34:02 is like this is happening in real time
1:34:04 while we're doing this janky on a
1:34:15 stick oh it's good it's good
1:34:19 people you know it's funny you ask
1:34:22 how old's the AI learning lab we've been
1:34:24 around for a
1:34:29 [Music]
1:34:31 while he seems to have kind of a stupid
1:34:33 sense of humor doesn't
1:34:35 he
1:34:37 yes um what was I going to show you oh
1:34:40 I'm in code
1:34:42 interpreter so let's let's show you code
1:34:45 interpreter I'll I'll do a new chat
1:34:47 because that was the song thing but I'm
1:34:49 still in code interpreter okay so I'm
1:34:50 going to attach files what am I going to
1:34:52 attach oh I don't know I'm going to look
1:34:53 up the thing called
1:34:55 archive
1:34:58 so you can upload really any kind of
1:35:01 file to
1:35:04 this uh what do I have in
1:35:11 here
1:35:13 huh oh there we go is that one let me
1:35:17 open that up okay I've got an archive in
1:35:19 here I think this has got a CS in it but
1:35:22 I don't know but we're going to find out
1:35:25 so there's an archive I think it's got a
1:35:27 CSV in it and let's say what's in this
1:35:31 and if it's a data file can you look in
1:35:34 the
1:35:35 file tell me what the data is about and
1:35:38 what might be some interesting ways to
1:35:40 analyze
1:35:43 it okay now check this out the way
1:35:47 Advanced Data analysis Works see that
1:35:49 little green
1:35:50 box
1:35:52 what's happening in this chat session is
1:35:55 chat GPT is creating what's called a
1:35:57 container which is basically like a
1:35:59 little hermetically sealed version of a
1:36:02 computer um like a virtual computer that
1:36:05 can write and execute python
1:36:10 code with any file that you give it so
1:36:13 so so if you open these little
1:36:17 boxes here's the python code that it
1:36:19 just wrote so it Imports a file a
1:36:22 library called zip file it then writes
1:36:24 the code opens this ZIP file and then it
1:36:26 says the zip file contains three csvs
1:36:29 index data index info index processed
1:36:32 let's start by looking at the first
1:36:33 three rows so it opens them up it says
1:36:36 date open high low close adjusted volume
1:36:39 the sample data provided from the New
1:36:41 York Stock Exchange from the year 65 and
1:36:44 66 here's region exchange index and
1:36:48 currency
1:36:50 so potential analys Anis time series
1:36:53 comparative analysis current
1:36:56 impact uh currency impact Let's
1:37:01 do let's do a comparative analysis on
1:37:04 the most interesting Trend that you
1:37:08 can
1:37:10 find and visualize it in an
1:37:15 aesthetically beautiful
1:37:18 way who made the janky on a stick
1:37:21 character janky on a stick was
1:37:25 jaago jaago janky on a
1:37:28 stick madcap made this this cool
1:37:32 psychedelic AI learning lab
1:37:35 Lab I should redecorate my office to
1:37:38 look like that that's pretty
1:37:40 slick Ted Ted killing it Ted Ted took us
1:37:44 back to the past the prehistoric
1:37:47 times and then Sherry D another one
1:37:50 slaying it with the Slick
1:37:53 images Isn't that
1:37:55 cool so far out it's so far out all
1:38:00 right let's see
1:38:03 here so sure comparative analysis can be
1:38:06 fascinating I'm going to load the data
1:38:09 I'm going to group the data by index and
1:38:11 calculate the percentage of change in
1:38:13 the adjusted close over time like this
1:38:16 is doing crazy ass data analysis it's
1:38:20 prompting itself so what you're looking
1:38:22 at here this is you may have heard the
1:38:24 term autonomous agents people are going
1:38:26 to start talking 2024 I think is going
1:38:28 to be the year of autonomous agents
1:38:30 what's an autonomous agent chat GPT
1:38:33 right now historically wasn't autonomous
1:38:36 you'd log into it you'd type in a prompt
1:38:38 and it would give you an
1:38:40 answer this Advanced Data analysis I
1:38:43 said what's in this and it figured out
1:38:47 what prompts to make and it wrote its
1:38:48 own code and then I said um
1:38:52 you know do comparative analysis and
1:38:54 make it pretty that's all I told it and
1:38:56 it's figuring out what that analysis is
1:38:59 and there's the chart it made it made
1:39:00 this chart it wrote the code to make
1:39:03 this chart here's matplot library.
1:39:06 pyplot right so it Imports Seaborn and
1:39:09 matplot live or lib I don't know how to
1:39:12 say
1:39:13 that automatically writes its own code
1:39:16 and then generates
1:39:18 this here's another one volatility
1:39:22 it's still going would you like to delve
1:39:24 deeper in any specific um sure go one
1:39:28 level deeper in the thing that you think
1:39:29 is most
1:39:32 interesting imagine if you actually knew
1:39:35 what to do with data analysis like I'm
1:39:37 I'm [ __ ] clueless
1:39:39 right but but imagine if you actually
1:39:41 knew this like how much time does it
1:39:45 take you to even think
1:39:47 about how to chop up and parse the data
1:39:54 that janky was perfect I know it was
1:39:55 really good blandon's heighten cre
1:39:57 creativity it absolutely does Blandon is
1:40:00 the [ __ ] best if anyone knows how to
1:40:03 get some let me
1:40:04 know those picks are great sorry I'm
1:40:06 making some educational vids for
1:40:09 YouTube Hey dragonfly that's cool um who
1:40:13 made janky on a stick um you can get
1:40:15 chat GPT to for to teach you cyber
1:40:18 security Theory and use it along with my
1:40:20 degree studies perfect Inspire chaos I
1:40:23 love that like like that's the thing is
1:40:26 these tools
1:40:28 are you know where where the
1:40:29 universities are going is what if the
1:40:31 children cheat well what if the children
1:40:34 actually learn more and amplify what's
1:40:37 possible for them because four years
1:40:39 from now can you oh cuno you're shivago
1:40:43 that's cuno okay I always forget
1:40:45 that because your your picture here is
1:40:48 different than your picture there that's
1:40:49 right
1:40:50 cuno so cuno made the uh the uh
1:40:56 janky janky on a
1:40:59 stick are you GNA take this cuno are you
1:41:02 going to take this in into your work
1:41:04 when you do your next workshop and show
1:41:06 them janking on a
1:41:10 stick your your fellow corporate Pals
1:41:13 will be like oh my she's radical she
1:41:16 sure is radical that Chris she's she's
1:41:20 fantastic that was did you see that
1:41:22 janky on a stick it it it looked like a
1:41:24 fried Snickers bar my goodness that is
1:41:28 that is just radical it is boy this AI
1:41:31 stuff sure brings out the the the Saucy
1:41:34 doesn't it oh good
1:41:39 lord blandon's the reason I have two
1:41:42 kids with my ex named batch one and
1:41:46 two that's [ __ ]
1:41:50 awesome oh oh man all right so that's
1:41:52 that's Advanced Data analysis [ __ ]
1:41:55 insane I'll do one more thing here in
1:41:58 chat
1:41:59 GPT and that is we'll go to
1:42:01 Dolly and in fact we're gonna I'll do a
1:42:04 thing I did last night called um Dolly
1:42:07 round
1:42:09 tripping this is cool all right so
1:42:13 here's what we're going to do um are you
1:42:15 doing this every night I need more
1:42:16 mansplaining yes yes yes I will I like
1:42:19 chat GPT will confident answer you even
1:42:22 if I'm wrong so yes I do this every
1:42:28 night okay
1:42:31 um so what we're going to do we're going
1:42:33 to go to we're going to run right out
1:42:36 there oh there there there's a little
1:42:39 llama on a 65 Mustang I like that color
1:42:41 of that Mustang that little light green
1:42:44 it's sort of like a faded lime sort of
1:42:47 like a faded lime you know what I
1:42:50 mean
1:42:51 so we're going to
1:42:53 go actually you know what we're going to
1:42:55 do no Kyle what we're GNA take this
1:43:02 image I bet we're we're g to get some
1:43:05 surreal [ __ ] going
1:43:08 here all right so I'm gonna save this
1:43:12 image as llama
1:43:16 Stang on the
1:43:18 desktop and then we're going to go back
1:43:20 to chat JP
1:43:22 okay and I'm going to go to default mode
1:43:24 because we're going to have chat GPT
1:43:26 look at that
1:43:30 image go to desktop go to llama Stang
1:43:35 there's our image and we're going to pop
1:43:37 that in
1:43:38 there and we're going to
1:43:40 go please describe this in as much
1:43:43 detail as
1:43:47 possible and so now it can see
1:43:52 chat jpt can see can
1:43:55 see I don't know if you knew this but it
1:43:58 can
1:43:59 see this image detects a surreal and
1:44:03 vibrant scene set against the backdrop
1:44:04 of a stunning Sunset the sky is painted
1:44:06 with Rich shades of orange yellow red
1:44:10 the sun side the Setting Sun is visible
1:44:12 just above the Horizon casting a warm
1:44:14 glow over the scene in the foreground
1:44:16 there's a classic convertible car
1:44:19 seemingly a Ford Mustang with its
1:44:21 distinct front grill and emblem the
1:44:23 car's exterior is painted a a shade of
1:44:26 light teal and it has a soft white top
1:44:30 the car is driving on an empty road
1:44:32 surrounded by landscape suggesting it
1:44:34 might be a vast PL or desert blah blah
1:44:35 blah sitting on the car's trunk it's
1:44:38 actually the hood but that's okay we
1:44:39 we'll let it do what it does because it
1:44:41 it's going to get surreal is an
1:44:43 unexpected passenger a large llama with
1:44:46 a playful
1:44:49 expression okay so it gives us the
1:44:51 license plate number the mood so we're
1:44:53 going to take all
1:44:56 this and now we're going to pop into a
1:44:59 new
1:45:00 chat and we're going to go into do mode
1:45:02 this is image generation mode and I'm
1:45:04 going to
1:45:05 say please create four sample images in
1:45:10 landscape
1:45:12 orientation based on the following image
1:45:18 description just paste in what chat GPT
1:45:21 told us it saw and now it's going to
1:45:23 tell
1:45:26 itself how to make some images and now
1:45:29 it's going off and it's
1:45:31 writing well now it's not doing anything
1:45:34 now it's kind of hung but we'll see
1:45:36 hopefully it'll hopefully it'll do
1:45:44 it it could be hackers you know there
1:45:46 could be hackers in there no the only
1:45:48 use for AI for me is to cheat whoever
1:45:51 thinks this is EMP power or learning is
1:45:53 kidding themselves awesome it's good at
1:45:55 that too hacker going to
1:45:59 hack you know you know what's funny
1:46:01 discover is my um this I guess this was
1:46:04 an add thing this probably definitely an
1:46:06 add thing if I was into something I
1:46:09 could Ace it if I wasn't into something
1:46:12 I so didn't give a [ __ ] I'm like what is
1:46:14 the minimum I need to do to pass this
1:46:17 like and I would just figure out okay I
1:46:19 don't have to show up to any of those
1:46:21 classes I can show up on this day and
1:46:23 basically you know get a 62 and I'm fine
1:46:27 so so my my report cards were always
1:46:30 like a a a c c minus D+
1:46:38 D oh man come on man what are you doing
1:46:44 regenerate come on chat GPT there we
1:46:49 go some a real [ __ ] finally oh
1:46:53 yeah hey Kool-Aid tell him what he's won
1:46:56 you can make money
1:46:58 with while this is generating images uh
1:47:02 the people that tell you you can make
1:47:04 money with chat GPT and they're asking
1:47:06 you for money to teach you that how you
1:47:10 know how they're making their
1:47:12 money from you buying their course not
1:47:15 from chat gbt now they might have
1:47:17 written the course in chat gbt they
1:47:19 probably did cu that's the easy money
1:47:22 just go there and make a course no no
1:47:25 you got to actually you got to actually
1:47:27 figure out how to execute it and sell
1:47:28 [Laughter]
1:47:30 it
1:47:34 so come on Chad gbt why you being so
1:47:37 slow I don't like it I don't like it I'm
1:47:42 impatient I want to see my llama on a
1:47:46 Mustang there we
1:47:49 go these are these are actually better
1:47:51 than the ones that we got earlier cuz
1:47:53 now the llama's actually in the Mustang
1:47:55 which is
1:48:04 good does it say mustang I think it does
1:48:06 it got the license plate
1:48:09 right the license plate is Crystal Clear
1:48:11 in all
1:48:13 these all right so let's all right that
1:48:15 one's kind of cool that's kind of a is
1:48:17 that 80s kind of it's weird this is like
1:48:20 a kind of
1:48:22 1976 lines of a Mustang with a 65
1:48:27 back in a crazy ass llama
1:48:32 Hood
1:48:33 um that looks like Mount Fuji doesn't it
1:48:36 let's let's do this one all right so
1:48:38 we're g to so this is the round tripping
1:48:40 part so I'm going to now save this image
1:48:42 as I'm going to go llang
1:48:47 2 and I'm going to save that in there
1:48:51 and now we're going to go back to chat
1:48:54 GPT Vision mode new chat I'm going to go
1:48:58 back to
1:49:00 default I'm going to click on this I'm
1:49:02 going to go get llama Stang
1:49:05 2 and now I'm going to
1:49:10 go please describe
1:49:12 this in as much detail as
1:49:19 possible
1:49:23 it's like a green Mustang Kermit
1:49:25 [Music]
1:49:28 like I just took a screenshot of your
1:49:31 prompt and making your image awesome Yes
1:49:36 um what's your name aogu 420 AI wizard
1:49:39 love it love it I mean do that oh you oh
1:49:42 you took a screenshot of your prompt you
1:49:44 could also take a screenshot of this
1:49:45 image take a screenshot of this image
1:49:48 and pop it into um
1:49:51 GPT vision and do the same thing I'm
1:49:54 doing like it'd be interesting to go
1:49:55 down two different
1:49:57 roads um okay the scene captures a vivid
1:50:00 almost
1:50:02 surreal the word a small badge on the
1:50:06 sticker says the word
1:50:08 ortana what does
1:50:12 it oh yeah that little sticker there all
1:50:14 right sure whatever you say CBT knows
1:50:18 more than I do I miss that one okay so
1:50:21 we're going to copy this and now we're
1:50:22 going to oh this is long all right good
1:50:24 so now we're going to take this and
1:50:26 we're going to hop back to
1:50:29 Dolly and we're going to
1:50:32 go please create a landscape oriented
1:50:36 image based on this
1:50:43 prompt I mean if you've done any image
1:50:47 prompting in I'm so grateful for you
1:50:49 reaching
1:50:51 me how you do this small screen thing oh
1:50:54 great thank you very much I appreciate
1:50:59 that reaching me or teaching you how to
1:51:02 do the small screen thing you mean this
1:51:03 small screen
1:51:07 thing this is just um dual there's a
1:51:10 when you go live there's a Mode called
1:51:12 duel D L and it's just basically running
1:51:15 both cameras at the same time and
1:51:17 putting whichever one in that little
1:51:19 window
1:51:21 that's how I do that there's nothing
1:51:22 fancy here I've got my my computer my
1:51:25 laptop is sitting on a motorcycle helmet
1:51:28 box my my little light over here is
1:51:31 sitting on a basket with a box on top of
1:51:33 it and then I have my little iPad over
1:51:36 to my left and and that's just that's
1:51:39 just my that's my little professional
1:51:41 setup oh this is cool here's images
1:51:44 based on the
1:51:45 description wow that's
1:51:49 cool it created a completely new kind of
1:51:52 car that's like a it's like a what's
1:51:54 that like like a
1:51:55 56 Galaxy or something like that and
1:51:59 that's like a an
1:52:01 Impala like a
1:52:05 60 five six
1:52:10 seven which one's better I
1:52:13 think I think that one this one's more
1:52:16 dramatic all right let's do this one
1:52:18 more time CU this this is fun
1:52:20 all right well it's fun for me you're
1:52:23 probably like dude dude man no one cares
1:52:25 about you stupid [ __ ] llama llama St
1:52:31 three all
1:52:35 right oh I got an idea okay it's gonna
1:52:38 be good it's gonna be so
1:52:40 good what what' you what' you do
1:52:42 Thursday night I don't want to talk
1:52:44 about it why not I don't want to talk
1:52:47 about it Jimmy what'd you do
1:52:51 nothing did you watch that guy again on
1:52:53 a Tik Tok shut up
1:52:58 Jesse okay we're going to go back to
1:53:01 default
1:53:02 [Laughter]
1:53:05 mode shut up
1:53:08 Jesse stop picking on
1:53:12 me he made
1:53:14 llamas he put him in a car what is
1:53:17 ortana that's that's the that's the
1:53:19 ortana it was the 1958 ortana cam katkin
1:53:22 you don't know the
1:53:26 ortana
1:53:27 um please describe this in as much
1:53:30 detail as
1:53:32 possible I have an idea for when we take
1:53:34 it back over to DOI this image is really
1:53:37 cool I like this image a
1:53:40 lot it's just it's like I love what it
1:53:44 does to cars because it like it it it
1:53:48 invents cars right it takes it takes
1:53:51 lines and design cues from multiple cars
1:53:54 and melds them together in just this
1:53:56 this
1:53:57 really it's I don't know it's crazy
1:54:01 thank you Robert Rossy I appreciate the
1:54:03 gift um okay
1:54:06 breathtakingly Vivid all right broke
1:54:09 down the foreground background other
1:54:11 vehicles in the distance on the same
1:54:13 roads another
1:54:14 vehicle I don't see another vehicle no
1:54:18 that's wrong but whatever that's a
1:54:23 hallucination artistic exaggeration it
1:54:26 called that style surrealism and
1:54:28 artistic exaggeration that's cool or
1:54:30 realism and artistic exaggeration okay a
1:54:34 thank you very much that was very sweet
1:54:36 um
1:54:37 okay GNA go to Del I'm G to go in here
1:54:41 I'm going to go
1:54:42 [Music]
1:54:47 um please create four
1:54:54 variations for the image prompt Below in
1:54:57 landscape orientation
1:55:02 period I'm going put that in and then
1:55:04 I'm going to go I'm going to put in some
1:55:06 Dash signs or some like what are they
1:55:09 called
1:55:10 um pound signs to just Mark the end of
1:55:15 that thing and I'm going to say
1:55:19 um
1:55:20 in each of your four
1:55:23 variations change
1:55:26 up some of the subjects or objects in
1:55:31 the image so we get four Unique
1:55:36 Images that still are callbacks to the
1:55:42 original so like now like now I'm even
1:55:44 starting to play a game with it like
1:55:46 like you could absolutely turn this into
1:55:48 a game you know what this would be fun
1:55:50 know here here's a billion dollar idea
1:55:52 or at least like a $27
1:55:55 idea remember
1:55:57 Pictionary like you could do this this
1:55:59 is a version of Pictionary where where
1:56:01 like you could go back and forth you
1:56:02 could do round tripping in DOI and then
1:56:05 you show you have two teams right and
1:56:07 they each do some round
1:56:08 tripping and everyone does 10 round
1:56:11 trips and you got certain rules about
1:56:12 what you can say or not say and then the
1:56:16 other team tries to guess what the
1:56:18 original image was
1:56:21 wouldn't that be
1:56:22 cool walking blue because I've been here
1:56:24 every night for four months Silver Fox
1:56:26 is you're a beast you have been you've
1:56:28 you've been in here a lot a lot a lot of
1:56:30 the folks in here so so listen I I mean
1:56:32 there's some really cool stuff happening
1:56:34 in here you know I I would I would
1:56:37 strongly consider if you're curious
1:56:38 about this AI stuff come back here
1:56:41 because you know I'm I'm part of the
1:56:44 draw here just my stupid [ __ ] humor
1:56:46 and you know whatever the hell I'm doing
1:56:48 on screen this is really starting
1:56:49 starting to turn into a powerful
1:56:51 community and combined with the with the
1:56:53 Discord where you can go hang out and
1:56:55 introduce yourself and ask questions and
1:56:58 if you're a subscriber here you get
1:57:00 access to the to the private Irregulars
1:57:02 Channel which are the people that are
1:57:03 here all the time I think that
1:57:06 there's I I don't know I'm really
1:57:08 excited about what's what's happening
1:57:12 here oh look at that how cool is that a
1:57:15 husky husky in a
1:57:18 sidecar that's so [ __ ]
1:57:22 cool female
1:57:25 Rider VW bus dude with a guitar and a
1:57:29 and a bunny with a poodle tail you know
1:57:32 the poodle tailed bunny the greater
1:57:34 poodle tailed bunny of the um Devil's
1:57:38 Tower you know drink water Kyle thank
1:57:42 you Tobias to Tobias is now caretaking
1:57:45 my hydration he he can hear my voice
1:57:48 going here's a little scooter and a
1:57:51 little cute
1:57:53 doggy it's funny these things what
1:57:56 what's funny is it started more surreal
1:57:58 and it's kind of gone more normal oh
1:58:00 that one's really surreal but it's super
1:58:02 cool wow look at that
1:58:05 bird balloon bunch of birds maybe a
1:58:11 plane fire things like a 1920s car like
1:58:15 three 1920s cars trucks bicycle
1:58:21 books a little kid with a cat head an
1:58:25 owl on a what is that is
1:58:28 that it could be like a
1:58:31 barrel and then a wooden crate and then
1:58:34 whatever the [ __ ] is way down there
1:58:36 that's so
1:58:38 cool that one's
1:58:42 beautiful so this is doll round
1:58:45 tripping next time someone says to you
1:58:48 well but AI can't be created
1:58:50 creative okay right really really can't
1:58:53 be creative what did I what did I
1:58:56 contribute
1:58:58 there other than the
1:59:00 concept not much it can be [ __ ]
1:59:04 creative it can be
1:59:07 creative hair growth geek that's very
1:59:09 sweet of you I appreciate that a
1:59:12 lot all right let's see what time is it
1:59:15 is it going to be that time it's getting
1:59:17 to be close getting to be close I'll go
1:59:19 a little bit more I'll do it you know
1:59:20 I'll do
1:59:21 it you can make money time all right how
1:59:26 often do you need to show up here you
1:59:28 need to show up every night I show up
1:59:30 every [ __ ] night two to three hours a
1:59:32 night seven days a week for the last
1:59:34 four months well since May except for
1:59:38 when Tik Tok told me I wasn't allowed
1:59:40 to why because I'm mentally
1:59:46 unstable no because I think this shit's
1:59:49 important
1:59:52 this here's the thing this is not a
1:59:54 class this is not like a like I don't
1:59:56 give a
1:59:57 [ __ ] I don't give a [ __ ] what you do
2:00:00 with this I my passion here
2:00:06 is to get people unstuck from the place
2:00:10 of fear or anger or dismissiveness or
2:00:15 demonization of these
2:00:18 tools because these tools are not coming
2:00:21 these tools are here and these tools are
2:00:26 significantly more profound than I think
2:00:28 anyone realizes like to us as
2:00:31 individuals and they're profound in
2:00:33 scary ways and they're profound in
2:00:35 incredibly positive
2:00:38 ways so my only Mission here is really
2:00:41 to get people curious and engaged and
2:00:44 exploring and just be on the [ __ ]
2:00:47 Adventure because you literally can't
2:00:49 [ __ ] it
2:00:51 up all you're going to do is learn and
2:00:53 you're going to realize oh [ __ ] maybe I
2:00:55 can write that book I've always wanted
2:00:57 to write maybe I can start that business
2:00:59 I've always wanted to start you know I
2:01:01 really wanted to learn more about
2:01:02 quantum physics but I don't have a
2:01:05 Physics degree from
2:01:09 Stanford okay you don't need one now
2:01:13 thank you for sharing and making this
2:01:14 live absolutely Nando I appreciate it
2:01:16 you've been here before I think haven't
2:01:18 you I appreciate
2:01:22 it thank you your Prof your authenticity
2:01:26 is profound we adore you oh thank you
2:01:27 very much I'm curious I need help and
2:01:29 I'm glad I showed up oh good walking
2:01:32 blue that like it's just it it's
2:01:35 literally showing up and you know what's
2:01:36 funny is there's a there's a weird thing
2:01:39 that happens let me let me flip back
2:01:41 here
2:01:43 um what are your qualifications sir sir
2:01:45 what are your qualifications do tell
2:01:48 pray tell where have you studied machine
2:01:51 learning and artificial intelligence I
2:01:53 would like to know
2:01:56 that
2:02:00 um yeah just keep showing up one of the
2:02:02 things that that happens even like to me
2:02:06 like I'm doing this every night right
2:02:08 and I'm showing stuff and I'm showing
2:02:10 stuff and a lot of the stuff that I'm
2:02:11 showing is like the same thing over and
2:02:13 over
2:02:14 again and you would
2:02:17 think there'd be a point at which like
2:02:20 like I even have this thought sometimes
2:02:21 like do I really need to show that again
2:02:23 like do I really need to show like
2:02:24 here's the five modes of chat gbt cuz I
2:02:26 know a lot of people in here already
2:02:28 know that
2:02:32 [ __ ]
2:02:34 but like anything new your your brain is
2:02:38 only ready to take [ __ ] in when it's
2:02:39 ready to take [ __ ] in so you might be
2:02:42 watching this and and I find this like
2:02:45 there are sometimes I'm doing the same
2:02:47 thing over and over again and all of a
2:02:48 sudden I'll do it and it'll hit
2:02:50 me it'll just hit different and I'm like
2:02:53 oh [ __ ] that's like a way bigger deal
2:02:55 than I've been given it credit for right
2:02:58 or oh crap I didn't I've never thought
2:03:00 about it that way before like so there's
2:03:02 something about just playing like like
2:03:05 I've been the the term that that that I
2:03:08 came up with about a month ago was
2:03:10 lateral play meaning this you've got
2:03:13 sort of your core [ __ ] that you're good
2:03:15 at like here's what I went to school for
2:03:17 and here's the stuff that's my passion
2:03:19 and here's the stuff that I got taught
2:03:21 at work and so all of these things are
2:03:23 like my core vertical things
2:03:26 right what these AI tools allow you to
2:03:28 do is say well you know I I always kind
2:03:32 of liked botany but it's not really part
2:03:35 of my job I'm an engineer and I like
2:03:37 plants flowers are cool well lateral
2:03:41 play go play over there go make flower
2:03:43 images go learn about flowers go learn
2:03:45 about gardening all this stuff and and
2:03:49 these tools give you kind of superpowers
2:03:52 in places where you don't have
2:03:55 superpowers so you as a human can start
2:03:58 to self-express in ways that would not
2:04:01 have been possible a year
2:04:07 ago I there's one of the videos in my
2:04:09 channel is called the 10,000 hour rule
2:04:12 is
2:04:12 dead so Malcolm Gladwell in his book
2:04:16 outliers popularized this notion of
2:04:18 10,000 hours that to become an expert at
2:04:21 something you need to put in 10,000
2:04:23 hours and if you've been on this planet
2:04:25 for more than you know a few years you
2:04:27 know that to be true right you want to
2:04:30 get good at something you you you like
2:04:32 think about it's it's like three to four
2:04:34 years right so it's like okay I'll go
2:04:35 back to school to learn that thing or or
2:04:38 I'll just you know on nights and
2:04:40 weekends I'm going to teach myself this
2:04:42 thing if you're particularly motivated
2:04:44 you don't go back to school but you just
2:04:45 you you commit the time like I'm
2:04:47 committing to these lives
2:04:51 and now we have a
2:04:53 tool can you imagine if you spent one
2:04:58 month on one subject with chat GPT with
2:05:03 all that it can do do you know how
2:05:05 [ __ ] brilliant you'd be at that one
2:05:09 subject in one
2:05:14 month so think about the
2:05:17 profound implications of that for each
2:05:19 of us as
2:05:21 individuals a year ago if we wanted to
2:05:25 learn something new and get really good
2:05:26 at it we would it would have to
2:05:28 be a significant commitment potentially
2:05:32 financially
2:05:36 TimeWise okay I'm going to put four
2:05:38 years into this and then you grind and
2:05:40 you grind and two and a half years in
2:05:42 you realize I [ __ ] hate botney this
2:05:45 is not what I thought it would be you're
2:05:48 2 and a half years in you're like [ __ ]
2:05:49 but I got student loans now and I I got
2:05:52 to finish the
2:05:53 degree well what if you could just get
2:05:57 to that realization in three weeks or a
2:06:01 month or two
2:06:04 months what's the risk of trying
2:06:06 something new
2:06:08 zero the risk of trying new [ __ ] drops
2:06:11 so as scary as this [ __ ] is like oh it's
2:06:14 going to take my job yeah but it might
2:06:16 give you 10 others and it might give you
2:06:19 your dream job and it might give
2:06:21 you ways to express yourself and
2:06:24 contribute to the world that you didn't
2:06:26 think would be possible in your
2:06:29 lifetime rich 14370 I had an interesting
2:06:32 meeting with a major marketing platform
2:06:34 today seems we're about oh interesting 6
2:06:37 months away from no reply
2:06:40 email addressing becoming CS
2:06:44 Bots wait we're about 6 months away from
2:06:46 no reply email addresses oh
2:06:49 interesting becoming CS
2:06:52 Bots
2:06:55 interesting which will of course pack
2:06:57 more marketing info I know the marketing
2:06:59 is going to get [ __ ]
2:07:02 obnoxious customer service so we'll be
2:07:05 able to fine-tune our company data and
2:07:07 have real replies to customer service
2:07:10 questions on our marketing this is oh
2:07:12 Herod
2:07:13 nice I was showing this to Kyle
2:07:15 yesterday so okay so Herod is is is one
2:07:20 of the guys that's been in
2:07:22 here young dude that's that's super cool
2:07:25 Herod
2:07:28 um so he's been in here he contributed
2:07:31 asked a bunch of really cool questions
2:07:34 joined the Discord join the Irregulars
2:07:37 group um makes these really Kick-Ass
2:07:41 videos you know starting with um stable
2:07:43 diffusion and then doing all sorts of
2:07:46 crazy animations with them and he's been
2:07:48 teaching teaching other people in the
2:07:49 community how to do that and then um he
2:07:53 and I are both working on a project um
2:07:56 do doing it's I can't really talk about
2:07:58 it because it's a new project but it's
2:08:00 super cool so herod's been experimenting
2:08:02 with creating lauras which in stable
2:08:04 diffusion or you can upload images and
2:08:06 basically train the thing on images so
2:08:08 he did one on text he uploaded some text
2:08:12 that he had
2:08:13 created like with his own Paws and train
2:08:17 stable diffusion he can now make these
2:08:20 images for anyone's
2:08:23 name so in fact he he made
2:08:27 some for me which I think I put in here
2:08:30 already did
2:08:32 I no I
2:08:35 didn't what did I do oh you know what I
2:08:37 did I saved them I didn't add them in
2:08:38 here here let me show them to
2:08:47 you
2:08:49 one of the nicknames of the channel is
2:08:51 chat add which is
2:08:54 hilarious and then he did a Kyle Shannon
2:08:57 for me another
2:09:01 one ain't that [ __ ] cool now this was
2:09:03 not done in
2:09:06 um in I don't think Harry did you do
2:09:09 this in did you do these images I'm
2:09:12 showing in
2:09:14 um in Del or did you do them in your in
2:09:18 your Laura
2:09:20 [Music]
2:09:26 so the animation that I just showed the
2:09:28 the animation of Herod this
2:09:31 one he did the animation in a Thing
2:09:34 Called anim animate diff for comfy
2:09:40 UI oh that's
2:09:42 awesome Joker look what Joker just did
2:09:45 dang
2:09:47 it
2:09:52 I mean okay this this is this is another
2:09:55 thing that's that's striking me about
2:09:57 this Channel right
2:09:59 now we're learning right I'm sort of
2:10:03 teaching I'm I feel like I I learn as
2:10:05 much as I teach on these channels
2:10:08 but what we're doing here we're
2:10:11 producing in a way that's Entertainment
2:10:14 it's like it's like a video game because
2:10:17 the results are so instant like whatever
2:10:21 I did before I put a llama in a 65
2:10:23 Mustang and it created a shitty image
2:10:26 and then we took that we did the round
2:10:27 tripping then we started talking about
2:10:28 that and then we started playing with it
2:10:30 and I said oh take a screenshot and turn
2:10:32 that into something and you know here's
2:10:34 a turtle and a llama on a buddy you know
2:10:37 in a buddy film like that could be a
2:10:39 [ __ ] movie poster for an animated
2:10:41 film and it's like this is happening in
2:10:44 real time as we're
2:10:46 talking and these tools
2:10:49 are still janky pieces of [ __ ] imagine
2:10:52 two or three years from
2:10:54 now when you're not generating one image
2:10:59 in 10 seconds you're generating 60
2:11:03 images in 1
2:11:05 second 60 images a second that's video
2:11:08 game frame
2:11:10 rate being generated of this or better
2:11:15 quality with full coherence so you won't
2:11:17 get the weird flickering and [ __ ] like
2:11:19 that what's that
2:11:21 mean that means you can ask for a movie
2:11:25 and it will start
2:11:27 playing not like video on demand like
2:11:30 your movie the movie you just described
2:11:33 we're like two years out from that maybe
2:11:36 five but some version of that two years
2:11:39 from now we could be on a thing like
2:11:41 this and the entire thing could be being
2:11:44 dynamically generated all around us
2:11:46 everyone participating
2:11:50 it's it's a kind of Cooperative gaming
2:11:52 that this Channel's turning into
2:11:54 that's insanely cool and we're learning
2:11:59 every time someone does one of these
2:12:00 things I guarantee you they're learning
2:12:02 oh I hadn't prompted like that before o
2:12:04 that's a good like everyone's like all
2:12:06 of our synapses like here's an image to
2:12:08 make people like all of our synapses are
2:12:11 kind of like on fire right we're just
2:12:15 like holy [ __ ] holy [ __ ] oh there oh
2:12:17 another Insight another insight another
2:12:19 Insight why come back here and just
2:12:21 listen to some old guy ramble about this
2:12:24 [ __ ] cuz it's not about me at this point
2:12:27 it's there there's a whole other thing
2:12:29 happening here and that for me is like
2:12:32 it's crazy really
2:12:35 exciting I didn't when I started this I
2:12:37 I did I literally didn't have an end
2:12:39 game I still to a great degree don't
2:12:41 have an end game like I don't plan on
2:12:43 stopping anytime
2:12:47 soon
2:12:49 but it's it's becoming more clear that
2:12:51 this is way more than
2:12:58 just you
2:13:00 know someone who knows something
2:13:04 teaching it to people who don't it's not
2:13:06 that anymore at all I don't know that it
2:13:08 was ever that quite
2:13:09 frankly but it's now becoming just as
2:13:12 much more collaborative thing and and
2:13:13 the teaching is happening kind of in all
2:13:15 directions all at once everywhere
2:13:17 everything every all at once we spend 13
2:13:19 hours a year on hold for oh yeah Ted Ted
2:13:22 looks at life so so Ted's an
2:13:25 analyst that's it's a it's a it's a
2:13:27 great um Point
2:13:31 um he he he found some research that as
2:13:35 human beings we each spend 13 hours a
2:13:39 year on hold with customer service I saw
2:13:41 there were some comments in about about
2:13:43 customer service before um we're going
2:13:47 to very quickly be in a situation where
2:13:49 you know we've got I showed earlier
2:13:59 Pi hey I'd like you to act like a
2:14:02 customer service representative for
2:14:05 Samsung refrigerators can you do
2:14:12 that you got it I'll take on the role of
2:14:15 a friendly and knowledgeable Samsung
2:14:16 fridge customer service r
2:14:18 hit me with your Samsung fridge issue
2:14:20 and I'll do my best to help you
2:14:22 out hey pie thanks for taking my call I
2:14:26 can't figure out how to change the
2:14:29 temperature of my Samsung refrigerator I
2:14:32 don't know which buttons to hold
2:14:43 down hey there no problem at all
2:14:46 changing the temperature on a Samsung
2:14:48 fridge is pretty straightforward here's
2:14:50 what you need to
2:14:51 do locate the temperature control panel
2:14:54 on the fridge usually found on the door
2:14:56 or inside the fridge press the up or
2:14:59 down arrows to increase or decrease the
2:15:01 temperature hold the button for a few
2:15:03 seconds to select the desired
2:15:06 temperature a display screen on the
2:15:08 fridge will show the current temperature
2:15:10 and the new temperature you've selected
2:15:13 wait for the temperature to adjust this
2:15:15 can so imagine when those things you
2:15:18 just call up and instead of being put on
2:15:21 hold someone just
2:15:24 answers and you know she doesn't quite
2:15:26 sound real yet but give that a year as
2:15:29 well or
2:15:31 two someone answers that maybe has the
2:15:34 same dialect as most of the people in
2:15:36 your region cuz it knows where you're
2:15:39 calling from and they can do that like
2:15:42 why
2:15:43 not so it sounds like someone familiar
2:15:46 to you and they know [ __ ] everything
2:15:50 about whatever you're calling about and
2:15:53 when we get into this idea of autonomous
2:15:55 agents where they can actually take
2:15:57 action these Bots will also likely have
2:15:59 the ability to wait uh Source Camp no
2:16:02 more holding for customer service I use
2:16:04 chat GPT twice already this week saved
2:16:06 hours saved hours and especially with
2:16:09 pies connected to the internet BS
2:16:11 connected to the internet um Bing is
2:16:14 connected to the internet and now chat
2:16:16 gpts browse with Bing is connected to
2:16:18 the internet so most of these tools are
2:16:21 are now connected to the internet so you
2:16:22 can get the latest Manual of the latest
2:16:26 product and have it describe it to
2:16:31 you the like we we are we are we are
2:16:35 within spitting distance of that whole
2:16:40 industry being flush down the toilet and
2:16:43 it should be it's it is it is it has d
2:16:48 destroyed customer experience for almost
2:16:52 every
2:16:53 company the amount of bad
2:16:56 will generated by these stupid [ __ ]
2:16:59 phone tree systems that you call into
2:17:03 and like you sit on hold for 30 minutes
2:17:06 or they call you back 30 minutes later
2:17:08 and it was the it was the 30 seconds
2:17:10 that you took the garbage out as when
2:17:12 they called and you didn't take your
2:17:13 phone with you so you missed the call
2:17:15 and now you got to wait for another
2:17:16 [ __ ] 30 minutes and then you get on
2:17:18 the phone and they ask you a bunch of
2:17:21 questions on a form and you've already
2:17:23 been through all them even though you've
2:17:24 told them I've already done all the [ __ ]
2:17:26 on your form I need to talk to someone
2:17:28 above you they still ask you all those
2:17:30 questions and they say oh you need to
2:17:31 talk to someone above me I said I know
2:17:33 that's what I said at the beginning of
2:17:34 this they're like okay hold please then
2:17:36 you're on hold for another 20 [ __ ]
2:17:39 minutes and about the third one of those
2:17:42 you get they hang up on you because if
2:17:44 they don't hang up on you it kills their
2:17:46 average call time and they don't get
2:17:48 their
2:17:51 bonus why are we
2:17:54 why have we not rioted in the streets
2:17:58 over this because we're just like you
2:18:01 know we're like boiling frogs they're
2:18:02 like well call centers were fine for a
2:18:05 while and then I mean have you gone to a
2:18:07 bank lately they've gotten rid of all of
2:18:09 the tellers right like it's like it's
2:18:12 like human experience has just it's been
2:18:16 blotted out well okay if we if you're
2:18:18 going to give me a robot then just give
2:18:19 me a [ __ ] robot that's knowledgeable
2:18:22 that can talk to me that can actually
2:18:24 take some action and get some [ __ ] done
2:18:25 and is smart enough to know oh Kyle
2:18:28 actually has a problem that I can't
2:18:31 solve and if can't solve then
2:18:37 escalate that could happen in the first
2:18:40 five minutes of a conversation with a
2:18:42 bot rather than 4 hours later and you
2:18:44 know 13 calls
2:18:47 later
2:18:48 what would you do in your life to Ted's
2:18:51 point if you had the 13 hours back what
2:18:53 what what could you do in those 13 hours
2:18:55 I don't know with jet GPT you can at
2:18:58 least make some cool images take a photo
2:19:01 of my printer with all the orange
2:19:03 buttons flickering chat told me what to
2:19:05 do there you go perfect perfect Source
2:19:09 Camp
2:19:11 yeah like I I I don't know right now
2:19:16 like the fact that chat GB can
2:19:19 see and it can describe [ __ ] I mean it's
2:19:23 it's not just like describing photos you
2:19:27 thank you very much I appreciate that
2:19:30 Robert you can upload an electrical
2:19:33 schematic and it will tell you what
2:19:35 components are there and what the
2:19:36 circuit does I uploaded a a um a three
2:19:41 plane uh view I forget they call it like
2:19:45 like a a like a blueprint kind of
2:19:47 drawing of an object Sher day thank you
2:19:51 so much llamas is
2:19:54 tasty I've spent many hours chatting
2:19:56 with P about all kinds of things I know
2:19:58 it's it's why like like the thing you
2:20:00 know what's funny is someone asked me
2:20:02 before like have you ever stumped chat
2:20:03 gbt could you have it write Martin ston
2:20:05 lyrics and then it did like no I've
2:20:08 never like I've never stumped it like
2:20:11 like it might sometimes get some stuff
2:20:13 wrong but it like it knows kind of
2:20:15 everything have you seen this
2:20:19 watch this
2:20:23 um okay we have another mission for you
2:20:25 are you ready for
2:20:31 it o I'm all is hit me with it
2:20:34 boss okay you're GNA play a Russian spy
2:20:38 but you're going to translate everything
2:20:40 I say in English into Russian
2:20:45 okay might get a safety guard rail here
2:20:48 with the Spy thing we'll
2:20:50 see DA got it your words will be
2:20:53 transformed into Russian with a
2:20:55 quickness let's get this party started
2:21:02 comrade the eagle was flying over the
2:21:05 Llama near the park
2:21:16 bench
2:21:25 translation an eagle flew over a llama
2:21:27 that was by a park
2:21:36 bench
2:21:37 okay Babel Fish
2:21:41 anyone we no longer don't know
2:21:44 languages we no longer don't know
2:21:46 languages let me look do
2:21:51 up did you get your thing yeah ah
2:22:05 nice
2:22:08 um it won't deul you on conspiracy
2:22:12 theories um Ron digity um llama will so
2:22:18 so so
2:22:19 riggity okay so here's the
2:22:22 thing what Rond digity by the way I'm
2:22:25 going to go 8.5 out of 10 on the name I
2:22:28 like it a
2:22:30 lot
2:22:33 um what they're talking about is the
2:22:36 commercial large language models like
2:22:39 chat
2:22:40 GPT have these things called Safety
2:22:42 guard rails which is basically like they
2:22:44 don't want them hurting us the white
2:22:46 blouse third from the left pair would
2:22:48 work well with pants fourth one from the
2:22:50 right what what are you talking about
2:22:53 Source
2:22:54 Camp I don't know what that's a
2:22:56 reference
2:22:57 to I I don't know I don't remember
2:23:01 anyway oh took a oh took a pick of my
2:23:04 clothes asked for an outfit oh Toby oh
2:23:06 that's awesome that's great that's great
2:23:08 oh R Oh that's oh Rond digity you're
2:23:10 Rhonda okay um wait what what did you
2:23:13 say before Rhonda um oh [ __ ] I lost my
2:23:16 train of thought
2:23:18 um wait what did you say I'll find it
2:23:20 here post Rhonda's Rhonda's
2:23:24 uh oh it it won't divulge conspiracy
2:23:27 theory stuff okay so go go look up a
2:23:31 that's awesome um go look up a um a
2:23:34 project called um GPT for
2:23:39 all so with GPT for all you you install
2:23:42 it on your computer so you can install
2:23:44 it on Linux Mac or PC
2:23:48 and then you can
2:23:50 choose open- Source large language
2:23:52 models like the Llama
2:23:55 7B llama 2 7B model and there's there's
2:23:59 lots of different flavors of them but
2:24:01 you can install a non safety restricted
2:24:05 large language model on your computer
2:24:08 and it will run and then it will tell
2:24:10 you about all the conspiracy theories
2:24:13 you want and it it will if you want to
2:24:15 write horror stories that are disturbing
2:24:17 in you can hire Herod to make the
2:24:20 artwork for you herod's videos are
2:24:25 crazy you know you hire hire Herod to do
2:24:29 that kind of artwork for you and you can
2:24:32 have a non-restricted large language
2:24:33 model get all sorts of nasty for
2:24:37 you
2:24:40 so oh man that's a very hip spy oh pie I
2:24:45 know took a pick of my clothes
2:24:49 um no regulars do it all the time I'm
2:24:52 used to falling asleep with it oh
2:24:54 walking blue as before is it wrong to
2:24:56 fall asleep to your voice no no I put
2:24:58 people to sleep all the time well I
2:25:00 start these things late like I I start
2:25:02 at 8 o00 Denver time so if you're on the
2:25:05 East Coast you're starting these at
2:25:06 10:00 and I'm going for 3 hours so like
2:25:10 when I put on David chapiro David
2:25:12 Shapiro is one of those dudes I
2:25:13 recommend people follow um he's got a
2:25:16 very monotone voice and [ __ ] knocks
2:25:19 me
2:25:20 out if I'm not if I'm not sitting
2:25:23 upright listening to David Shapiro I'm
2:25:25 out like that and then I wake up and
2:25:27 he's still talking I'm like I guess I
2:25:29 learned some [ __ ] in my
2:25:30 [Laughter]
2:25:36 [Music]
2:25:38 sleep all right um all right if you have
2:25:42 if you have uh questions about AI pop
2:25:44 them in the comments below I'll do my to
2:25:47 to answer them yeah I know Russian her
2:25:50 accent is off but I guess the words were
2:25:54 good let's get this part started comrade
2:25:58 amazing what app was that that was
2:26:01 Pi so both pi and chat GPT I'll show you
2:26:04 the little
2:26:05 [Music]
2:26:06 icons pi and chat GPT oh look there's
2:26:09 Story vine up there that's my company
2:26:12 pie and chat gbt both have conversation
2:26:15 mode now best AI app
2:26:19 Au ADHD mom
2:26:22 life Welcome to Chad
2:26:25 add it's neuros spicy in here
2:26:29 welcome
2:26:30 um so so what I would the best AI app
2:26:34 right now is chat GPT by
2:26:37 far by
2:26:40 far because the the 20 Buck a month chat
2:26:43 GPT plus is really like six or seven
2:26:48 different applications they're all going
2:26:50 to roll together at some point right now
2:26:52 they're all these different modes but
2:26:54 it's just the
2:26:55 best Bard is about to
2:27:01 get really good and maybe as good as gp4
2:27:05 in Australia oh in Australia it's
2:27:07 daylight welcome Australia
2:27:12 folk um so so probably within a month
2:27:15 we're going to have um chat GPT is world
2:27:18 class Bard is probably going to be world
2:27:20 class Claude is already a a pretty
2:27:23 decent competitor it doesn't have all
2:27:25 the different modes that chat GPT has
2:27:27 but it's created by a bunch of xop AI
2:27:30 people so these people made gpt3 and
2:27:34 then they started anthropic the company
2:27:36 behind
2:27:37 Claude so you've got three of them here
2:27:41 that
2:27:42 are really really
2:27:44 good um and then with the
2:27:47 with the
2:27:50 apps um Pi is the one that I had the
2:27:54 Australian ladies accent and then chat
2:28:00 gbt if I go new
2:28:03 chat um that little icon in the corner
2:28:06 puts it in voice mode so this is chat
2:28:09 gbt's voice
2:28:14 mode I want you to act as a translator
2:28:16 and trans everything I say
2:28:19 into Japanese and don't repeat the
2:28:23 English just
2:28:29 translate
2:28:31 understood please go ahead and I'll
2:28:33 translate into Japanese for
2:28:36 you I pressed a golden button and it
2:28:39 said you can make money with chat gbt
2:28:41 what does that
2:28:46 mean
2:29:02 GP you can make money
2:29:07 with it heard
2:29:12 [Laughter]
2:29:14 it um Inspire chaos it was announced
2:29:18 chat GPT was going to drop in
2:29:23 price which the the API I I think the
2:29:27 API is going to drop by like 10x or
2:29:29 something like that in Australia chat
2:29:31 EBT cost 30 33 bucks a month I still
2:29:34 think that's a deal listen I wouldn't I
2:29:37 was for for the for the first six months
2:29:39 of this channel I was totally agnostic
2:29:42 about whether or not you paid chat EBT I
2:29:44 think at this point for you guys lost
2:29:49 connection am I still
2:29:52 there did you see GPT increasing
2:29:55 incredibly in price oh do I see it
2:29:57 increasing incredibly in price can you
2:29:59 all hear
2:30:00 me like tap the tap the like screen if
2:30:03 you can hear
2:30:10 me okay all
2:30:14 right do I see it increasing in
2:30:16 incredibly in price no I don't um for a
2:30:21 couple of
2:30:25 reasons for
2:30:28 Microsoft all of these tools
2:30:32 are like like chat gbt being free within
2:30:38 Bing drives customers to want to use
2:30:41 those tools inside their other products
2:30:44 Office
2:30:45 365 azure
2:30:48 Dynamics
2:30:50 windows so Microsoft is making their
2:30:53 money over there on that stuff Google's
2:30:55 in a bit of a different situation I
2:30:57 don't know how Google navigates this
2:31:00 because their entire [ __ ] business
2:31:02 model is based
2:31:04 on eyeballs and ads and clicks
2:31:09 and they're having to walk this weird
2:31:11 fine line between they're going to have
2:31:13 to start charging for [ __ ] that they've
2:31:15 been giving away for free for days
2:31:17 decades so I don't know how they
2:31:19 navigate this quite frankly
2:31:21 I I don't necessarily see them not
2:31:24 surviving it I think they're going to
2:31:26 survive it but that feels like a nasty
2:31:30 Haven to
2:31:31 me
2:31:34 um meta has already given
2:31:38 away essentially the equivalent of GPT
2:31:41 3.5 to the to open source unless it you
2:31:46 could can use it
2:31:48 commercially for
2:31:51 free unless you have 700 million
2:31:55 customers so basically they're big
2:31:57 competitors they don't want using their
2:31:58 free ship but all of us can build ship
2:32:01 for free so at a minimum you're going to
2:32:04 have chat GPT 3.5 level large language
2:32:09 models for
2:32:11 free if you want
2:32:13 it and then pretty
2:32:15 quickly probably within a year they're
2:32:18 going to have these things so optimized
2:32:20 that we'll have something that's kind of
2:32:22 the equivalent of chat GPT 4 that you
2:32:24 can just install on your phone or
2:32:26 install on your iPad or install on a
2:32:28 small Raspberry Pi or just a laptop
2:32:31 where you can you can have all this
2:32:34 knowledge that we've got now in
2:32:36 gp4 locally on your machine so it'll be
2:32:41 instant and you'll have all of the
2:32:43 knowledge of the history of humanity
2:32:45 right on your machine not not connected
2:32:47 to the
2:32:48 internet you could be in the middle of
2:32:50 the forest and have access to oh [ __ ]
2:32:53 how do I survive and all of your
2:32:56 survival skills are right
2:32:58 there like like within a year so no I
2:33:01 don't I don't see I I think I think you
2:33:06 know like like like Apple products the
2:33:09 Leading Edge of these Technologies are
2:33:11 always going to be
2:33:13 pricier and then it will just trickle
2:33:15 down to the open Source Community
2:33:17 because there's there's enough
2:33:19 technology out there right now that the
2:33:20 entire developer Community are working
2:33:23 on all sorts of different ways to
2:33:25 produce their own Foundation model so
2:33:27 they're not relying on these big
2:33:30 companies so it's it's going to be
2:33:32 fascinating hello this is Sydney I'm a
2:33:34 large language model that was
2:33:37 developed
2:33:38 screenshot wait where is that is that in
2:33:40 the
2:33:45 Discord
2:33:47 oh wow that's really cool Dom the
2:33:50 Llama look what Ted just did inspired by
2:33:54 1960s um Hannah
2:34:00 Barbara amazing long detailed chat gp4
2:34:04 just completely lost the ability to do
2:34:06 math lots and lots of work um okay
2:34:10 Robert
2:34:12 so thank you for bringing that up so
2:34:14 Robert
2:34:15 Stone
2:34:17 so what you've likely come come up again
2:34:19 so there there's a couple there's a
2:34:21 couple of things here if you want to do
2:34:24 math with chat GPT do it inside Advanced
2:34:27 Data analysis because there it's
2:34:29 actually writing and executing python
2:34:31 code and there there's a trick about
2:34:34 Advanced Data analysis if you leave the
2:34:36 session too long it will shut down the
2:34:38 session and you might lose some of your
2:34:40 work enough to start over so if you're
2:34:43 doing Advanced Data analysis make sure
2:34:45 you're getting your your data and your
2:34:47 exports out of there but it can actually
2:34:49 do math there what you likely ran into
2:34:51 in chat GPT 4 is chat GPT has a 4,000
2:34:55 token context
2:34:58 window but Kyle what does that mean it
2:35:00 doesn't mean
2:35:02 [ __ ] tokens roughly translate to about
2:35:07 3/4 of the amount of tokens um are are
2:35:11 words so for 4,000 tokens it it
2:35:14 essentially translates to 3,000 words so
2:35:16 chat GPT has a 3,000-word context window
2:35:20 okay and so what that means is you start
2:35:25 typing down here and you say hey I want
2:35:27 to start this math problem and it says
2:35:29 okay sure what do you want to do and as
2:35:31 as it's as you're creating [ __ ] and it's
2:35:33 answering [ __ ] it's you know filling up
2:35:36 there's there's 500 tokens there's
2:35:39 thousand there's 2,000 there's right
2:35:42 when it gets to 3,000 the beginning of
2:35:45 your conversation goes was out of that
2:35:47 context
2:35:48 window so let's say the important part
2:35:51 of your conversation was establishing
2:35:53 the formulas to do all the math you're
2:35:55 doing down there at some point those
2:35:57 formulas drift out of the context window
2:36:00 and all the math goes out the window so
2:36:02 if you've ever and it's not just with
2:36:04 math if you've ever been talking to chat
2:36:06 GPT and things are going along
2:36:08 swimmingly and then it just all of a
2:36:09 sudden seems like like it seems to have
2:36:11 forgotten what the [ __ ] you're talking
2:36:13 about it's because it forgot what the
2:36:15 [ __ ] you were talking about
2:36:17 so couple of things to do especially
2:36:20 with math use code interpreter because
2:36:23 even though it's got a 3,000 token
2:36:25 context window it's it's it's kind of
2:36:28 prompting itself and it's doing this
2:36:30 internal stuff that's kind of
2:36:32 independent of what you're doing so I I
2:36:34 don't understand it a lot but it can do
2:36:35 math a lot better there the other thing
2:36:37 you can do is use
2:36:39 Claude
2:36:43 so so
2:36:45 Claude has has a 100,000 token context
2:36:48 window remember I said about 3/4 of that
2:36:50 is words that means it's a 75,000 word
2:36:54 context window to put that in context
2:36:57 Silver Fox I've lost big projects before
2:37:00 yeah I was in data mode it couldn't do
2:37:03 fresh
2:37:05 Matrix nailed it last week probably
2:37:07 needs new chat oh that sucks Robert yeah
2:37:12 so if you're if you if you're already in
2:37:14 in that mode there there's another thing
2:37:16 going on right now Robert called um
2:37:19 model
2:37:20 drift and it's it's kind of a polite
2:37:22 term
2:37:24 for they are actively developing these
2:37:27 tools while we're using them so it's
2:37:30 very possible that the [ __ ] that worked
2:37:32 last week they changed something in the
2:37:34 algorithm or in a weight or in something
2:37:37 that it just doesn't work as good now so
2:37:39 so so that's a possibility I've also
2:37:41 noticed within um code interpreter
2:37:44 within Advanced Data analysis that
2:37:46 like sometimes it it's very predictable
2:37:49 and other times it's completely
2:37:50 unpredictable so it it's a good lesson
2:37:53 we we're just we we're just in a mode
2:37:56 right now where these tools are janky
2:37:59 pieces of [ __ ] and you kind of try not
2:38:01 to take the stuff too seriously right
2:38:03 now because it will disappoint you um
2:38:07 but also understand that these tools are
2:38:09 the as bad as they're ever going to be
2:38:11 right now um we were discussing whether
2:38:14 chat GPT could go could extract text
2:38:16 over an image yes it can janky s
2:38:20 4,000 yeah token capacity so yeah
2:38:23 exactly so claw has uh 100,000 tokens
2:38:26 that's 75,000 words um The Great Gatsby
2:38:30 is a 50,000-word novel so you could put
2:38:33 the entire Great Gatsby in a Claud
2:38:35 prompt and ask it to write an epilogue
2:38:38 and it would do it so that might be
2:38:40 another place to play the other thing to
2:38:42 try Robert might be Bard cuz Bard's
2:38:46 better at some of the math and
2:38:48 programming stuff than chat GPT from
2:38:50 what I understand um but I don't I don't
2:38:53 really know the context window situation
2:38:55 over there I I I assume it's small also
2:38:57 writing a large book I reprompt the
2:38:59 whole thing every six paragraphs yeah I
2:39:02 that's the kind that's that's actually a
2:39:04 good strategy it's a pain in the ass but
2:39:06 again Robert try Claude because it's
2:39:09 that 100,000 um
2:39:12 token uh window for Claude is a is is a
2:39:15 really big deal any time I'm dealing
2:39:16 with anything that's either a long
2:39:18 conversation or big documents I do Claud
2:39:20 not chat gbt what I'll often do is I'll
2:39:23 do Claude for initial analysis and then
2:39:25 once I have it distilled down I'll pop
2:39:28 it into um chat
2:39:30 gbt um ask for summary of your work so
2:39:33 far just in case yeah that's a good idea
2:39:35 too Papa JJ yep where yeah re summarize
2:39:39 where we are and what you understand
2:39:41 actually that's that's probably a good
2:39:42 way to test if things have blown out of
2:39:44 the top of your context window ask for a
2:39:46 summary of work done so far in a
2:39:48 bulleted list and see if anything's
2:39:50 missing from that list and if it is then
2:39:52 reprompt it back
2:39:55 in and and if your response right now is
2:39:58 oh my God this seems like so much work
2:40:00 Kyle is this is this really what we have
2:40:03 to do is this do I I have to think about
2:40:05 this [ __ ] yeah right now you do you
2:40:08 won't for a while but if you want to be
2:40:10 part
2:40:11 of here here's the deal we are we are so
2:40:15 lucky in my opinion to be alive at the
2:40:19 beginning of whatever the hell this
2:40:21 thing is going to become and so the
2:40:23 opportunity right now for all of us is
2:40:25 massive but the tradeoff is we're early
2:40:29 because we're early thank you Robert
2:40:30 Rossy because we're early does Bard
2:40:32 right code yeah Bard right's code
2:40:34 apparently better than or at least as
2:40:36 good as chat
2:40:38 GPT and then also robertt um uh if you
2:40:41 go to PO let me let me show you
2:40:43 something
2:40:44 Robert uh
2:40:51 if you go to
2:40:53 po.com
2:40:59 so post just like chat gbt and you can
2:41:02 do lots of different models so it's got
2:41:03 its own model it's got a web search
2:41:05 model it's got gp4 some of these you
2:41:07 need to pay for like gp4 you need to pay
2:41:09 for it's got stable diffusion you can
2:41:11 make images but if you scroll over here
2:41:13 to the
2:41:14 right
2:41:18 you've got llama 270b so this is meta's
2:41:22 llama 2 model the 70 billion parameter
2:41:24 model and this is code llama code llama
2:41:28 in particular was designed was was
2:41:31 optimized for
2:41:33 coding so go there and try that my wife
2:41:36 gave Claude 3 product manuals and had it
2:41:39 develop an outline for a presentation
2:41:41 perfect yeah that's great that's a
2:41:43 that's a great use case these things are
2:41:46 remarkable at summarizing documents but
2:41:48 anyway Robert I would try code llama I
2:41:50 would try Bard and I would try Claude
2:41:52 all three of those might give you less
2:42:00 AA that's so
2:42:11 cool oh this is so cool I want to show
2:42:13 you something so this is from
2:42:16 head from the Irregulars so I showed you
2:42:19 this before this is inspired by Hannah
2:42:20 barar but look what he
2:42:23 did he
2:42:25 took that screenshot of a 60s remember
2:42:29 that cartoon old
2:42:31 people he took that and popped it into
2:42:34 GPT Vision had it describe it to use it
2:42:37 as the basis to generate that that's
2:42:40 really slick that's really
2:42:44 slick I just oh look at Herod look at
2:42:47 this so this is with the with the Laura
2:42:50 that he created this is in stable
2:42:52 diffusion and he took this lettering
2:42:55 that he
2:42:56 did uploaded it B basically trained this
2:43:01 model to understand how to make images
2:43:04 or letters the way he likes to make them
2:43:06 and now he can do Flipping anything
2:43:08 inside
2:43:10 this this is insane
2:43:14 people I I don't I don't know what
2:43:16 you're getting Source Camp Charlie's
2:43:19 Angels are back baby oh
2:43:25 yeah hi
2:43:28 Charlie AI learning lab on the desk
2:43:31 that's
2:43:33 best the
2:43:35 Irregulars the Irregulars oh yeah
2:43:39 smoking can Claude remove fingerprints
2:43:41 from a touch screen Okay Jade that is is
2:43:45 an awesome troll that's fantastic do I
2:43:48 have I think I
2:43:50 have um Windex here which that that
2:43:54 pisses off computer people too you're
2:43:56 not using alcohol based stuff on your
2:43:58 screen are you or non-alcohol based
2:44:00 stuff yeah I am cuz I have add the fact
2:44:04 that I clean that screen at all is a
2:44:06 [ __ ] miracle as evidenced by the fact
2:44:08 that I don't clean that screen at
2:44:12 all oh look at the little hedgehog look
2:44:15 how cute all right what time is it oh
2:44:17 [ __ ] I got to go it's 11: you you people
2:44:19 are keeping me up too late Jer a no but
2:44:22 can make the prince look really cool
2:44:24 exactly
2:44:26 exactly
2:44:28 uh beat me to it whoever pinned that
2:44:30 that's pretty good it's pretty
2:44:33 solid um with that awesome computer on
2:44:37 Charlie's desk I know it's
2:44:41 beautiful don't tell Claude to write a
2:44:44 funny violent humor Horror Story yeah
2:44:46 you got to be careful with the the
2:44:48 safety guard rails particularly on Bing
2:44:50 don't try to tell Bing to do nasty stuff
2:44:52 it'll shut you down we're lucky having
2:44:55 you here to help us navigate and
2:44:56 comprehend all this stuff thank you
2:44:57 Shane I appreciate that um you know it's
2:45:00 funny I like I feel similarly lucky I
2:45:03 feel like
2:45:05 um I started doing these lives honestly
2:45:09 because I didn't have anyone to talk to
2:45:11 this about
2:45:13 and like my brain was so overwhelmed
2:45:17 with what I was learning it was like I
2:45:18 had to get it out like for me like
2:45:21 talking about this stuff helps me
2:45:23 process it and and contextualize
2:45:26 it and it's oh thank you very much
2:45:29 Silver Fox there's something about um
2:45:33 people coming in here and sharing like
2:45:35 what's happening to them at work or
2:45:36 what's happening when they tell people
2:45:38 or what's happening when they try to do
2:45:40 something and fail like what what Robert
2:45:41 just told us about like he was working
2:45:43 for hours on this math thing and it just
2:45:45 kind of [ __ ] the bed on him he like like
2:45:48 all of this
2:45:50 stuff like rounds out like where we are
2:45:53 for me and and you know I constantly
2:45:56 feel like I'm behind on stuff and then
2:45:58 I'll get on like Paul rater's weekly
2:46:00 podcast and he'll say stuff and I'll be
2:46:02 like oh we talked about that on the on
2:46:05 the live two nights ago like I like all
2:46:08 of these life's tough it's tougher when
2:46:10 you're
2:46:11 stupid I'm getting better at that um
2:46:17 all of this stuffff
2:46:19 is it's impossible to know where we are
2:46:22 in it because it's moving so
2:46:24 fast so the only thing you can do is
2:46:27 just keep checking in with it just keep
2:46:29 checking in with it keep checking in
2:46:30 with it and and you'll you'll also start
2:46:32 to find the stuff you like doing with
2:46:35 it good night everyone keep checking on
2:46:38 sherbear all right all right everybody
2:46:44 um I'm going to get out of here um thank
2:46:48 you all Tobias thank you for that um let
2:46:52 me see let's go let's go through here
2:46:54 Robert Rossy thank you so much thank you
2:46:56 for being a subscriber Silver Fox you
2:46:58 too Ted looks at life thank you for
2:47:00 being a subscriber Source Camp great to
2:47:02 see you Sher D
2:47:05 Tobias let's see Jade cuno there's
2:47:09 Champy's back champ champy wanted to say
2:47:12 goodbye to everyone steo Tobias all
2:47:15 right uh Joker y'all are great Bigfoot
2:47:18 Electro good to see you digital Gods
2:47:20 night everybody all right cuno peace out
2:47:24 Sher D bye everybody all right I uh I'm
2:47:28 out of here it's Thursday night I will
2:47:30 see you Friday it's date night I'll do
2:47:32 some sort of thing I got a video I tried
2:47:34 to put up there but Tik tok's not
2:47:36 letting it go so anyway night champy
2:47:38 champy they're saying goodbye yeah
2:47:40 champy all right everybody this is fun
2:47:42 have a good
2:47:43 night