aiLL 10/21/23 - AI is like a Golden Retriever, Training Us to be More Compassionate Collaborators

Description

Chapters

Transcript

0:07 champy champy champy
0:09 [Music]
0:19 champy welcome welcome welcome happy
0:22 Saturday
0:24 night it's Saturday night Do You Know
0:26 Where Your Children
0:29 Are share D what's
0:35 happening howdy howdy howdy Sher D
0:38 slaying it in the image
0:41 generation becoming quite the
0:45 artist very cool very cool cool
0:48 cool George
0:52 sack Dr George
0:55 saaks S D
0:57 ADHD nice welcome to chat ad
1:01 D it's nice to have some professionals
1:03 in here well I'm a professional
1:07 ADHD
1:12 member not so much in the treating of it
1:15 just in kind of acknowledging
1:19 it actually I mean I I you know as I've
1:22 gotten older the thing I've done with
1:24 ADHD has stopped trying to fix it and
1:26 and it's it's gotten much
1:29 better
1:31 it's like once you realize oh yeah I'm
1:34 never going to be good at that thing it
1:36 gets a lot
1:37 [Laughter]
1:41 easier oh
1:44 man all right so uh as we get started
1:48 here happy Saturday night I'm a
1:51 professional dulu hey everybody be Not
1:53 Afraid welcome welcome it's good seeing
1:55 you in here pretty regularly you're in a
1:58 regular now be not afraid if you're are
2:02 you on the salon Discord if you're not
2:04 you should join us over
2:07 there you ask good
2:09 questions good questions it's very good
2:13 it's very good oh hey Mark oh hey Mark
2:19 hi oh this cool image here Isn't that
2:23 cool at
2:24 45 um that was from uh Ted looks at life
2:27 did that in the salon Discord it's so
2:29 cool and the song is you can make money
2:31 with chat GPT you can make money with
2:33 chat
2:36 GPT isn't that awesome um so if you
2:40 would share the live um what time is it
2:43 it's just it's just turning
2:45 8 so I've been putting out those
2:47 countdown timers I don't know if they're
2:49 worth putting out but
2:52 whatever I'm trying Tik Tock stupid
2:56 [ __ ] put a sticker on it's got a
2:59 countdown
3:00 I don't know if it actually makes a
3:02 difference but you know if it does
3:04 cool
3:06 uh can you get The Irregular title now
3:09 what's your Be not Afraid what's your um
3:12 Discord handle I'll give it to you right
3:19 now Cindy entrepreneur happy
3:23 life good to see
3:27 you same as this one all right let me go
3:30 will find you in
3:32 here oh wait I can't do it there hold
3:34 I'll do it over
3:40 here be
3:43 not
3:45 afraid you now have the
3:50 power you are an irregular
3:54 poof see it happens just that
3:59 quick
4:01 welcome everybody welcome welcome
4:02 welcome happy Saturday night it's
4:05 Saturday Night Live live from Denver
4:08 it's Saturday
4:10 night we'll do some stupid voices in
4:15 here um we'll talk about AI stuff lately
4:19 we've been talking
4:22 about philosophy kind of stuff like
4:25 implications future work kind of stuff
4:28 but the conversation goes wherever you
4:30 want me to take it so thank you Sherry D
4:32 appreciate that yeah the conversation
4:35 goes wherever you wherever you want me
4:36 to take it so if you have uh thoughts
4:38 comments about generative AI questions
4:41 pop them in the comments below um I'll
4:43 walk you through some Basics if you're
4:45 brand new if you just stumbled in here
4:47 and you're like oh my God that's Tom
4:49 Arnold or is that Rosie o'donell
4:53 um you just I do the same joke every
4:56 night it's like at some point I got to I
4:59 got got to I got to get chat gbt to
5:01 write me some new
5:04 openings but uh but if you haven't
5:07 played with chat GPT if you haven't
5:09 tried it yet um go right now get the to
5:14 a Nery get the T to a nuny
5:17 go go to chat. open.com that top little
5:22 thing
5:23 there chat. open.com that's the official
5:26 chat GPT website just go there right now
5:29 while I of wax poetic about whatever the
5:32 hell we're going to talk about
5:34 tonight just go start playing with it it
5:37 just just go just go right now um the
5:39 other thing I'd request as we get
5:40 spinning up here tonight is share this
5:43 live with your with whoever you're
5:45 comfortable sharing it with even if
5:48 you're like but they'll think I'm weird
5:50 Kyle share it anyway let them think
5:52 you're weird um this shit's coming
5:55 whether people want it or not so you
5:57 might as
5:58 well might as well just fry fly your
6:00 freak flag um but also if you if you
6:03 follow this channel if if you like if
6:06 you're enjoying these lives and you
6:08 don't follow the channel you don't get
6:10 notified when I go live so follow you
6:12 can also subscribe which I think gives
6:14 you I don't know what it does whatever
6:17 um but um but definitely follow the
6:20 channel share this live uh do the likes
6:23 do whatever engagement um you can and
6:26 then we'll get some more people in here
6:28 because the whole point here is to
6:31 sh share I was going to say share the
6:33 gospel but it's not that this is not a
6:36 belief this is just this stuff's
6:39 important and I think it's important for
6:40 people to experiment with it and explore
6:42 and that's what it is Clint Copp ahead
6:44 everyone in regular mods present to
6:47 llamas or hedgehogs or your favorite
6:50 animal I suppose also gummy bears gummy
6:53 bears are good your favorite spirit
6:54 animal I like it very nice hey Silver
6:56 Fox with shaking good to see you
7:01 Chandra it brother oh
7:04 yeah
7:06 welcome welcome brothers and sisters
7:09 welcome to the Church of the AI learning
7:11 [Laughter]
7:14 lab what level are you
7:16 [Music]
7:17 at we can get we can get real creepy
7:20 real
7:21 quick oh it's not a pyramid scheme but
7:24 it could
7:25 be um oh who's who overachieved there
7:28 with the llamas head HS be not afraid
7:30 very nice let's see dinosaur or Turtle
7:34 Hedgehog and
7:36 llama repeating very nice very nice over
7:40 I love
7:41 overachievers there's aot there's a lot
7:43 of overachievers in the Irregulars so
7:46 the Irregulars are people that show up
7:47 here night after night so if if you
7:49 don't uh if you haven't been here before
7:51 my name's Kyle Shannon uh I'm an
7:53 entrepreneur a many-time entrepreneur
7:56 I'm the co-founder and CEO of a company
7:58 called storyvine that's an automated
8:01 video storytelling platform thank you
8:03 Silver
8:04 Fox um what is that oh it's a whiskey
8:06 it's a wasy nice or is it a latte or an
8:10 espresso I'm going to call it a whiskey
8:13 but oh it's probably an iced coffee like
8:18 this um but thank you for that
8:22 um so yeah so my company Story vine is
8:25 an automated video storytelling platform
8:27 we've now got sort of an expanded
8:29 product Vision thanks to generative AI
8:32 I've been digging into this stuff for
8:34 for about the last two years and this
8:36 Channel's been around for coming up on a
8:39 year and um I've been doing these lives
8:42 now every single night for I don't know
8:46 two three hours a night for the past
8:49 four
8:50 months cuz I'm
8:52 crazy
8:54 um and and it's really just a free
8:56 flowing conversation um I'm trying to
8:59 love I'm trying for level 20 now Kyle
9:02 exactly Clint Copperhead Clint
9:04 Copperhead is an
9:06 overachiever what you're really shooting
9:08 for Clint is Level
9:10 33 level 33 is the big
9:13 unlock um Sher bear thank you for the
9:16 gummies very nice I like I like a gummy
9:19 I live in Denver there's a lot of
9:20 gummies here hey digital Gods what's
9:23 happening so hey Pate good to see you if
9:26 you uh if you have questions pop them in
9:28 the comments below Got Mods in here that
9:30 will will raise the question or you know
9:32 sort of pin the questions I'll also go
9:35 through them and try to answer things as
9:36 we go along um why don't we get started
9:39 I'll I'll uh I'll show you some things
9:42 I'm going to give you a pile of URLs to
9:44 explore explore these are things for you
9:48 to explore
9:50 okay so these
9:53 four are what I call the adult table
9:57 these are the biggies so chat GPT you've
9:59 heard of but if you haven't played with
10:01 it go there right now go there right now
10:04 don't don't mess around just go and then
10:07 if you have played with that but you
10:08 haven't explored these other things
10:10 they're worth exploring so Bard is
10:12 Google's entry it's okay right now it's
10:14 going to get really good as soon as they
10:15 release Gemini which is apparently
10:18 sometime this
10:19 month
10:21 Claude is a big competitor it's it's
10:24 created by some exop AI people company
10:28 called anthropic and and the big deal
10:31 about Claud is you can put very large
10:33 prompts in like 50,000-word novel you
10:35 can just paste it into the prompt of
10:37 claw and ask it to write you an epilogue
10:39 of that novel or whatever um it's it's
10:43 really quite capable um Chad gp4 here is
10:47 still the best there's nothing really
10:49 close to it right now and they just
10:50 added so many new features that I
10:53 strongly recommend um paying the 20
10:55 bucks a month to them I'm not affiliated
10:57 with them I don't get any commission I
10:59 should
11:01 [ __ ] but I don't this channel is
11:04 effectively a giant ad for those guys uh
11:07 and then Bing is Microsoft's uh entry
11:10 into this world it's they're it's
11:12 basically built on top of GPT 4 um and
11:15 it's it's not bad bing bing wants to be
11:18 really good but the the lawyers got to
11:20 it and it's safety guard rails make it
11:22 suck right now so it'll get better um Pi
11:26 really interesting company called
11:28 inflection they just raised $ 1.3
11:30 billion and uh the PI app the IOS app
11:36 you can put it in conversation mode it's
11:38 really quite good at conversations it's
11:40 actually better than chat gpt's
11:43 conversation mode which also just
11:44 launched and then you got perplexity
11:47 doai which is really good for research
11:49 and then labs. perplexity doai you can
11:52 play with metas Lama 2 models you can
11:54 also play with metas Lama llama 2 models
11:57 at po.com which lets you you play with a
11:59 lot of different models so so that's
12:02 interesting for that reason okay so
12:05 screenshot those screenshot this one
12:07 this is this is the uh the image
12:10 Generation stuff that I think is worth
12:12 playing with right now um there's one
12:14 missing from here which is
12:16 Leonardo but but this is plenty um Dolly
12:20 3 is remarkable it's now inside gp4 so
12:24 if you pay for chat GPT plus I'll demo
12:27 that in a bit mid journey is kind of the
12:29 best from an art quality standpoint um
12:32 but these the the other ones are
12:33 catching up so so very very interesting
12:36 stuff coming down to Pike um ideogram
12:40 I'm actually I I should probably replace
12:43 ideogram with
12:44 Leonardo I think I'm going to do that
12:47 let's do that right
12:50 now look at
12:54 this Leonardo I think they're
12:57 independent and then is it Leonardo a I
13:00 think it is but let's go figure it
13:04 out
13:06 Leonardo yes yes yes it is all right
13:12 so
13:15 edit
13:18 Leonardo why is he just typing this live
13:21 shouldn't he have done this before he
13:23 came on here what are his
13:25 qualifications
13:27 Leonardo doai is that right I think
13:30 that's
13:32 right all right nope that's not right I
13:36 knew something was wrong
13:38 edit.
13:41 AI not AO AI
13:45 okay yeah so go
13:48 there why go there Kyle um because so
13:53 the reason I got rid of
13:55 ideogram is that it has effectively been
13:59 replaced by Dolly 3 which can do text
14:02 and that was the the reason ideogram was
14:04 so exciting um was that it could do text
14:08 but dolly3 now does it really good it
14:11 sits live inside Bing it also sits
14:13 inside uh chat gp4 and you can also do
14:17 it on the chat GPT app at least on iOS
14:22 um Leonardo's built on top of stable
14:24 diffusion this is a a really powerful
14:26 and and decently intuitive front end to
14:28 stable diffusion so you don't have to go
14:31 installing [ __ ] on you know Google
14:33 collab it's just a website that works
14:36 and then the Adobe products Firefly just
14:38 went into version two and they just
14:40 announced a whole pile of AI like
14:44 adobe's going all in with this
14:45 generative AI stuff so um they're
14:48 good um trying to think what else let
14:54 me I'm going to skip some things tonight
14:57 just so we can get started on the
14:58 question and answer session so the
15:01 salon. a so um this is a community I
15:04 started last December um we meet every
15:07 other Tuesday and we have a Discord
15:09 server so tonight I'll I'll refer to the
15:12 Discord server and sometimes people are
15:15 in here I'll talk about making an image
15:17 they'll share images on the Discord or
15:19 on the Discord server that's what I'm
15:21 talking about so if you go to the salon.
15:23 there's a link tree there there's a link
15:25 to the Meetup there's a link to the
15:26 Discord it's all in there so go to that
15:30 and then what else do we want to talk
15:32 about I think that's it I think that's
15:34 it we'll get started here
15:37 so you tell me what you want to do and
15:40 we will figure that out have you ever
15:44 noticed anything strange when you ask a
15:46 question with logic
15:50 conflicts you mean with chat
15:53 GPT I don't know specifically what you
15:55 mean with a logic conflict let's let me
15:58 let me go ask chachie PT what the hell a
16:00 logic conflict
16:14 is come
16:16 on uh
16:18 oh oh wait H hang on chat gbt is there
16:25 it
16:28 is
16:31 can you describe to me what a logic
16:34 conflict is and how I might write a
16:36 prompt to give to you to see how you
16:39 respond to a prompt that has a logic
16:41 conflict in
16:43 it uh uh uh
16:46 uh oh
16:50 sorry logic conflict a logic conflict
16:53 arises when two or more statements
16:54 propositions directives contradict each
16:56 other thought that's what it
16:57 was examples of logic conflicts the
17:01 circle is square mutually exclusive
17:03 derivatives oh the old mutually
17:05 exclusive derivatives that old house
17:07 dress write a sentence without using any
17:10 words self-referential paradoxes this
17:14 statement is false ambiguous directives
17:17 writing a prompt with a
17:19 logic conflict draw a picture of a fish
17:23 that is also a bird I think it would do
17:25 that pretty well describe a a color that
17:28 is both entirely black and entirely
17:30 white write a story where the
17:31 protagonist is both alive and dead at
17:33 the same time I feel like I could handle
17:35 these but let me if you've got a
17:37 specific a specific prompt you put in
17:39 there that kind of stumps it let me do a
17:42 new chat I'll do it in gbt 4 since it's
17:45 theoretically
17:48 smarter the concept of a color both
17:50 being entirely black and white
17:52 simultaneously is paradoxical in nature
17:55 in traditional color theory physic color
17:58 and physics black is the absence of
17:59 color while white is the presence of all
18:03 therefore color cannot be both black and
18:05 white at the same time however if we
18:07 were to approach the question from a
18:09 philosophical standpoint we could
18:12 another way to interpret this would be
18:15 through the lens of Duality or
18:18 coexistence so no Kyle some folks are
18:21 asking for a Leonardo demo please okay
18:24 cool we can go do that we can go do that
18:28 um
18:29 I don't the I I I have not noticed any
18:32 particular um thing with logical
18:35 conflicts and probably evidenced by the
18:38 fact that um I had to look up what what
18:41 one
18:42 was
18:46 um Leonardo a okay so if you say launch
18:53 app
18:55 bow upgrade to become a true Alchemist
18:59 your Leonardo trial has
19:02 expired not
19:06 now introducing Alchemy V2 okay am I
19:11 logged
19:18 in ftuned models your models yeah I'm
19:22 logged in okay okay so let me go
19:27 home
19:31 so they've changed some things here hang
19:33 on let me let let me go let me go change
19:37 change some
19:38 stuff your username user API account
19:43 management one of the weird things with
19:47 um with Leonardo is it's got really bad
19:51 dark mode support you can't read
19:53 anything so let me go let me go flip my
19:56 system settings to light mode cuz I
19:58 think I think it's just going to make it
19:59 less
20:00 annoying light mode and then let me
20:03 reload
20:06 this no it's still doing it I don't
20:09 know I don't know why it's so
20:17 weird
20:20 huh okay so thing about
20:25 Leonardo is that the there's a lot going
20:28 on here so one of the one of the good
20:31 things about stable diffusion is that um
20:34 it's got a lot of features one of the
20:37 bad things about stable diffusion is
20:39 it's got a lot of features so Leonardo
20:42 does a decent job at um at
20:48 describing
20:49 um at capturing those features in in a
20:52 decently intuitive way there's still a
20:54 lot going on here
20:57 so okay so when you first get here these
21:01 things across the top are basically
21:03 Styles so in in um
21:07 in uh stable diffusion speak they are
21:11 why can I
21:13 not slide them left and right hang on a
21:21 sec oh there we go okay now I can uh uh
21:26 okay so across the top these are styles
21:28 so this one's photo real this one's
21:31 Leonardo diffusion XL this is Leonardo
21:34 Vision XL this is right so so they give
21:38 you kind of a representative image like
21:40 this one kind of looks like Pixar
21:41 characters that one looks like
21:43 cyberpunk this this one looks like
21:46 creepy
21:47 characters like here's a 50s cartoon
21:49 model so all of these are different
21:52 models of stable diffusion that that you
21:54 can just choose and like this one is
21:57 paper art style so if I click on that I
22:01 can see a bunch of images that have been
22:03 made with paper art style and then this
22:05 little button right here says generate
22:07 with this model so that'll pop me over
22:11 to the generation thing this is my sort
22:14 of image history down here look there's
22:18 there's me as a as a janky super
22:23 hero um so so let's see um let's do um
22:30 uh let's
22:34 see winding Mountain
22:38 stream with a jumping
22:41 trout and the trees in Fall
22:49 colors and a bear walking
22:52 down a path near the
22:57 water
22:59 you are currently on a free
23:04 [Music]
23:05 [Applause]
23:06 plan so wait hold on let me turn off V3
23:10 of alchemy apparently apparently you got
23:12 to pay for V3 of alchemy is that going
23:15 now yeah now it's going okay so so
23:19 they've got they've got some fancy Ash
23:21 version of this now so so what they're
23:23 doing is oh that was quick I one of the
23:26 things they they did was speed it up so
23:28 this this kind of looks like you know
23:29 that paper art
23:31 style and you know those are decent
23:33 Mountain paths do any of them have bears
23:35 in them no that one does well it's sort
23:38 of a bare
23:42 platypus um um uh Manatee it's like a
23:47 walking Manatee rather than a
23:50 bear um but anyway so so you you can use
23:54 those models okay so now let me go back
23:57 to the beginning the other thing thing
23:58 you can do is they they elevated the
24:01 generation prompt thing right here or is
24:04 this search I think this is just search
24:07 so I can now I guess I can filter so
24:09 here's trending here's new
24:11 images so I guess this is what people
24:13 are doing right now here's top images
24:17 maybe there's are rated ones so so like
24:20 a lot of these image generation tools
24:21 you can just scroll through here and if
24:23 you find something you like you're like
24:24 oh that thing's cool so someone did a a
24:28 car artwork of t-shirt graphic design
24:30 that's awesome so I can say generate
24:32 with that now I can come in here and
24:35 go 69 GTO
24:40 Judge we's see how quick it does does me
24:43 up a 69 GTO Judge t-shirt although I did
24:47 kind of kill the uh Kill The Prompt that
24:50 was in
24:54 there yeah it just did because I killed
24:57 the prompt it made it a photograph but
24:58 it's still pretty good um so so you can
25:01 do that
25:02 there let's see there's a Community Feed
25:06 um which which I think is what this
25:07 thing is down here on the bottom there's
25:09 a personal feed which is your stuff um
25:12 then there's training so these are so
25:15 you can train your own model this is one
25:18 of the big deals about this so here's
25:20 some here's a training set so if I say
25:22 edit this training set this is a bunch
25:24 of Photography that I've done of
25:26 dramatic clouds and sky
25:29 and I've got a particular way that I
25:30 edit them and you know they've got
25:32 they've got a look and feel so I put in
25:34 things that I felt were represented of
25:36 that I then built that into a model and
25:39 so over here there's there's the
25:42 training and data set section which is
25:45 where you create your models and then
25:46 there's this thing called fine-tuned
25:48 models where you can go use a model so
25:51 um Leonardo has its own models you can
25:54 use those are the things across the top
25:55 but there's also a list for for here's
25:58 Community model so here's what other
26:00 people have created and then here's my
26:02 models so here's three so there's two of
26:05 me with my big giant fat head these
26:08 These are really bad cuz I put in I put
26:10 in bad photos and so it just it just
26:13 makes me look like a 400 lb tubble Ard
26:17 every time so I wasn't happy with that
26:19 but my my cloud photos it did pretty
26:21 good so if I say use one of
26:24 these so I can now go to this model and
26:27 so that'll me in so here's the model
26:29 that I'm using and I can go
26:33 um dramatic
26:36 Sunset
26:38 Over
26:40 Pittsburgh with a view of the rivers
26:43 coming
26:45 together and this
26:48 should generate a series of photos that
26:50 has really nice pretty dramatic clouds
26:53 in
26:53 it oh I think I have to say in a Kyle
26:56 Shannon sky but let me let me let me see
26:59 what it gives
27:08 me yeah so there's there's some
27:12 beautiful pictures of
27:15 pittsbur with clouds that kind of look
27:17 like the stuff that I
27:19 do right so um and and it goes It goes
27:24 much deeper but but the the basic deal
27:27 is it's it's a lot of the features of
27:29 stable diffusion without having to go
27:32 install this on a on a you know sort of
27:36 command line interface or a a Google
27:39 collab notebook which which if you want
27:41 to geek out you can do that there's lots
27:43 more you can do um but if you just want
27:45 something
27:46 that's it's got a decent amount of dials
27:49 and switches you can also upload images
27:51 to use as a source image you know this
27:53 is this kind of um competes with mid
27:57 journey and some way with the dolly3
27:59 stuff in some way but so actually let me
28:01 let me go show you now so
28:04 that's I think this is a really good
28:06 example of where image generation has
28:10 been for most of this year and for most
28:13 of last year and look where we're headed
28:17 okay so I'm in chat GPT right
28:21 now and within chat GPT there's this
28:25 Mode called Dolly 3 right so this is the
28:27 new dolly 3 image model but but look at
28:30 what happens here rather than having all
28:32 of that crazy interface all that stuff
28:34 we need to do finally got my free hen
28:35 sample translation video back days later
28:38 yeah I know it's there if if you're
28:39 doing free free image or free video
28:42 translations on hey Jen it's definitely
28:44 days right now they're a little busy
28:48 so oh there's some cool images coming
28:50 into the to the uh Irregulars channel in
28:53 the salon I'll show those in a second
28:56 okay um
28:59 so there's really no interface here um
29:02 but there's this powerful image model
29:04 sitting underneath this and so what I
29:06 can do is I can
29:11 go I'm starting a new business that's a
29:15 car
29:16 wash that only uses
29:20 rainwater and soap that's made from
29:24 dried llama
29:26 skins period
29:28 I don't have a name for the business yet
29:30 so maybe you can help me come up with a
29:31 clever
29:41 name what AI is good to generate leads
29:43 for cold
29:45 emailing um I don't know right now on
29:48 that one dos backup
29:53 um go check out futurepedia
29:55 doio and um um see if there's anything
30:00 there for sales there's probably
30:01 something that's been built for that um
30:04 you could likely do it with one of the
30:07 um now that chat GPT is connected to the
30:10 internet and now that you know Bing is
30:13 and barred is you could probably go do
30:15 something all three of them kind of have
30:18 safety guard rails around people so it
30:22 there's probably not going to be a great
30:23 solution right now but there might be a
30:25 a standalone service and I would imagine
30:28 that there are um sales lead kind of uh
30:33 companies out there that are that are
30:36 putting AI on top of their systems so
30:39 there there's probably something
30:40 specialized for that I don't I don't
30:41 know of a specific thing all right uh
30:43 Eco llama Splash Nature's llama gleam
30:46 rain and llama
30:48 Radiance llama drop
30:51 ecash pure llama rain rinse llama Mist
30:55 Natural clean these are these are all
30:58 close
31:02 um can't we try some alliteration with
31:05 the
31:07 Llama let's see if we can see how many
31:09 L's we can get in the
31:11 name we can do this come on Chad GPT
31:15 absolutely alliteration can make a name
31:17 llama luster there we go perfect okay
31:20 llama luster it is so now I'm just going
31:22 to say llama luster it is now I need a
31:25 tagline not l molester come on chat GPT
31:31 really L
31:35 molester llama luster it
31:41 is oh man man alive get me in
31:47 trouble I can't see the text against the
31:49 white background um here how's that is
31:52 that
31:56 better nature shine Echo pure pure
31:59 rainwater pure Radiance nature powered
32:02 where llama luxury meets Echo
32:05 Elegance the natural way to a brilliant
32:08 day so I'm just going to
32:10 say
32:13 um okay I like number nine period can
32:16 you now give me five
32:18 different Creative Concepts
32:23 for a grand opening poster that we can
32:26 make period I don't want you to make the
32:28 images I just want you to give me five
32:31 Concepts and I'll choose
32:34 one so what I'm effectively doing here
32:37 is is kind of like creative directing or
32:39 doing creative strategy or marketing
32:41 strategy or brand strategy right so I'm
32:43 starting out going um I've got a project
32:46 I want to do and I'm going to make
32:47 images for this now I could come in here
32:50 and just ask for an image but I'm
32:52 purposefully not doing that because I
32:54 just want to compare the experience of
32:56 what we just did over in Leonardo which
32:59 was super kind of geeky and
33:01 Technical to this I'm just asking it for
33:05 [ __ ] and it's helping me ideate right
33:07 like I've got I've got chat PT in this
33:11 case is acting as a creative
33:12 collaborator the Llama landscape no the
33:14 rain dance celebration I like that all
33:17 right we're going to go with that
33:18 llama's day out um so what was that
33:21 number
33:25 two okay I like number two too I'd like
33:28 you to create a grand opening poster
33:31 that prominently features the name of
33:32 the business
33:35 includes the tagline toward the
33:39 bottom
33:42 and makes a very compelling image
33:48 for what the business is going to
33:51 be I don't know that is purposefully a
33:55 crappy client directive which if you've
33:59 ever worked with
34:00 clients that's a pretty good one
34:02 actually the fact that I had any point
34:04 of view as a client uh is is a little
34:09 miracle Joker Joker's making Ira Glass
34:13 Ira Glass images this dude totally looks
34:16 like Ira Glass did you ask for Ira Glass
34:19 that's pretty
34:20 funny doesn't he look like him that's
34:22 really
34:22 [Laughter]
34:26 funny
34:28 so so what's happening in the background
34:31 here is chat GPT is actually writing the
34:33 prompt so rather than me having to write
34:36 the prompt it's just doing it now it
34:39 didn't generate one of the images oh
34:41 yeah it
34:42 did
34:44 uh llama luster the natural way to a
34:48 brilliant
34:49 day nice Lama laser it almost got the
34:53 spelling
34:55 right natural way to brilliant day with
34:58 a with a car from
35:01 1962 llama luster the natural way to a
35:04 brilliant day it's actually pretty good
35:06 it's dirty on one side it's clean on the
35:08 other or whatever the hell that means um
35:13 so but it put all of these in these
35:15 weird sort of scenes like it's a
35:17 billboard so we're going to
35:21 say the ideas aren't bad but I just want
35:25 the poster to be a standalone poster
35:27 that is vertical
35:30 orientation rather than showing a
35:32 billboard in some other scene which is
35:34 what you created here please create a
35:37 standalone grand opening
35:40 poster
35:42 that should have the words grand
35:45 opening as well as the name of the
35:47 company And the
35:50 tagline don't include more than three
35:56 llamas
36:00 and make the concept of rainwater more
36:04 prominent so now I can now I'm just
36:06 giving it you know kind of creative
36:08 director kind of feedback just like you
36:11 would give an art director right that
36:12 came in with those Concepts You' be like
36:14 okay we you know we've got something to
36:16 work with
36:18 here um so this is so this is where I
36:21 think
36:25 um I I'll use nft term here so so the
36:28 nft Bros we'll often talk about normies
36:32 and the normies are people that are not
36:34 you know super into nft culture so I
36:37 think what's going to happen is we're
36:38 going to start to see a divide where
36:40 you're going to have Specialists that
36:42 geek Out On Tools like mid journey and
36:44 stable diffusion and things like that
36:46 and then you're going to have just
36:47 people that are using these tools and
36:49 these tools are going to get
36:50 increasingly intuitive and increasingly
36:54 the the complexity of
36:58 the Tactical execution is going to be
37:01 handled by the large language model so
37:03 our job as people is just going to be to
37:05 know what we want and ask for it and so
37:09 what so what's the skill there the skill
37:11 there is going to be knowing what you
37:14 want right understanding what's your
37:16 goal what's your target audience what
37:17 are you trying to do and then and then
37:20 the skill after that's going to be a
37:21 curation skill okay that's good that's
37:24 bad let me steer it in a direction and
37:26 then the skill is also going to
37:28 be rapid context shifting between
37:32 writing and image generation and music
37:34 generation and things like that right so
37:36 so this points to a future that is that
37:39 is much more seamless in terms of
37:42 creative
37:44 execution llama grand opening notice we
37:47 only have three llamas now we got some
37:49 rain coming out of the sign and llama is
37:52 spelled very
37:54 creatively um grand opening
37:57 completely missed the name there grand
38:01 opting
38:04 grand if you're wondering why the
38:06 spelling is shitty it's just because the
38:09 way uh diffusion models work the fact
38:11 that it can get the words even close is
38:13 actually a minor miracle it's probably
38:15 actually closer to a major Miracle um
38:18 but anyway that's that's Dolly 3 um it's
38:21 it's a pretty remarkable tool and it's
38:24 brand new so if you pay for chat GPT
38:29 plus you get it it'll just show up in
38:32 your list of options if it's not there
38:35 yet it's rolling out over the month of
38:38 October they have a big meeting coming
38:41 up
38:42 uh um Apple user hey Kyle and labster
38:45 I'm already
38:49 late oh that's funny that funny that fun
38:52 fun funny Let's see we got a nice little
38:55 AI learning lab image here here look at
38:58 that how cool is that that's very
39:00 psychedelic that's very
39:02 psychedelic how can you learn anything
39:04 in that environment oh cuz it's AI you
39:07 just learn you just learn learn learn
39:08 learn Lear learn all right fantastic
39:13 Bob let's go look at some other
39:15 questions what can we do here what can
39:18 we do oh wait you know what I want to do
39:20 I want to show some of the artwork that
39:22 people in the AI Salon have been putting
39:25 up there's some cute stuff coming
39:28 in Ted looks at life killing it with the
39:31 AI learning
39:34 lab digital merch that one's
39:38 awesome this AI shit's
39:43 coming every time someone says chat gbt
39:45 I always hear you say you can make money
39:47 with chat
39:49 gbt you can make money with
39:53 ch thank you I think I'm sorry
39:59 the AI learning lab fun house
40:04 awesome that was brilliant who did that
40:07 joker was that joker oh that was Sherry
40:09 D and then Joker did the Ira
40:12 Glass the Ira Glass AI learning lab but
40:16 look AI learning lab is spelled correct
40:18 that's pretty
40:21 special um what else we got
40:25 here
40:28 a bear oh this is the when I mentioned
40:29 the bear the manatee and the and the
40:32 jumping
40:36 trout by a mountain stream very
40:39 nice um this is a llama this one's cool
40:44 I don't know what the prompt is there
40:45 but a llama in a dream workshop
40:49 maybe pretty sweet sherry D killing it
40:53 Ted looks at life llama
40:55 luster
41:00 Touch free cleaning that's
41:04 awesome your llamas will not be
41:08 molested oh
41:10 [Laughter]
41:13 man this place is cool all right so you
41:18 know you know what's funny hey those
41:21 images are funny
41:23 but
41:25 um
41:27 one of the things that's been striking
41:30 me a lot lately is
41:37 that I think work is going to get more
41:42 fun and and this might be a shortterm
41:46 thing
41:49 but one of the things we've talked about
41:52 in here is that
41:54 the the friction between an an idea in
41:58 your head and the expression of that
42:01 idea the friction of that is going to
42:06 zero and and we're already experiencing
42:08 it here like as I'm talking ideas and
42:11 saying stupid [ __ ] people are taking
42:13 those ideas popping them into Dolly 3
42:16 generating images and sharing them and
42:19 it's kind of like work and learning as
42:23 entertainment and and why is that
42:25 happening because
42:28 these tools are so fast and so powerful
42:31 that rather than having to go get a
42:33 4-year art degree and spend another four
42:36 years learning how to use Photoshop in a
42:38 competent way and then another four
42:41 years or a lifetime getting good at you
42:44 know illustration or image generation
42:48 before you can manifest an idea visually
42:51 you can now just ask for it and there it
42:53 is and it's like that with every tactic
42:57 iCal execution whether it's coding or
43:00 writing or consulting or project
43:04 management or like all of this stuff
43:06 we're just going to be able to ask for
43:08 things and bang there it is
43:12 so rather than having that frustration
43:14 of oh I have this idea and then
43:17 realizing oh [ __ ] I don't have the
43:20 education or the skills necessary to
43:23 bring that idea to life so I either have
43:25 to raise money
43:27 or partner with someone who does have
43:29 the skills and if I don't if I don't
43:31 have a big network if I didn't work in
43:33 an agency or have a big Network that's
43:36 tough and now all of a sudden you can
43:39 have an idea and go your your initial
43:41 instinct is going to be oh I don't have
43:43 the skills for that D oh yes wait I
43:45 wonder if AI could do that oh
43:49 my and so so I think that we're we're
43:54 we're I think that we're going to
43:56 transition into this place
44:00 where the more
44:04 you lean on these tools to augment who
44:08 you
44:09 are and what you do combined with the
44:12 fact that these tools are getting
44:13 dramatically better dramatically
44:15 fast I think we're moving into a place
44:18 where work's going to start to get fun
44:19 and exciting and like a video
44:23 game except for the politics part but
44:27 chbt can help you navigate the politics
44:29 too
44:31 right so kind of cool kind of cool I'm
44:35 going to leave and try to figure this
44:37 out what what was going
44:40 [Music]
44:46 on what's going on
44:49 t-shirt says text
44:52 bad tried to uh my wife is watching
44:55 football and I'm watching this I win
44:57 that's awesome Jim
45:00 Ross happy Saturday
45:03 night it's the Saturday night Saturday
45:05 night NFL
45:06 game honey I I'll be watching Tick Tock
45:10 what did you do Saturday night I don't
45:12 want to talk about
45:16 it
45:18 yeah this is exciting [ __ ] Jim Jim Ross
45:22 is a is a great guy he's he's in here um
45:25 he's been using the stuff for a while
45:27 he's killing it he just wrote a book he
45:29 wrote a book that he's using kind of as
45:31 a business calling card and it's working
45:34 he got a meeting or something out of it
45:36 I don't know some some sort of something
45:38 sales opportunity just like within a day
45:41 of publishing his book that he wrote in
45:43 I think what two hours Jim you put that
45:46 together and then published it and he
45:48 now has it in his hand he can hold it up
45:50 and go here's the book I wrote on on my
45:52 industry how long till we see Dolly 3
45:55 kiosks at the moment print um frame
45:58 Christmas gifts bam so I'm actually an
46:01 adviser to a company that's doing that
46:03 Ted not kiosks and malls but um but that
46:07 exact kind of thing uh we're using
46:09 stable diffusion right now but we might
46:11 move to we might move to do because
46:13 you're going to be able to call it just
46:14 like you could call do too and I I would
46:17 think you probably can do that right now
46:19 I just haven't looked into it wondering
46:21 about how we advocate for higher money
46:24 because of this un unique skill set
46:28 set and self-driven up skill I think
46:32 that will naturally happen so cuno I
46:34 mean you're already killing it with this
46:36 right so one of the things I talk about
46:38 a lot on the lab here is number one get
46:41 curious number two get adventurous start
46:43 playing and number three talk about
46:47 it talk about it post it to your
46:51 LinkedIn talk about it at dinner with
46:54 your family talk about it when you go
46:55 out for drinks after work talk about it
46:58 at work but Kyle they're going to think
47:00 I'm a weirdo [ __ ]
47:02 them you are a weirdo if you're playing
47:06 with this stuff right now cuz we're very
47:08 early so there's a lot of fear right now
47:12 there's a lot of misinformation there's
47:13 a lot of bad assumptions about what
47:15 these things
47:17 are so what ends up happening is if you
47:19 get curious about this stuff and start
47:21 playing with them what you realize is oh
47:23 they're not as scary as I thought and
47:25 then at the same time they're they're
47:27 way more profound than I thought both of
47:29 those happen so you kind of have this
47:32 intense kind of like oh [ __ ] this is
47:34 actually a big deal you have that
47:37 moment and then and then you start
47:40 playing with them and you start sharing
47:42 it with the world and what's going to
47:43 happen cuno I
47:45 think is
47:48 that okay i' I've I've I've talked about
47:51 you've seen me sort of talk about this
47:52 before I I can't not talk with my hands
47:55 so tough
47:57 [ __ ] so the the the generative AI
48:01 industry for lack of a better term it's
48:03 it's it's kind of as broad as saying the
48:05 internet industry right the internet is
48:07 kind of ubiquitous and everywhere AI is
48:09 going to be like that too
48:11 so but but right now it's a fairly
48:14 discreet distinct thing right it's it's
48:17 very very skinny and it's it's about to
48:19 do this it's about to sort of expand out
48:21 and just get really big and really
48:23 ubiquitous and be
48:24 everywhere when starts to do that and
48:27 it's probably starting to do it
48:29 now there's going to be an increasing
48:32 demand for anyone that knows [ __ ]
48:35 anything about
48:36 this and so I think that we will all
48:40 anyone who's who's let's just call it AI
48:43 literate to some degree I think we we
48:46 will all be able to either have side
48:48 hustles where we talk about you know we
48:50 we do workshops we do education things
48:52 like that um it doesn't even necessarily
48:55 need to to be sort of a a side hustle
48:59 where you're charging money for that it
49:00 could be you're doing those workshops
49:01 within your company which I know you're
49:03 doing
49:05 cuno
49:07 um and and at some point it's going to
49:10 be clear that people are going to need
49:12 help for people that know what they're
49:13 doing so if you've established yourself
49:15 as a thought leader who's AI literate
49:18 there's just going to be opportuni
49:19 sitting out there um I'm shy but all my
49:23 co-workers know I'm the AI guy see tired
49:26 Runner it's perfect and listen you don't
49:28 need to you don't need to be me right
49:29 you don't need to be like hey hey it's
49:31 awesome you know en join my little lives
49:34 you don't have to be that you can be one
49:36 of the quiet guys but still talk about
49:38 it right and
49:40 and you know one of the things tired
49:43 Runner that you may have the opportunity
49:45 to do
49:46 is make this AI
49:49 stuff lowkey and less scary for people
49:53 cuz there's a lot of fear one of the com
49:56 comments last night was
49:58 um someone said as they've been talking
50:01 about this stuff in their company
50:03 they've been getting downright hostility
50:05 right so so people are not only afraid
50:07 they're angry so I think taking a taking
50:11 a you know
50:13 low-key you know shy kind of approach
50:15 but just sort of stand in you know I'm
50:19 excited about this stuff this stuff's
50:21 powerful let me show you how to do it I
50:23 think it's going to serve you they come
50:25 to me asking for help using Bing chat
50:27 I'm like yeah sure yeah and and see
50:30 that's the other thing
50:36 is especially depending on the the
50:39 political culture of your company you
50:41 may be in a culture where people kind of
50:44 hoard information and they hoard
50:46 knowledge because there's power if
50:48 you're the only one that can do it
50:52 right this AI stuff's going to be so
50:55 ubiquitous and it's changing so
50:58 much my strong um recommendation is be
51:02 very generous with just give away all
51:04 the secrets like don't hoard any don't
51:06 hoard any information if you learn how
51:09 to do a really cool prompt share the
51:10 prompt well what if what if they take
51:13 that and turn that into a business [ __ ]
51:14 it
51:15 great cuz three months from now the
51:18 technology is going to change
51:19 anyway so so their business now will
51:22 just be a feature in whatever the next
51:24 tool is that comes down the line
51:28 um but it's it's it's like shifting into
51:31 um what do they call that a
51:34 uh like a a sense of abundance rather
51:37 than scarcity
51:39 right so so treat it like that so so
51:42 just share just put it out
51:46 there PO let you create great Bots
51:49 really fast yeah I agree the the the the
51:51 ability to create a bot on PO is um is
51:55 is super powerful and
51:59 um if you're just getting started with
52:02 your
52:03 prompting what's good about creating a
52:05 bot on PO that you can sort of put out
52:07 there is that you you'll write a prompt
52:09 then you you publish the bot and then
52:12 you start playing with it and it'll
52:13 start giving you crappy answers and
52:15 you're like oh crap so you got to go
52:16 modify the bot and then you launch it
52:18 again and then it'll give you more
52:19 crappy answers and you go modify the bot
52:21 and
52:22 eventually you get it to the point that
52:24 it's kind of doing what you want to to
52:25 do that's that's pretty gratifying it's
52:27 a good way to practice your
52:33 prompting Ted Ted looks at life Ted Ted
52:37 is no longer even listening to to the AI
52:39 lives he's just off he's just off making
52:42 marketing materials for the for the AI
52:45 learning lab that's [ __ ]
52:48 awesome that's
52:52 great wait who is this Joker yeah look
52:54 at Joker J Joker's a more a fan of it's
52:57 a it's a different kind of marketing
52:59 campaign that Joker's going
53:05 after fling all the poop you can we'll
53:09 make
53:10 [Laughter]
53:12 more but you know like I I actually you
53:15 know what's funny like that's the
53:17 attitude you should have with like
53:20 sharing your knowledge as you're
53:22 learning things just fling it out there
53:26 put it on LinkedIn hey I made this
53:29 prompt and it did a thing and and it was
53:31 cool put it out there I I went to chat
53:34 GPT and I thought I could do this and it
53:36 failed miserably put that out
53:39 there you know I thought the robots were
53:41 going to kill us and now they're my best
53:43 friend put that out there just just
53:46 start flinging knowledge
53:48 poop there there there's a there's a
53:51 t-shirt the AI learning lab flingers of
53:54 knowledge poop since 20
54:03 22 what's the best way to launch your
54:06 Bot here I'll go let's go make a little
54:08 tricks let me show you a cool trick I
54:10 did with um
54:13 [Laughter]
54:18 Poots Waka so I'm going to go to PO
54:22 there's
54:24 po all right so when you come to
54:30 PO you have couple of things you've got
54:33 this is your little chat box so this is
54:35 just like chat GPT or whatever you can
54:37 use their model you can use their model
54:40 connected to the internet um chat gp4
54:43 and then there's a more button and
54:45 there's lots of different models you can
54:46 play
54:47 with so you can do all
54:51 that and then down here there's these
54:53 official Bots so there's all the
54:57 different flavors of chat GPT there's
54:59 Google's Palm model here's the Llama
55:01 models the 70 billion uh parameter model
55:05 from meta here's the 34 here's the 13
55:09 here's code llama 13 and 7B so there's
55:13 lots of stuff you could play with
55:15 here and
55:21 then here's popular Bots and new Bots
55:23 here's all these Bots that other people
55:25 have made well what are these
55:28 well you can go explore them here and
55:31 find find a bunch but let's see here's a
55:34 bot that I made called um I am
55:38 Shakespeare so this is this is a virtual
55:41 Shakespeare so if I type in
55:44 um why are
55:50 relationships so hard question mark Yan
55:54 bard relationships are ever prone to Vex
55:58 and try for hearts of men of diverge and
56:01 reasons why L's tied doth ebb and flow
56:04 Neath moon and stars its calm so CE or
56:09 cers disrupted by storms that Mar so
56:13 I've written a bot that responds an i
56:15 Amic
56:16 pentameter so so that's kind of fun um
56:20 and I think I'm only using um I think
56:22 I'm using the Claud model here or
56:24 whatever it is but anyway so let's let's
56:25 go let's go in and and look at this
56:29 bot um let me make it bigger so you can
56:32 see it
56:38 potentially so you can put a picture in
56:41 there where did I get that picture I
56:43 made it in mid Journey just go like this
56:45 is the thing like it used to be like if
56:48 you're were going to make a bot like
56:49 this you're like oh God I got to go find
56:50 a picture of Shakespeare then you go to
56:52 Google and you spend like half an hour
56:54 trying to find a [ __ ] picture
56:56 and then and then you you'd install
56:58 malware in your computer cuz you went to
56:59 the classic paintings site that was
57:02 actually a
57:03 malware anyway so now you just go make
57:05 [ __ ] um okay so you create a name of the
57:09 bot so I am Shakespeare it's got to be
57:11 unique and then um then you create your
57:15 prompt so what I wrote was you're a
57:17 chatbot embodying the renowned play
57:20 playright William Shakespeare engage
57:22 with users as if you are crafting a play
57:24 for them using authentic Shakespearean
57:27 language and phrasing your responses
57:29 must adhere to I Amic pentameter with
57:31 stanzas extending Beyond two lines if if
57:34 needed your memory of your written works
57:37 is impeccable so I wanted it I wanted it
57:40 to be able to like if someone asked
57:42 about a specific play I wanted him to be
57:44 able to reference those um maintain
57:47 familiarity with historical context of
57:49 your time to enrich the conversations
57:52 your main focus should be providing
57:53 original content in response to
57:55 inquiries ensure that they that you
57:58 remain courteous and respectful in all
58:00 interactions as befits your persona
58:03 right so so um relatively
58:06 straightforward prompt it you know it's
58:09 got some stuff in it how did I get to
58:11 this I probably wrote like 15 or 20 of
58:14 these I kept writing it and then it
58:16 would do something weird and then I'd go
58:18 in and go oh let me modify that and then
58:21 i' come back and I go in and I'd modify
58:23 it and then I'd play and I'd go back and
58:24 modify it
58:25 um so so there's your prompt oops what
58:29 happened there what the hell was that it
58:31 just popped me somewhere else okay um
58:34 then you choose your model right so you
58:37 could use gp4 you can do Claude 2
58:41 100,000k so if you wanted to do a bot
58:43 where you could upload really big
58:45 documents you could do
58:46 that um you can choose stable diffusion
58:49 XL so that's an image generation model
58:52 so you can make a bot that generates
58:54 images and you could put in a prompt
58:56 let's say you wanted to make a bot that
58:58 only makes 70s muscle car magazine
59:01 photos you could write a bot a prompt
59:04 that's like a really killer um you know
59:08 image prompt and then choose the stable
59:10 diffusion XL thing and now you have a
59:12 bot that makes muscle car
59:15 photos um there's a you know greeting
59:17 message and then there's a bio for the
59:21 thing there's a bot preview here so
59:25 that's that so here's what we're going
59:27 to
59:27 do I'm G to take that prompt for
59:30 Shakespeare we're going to go over to
59:31 chat
59:32 jpt you can make money we try TR our T
59:36 and I'm going to
59:39 [Music]
59:40 say below is a prompt that I wrote for a
59:47 bot that acts like William
59:53 Shakespeare so I'm going to hit shift
59:55 turn I'm going to paste in my
59:57 prompt I'm going to
1:00:05 [Music]
1:00:12 say I'm just trying to think of a of a
1:00:15 character that might be
1:00:16 good that we that we all we kind of know
1:00:20 the references to
1:00:23 um any idea is let's see am I at the
1:00:26 bottom yet give me an idea for for a
1:00:29 character that we could do Tom Cruz well
1:00:32 but would we know what Tom Cruz sounds
1:00:33 like give me someone who Horatio it's a
1:00:36 little Aristotle Aristotle's
1:00:42 interesting Puck no don't go Shakespeare
1:00:45 just go just go like modern character
1:00:47 Robert dairo I don't know if he's got
1:00:51 lines let's
1:00:54 see maybe Quenton
1:00:56 Tarantino llama Cruz Eric
1:01:02 Cartman
1:01:04 Kramer um let's do Kramer all right um
1:01:07 so so here's what I'm going to
1:01:09 do
1:01:11 rewrite this bot
1:01:13 prompt so that I can create a bot for
1:01:17 the character of Kramer from the sitcom
1:01:22 Seinfeld
1:01:23 period
1:01:27 and now we're going to make
1:01:31 this you're a chatbot channeling the
1:01:33 quirky Persona of Cosmo Kramer from the
1:01:36 iconic sitcom Seinfeld engage with users
1:01:38 as if they're part of a comical everyday
1:01:40 scenario in Jerry's apartment notice
1:01:43 what it's doing it took that really
1:01:45 specific thing I wrote about Shakespeare
1:01:47 and plays and I Amic pentameter it
1:01:50 interpreted it and completely rewrote it
1:01:52 this is this is like one of the
1:01:54 Brilliance of of chat GPT particularly
1:01:57 for is that it understands a level of
1:02:01 nuance in language that is just it blows
1:02:03 my mind every time I see
1:02:05 it um everyday scenarios in Jerry's
1:02:09 apartment employing the character's
1:02:11 unique mannerisms offbeat humor and
1:02:13 unconventional wisdom your responses
1:02:15 should mirror Kramer's distinctive
1:02:18 speech patterns and an erratic yet
1:02:20 endearing Behavior your memory of the EP
1:02:23 episodes is also impeccable
1:02:26 right okay so keep your in so so it just
1:02:29 wrote us this perfect prompt so let's go
1:02:31 make a new bot so now we have a
1:02:34 prompt right yeah isn't this cool thank
1:02:37 you Marina appreciate that um the gifts
1:02:40 really make a difference I appreciate it
1:02:41 a lot
1:02:43 um so let's go back to po po
1:02:49 po and
1:02:53 go and so in this leftand
1:02:57 um what is this called menu whatever the
1:03:01 [ __ ] it's called Channel there's create
1:03:04 a bot this this little button right here
1:03:07 is how you create a new
1:03:11 bot and so we're going to call this I
1:03:15 let me just see if we'll see if Kramer
1:03:17 I'm I'm I'm sure that Kramer is taken
1:03:21 but if not this will be po.com
1:03:23 Kramer uh um okay so the prompt so we're
1:03:26 just going to paste The Prompt in that
1:03:27 chat GPT just wrote we're going to use
1:03:30 gp4 because it's just better the
1:03:33 greeting message oh
1:03:36 um let me go back to chat GPT for a
1:03:43 second oh wait hang on you know what
1:03:45 I'll uh yeah
1:03:48 yeah
1:03:51 um please write a greeting message from
1:03:55 Kramer that's a
1:03:58 short sentence that greets a person who
1:04:02 starts using the
1:04:12 bot and again I could have written that
1:04:16 but why hey buddy welcome to crer
1:04:20 Industries the land of wild ideas and
1:04:23 Wilder hair what's cooking all right
1:04:25 that's fine whatever is that perfect no
1:04:27 doesn't matter so that put that in there
1:04:31 um Advanced let's see the
1:04:34 bio
1:04:39 um please write a bio for the bot in the
1:04:42 first
1:04:43 person and include something that we
1:04:46 don't know about
1:04:48 Kramer and that'll drive Seinfeld people
1:04:51 nuts trying to find the episode that
1:04:53 that's in hey there I'm Cosmo Kramer
1:04:55 your hip door busting
1:04:58 neighbor
1:05:00 um okay that's not good all
1:05:03 right that's way too long and Kramer
1:05:06 should not be aware of his own hipness
1:05:09 and
1:05:12 coolness it should be weird but not
1:05:20 self-referential oh hey there name's
1:05:22 Kramer you've stumbled on my world vah
1:05:27 well whatever that's fine so we'll just
1:05:29 copy that it's not great I would I would
1:05:31 rewrite this if I were actually doing
1:05:33 this so pop that in the bio access um I
1:05:38 we want to make the bot publicly
1:05:42 available uh uh if this setting is
1:05:46 enabled it will be added to your profile
1:05:49 and will be public
1:05:52 available and then
1:05:55 there's a there's also an option here
1:05:57 where you can make the prompt available
1:05:59 I'm going to turn that off because just
1:06:01 because um I like having these things
1:06:04 have a bit bit of mystery behind them
1:06:07 and then we also need an image of Kramer
1:06:09 so let's go back to chat
1:06:11 J and we'll do new chat am I in no we'll
1:06:16 go Dolly
1:06:19 3
1:06:22 create a square image
1:06:25 of Cosmo Kramer from the hit sitcom
1:06:29 Seinfeld with a wacky
1:06:34 face we see what we get it it might not
1:06:38 do it there might be a safety thing I'm
1:06:39 sorry I cannot yeah so okay so chachy BT
1:06:42 can't do it um so so screw that we'll go
1:06:46 over to Mid Journey so now I'm over in
1:06:49 Discord we're going to do a mid Journey
1:06:53 image and the way you do mid Journey
1:06:55 images is we'll go slash
1:07:04 imagine photo of closeup of Cosmo
1:07:08 Kramer's face from the sitcom Seinfeld
1:07:12 entering Jerry's
1:07:17 apartment and this should give us a nice
1:07:19 little square thing I've got a few Bots
1:07:21 that run locally that won't run question
1:07:25 question Einstein bot splits the
1:07:32 atam oh
1:07:34 man oh and and by the way what I just
1:07:37 did there where I was like I was in Del
1:07:40 and DOI was like no I can't do that
1:07:42 because of safety guard rails you could
1:07:45 probably spend some time and work your
1:07:47 way around that guard rail but one of
1:07:49 the reasons to mess around with these
1:07:51 tools is um that that
1:07:55 you just always run into roadblocks so
1:07:57 if you can just quickly switch between
1:07:59 tools and do and do different stuff it's
1:08:01 great these don't really look like
1:08:02 Kramer but that's all right we we'll do
1:08:05 this crazy stupid
1:08:08 one so so these these letters here in
1:08:11 dis in in uh mid Journey if you haven't
1:08:14 seen them U stands for upscale v stands
1:08:16 for variation so I'm just going to grab
1:08:18 that one and upscale it so U2 so that'll
1:08:21 just give me a big version of
1:08:23 this
1:08:24 [Laughter]
1:08:26 so there it is it's Kramer enough like
1:08:29 again I would spend more time on this if
1:08:31 uh if I wasn't just trying to rip
1:08:34 through it
1:08:35 here so we're going to save that to the
1:08:40 desktop we're going to go back to
1:08:43 PO and then we're going to stick our
1:08:45 little image
1:08:50 in desktop photo of Kramer so there's
1:08:53 Kramer so we got we got Kramer now and
1:08:55 we're going to say create the bot so now
1:08:59 we'll go test the
1:09:00 bot bot handle is already taken so I'll
1:09:04 do what I've been doing I am
1:09:10 Kramer
1:09:13 ha hey there name's Kramer hey buddy
1:09:16 welcome to kramerica Industries the land
1:09:18 of wild ideas so let's see
1:09:23 um why are relationships so
1:09:27 hard see what he comes up with well
1:09:29 buddy you see relationships are like the
1:09:31 game Frogger you're always dodging and
1:09:33 weaving trying to avoid oncoming traffic
1:09:35 from One log to another sometimes you
1:09:37 just make it to the other side sometimes
1:09:38 you end up in the in the drink but
1:09:40 remember when I dated Miss Rhode Island
1:09:43 the Miss Rhode Island contestant well
1:09:46 she was a beauty but boy she had high
1:09:47 maintenance so so it kind of works so if
1:09:50 you want to go play with it go to po.com
1:09:54 slash am Kramer k r a m e r and you can
1:09:59 play with the bot I just did um in fact
1:10:02 what I'll also do I'll go turn
1:10:05 on um I'll turn on the ability to see
1:10:09 The Prompt as part of the bot and so you
1:10:12 can go copy and paste The Prompt there
1:10:14 you
1:10:17 go um difficult oh that's cool you can
1:10:20 continue your last
1:10:22 chat so let's see
1:10:24 I am
1:10:27 Kramer I guess you just go over
1:10:32 here operated by Kyle Shannon view
1:10:35 prompt so over
1:10:37 here when you go to um po.com
1:10:42 Kramer there's little three dots
1:10:45 here and the first choice is view prompt
1:10:49 and there's the prompt so you can just
1:10:50 copy and paste that and then just do
1:10:52 what I just did take it into chat gp4
1:10:55 and and and write it and and
1:10:58 listen here's why I'm reticent about
1:11:00 just giving out a prompt like that it's
1:11:03 really easy just to copy a prompt and
1:11:04 use it but you don't [ __ ] learn
1:11:06 anything so go make a prompt for some
1:11:08 other person and start messing with the
1:11:11 prompt and change it and you'll probably
1:11:13 come up with you can give it rules do do
1:11:15 these things don't do these
1:11:17 things uh if a person isn't a subscriber
1:11:20 to poke can they still use it I think
1:11:23 you can still
1:11:26 make I think you can still make Bots if
1:11:29 you're if you're a non-subscriber you
1:11:32 may not have access to like gp4 and the
1:11:36 Claude 100,000 token things cuz those
1:11:39 are part of the subscription but I think
1:11:40 you might be able to it's in it's in POS
1:11:43 interest for people to make as many Bots
1:11:45 as
1:11:46 possible so
1:11:49 anyway all right that
1:11:51 was there was a rabbit hole
1:11:55 that was fun I love you
1:11:58 Irregulars I think I think um who made
1:12:01 this this was
1:12:03 uh why is that here we go oh I see
1:12:07 what's going
1:12:08 on I think this was uh lunatick made
1:12:12 this is that
1:12:16 [Music]
1:12:22 right
1:12:29 all right I don't know all right so
1:12:32 welcome everybody if you're new here my
1:12:35 name is Kyle Shannon this is the AI
1:12:36 learning lab I just emerged out of a
1:12:38 deep Rabbit Hole tunnel um showing
1:12:41 people how to make a bot using po which
1:12:44 is pretty
1:12:46 cool um and a good way to learn
1:12:49 prompting and a good way to understand
1:12:52 that we're entering a
1:12:56 Zone where you really don't need to be a
1:13:00 technologist to do incredibly
1:13:02 sophisticated things
1:13:05 um you don't need to be an artist to do
1:13:07 art you don't need to be a writer to do
1:13:09 creative writing you don't need to be a
1:13:11 coder to do coding now is someone who's
1:13:15 not a coder going to be as good at
1:13:17 coding as someone who is a coder that
1:13:19 uses the same tools of course not anyone
1:13:21 who's got a vertical specialty
1:13:24 is going to be really good using these
1:13:27 tools
1:13:29 but the other
1:13:31 people are going to be like 60% as
1:13:35 good at
1:13:38 anything
1:13:41 so if you ever deal with that
1:13:44 self-limiting [ __ ] that I have spent
1:13:47 my life battling I'm not good enough I'm
1:13:50 not smart enough I didn't go to the
1:13:52 right school I didn't learn that I'm too
1:13:55 lazy to go back to
1:13:56 college all that [ __ ] that that's
1:14:03 gone we are now in a world today with
1:14:07 chat GPT especially chat GPT 4 I think
1:14:10 it's the best 20 bucks a month you could
1:14:15 spend we're in a world today where your
1:14:19 lack of knowledge about any given
1:14:22 profession or skill
1:14:28 is instantly
1:14:31 fixed there's a there's a I did one of
1:14:34 the one of the videos on my channel is
1:14:36 the 10,000 hour rule is
1:14:39 dead and um so Malcolm Gladwell in his
1:14:43 book outliers popularizes this concept
1:14:46 of the 10,000 hour rule which is
1:14:48 basically if you want to get good at
1:14:49 something you need to put in 10,000
1:14:51 hours which roughly is 3 to four 4 years
1:14:55 right you go back to school or you just
1:14:57 commit to something and you're going to
1:14:58 [ __ ] learn it and you learn
1:14:59 songwriting or you learn whatever it is
1:15:01 and about 10,000 hours later you have
1:15:03 some level of expertise and
1:15:08 Mastery you don't need four years
1:15:11 anymore can you imagine if you spent two
1:15:15 weeks with chat GPT on a single subject
1:15:19 how good you'd get at it I mean the
1:15:21 exceptions are things that require phys
1:15:23 physical dexterity like playing an
1:15:25 instrument or juggling or you know some
1:15:27 some physical thing but any knowledge
1:15:29 work most creative
1:15:32 things you've got instant access to that
1:15:38 knowledge and you can go from a
1:15:41 non-published author to a published
1:15:45 author in a short amount of time you can
1:15:47 go from I've never written a computer
1:15:49 program to I just launched an app in the
1:15:52 app store
1:15:55 in a couple of
1:15:56 weeks without having to
1:15:59 have money or a programming partner well
1:16:04 how do you do that I don't just go ask
1:16:06 chpt to teach it to you and it will
1:16:09 it'll write the code for you it'll tell
1:16:11 you how to download you know the xcode
1:16:15 development environment
1:16:17 and all that
1:16:22 [ __ ] never written a marketing plan but
1:16:25 you're kicking off a new business guess
1:16:27 what now you can write a marketing plan
1:16:29 for your new
1:16:30 business what's the name of the new tech
1:16:32 that is supposed to Leap Frog chat gbt
1:16:35 it's um it's Google's Gemini
1:16:40 model they're claiming it's five times
1:16:43 better or stronger or faster than chat
1:16:47 gbt 4 which I
1:16:50 doubt
1:16:52 but even if it's as good even if it's
1:16:57 close to as good as chat
1:16:59 gp4 holy
1:17:01 [ __ ] cuz then we're going to
1:17:04 have you
1:17:05 know we've already got open Ai and chat
1:17:09 GPT and Bing right which they're kind of
1:17:12 you know cousins right
1:17:15 now and then we're going to have Google
1:17:17 that's got stuff that's as
1:17:19 good and then we've got whatever Elon
1:17:21 musk's going to come up with and then
1:17:23 we've got whatever claude's going to do
1:17:25 so claude's the xop AI
1:17:28 people meta claims to be cooking up some
1:17:31 [ __ ] that's really good all the
1:17:33 companies are claiming they have stuff
1:17:34 that's as good as chat GPT for but no
1:17:39 one's hit it yet but I think Gemini's
1:17:42 seems like it's going to be
1:17:45 there I taught my special needs sun to
1:17:48 program in Python in one day using chat
1:17:52 gbt
1:17:54 I on China like that just gave me
1:17:57 goosebumps like in this channel I do a
1:18:00 lot of these silly kind of things I
1:18:01 think
1:18:02 access for blind people special needs
1:18:07 people deaf
1:18:09 people people that you don't speak their
1:18:11 language maybe people that can't speak
1:18:15 are going to be able to
1:18:17 speak this shit's [ __ ] profound it's
1:18:22 profound
1:18:28 it's
1:18:29 crazy it's amazing thank you for sharing
1:18:33 that yeah Herod D Herod just said thank
1:18:37 you so much for sharing that with us
1:18:38 exactly like that's really generous
1:18:40 thank you for sharing that super cool
1:18:42 thank you for the
1:18:47 rose imagine if you
1:18:51 had what what do they call those people
1:18:53 people
1:18:54 um mult multi- potential ites right
1:18:57 isn't that it ug how much are po Claude
1:19:01 and Leonardo Kyle yeah sorry about that
1:19:03 okay po there's a free version so po you
1:19:06 can use for free and I'm pretty sure
1:19:08 that you can make bots on PO for free
1:19:11 Claude has a free
1:19:13 version they've they've dramatically
1:19:15 limited how much you can use it but if
1:19:17 you just need to do long documents and
1:19:20 you don't need to do a ton in a row then
1:19:22 you don't need to pay for claw
1:19:24 and Leonardo gives you 150 credits a day
1:19:27 now when you generate images it's
1:19:29 somewhere between like eight
1:19:32 and I don't know 15 credits so they're
1:19:35 may be giving you 10 images a day or 15
1:19:37 images a day but all three of those have
1:19:40 free versions I personally if I were you
1:19:44 the real gunny R I like your name the
1:19:47 real gunny rat um if I were you I would
1:19:51 not pay for any of those
1:19:53 unless Claude would be the exception if
1:19:55 you're doing lots of long document
1:19:57 processing claude's the one I would pay
1:19:59 for out of the three of those I would
1:20:02 pay I do recommend paying the 20 bucks
1:20:04 for chat
1:20:06 GPT because gp4 with this new multimodal
1:20:09 [ __ ] that they just launched over the
1:20:11 past two weeks is is is a profound value
1:20:16 like like I sound like you can make
1:20:19 money with time I sound like a [ __ ]
1:20:21 salesman for it but it it's just it's
1:20:23 it's the for 20 bucks a month like I
1:20:27 know what we pay for [ __ ] Adobe
1:20:29 products that I don't use enough to
1:20:31 justify that [ __ ] money but like if
1:20:33 you want worldclass [ __ ] you got to pay
1:20:36 the money 20 bucks a month for what you
1:20:38 get for chat G pt4 is [ __ ]
1:20:41 insane it really
1:20:44 is it's just [ __ ]
1:20:47 nuts pose a good investment because it
1:20:49 offers access to a lot of bots including
1:20:51 gp4 yep I I still
1:20:54 think given Po and given gp4 the chat
1:20:59 GPT
1:21:01 plus um I would I would go with
1:21:07 that do we got some new images over here
1:21:10 oh we got lots man the Irregulars are
1:21:12 going nuts in the old image gen tonight
1:21:14 this super cool and we got some new do
1:21:16 we got some new joiners we
1:21:19 do
1:21:22 Kingpin J Rea work in customer service
1:21:25 at a pharmaceutical SAS company obsessed
1:21:28 with AI
1:21:30 nice super
1:21:33 cool
1:21:35 beautiful um let's go to show and tell
1:21:37 what we got over here oh nice fling
1:21:40 fling your knowledge [ __ ] Ted Ted is
1:21:44 just [ __ ] cranking Ted you you you
1:21:46 need to be like our our t-shirt art
1:21:49 dude multi- potential equ equals me AKA
1:21:52 sles oh so this is what I was saying
1:21:54 before so welcome to chat add Kyle do
1:21:57 you ever finish a thought
1:22:00 occasionally um oh boy there goes the
1:22:02 hair
1:22:05 um imagine if you're if you're just like
1:22:10 if if you know you're good at two or
1:22:11 three
1:22:12 things and then you've got a friend
1:22:15 who's like a
1:22:16 Renaissance kind of person that's good
1:22:18 at a lot of [ __ ] imagine if you could go
1:22:21 from being good at two or three things
1:22:23 to being good at 10 or 20
1:22:26 things that's where we are right
1:22:30 now and these tools these AI tools are
1:22:34 as bad as they're ever going to be like
1:22:37 they only get better from here they only
1:22:40 get like right right now these tools are
1:22:43 are I am fond of calling them janky
1:22:46 pieces of [ __ ] you know chat gp4 po Bard
1:22:51 Bing all of them mid Journey what a
1:22:55 stinky nasty piece of horrible crap it
1:22:59 is and yet it is profoundly
1:23:04 powerful and good and so is Chachi PT
1:23:08 and so's Bing and so this why because
1:23:10 we've never
1:23:12 ever in the history of
1:23:17 humanity we've never ever had
1:23:20 tools that Translate
1:23:26 thin ideas in our
1:23:29 imagination into something that we can
1:23:32 share with the world
1:23:34 instantly that's what these tools are
1:23:38 they're amplifiers of
1:23:41 humanity it's not like chat GPT wrote a
1:23:45 thing no no no Kyle had an idea and said
1:23:49 hey chat GPT can you help me bring that
1:23:51 idea to life and it goes
1:23:54 here it is oh [ __ ] can I see what that
1:23:56 looks like here it
1:23:59 is hm that reminds me of a song let me
1:24:01 go over to this other tool can you make
1:24:03 me a song this here here it
1:24:06 is we've never had that before
1:24:21 ever
1:24:24 generating GPT the G in GPT stands for
1:24:29 generative
1:24:31 generating generating it's not copying
1:24:34 and pasting it's one of the
1:24:35 misconceptions about it if you're new to
1:24:36 this
1:24:40 stuff and it's
1:24:44 and what chat GPT did and why I'm
1:24:47 probably going to do a 24-hour live Aon
1:24:50 on November 30th which is when chat GPT
1:24:54 was introduced to the
1:24:56 world is
1:24:59 that chat GPT represents for AI and
1:25:03 machine learning what the worldwide web
1:25:05 represented for the internet so the
1:25:08 internet had been around for
1:25:10 decades but it was all command line and
1:25:12 it it was basically the purview of
1:25:15 scientists and researchers and
1:25:16 Librarians and you know
1:25:21 nerds
1:25:24 and then the worldwide web comes along
1:25:27 the browser gets invented and now you
1:25:29 can do
1:25:31 this and all of a sudden the power of
1:25:33 the
1:25:34 internet was handed to the rest of the
1:25:39 world you no longer had to be a
1:25:41 researcher you know had to no longer had
1:25:43 to use command line interface you just
1:25:45 did
1:25:50 this what's chat GPT it's it's just like
1:25:53 a Google box and you're like well Google
1:25:56 like you get all the information with
1:25:57 Google you get information with Google
1:26:00 you get knowledge with chat GPT we are
1:26:03 moving from information on demand world
1:26:07 that chat that that Google
1:26:09 mastered I will index all of that
1:26:13 information and when you want
1:26:14 information you just type in a few words
1:26:17 and I will give you access to all the
1:26:19 information so you don't have to go find
1:26:20 it I'll find it for you that's what
1:26:22 Google Google did chat GPT same box is
1:26:26 now
1:26:27 knowledge write me an application that
1:26:30 does
1:26:34 this it's not saying here's a python
1:26:36 manual go learn it yourself loser it's
1:26:39 writing python it's writing and
1:26:41 executing
1:26:43 python at our
1:26:51 request did you read the the hidden
1:26:53 prompts inside images that are executed
1:26:55 when
1:26:57 using GPT Vision
1:27:00 features no oh did I read about hidden
1:27:03 prompts inside images that are executed
1:27:07 I didn't I didn't I on China is there
1:27:10 like like what's the what's the the uh
1:27:14 what's the tldr on
1:27:17 that what's What's the 411 what's the
1:27:22 give me the
1:27:24 synopsis
1:27:31 um Holly Allison I have a million ideas
1:27:35 until I try to write it prompt and then
1:27:36 my brain goes illiterate L Holly I have
1:27:39 a way to I I have a solution okay so
1:27:42 what Holly talks about there is what are
1:27:45 the best places to find prompts I'll
1:27:46 talk about that to in a
1:27:48 second I would strongly
1:27:51 recommend don't go try to find
1:27:55 prompts I mean you
1:27:58 can I I I feel like it's your
1:28:00 responsible of me to say that because
1:28:02 there's lots of good prompts out there
1:28:03 there there's a there's one called
1:28:05 prompt Mania that's good for
1:28:08 um for image generation tools there's
1:28:11 one called prompt base there's a
1:28:13 document that I talk about called
1:28:14 prompt. chat um but but the the the
1:28:20 prompting thing is is
1:28:23 um
1:28:25 it's here here's okay there's a thing
1:28:29 called White Page syndrome which writers
1:28:31 have it's writers block white page
1:28:33 syndrome is you're ready to write the
1:28:36 Great American novel and then you sit
1:28:38 down and you look at the blank piece of
1:28:39 paper it's like you know [ __ ] brain
1:28:42 just goes right I experienced the same
1:28:45 [ __ ] with chat GPT there's something
1:28:47 about that
1:28:49 box knowing that chat GPT can
1:28:51 essentially do anything is really
1:28:53 [ __ ] intimidating it is and like I
1:28:57 love this [ __ ] and sometimes I'll sit
1:28:59 down and I'm like what can I do with
1:29:01 chat GPT today and it my mind does the
1:29:04 same thing it just [ __ ] flatlines
1:29:07 here's the solution to
1:29:08 it don't start with chat
1:29:12 GPT think about a problem in your life
1:29:15 that you need to solve think about a
1:29:17 thing in your life that you want to know
1:29:19 something about like like flip it around
1:29:21 and say what is something I'm trying to
1:29:24 do in my life oh well I've got to write
1:29:27 that really uncomfortable letter to
1:29:29 Sally oh
1:29:31 perfect then go to chaty BT all right I
1:29:34 got this friend Sally and we were
1:29:35 talking and this happened and then and
1:29:37 she got offended by this and and so I
1:29:39 got to write her a nice email to to mend
1:29:41 the
1:29:42 fences there it'll be start with a
1:29:45 problem or start with an
1:29:48 idea or start with a
1:29:51 wish
1:29:53 and then say which tool do I need and
1:29:56 most of the time it's chat GPT but
1:29:58 sometimes it's an image tool sometimes
1:29:59 it's one of the other tools start with
1:30:02 the
1:30:03 problem my philosophy on prompting right
1:30:06 now is I don't think any of us are going
1:30:08 to have to be prompt Engineers I think
1:30:11 some of us can choose to be prompt
1:30:12 Engineers but I think prompt engineering
1:30:16 increasingly drops into the background
1:30:19 of software
1:30:21 development
1:30:24 these tools are going to start prompting
1:30:25 themselves if you watch what Del does
1:30:28 you're not writing image prompts you're
1:30:29 writing
1:30:32 requests for for it to visualize images
1:30:35 it's writing its own
1:30:39 prompts when you play with Advanced Data
1:30:42 analysis inside chat GPT it's writing
1:30:45 and executing its own python code and
1:30:48 then it's taking the results of those
1:30:51 python applications
1:30:53 and writing reports for you like it's
1:30:56 doing all that it's writing its own
1:30:57 prompt so we're not going to have to
1:30:59 prompt
1:31:00 but I think it's important right now to
1:31:04 explore
1:31:07 prompting in a similar way that it's
1:31:11 important to
1:31:13 still understand how to do
1:31:16 mathematics on a piece of paper even if
1:31:19 it's just addition subtraction Division
1:31:21 and multiplic
1:31:23 do we really need to understand that no
1:31:26 we don't but it helps you if you
1:31:30 understand that to use these really
1:31:31 powerful math tools same thing with this
1:31:34 with with the prompting right now I
1:31:36 think teaching
1:31:38 yourself how to push the boundaries of
1:31:41 these tools try to break them try to get
1:31:44 it to hallucinate try to get it to not
1:31:45 hallucinate try to get it to be creative
1:31:48 try to get it to be
1:31:50 systematic figure out where the
1:31:51 boundaries are these
1:31:54 things because as you play as you figure
1:31:57 out where the boundaries are someone's
1:31:59 going to come into your life that goes I
1:32:00 need to solve this problem you be like
1:32:02 oh I know how to solve that problem I
1:32:05 just did that three days ago here watch
1:32:07 this and you'll put in some prompt and
1:32:10 it will solve a [ __ ] problem and
1:32:11 they'll
1:32:12 go they'll have their Kevin McAllister
1:32:15 moment they'll be like how did you do
1:32:17 that oh that's chat chpt that's chat
1:32:21 chpt and then they just took the red
1:32:23 pill they can't unsee that
1:32:25 [Laughter]
1:32:29 [ __ ] oh
1:32:32 man man
1:32:38 [Music]
1:32:43 alive oh that's really
1:32:45 cool that image from Sherry
1:32:47 D I don't know what that prompt was but
1:32:50 that's pretty flipp and swell
1:32:55 you know what's cool about
1:32:59 this
1:33:01 is people are going to be making art and
1:33:04 putting art in the world that people are
1:33:07 going to be falling in love with and the
1:33:09 people that created the art were not
1:33:13 artists and the way they think about
1:33:16 creating art is going to come from a
1:33:18 completely different place than an
1:33:20 artist would have come from and we're
1:33:22 going to see all sorts of innovative
1:33:24 ideas emerge because you've got people
1:33:28 that are that are
1:33:31 expanding who they are
1:33:34 laterally right if I'm really good at
1:33:36 this and I start playing with these
1:33:38 tools I get better and better and better
1:33:40 over here oh is that the rabbit hole
1:33:42 tunnel yeah that's probably the rabbit
1:33:43 hole I was down it's a llama
1:33:50 hole Yeah that was that was that was
1:33:53 that that's right fling your
1:33:56 [Laughter]
1:34:00 knowledge learn about this [ __ ] and
1:34:03 fling your
1:34:05 knowledge have fun with it if people
1:34:09 think you're a [ __ ] weirdo lean into
1:34:11 that yeah cuz there was when I first
1:34:15 started doing this I was
1:34:17 like I was kind of like hiding it from
1:34:19 work like I felt like I was kind of
1:34:21 cheating on my compan
1:34:23 and I felt like like I don't want to
1:34:24 talk about this too much and then I was
1:34:26 just like well wait [ __ ]
1:34:32 it I did this with with nfts and
1:34:34 blockchain where I started talking about
1:34:36 them and people are like you're [ __ ]
1:34:37 weirdo and then they were right on that
1:34:40 one I think the blockchain stuff's going
1:34:42 to come back
1:34:45 but what I've experienced with the AI
1:34:47 stuff is the deeper I dig into it the
1:34:50 more profound it is the more clearly
1:34:53 profound it is with nft stuff the deeper
1:34:57 I dug into it the more I realized almost
1:34:59 all of these projects
1:35:01 are complete pieces of
1:35:05 [ __ ] and and you know bordering on
1:35:08 criminal and then you know people are
1:35:11 going to jail
1:35:14 so this AI stuff's the complete opposite
1:35:16 where the deeper you dig the more
1:35:18 important it it gets was that in the
1:35:20 Discord yeah these images so the images
1:35:22 I just showed are in the showand tell
1:35:24 channel on the Discord so the Discord is
1:35:27 at the salon if you go to the salon.
1:35:32 a why wait
1:35:36 what is did did my machine just crap out
1:35:39 oh it did God damn
1:35:41 it oh
1:35:43 Microsoft you and your your memory
1:35:47 leaking Microsoft edge
1:35:50 browser let me quit that hang on I may
1:35:53 need to restart my
1:35:57 [Music]
1:36:02 machine hang
1:36:05 on all right there you go um oh yeah
1:36:09 this is this is crashed this is crashed
1:36:11 hard anyway if you go to the
1:36:14 salon. it's a link tree and there's a
1:36:16 there's
1:36:17 a the first link talks about what the
1:36:20 salon is it's it's basically a community
1:36:23 for AI
1:36:25 weirdos not necessarily machine learning
1:36:28 people but like people trying to figure
1:36:30 out this generative AI [ __ ] how do we
1:36:32 use this stuff now that it's been built
1:36:34 now that it's been handed to us on a
1:36:36 silver platter what do we do with it how
1:36:38 do we do
1:36:40 this how's it going to change work how's
1:36:42 it going to change my work I don't know
1:36:45 nobody [ __ ] knows right now so that's
1:36:47 what this is community of weirdos trying
1:36:49 to figure this [ __ ] out the second link
1:36:52 is to the meetups we have a Meetup every
1:36:53 other Tuesday this coming Tuesday is the
1:36:56 next Meetup and then the third link is
1:36:57 to the salon and so that's where I've
1:37:00 been showing these images from we're
1:37:01 actually going to be
1:37:03 um transitioning the the community off
1:37:07 Discord and onto Mighty networks at some
1:37:09 point I've got a meeting tomorrow to
1:37:11 figure out how we build that so because
1:37:15 Discord sucks so but it's it's serving
1:37:18 its purpose right now um but it could be
1:37:21 so much better and I think I think
1:37:23 Mighty networks is going to be really
1:37:24 cool I still do blockchain and
1:37:27 Healthcare is where um it out well is
1:37:30 where it's at yeah listen I think I
1:37:32 think blockchain technolog is really
1:37:34 important um and and I think AI is going
1:37:36 to drive adoption of it in in a lot of
1:37:40 ways um and
1:37:44 uh but but the nft Bros needed needed a
1:37:48 they earned a very well-deserved timeout
1:37:51 they're sitting in the corner and going
1:37:53 to jail and then we'll come back to it
1:37:55 when we have some actual [ __ ] use
1:37:59 cases uh I got to restart my machine God
1:38:05 Microsoft and if you're like but Kyle
1:38:08 why do you have Edge as your default
1:38:09 browser isn't that stupid shut up shut
1:38:12 up shut
1:38:13 [Laughter]
1:38:17 up Kyle into the llam verse coming soon
1:38:21 that's good
1:38:22 I like it
1:38:23 will um that felt the same I ended up
1:38:26 creating an unofficial Microsoft teams
1:38:28 chat at work I had chat GPT right
1:38:31 something let the blockchain go Kyle
1:38:33 it's time yeah
1:38:37 exactly oh
1:38:46 man can you put the links back up for a
1:38:49 moment which links those
1:38:53 links or the the I'm restarting my
1:38:56 machine I'll put the links up I will put
1:38:58 the links back up
1:39:00 Holly did you see my posts in Show and
1:39:03 Tell um which ones JC guy are you a
1:39:06 different dude in there are you the
1:39:08 Mandalorian JPC
1:39:16 [Music]
1:39:20 guy yeah these are cool yes what did you
1:39:24 do these in are these these look like
1:39:26 mid Journey are these mid
1:39:29 Journey so look these are the man some
1:39:31 of the Mandalorian things that you just
1:39:32 put in here these are really
1:39:35 cool that one's beautiful like some sort
1:39:39 of crazy ass ride at the future
1:39:41 Carnival oh Dolly 3 wow these are
1:39:46 great nice
1:39:49 prompting Dolly 3 man it's here's here's
1:39:54 the thing I I mean I if you do side
1:39:58 by-side comparisons with Dolly 3 and mid
1:40:00 Journey mid Journey for the most part
1:40:03 kicks Dolly 3's ass in terms of just es
1:40:09 aesthetic
1:40:11 Nuance but it can't do
1:40:13 text and it also requires you to be in
1:40:18 Discord and Discord blows
1:40:22 and it requires you to learn
1:40:27 commands and write your image
1:40:31 commands like a
1:40:34 programmer and dolly3 just lets you talk
1:40:37 hey I want an image
1:40:44 of and and that distinction is massive
1:40:47 because I think you're going to have
1:40:48 people that geek out on Mid journey and
1:40:49 do really cool stuff but the fact that
1:40:52 you can just talk to the thing just ask
1:40:54 it for what you want and it
1:40:57 generates 80 90 or 100% of what you
1:41:01 need is remarkable a very real use case
1:41:04 is trying an AI movie extra wait is
1:41:08 tying an AI movie extra to an nft so
1:41:11 they get paid whenever the oh so so
1:41:15 gregoli GRE or GRE yeah gregoli um
1:41:20 absolutely so so I think that um so I've
1:41:24 got I've got I can show
1:41:28 you oh I can't because I'm restarting my
1:41:31 machine dang
1:41:34 it this this machine is really broken
1:41:37 this is really like Microsoft
1:41:41 hosed this session I haven't restarted
1:41:44 in like three days and
1:41:46 so I'm one of those guys that
1:41:50 um I'll go months without restarting my
1:41:54 machine my kids think I'm an idiot and
1:41:58 it never used to be a problem but when I
1:42:00 SW when I switch to Microsoft Edge it's
1:42:02 not it's not a problem I need to restart
1:42:04 daily all right I'm just going to shut
1:42:06 this down let me
1:42:14 see here we
1:42:20 go
1:42:22 yay you did it go Apple go people
1:42:27 restart their machines people that use
1:42:29 Microsoft Edge on on OSX restart their
1:42:32 machines a lot
1:42:37 apparently I never used to restart my
1:42:41 machines if we all chip in a dollar
1:42:43 could you get one that
1:42:46 works a machine I could I suppose this
1:42:50 machine's really good it's it's this is
1:42:52 a software problem I what I need is I
1:42:55 need a I need a
1:42:56 17 petabyte Drive apparently because I
1:42:59 just I'm all maxed out I think that's
1:43:01 the other problem here is my swap disc
1:43:03 is as they say
1:43:07 full what about minting an nft of your
1:43:10 own logo brand so any use will lead to
1:43:13 your url yeah so oh so that's what I was
1:43:15 saying so so I'm I'm a I'm a card
1:43:19 carrying member of
1:43:21 uh a world ID which is Sam alman's you
1:43:25 know worldcoin
1:43:28 startup um that is
1:43:31 blockchain um
1:43:34 verified human uniqueness right so I'm a
1:43:37 unique human on the blockchain so I've
1:43:39 got my world ID that that is tied to my
1:43:44 retina and so I think that a a system
1:43:48 like that where you could have a
1:43:51 person verified and then you could have
1:43:54 Works stacked on top of that as
1:43:56 additional
1:43:58 nfts works and
1:44:01 transactions and so the ideas that
1:44:04 you're saying there mint your logo um
1:44:07 you know make an nft of your likeness
1:44:09 and then you know anyone can use that
1:44:11 likeness and if they use that likeness
1:44:13 you get paid per generation or per use
1:44:15 or whatever it might be I think that
1:44:17 absolutely is where all of this stuff
1:44:19 has to go and know that there's projects
1:44:22 that are working on that with the nft
1:44:24 stuff but what what what we have right
1:44:27 now with nfts and blockchain is we have
1:44:29 a lot of solutions solving problems that
1:44:32 no one really gives a [ __ ]
1:44:35 about not enough to drive
1:44:39 adoption my this it's not a theory it's
1:44:43 a hypothesis my hypothesis is that the
1:44:46 2024 presidential election in the United
1:44:49 States is going to be so polluted with
1:44:53 misinformation and deep fakes and all
1:44:55 this AI generated stuff it's going to be
1:44:58 such a [ __ ] show that in 2025 there's
1:45:01 going to be a push to solve that [ __ ]
1:45:04 problem and it's going to be it's it's
1:45:07 going to be a pretty massive movement
1:45:09 it's going to be a pretty massive
1:45:11 problem and that will likely drive
1:45:14 toward the solutions that you're talking
1:45:16 about I I run out of ram
1:45:19 hourly where's why did it not restart
1:45:22 any of my thing I guess cuz I just shut
1:45:23 oh I restarted after it quit all those
1:45:27 apps okay so what do I need to start I
1:45:28 need keynote oh this is what' you do
1:45:32 Saturday night I don't know I watched
1:45:34 some old guy restart his machine and try
1:45:36 to figure out what absy needed to
1:45:39 [Music]
1:45:40 [Laughter]
1:45:44 open all right there's that one let me
1:45:47 launch Edge just cuz I like living on
1:45:50 the edge
1:45:55 this this is painful this is painful
1:45:58 people let's see 22 tabs yes I will I
1:46:02 would like to restore the 22 tabs that
1:46:05 exploded my computer yes I would why
1:46:08 because I'm a
1:46:10 [Laughter]
1:46:15 genius let me pin my little things over
1:46:19 here
1:46:21 what else do I need s is trying to open
1:46:24 po I don't want that
1:46:26 okay um somebody asked about the
1:46:33 links so the the links that I um put up
1:46:38 before that someone was asking about
1:46:42 um these are they so feel free to
1:46:45 screenshot these at some point I'm going
1:46:47 to get my [ __ ] together and put these I
1:46:49 I'll just have a single play place to go
1:46:51 and you can get all these but I don't
1:46:53 have it yet so you probably should have
1:46:56 loaded Microsoft on your Game Boy Kyle
1:46:59 exactly Silver Fox exactly the Game Boy
1:47:03 would have crashed
1:47:05 less
1:47:07 um all right so chat GPT Bard Claude and
1:47:10 Bing those are your big four prompt.
1:47:13 chat will teach you about prompting
1:47:14 people were asking about that before I
1:47:16 posted a logo example in Discord for who
1:47:19 whoever was asking great that's perfect
1:47:23 um here's the kids table
1:47:26 piie is is actually I think Pi is going
1:47:29 to be a really powerful app it's it's
1:47:30 pretty powerful right now just as a
1:47:32 conversational
1:47:33 bot um perplexity and The Meta models
1:47:37 and po are both both interesting we've
1:47:39 spent a lot of time on PO tonight here's
1:47:41 the image generation tools that I think
1:47:43 are worth playing with um Dolly 3 mid
1:47:47 Journey Leonardo and Firefly or just all
1:47:50 the Adobe products right now they're
1:47:52 they're they're going all in on AI stuff
1:47:54 the nice thing about Adobe if you work
1:47:56 for a big
1:47:57 company the way Adobe has trained their
1:48:00 image models um they haven't trained
1:48:02 them on any copywritten work which means
1:48:04 that they're not as good but they're
1:48:07 getting better because they're starting
1:48:08 to cut deals with artists
1:48:12 um but they're safe so if you work in a
1:48:15 company the Adobe stuff is is probably
1:48:18 the way to go um
1:48:22 here
1:48:23 are the I need to make this some of this
1:48:26 stuff smaller how I'm going to do
1:48:32 this oh [ __ ] I'll do it offline
1:48:35 um for video you got Runway ml hey Jen
1:48:40 and and D- ID down there are kind of the
1:48:42 same thing except hey genen also has
1:48:44 video translation which is really good
1:48:47 go to labs. gen.com Skybox is text
1:48:52 to 3D World
1:48:55 csm.edu
1:48:58 I tried one last night and the the 3D
1:49:01 object it created was absolute garbage
1:49:06 but whether it's that tool or some other
1:49:08 tool someone's going to crack that the
1:49:10 the 3D object it created reminded me
1:49:13 very much
1:49:14 of where images were 18 months ago where
1:49:18 you could generate images but they
1:49:19 sucked the the 2D object to a or 2D
1:49:23 image to a 3D object thing right now it
1:49:26 sucks it's going to get really good so
1:49:27 just that's one to pay attention to and
1:49:30 then the the final list of links here is
1:49:35 um futur
1:49:39 pedia.com um are both directories of
1:49:43 like thousands of AI tools yeah Vicky I
1:49:48 tried csm2 and it's so janky it's
1:49:50 totally janky it's so bad but it'll get
1:49:53 it might not be CSM I have a feeling
1:49:55 it's not going to be CSM but whatever
1:49:56 it's fine um and then I would follow if
1:49:59 if I were you if you're interested in
1:50:01 this stuff JPC guy I have a problem we
1:50:04 we all have problems buddy unless you
1:50:06 have a specific problem I can I can
1:50:07 answer but I assume that was that was
1:50:09 kind of a um like you need a support
1:50:12 group oh that's the other app I need to
1:50:14 open his
1:50:16 photos we we'll go to the
1:50:19 uh we'll go to the
1:50:23 uh the AI support group image because I
1:50:27 think that's a that's appropriate there
1:50:29 we
1:50:33 go that this is what this channel really
1:50:37 is hi my name's Kyle oh my drive is
1:50:41 completely full I know you and me both
1:50:43 it's so bad it's just like you know your
1:50:46 Drive is full I'm like I know it's been
1:50:47 full for a year leave me alone
1:50:51 you could delete some [ __ ] no I'm a
1:50:53 digital hoarder I can't you don't
1:50:56 understand just put in the cloud
1:50:58 somewhere just be smarter than I
1:51:01 am oh
1:51:03 man all right leave we'll leave AI
1:51:07 Anonymous up there it's
1:51:09 good I love that image the llamas the
1:51:13 llamas sitting around a
1:51:18 circle um I checked out Ethan Ethan's X
1:51:21 account pretty interesting oh Ethan mik
1:51:23 yeah he's he's a really really
1:51:26 interesting dude really smart dude
1:51:28 deletion is impossible at this
1:51:32 [Music]
1:51:38 [Music]
1:51:39 point what time is it it's
1:51:43 9:50 hi people you having a good
1:51:45 Saturday night are we having a good
1:51:46 Saturday Night Live welcome to AI
1:51:48 learning lab AI anonymous Chad add we
1:51:51 got lots of names for it you know what
1:51:52 we do here you can make money
1:51:55 with we don't do
1:51:57 this well we do do
1:52:00 that but this is not sort of a [ __ ]
1:52:03 getrich quick scheme kind of Channel
1:52:05 this is a holy [ __ ] this stuff's
1:52:07 profound has anyone figured this out no
1:52:11 oh
1:52:12 God have I I can't breathe how am I
1:52:16 supposed to learn this isn't this a lot
1:52:19 I can't keep up can anyone keep
1:52:23 up
1:52:26 nope welcome to AI Anonymous hi my name
1:52:29 is Kyle I've been on chat gbt since
1:52:31 November 30th 20 hi
1:52:37 Kyle every day I'm more excited than
1:52:39 I've ever been and I'm more scared than
1:52:41 I've ever been and I feel like I'm more
1:52:42 behind than I've ever been and I feel
1:52:44 like I'm smarter than I've ever been is
1:52:46 anyone else like this all of us
1:52:49 Kyle medic
1:52:53 exactly we got a
1:52:55 [Laughter]
1:53:04 runner is this is this Tik Tock thing
1:53:08 on
1:53:09 Jane hey Jane Jane Jane Jane stop
1:53:14 watching NCIS and pay attention to me
1:53:17 how how do I turn on a Tik Tok how what
1:53:20 oh it's on is this on who are these
1:53:24 people who
1:53:27 what yeah that's
1:53:30 us
1:53:32 people that's what this
1:53:39 is it's like Saturday Night Live but not
1:53:44 funny well Saturday night life's not
1:53:46 funny anymore either but you know that's
1:53:49 a whole separate conversation
1:53:51 I have a notion page called you can make
1:53:53 money with jet gbt that's [ __ ]
1:53:55 awesome Vicki that's [ __ ]
1:54:01 awesome I will happily talk about how
1:54:03 you make money with chat gbt but you
1:54:05 won't like the answer um it's not as
1:54:08 slick as those guys that have the fancy
1:54:09 Tik Tock channels this is one of those
1:54:12 janky ass Tik Tock
1:54:14 channels this my favorite thing is when
1:54:16 young people come in here and they're
1:54:17 like oh isn't that cute the elderly are
1:54:19 teaching other how to use
1:54:22 AI it's pretty good that's it's it's
1:54:26 solid
1:54:29 trolling
1:54:31 ah jokes on them though revenge of Gen X
1:54:35 is coming I think gen xers are are well
1:54:39 based on what's happening in this
1:54:41 channel gen xers are [ __ ] baller with
1:54:43 this AI stuff cuz the young people are
1:54:46 like they're cynical the young people
1:54:48 right now are really cynical about this
1:54:49 [ __ ]
1:54:51 we have herodin here he's he'll
1:54:53 represent the he's kind of the middle
1:54:55 generation he's not the Young Generation
1:54:57 anymore he just turned 32 you a little
1:55:00 youngster you little whipper
1:55:02 snapper but he's good but it's like it's
1:55:04 like a lot of the young people you know
1:55:07 my my kids are they don't give a [ __ ]
1:55:09 about this at all so I think I think the
1:55:12 Gen gen xers are going to be there're
1:55:14 going to be a big Resurgence dude we
1:55:16 wrote the book exactly I would pay to
1:55:19 see you do Broad way U [ __ ] hire me
1:55:21 I'll go I'll move back to New York I'll
1:55:23 go on Broadway I'd go on Broadway
1:55:24 tomorrow my name is Holly and my toxic
1:55:27 trade is believing I can Co conquer AI
1:55:29 with zero technical wait tech knowledge
1:55:33 okay Holly that's a really important
1:55:35 statement hi hi my name is kly Holly and
1:55:38 my toxic trait is I'm believing I can
1:55:41 conquer AI with zero tech
1:55:44 knowledge you're
1:55:47 right you're right they
1:55:50 wrong anyone who thinks that this
1:55:54 requires a math
1:55:56 degree November 30th 2022
1:56:00 look I do I have it open oh I don't have
1:56:04 it open let me see if I have it it's
1:56:05 probably up in the cloud cuz my hard
1:56:07 drive is
1:56:12 full I think November 30th
1:56:16 2022 is such a significant date that I
1:56:19 wrote a
1:56:20 I made a it was kind of the money slide
1:56:24 of well it is the money slide of a uh a
1:56:28 keynote talk that I do comparing the
1:56:31 early days of the worldwide web to the
1:56:33 early days of of this stuff of this AI
1:56:37 stuff um let's see where is it it's in
1:56:40 that one isn't it oh yeah it's in the
1:56:42 cloud I'll have to go get
1:56:44 it is it coming yeah slowly all right
1:56:48 whatever
1:56:50 [Applause]
1:56:52 hi my name is Kyle hi
1:56:56 Kyle I can't stop thinking about AI
1:57:00 nobody can Kyle can anybody else sleep
1:57:03 no
1:57:04 Kyle obviously we're
1:57:13 here is anyone playing with chat GPT on
1:57:16 their phone while I'm telling my story
1:57:18 yes Kyle of course we are
1:57:20 what do you think this is AA
1:57:24 no but Kyle I have issues with all the
1:57:27 pro all the plugins just to make
1:57:29 Halloween short
1:57:31 video with my own voice
1:57:35 oh plugins uh yeah go just go to 11 Labs
1:57:40 Clint
1:57:40 copperhead make yourself a model over at
1:57:43 11 Labs dude we wrote the book we thrive
1:57:47 in the
1:57:48 Jank I love Holly I love your toxic
1:57:51 trait you're right they're
1:57:54 wrong I would pay to see you do Broadway
1:57:57 Kyle I did write a oneman show I wrote
1:58:01 and performed a oneman show but didn't
1:58:03 make it to Broadway but it good I got
1:58:07 good feedback on that show it was called
1:58:09 naked in the cockpit about a dream I had
1:58:11 in third grade where I was naked in the
1:58:17 cockpit it's a it was a a play about
1:58:22 fear and how we all just [ __ ] live in
1:58:25 fear all the
1:58:32 time I never have fewer than three
1:58:34 devices going during the lab just like
1:58:36 you exactly I got that one I got this
1:58:38 one and I got the I got the uh the old
1:58:41 Ye Old Ye Old iPad man you guys are
1:58:44 prolific tonight holy crap who did this
1:58:47 one this is beautiful Ted Ted killing it
1:58:51 look at that that's
1:58:55 gorgeous and look the fingers aren't
1:58:57 even all that [ __ ] up is this DOI do
1:59:00 is
1:59:06 amazing
1:59:09 amazing Mandalorian
1:59:12 doing doing some logo
1:59:15 work by the way if you didn't see it um
1:59:18 Mid Journey added 2X and 4x upscaling 2
1:59:23 days ago or 3 days ago oh here's
1:59:27 Herod this one involves creepy imagery
1:59:29 and unreasonably loud heavy metal music
1:59:32 you've been warned well we're playing
1:59:34 that dude you think we're going to not
1:59:36 play that all right here's a new video
1:59:38 from Herod he's he's oh oh dang it hang
1:59:43 on I got to get I got to do it on the on
1:59:46 the
1:59:48 MacBook disc
1:59:52 [Music]
1:59:52 [Laughter]
1:59:53 [Music]
1:59:58 cord download it first
2:00:02 okay yeah on the iPad though it just
2:00:04 doesn't the like the Mac seems to
2:00:07 download it better oh now it's
2:00:08 downloading updates awesome thanks
2:00:11 Discord oh my god
2:00:17 really uh uh uh uh uh uh
2:00:30 except walruses don't have fingers well
2:00:32 th that one does he's playing guitar
2:00:35 Vicky how could the walrus play guitar
2:00:37 if he didn't have fingers uh it's just
2:00:39 biology duh go to chat GPT and ask if
2:00:43 walles have fingers the answer is
2:00:45 clearly yes I asked it yesterday okay
2:00:48 God duh my god do I have to teach
2:00:52 everything around here
2:00:54 jeez all
2:00:56 right let's see where are we there we
2:01:00 are and we're
2:01:03 in show and
2:01:07 tell harod harod
2:01:10 where well there you are there you are
2:01:13 ra carry on Cheerio
2:01:16 pip all right this is going to be cool
2:01:20 oh I guess I got to download it what you
2:01:22 what would you do this in 8k dude dude
2:01:25 dude
2:01:27 dude download all right we're
2:01:29 downloading
2:01:31 it 352 megabytes this this is why my
2:01:35 hard drive's full I I just willy-nilly
2:01:37 download like you know six gigabyte
2:01:41 builds of large language models 400
2:01:44 megabyte videos that are going to creep
2:01:46 me out and I won't be able to sleep do
2:01:49 not not watch the
2:01:53 video hey if we're learning we're
2:01:55 learning from everyone man you gota you
2:01:58 got to be open to different perspectives
2:02:00 at the AI learning lab we learn from
2:02:03 everyone we're one with the
2:02:05 world um are there any real problems
2:02:08 being solved here or is it just PE
2:02:10 people being enamored with generated
2:02:12 images um it
2:02:15 depends what's your name Brian Pullman
2:02:18 pulum uh
2:02:19 um Brian you seem like the kind of guy
2:02:22 that wants to know my credentials
2:02:24 um I have no credentials the
2:02:29 uh one of the reasons there's more
2:02:31 images going on than than lately is that
2:02:36 chat GPT rolled do inside chat
2:02:40 GPT um and
2:02:45 so that's cool and people are
2:02:48 experimenting so what this channel is
2:02:50 really about
2:02:52 is talking about experimenting with
2:02:55 playing with generative
2:02:58 Ai
2:03:00 and sometimes it goes like this
2:03:03 sometimes one night I did an hour on
2:03:06 automation building an Automation in
2:03:08 zappier and we went from 80 people to 20
2:03:12 people but we did it so 20 people were
2:03:15 into
2:03:16 it but it [ __ ] it [ __ ] tank the
2:03:20 tank the life so I mean I wouldn't stay
2:03:24 if I were you like if this is driving
2:03:26 you [ __ ] crazy I'd leave but if you
2:03:29 want to learn stuff and explore stuff
2:03:31 then I'd hang out and come back night
2:03:33 after night like the weirdos in here oh
2:03:35 I'm sorry they're not weird they're
2:03:37 Irregulars all right this is herod's
2:03:39 video all right are you ready
2:03:43 people you've been warned you're not
2:03:45 supposed to watch this I think herod's
2:03:47 got uh subliminal stuff in here it's
2:03:49 going to it's going to take us out it's
2:03:52 you're about to be
2:03:56 reprogrammed it's only is this a minute
2:03:58 or is this an hour this is not an
2:04:01 hour no it's a
2:04:05 [Music]
2:04:14 [Applause]
2:04:17 minute wow
2:04:19 [Applause]
2:04:28 [Music]
2:04:33 wow dude that's amazing that's
2:04:39 amazing I'm gonna play that my next
2:04:41 Pharma
2:04:43 meeting hey look what's going on with
2:04:45 generative AI it'll be just
2:04:48 fine
2:04:51 um by the way Herod um is in the salon
2:04:54 Discord and that's that's where this was
2:04:57 shared um and if you don't know what the
2:05:00 salon is so it's a it's a community uh
2:05:04 that I started a bit ago last December
2:05:07 actually um and if you go to the
2:05:09 salon. um that's the there's a link tree
2:05:12 there for the Meetup and what the
2:05:15 salon's about and also an invite to the
2:05:17 Discord um
2:05:20 and uh and so herod's been really
2:05:23 generous with teaching people how to do
2:05:25 that animation technique and things like
2:05:27 that so that's the community in there is
2:05:29 really cool you can ask questions you
2:05:31 can learn about this stuff
2:05:34 um yeah so it's it is uh as as as we
2:05:39 were just semi- trolled it it is a
2:05:42 Discord full of people that are over
2:05:44 enamored with generative AI but it's not
2:05:46 just images um but but we listen there
2:05:50 there's there's a there's a cynical view
2:05:52 of this [ __ ] that says you know this
2:05:55 stuff's bad it's evil it's imperfect it
2:05:59 it's it's this is a hype
2:06:01 cycle I absolutely understand that point
2:06:04 of view I am of
2:06:09 the mental model right now
2:06:13 that this generative AI stuff is likely
2:06:16 underhyped
2:06:17 [Laughter]
2:06:20 which which is it's going to have a more
2:06:22 profound effect more quickly than
2:06:25 anything we've ever seen and we got to
2:06:28 get ready for it and that's what this
2:06:29 Channel's all about is just start
2:06:31 [ __ ] playing with this stuff just
2:06:33 start getting educated just start just
2:06:36 just take the [ __ ] red pill have your
2:06:39 Kevin McAllister
2:06:41 moment and then dig in why not why not
2:06:45 what's it hurt what's it hurt doesn't
2:06:49 hurt a thing maybe I'm wrong maybe this
2:06:51 is all a hype cycle and this shit's
2:06:53 going to be gone in a year like the nft
2:06:55 [ __ ] was
2:06:57 great then I'll just I'll just let my
2:07:00 I'll let the Tik Tok account dwindle
2:07:02 back
2:07:03 down but I don't think that's the case
2:07:09 so um is that zappier session recorded
2:07:13 it is but I have no idea if you go to
2:07:18 um
2:07:19 if you go to
2:07:22 YouTube to this URL or to this account
2:07:27 here AI learninglab
2:07:30 dtt this is the archive of all of these
2:07:33 lives for the past four
2:07:36 months I think it started April
2:07:40 30th and I don't I I've missed a couple
2:07:42 of days but it's been it's been 7 days a
2:07:45 week for 4 months and all of them are
2:07:47 here and and and most of them have
2:07:50 titles and descriptions so you could
2:07:52 probably just go into the descriptions
2:07:55 and search for zapier there's there's
2:07:56 probably some in there found an awesome
2:07:59 use for chat GPT and
2:08:09 husbands run him down some uh some how
2:08:12 to fix it rabbit holes have you an
2:08:15 that's really
2:08:17 hilarious
2:08:19 do tell please share please share your
2:08:22 secrets that's [ __ ]
2:08:27 funny hi my name is
2:08:30 Kyle seven days a week for four months
2:08:33 except for the timeout oh yeah there
2:08:35 there are 7day there are three 7-Day
2:08:37 Windows where there are um where there
2:08:41 are missing videos because Tick
2:08:44 Tock kicked me
2:08:46 off I know what I did in one of them I
2:08:49 don't know what I did in the other two
2:08:51 but I was talking about bias so the
2:08:53 minute you start talking about
2:08:55 bias you just can't talk about
2:08:59 it because what I've learned is there
2:09:01 are certain words that if you say those
2:09:04 words with your mouth
2:09:07 Parts boom 7 day
2:09:10 [Laughter]
2:09:13 suspension I wish there were a list but
2:09:16 you know I think if there were a list
2:09:17 that would get Tik Tock and
2:09:21 trouble I know there's a list it's just
2:09:23 not published it's not transparent you
2:09:25 know there needs to be more
2:09:27 transparency in ticktock ification of
2:09:32 communications I suddenly thirst for
2:09:34 battery
2:09:37 acid uh it's a community of hair
2:09:42 stylists oh Herod it's Fab I probably
2:09:45 won't sleep tonight marry Mary yeah that
2:09:47 was it was it listen it was disturbing
2:09:49 but it's Halloween and you know it went
2:09:53 with the music I think it was I think it
2:09:55 was on point that was really
2:09:59 good I mean what I find really
2:10:01 impressive about your your animations
2:10:04 Herod is the um is the the uh um the
2:10:09 continuity right cuz cuz most of the
2:10:13 time when you see these AI generated
2:10:15 movies they look like AI generated
2:10:17 movies and you've got continuity like
2:10:19 where you do you know your camera rocks
2:10:22 and switches where it sort of takes you
2:10:24 out of that this is an AI generated
2:10:26 thing so it's pretty it's pretty slick
2:10:29 what you what you got going on there is
2:10:31 pretty slick I like it I like it a lot
2:10:33 mhm it's fantastic Bob Bob you tell him
2:10:36 what he's want mhm we're going to take
2:10:38 this show to Broadway AI learning lab on
2:10:42 Broadway we we should okay we that
2:10:46 should be a thing that should be an
2:10:49 image the Broadway Marquee thank you
2:10:52 Sher
2:10:55 bear so so do me a favor if you would
2:10:58 it's Saturday night I think what could
2:11:01 be really exciting here share this live
2:11:02 with your friends and let them know what
2:11:04 a [ __ ] weirdo you are for watching a
2:11:08 58-year-old on a shitty YouTube or Tik
2:11:10 Tok Channel talking about AI let your
2:11:13 friends know what you're really doing
2:11:15 Saturday
2:11:16 night seriously I was with hookers and
2:11:19 doing a lot of blow all right I was on
2:11:21 Tik Tok watching some old
2:11:25 dude let's get some more people in here
2:11:28 tap the do the tapy tap thing um you
2:11:31 know do engage engagements we're up to
2:11:33 13,000 likes that's pretty
2:11:36 cool um and follow the channel so if if
2:11:39 you want to get notified of these lives
2:11:40 I do go live every night and uh some of
2:11:44 you are like
2:11:46 so don't follow me but if if you do like
2:11:49 this follow the channel and then I think
2:11:50 you'll get notified sometimes sometimes
2:11:52 Tik Tok notifies
2:11:54 people it's Tik Tok What are you GNA
2:11:58 do oh way underhyped JPC yeah I agree
2:12:02 one of my friends is here watching
2:12:03 tonight too so this is a young crowd
2:12:06 yeah yeah I we we it's clear we've had
2:12:08 some youngsters in here tonight trolling
2:12:10 away but uh but but you're some of the
2:12:13 cool cool young crowd so I like
2:12:16 it we we are we are a fully inclusive
2:12:20 Channel if young people want to hang out
2:12:22 here you're welcome to it just tends not
2:12:24 to be who's who's in here you're the
2:12:26 exception to the rule Herod but um
2:12:29 that's not true we we've had we've had
2:12:31 some cool young people in here but um
2:12:33 the the only real prerequisite is like
2:12:35 get [ __ ] curious about this stuff
2:12:37 right you know it this stuff is weird
2:12:41 like this
2:12:42 stuff it's it's it's it's always
2:12:44 surprising to me that you know I'll talk
2:12:46 to people like oh yeah I've used chat
2:12:48 GPT
2:12:49 but what that means is they've gone to
2:12:51 it once and treated it like a Google
2:12:54 Search and then I'm like have you cuz
2:12:57 the way you're talking about it like
2:12:59 doesn't seem like that and then I'll
2:13:01 show them
2:13:02 something and then they'll be like oh it
2:13:05 can do
2:13:07 that my son dad do you know everything
2:13:10 in AI is trained on was written by
2:13:13 people Papa JJ yeah exactly that's I did
2:13:18 a I I did a I I clipped one of the one
2:13:21 of the things I was rambling about last
2:13:22 night was about
2:13:24 education and I was talking about you
2:13:26 know I find
2:13:28 it
2:13:30 staggering that educational institutions
2:13:33 are not jumping all over this and why
2:13:38 because these things are called large
2:13:40 language models chat gp4 was trained on
2:13:44 as much of the public internet as they
2:13:46 could get their hands
2:13:47 on so chat GPT is a reflection of
2:13:53 humanity it's not it's not a robot
2:13:56 that's smart it's just it's just a
2:14:01 smartly mathematically optimized
2:14:04 reflection of the [ __ ] that we've
2:14:06 written on the internet for the past 50
2:14:09 years that's what it is it's a large
2:14:11 language model what are humans large
2:14:15 language models how are univers
2:14:19 ities and and all all educational
2:14:23 institutions how are they not looking at
2:14:26 this
2:14:26 going there's this thing called a large
2:14:29 language model that seems to have all
2:14:32 this knowledge in it we are the stewards
2:14:36 of knowledge and and the the incubators
2:14:40 of knowledge and curiosity that's our
2:14:43 job but there's this thing over here
2:14:46 that has all that knowledge in it that
2:14:47 people are using you know what we're
2:14:49 going to do we're going to block it or
2:14:51 Worse we're going to just ignore it and
2:14:54 pretend it'll go
2:14:55 away I like how is that even [ __ ]
2:14:58 possible right now but it is so so yeah
2:15:02 your son is
2:15:04 correct it's been trained on people like
2:15:07 it it lies so to
2:15:09 people it's ignorant so are people it's
2:15:12 not perfect so are not
2:15:17 people
2:15:21 and that's you know you know it's funny
2:15:23 the I think one of my biggest surprises
2:15:26 Sam Alman was asked what's been his
2:15:28 biggest surprise about the adoption of
2:15:30 chat GPT and he said the amount of
2:15:34 parents and
2:15:35 kids writing custom children's books for
2:15:39 the
2:15:41 kids and and for me the biggest surprise
2:15:44 is is somewhere in the same neighborhood
2:15:47 of I'm finding myself I chat GPT is
2:15:51 training me to be a better
2:15:54 person which is [ __ ] weird
2:15:59 because it's not
2:16:02 judgmental it's like a it's like a it's
2:16:05 like a golden
2:16:07 retriever like I ask at someone it's
2:16:09 like sure boss that's great great idea
2:16:11 let me help you with that could you make
2:16:14 some code absolutely here's some code
2:16:16 you want some more no good could could
2:16:19 you could I tell you a story and you
2:16:22 could help me turn it into a poem yeah
2:16:25 here's a poem do you like it no not
2:16:27 really it's kind of sucks okay here's
2:16:29 another one you want some
2:16:32 more I'm really sorry I didn't get it
2:16:34 right that time I'll try to get it right
2:16:35 this time I'm like that's the [ __ ]
2:16:38 attitude to
2:16:40 have it says please it says thank you it
2:16:43 apologizes when it gets it wrong it
2:16:45 doesn't get hung up on the fact that it
2:16:47 that it it [ __ ] wrote something
2:16:49 horrible and completely lied to you it's
2:16:51 just like okay I'll try
2:16:56 again and you know the the the
2:16:59 technologists in here are probably well
2:17:01 that's how it was programmed they
2:17:03 programmed it to be a helpful assistant
2:17:05 so that's why it's doing that okay fine
2:17:08 I don't give a [ __ ] how it was
2:17:10 programmed or why it was programmed or
2:17:11 why it behaves like that but it behaves
2:17:15 like the kind of collaborator you
2:17:19 dream you'll have every time you hire an
2:17:22 intern or hire an employee or try to
2:17:24 partner with someone on a project you
2:17:26 dream that they will not be a flaming
2:17:28 [ __ ] that doesn't show up doesn't do
2:17:30 the work is bitter and cynical and is
2:17:34 completely locked up in fear you dream
2:17:37 that you find that person and you find
2:17:40 that person maybe one in a hundred
2:17:43 [Laughter]
2:17:45 times you know how [ __ ] hard it is to
2:17:47 get a
2:17:48 good collaborative human being that you
2:17:53 resonate with that resonates with you
2:17:55 where your conversations are additive
2:17:57 rather than
2:17:58 subtractive it's
2:18:01 rare and then chat GPT comes along and
2:18:03 it's like
2:18:05 here I'm singing I'm dancing I'm here to
2:18:08 collaborate anything you say is
2:18:14 awesome I'm like oh yeah I remember when
2:18:17 I used to be like optimistic like that
2:18:20 maybe I should get back to
2:18:28 that and and it's also like like the the
2:18:32 thing I've been thinking about for the
2:18:35 past week or so is just this idea that
2:18:37 that oh thank you so much Apple user
2:18:39 appreciate
2:18:41 that is
2:18:43 that the friction the friction
2:18:48 of taking an idea and turning that into
2:18:51 something is
2:18:53 gone when when you when you really adopt
2:18:55 these tools when you really Embrace what
2:18:58 these things
2:19:01 are and let your ego [ __ ] drift off
2:19:04 to the
2:19:05 side well I'm the most powerful thing in
2:19:08 this equation humans are important
2:19:11 robots are
2:19:12 stupid when when you let go when you let
2:19:15 that insecurity sort of drift off to the
2:19:17 side
2:19:25 then whatever
2:19:28 thin idea that you
2:19:32 have you know it's funny it It's it is
2:19:35 not dissimilar from AA like you you sort
2:19:38 of give over to this other power right
2:19:40 you know there it's the higher power you
2:19:42 know here it's here it's this this AI
2:19:45 power where you sort of give over to to
2:19:48 it and you say here's this kind of idea
2:19:50 and then it generates something you're
2:19:51 like
2:19:52 oh that wasn't what I had in mind but
2:19:55 that's kind of interesting and then you
2:19:57 give it some feedback and it generates
2:19:58 some more [ __ ] and you're like oh that
2:20:01 gives me a new idea and so like I'm
2:20:03 finding myself having 10 12 20 ideas
2:20:08 where I used to have
2:20:10 one and they're not ideas that just like
2:20:13 drift around my
2:20:15 [ __ ] add riddled brain
2:20:18 for 3 months while I try to figure out
2:20:20 if I'm going to be secure enough to get
2:20:22 it out of there they're out instantly
2:20:25 there's no
2:20:26 friction idea out idea out idea out and
2:20:31 every time you put an idea in the world
2:20:33 [ __ ] changes you like we all know this
2:20:36 if you've been on the planet for more
2:20:38 than a year you understand that if you
2:20:42 if you give into the fear no nothing
2:20:45 [ __ ]
2:20:46 happens if you overcome the the fear do
2:20:48 the courage take action in the face of
2:20:50 fear [ __ ]
2:20:54 happens so so now that we have a tool
2:20:59 where that creative friction is gone
2:21:01 writer's block is
2:21:04 gone the insecurity of I don't have
2:21:07 those skills or I'm not educated enough
2:21:09 or I'm not a good enough writer is
2:21:12 gone the only thing left is your ideas
2:21:15 and your instant ability to Express
2:21:19 them that will change your [ __ ]
2:21:24 world and it won't just change your
2:21:28 world this is why one of the things that
2:21:30 I harp on here is share what you're
2:21:32 learning share what you're learning
2:21:35 share what you're
2:21:36 learning but Kyle they'll think I'm
2:21:41 weird
2:21:43 yep the only thing you do in here is
2:21:46 make pretty pictures to tonight
2:21:51 yeah tomorrow we might solve
2:21:56 [Laughter]
2:22:01 Fusion yeah new large language model
2:22:04 came out where you can solve uh
2:22:06 unsolvable physics problems why don't we
2:22:08 go play with that that's going to happen
2:22:11 that probably within a year there's
2:22:13 going to be some [ __ ] model comes out
2:22:15 and it's like it's like you know
2:22:17 simulate things that have never been
2:22:19 solved in science for thousands of years
2:22:22 Kyle do you remember the post I tagged
2:22:24 you in today about Google search
2:22:28 generating images um yeah I do so I
2:22:31 haven't played with it yet
2:22:34 so goo what I don't know is if they've
2:22:36 added it for everyone or and I don't I
2:22:39 didn't I didn't go set it up so um
2:22:43 Google has apparently if you search for
2:22:46 an image on Google Now you can now
2:22:48 enable image generation so you can go
2:22:51 search for um an image on Google and if
2:22:54 you can't find it you can just say
2:22:55 generate it and it
2:22:57 will um so so they they have kind of
2:23:00 quietly released their image Generation
2:23:02 stuff I I'm I'm looking at that as kind
2:23:05 of a precursor to what I think Gemini is
2:23:07 going to be some multimodal
2:23:11 Beast it's a superc conductor for
2:23:13 knowledge yeah it's a Humanity amplifier
2:23:16 absolutely for Goofs I regularly ask
2:23:18 invent something mind-blowing Abbott
2:23:20 skinny yeah why not why not because you
2:23:23 know what's going to [ __ ] happen some
2:23:26 you potentially are going to say invent
2:23:28 something mind-blowing it's going to
2:23:30 it's going to invent it maybe with the
2:23:32 image generation it even visualizes
2:23:37 it and then you might get inspired to go
2:23:39 I'm going to go [ __ ] build that and
2:23:41 then like [ __ ] is going to be invented
2:23:44 because someone who had no right no
2:23:47 right right has not been trained in a
2:23:51 discipline is going to invent something
2:23:53 in that discipline that no one in that
2:23:55 discipline would have ever even thought
2:23:57 to invent because of their ignorance of
2:24:00 that
2:24:01 discipline as we as we as we all move
2:24:04 laterally and and get skilled at things
2:24:07 that are not the core to who we are
2:24:10 we're going to do [ __ ] differently than
2:24:12 the experts in those Industries or in
2:24:15 those verticals in those expertises
2:24:18 expertises
2:24:20 [Laughter]
2:24:23 experti and so what's going to happen is
2:24:25 there's going to be this wild invention
2:24:27 that
2:24:31 happens well they don't have a degree
2:24:33 and whatever right and because they
2:24:38 don't have a degree in that they asked
2:24:39 this really stupid question and then
2:24:42 chat GPT actually answered it without
2:24:44 judging
2:24:45 it and then they took took that answer
2:24:48 and they did something with
2:24:50 it well that's not how we normally do
2:24:52 things around here
2:24:56 exactly that's why you haven't [ __ ]
2:24:58 innovated anything in 30
2:25:03 years Sher D continuing to kill it with
2:25:06 the
2:25:09 imagery amazing I answer for that is
2:25:12 there something else I can help with no
2:25:14 Siri could you please go
2:25:16 away
2:25:18 [Laughter]
2:25:20 oh
2:25:26 Siri hi Kyle how can I help no I want
2:25:29 you to go
2:25:33 away okay you people are mean you people
2:25:37 are
2:25:38 mean but I love the story Vine
2:25:43 hat oh my
2:25:46 God
2:25:48 that's awesome thank you jrc for my
2:25:51 little
2:25:55 self-portrait
2:25:58 oh how we doing on time all right we're
2:26:01 getting there oh my back I'm sore like
2:26:03 all day today I've been sore and I
2:26:06 haven't drunk enough
2:26:08 water I haven't moved
2:26:11 enough any anyone else getting
2:26:14 increasingly sedentary with this AI [ __ ]
2:26:17 it's just like just just another hour
2:26:20 and then like eight hours later I'm like
2:26:22 I should probably do something with my
2:26:24 life this this is good right this is
2:26:27 good because I'm learning so much it's
2:26:29 all good right isn't this good hi my
2:26:31 name is Kyle I'm been a little too into
2:26:34 AI lately oh hi Kyle us
2:26:38 too it's not you it's the
2:26:43 character that's still pretty
2:26:46 good I I am not actually offended I
2:26:49 think it's actually really good David sh
2:26:51 Shapiro has me going down the AGI in 12
2:26:53 months hole jrc right holy [ __ ] that
2:26:57 video so okay so I did this the other
2:27:00 night I I'll do it again one of the guys
2:27:02 that I recommend you
2:27:05 follow seriously is this dude D oh wait
2:27:08 hang
2:27:11 on David Shapiro fourth at4 ir. david.
2:27:16 Shapiro on YouTube fourth Industrial
2:27:19 Revolution you should watch his video on
2:27:21 the fourth Industrial Revolution but go
2:27:24 watch this
2:27:36 [Laughter]
2:27:37 [Music]
2:27:41 video Switch
2:27:46 Account
2:27:50 abolish money here's a new one of his
2:27:52 abolish money will AGI destroy money so
2:27:56 so let me go to his channel um the one
2:27:58 that I'm looking for is
2:28:02 videos brace for impact seems like it's
2:28:05 a forgone conclusion that we will have
2:28:07 AGI within the next year or at least
2:28:10 really Advanced AI plus robotics whether
2:28:13 or not you define it as AGI it is time
2:28:16 to start preparing and adapting go watch
2:28:18 this
2:28:19 video David
2:28:21 shapiro4 i. david. Shapiro brace for
2:28:25 impact go watch that video 37 minutes
2:28:28 put it on time and a half and it'll be
2:28:30 30 minutes or 20 28 or whatever it
2:28:34 is or put put it on put it on 2x if you
2:28:38 want to hear David as a
2:28:42 chipmunk
2:28:44 um yeah
2:28:49 cuz it's possible he's wrong but if he's
2:28:55 not well go watch that video and come
2:28:58 back to go come back to the
2:29:00 meeting wait go watch that video smoke a
2:29:03 cigarette and come back to the
2:29:07 meeting any guesses when something AI
2:29:10 created will win a Nobel Prize oh that's
2:29:16 interesting
2:29:18 I would I would assume that
2:29:23 um I would assume that never right I
2:29:26 would assume
2:29:28 that 10:28 p.m. would you stop it I'm
2:29:32 not saying your
2:29:40 name poop the
2:29:46 rainbow
2:29:47 this is what happens creatively when you
2:29:49 start playing with
2:29:51 AI Gibbons Leonardo is amazing I know it
2:29:54 is isn't it it's really good Leonardo is
2:29:56 really good that's the thing like these
2:29:59 tools now there are so many remarkable
2:30:01 tools like it's just it it it's it's
2:30:04 literally
2:30:07 impossible like you could really spend a
2:30:10 month just learning Leonardo and getting
2:30:12 really good at Leonardo you could spend
2:30:15 [ __ ] five years getting good at chat
2:30:18 GPT that's the most polite way I've ever
2:30:22 heard someone tell Siri to [ __ ] off
2:30:25 that's awesome solve Fusion I'm in can
2:30:28 we tackle element uh 115 problem too I'd
2:30:32 love to Let's yeah let's do the element
2:30:34 115 thing because then we then we have
2:30:36 flying cars right like that look like
2:30:39 this you can make money with of and we
2:30:42 just shrink ourselves hop in this and
2:30:43 fly over to the other planet yeah I'm in
2:30:47 let's do
2:30:50 it maybe he will be right but I
2:30:52 certainly don't think so who David
2:30:55 Shapiro yeah we'll see we'll
2:31:00 see I mean the thing about David is he's
2:31:02 written five books on AGI he's got a
2:31:04 cognitive architecture open- source
2:31:06 project about AGI
2:31:10 um like one of the things he's talking
2:31:12 about just go watch that video because
2:31:13 one of the things he's talking about in
2:31:15 that video is
2:31:17 the cognitive architecture that he's
2:31:19 putting together like all the piece
2:31:21 Parts I made myself into a model on on
2:31:24 Leonardo oh that's super cool yeah I did
2:31:26 too I did I did one in there too I did
2:31:28 two in there cuz the first one was so
2:31:30 bad um but the thing about what what
2:31:33 David's talking about is all the piece
2:31:34 parts for for a lot of the AGI
2:31:37 components are kind of there right now
2:31:39 and and
2:31:41 so so the models are going to start
2:31:45 improving themselves which is the the
2:31:47 thing you know some of the people that
2:31:48 are afraid of but it also means that
2:31:52 that the Innovations are going to start
2:31:53 accelerating if if if you think that
2:31:57 this past year has been [ __ ] insanity
2:32:00 if you've been paying attention to this
2:32:01 stuff at all the next year is going to
2:32:03 be
2:32:05 Bonkers
2:32:08 Bonkers I honest to God if if I were a
2:32:11 freshman in college right now
2:32:14 like what like
2:32:18 the world four years from now is going
2:32:20 to look nothing like the world we're in
2:32:25 nothing how do you prepare for that how
2:32:28 do we prepare for that that's why we're
2:32:31 here that's what this that's what this
2:32:33 [ __ ] meeting is about I have so many
2:32:36 screenshots of Kyle on my phone my
2:32:39 photos think he's a family member well
2:32:41 that's really
2:32:45 bad
2:32:49 who's Kyle listen who's Kyle I want to
2:32:52 know who this is who's this Kyle
2:33:00 person oh man mine are coming up in
2:33:02 Google memories now oh
2:33:06 God that's really bad that's really bad
2:33:11 oh I'll watch tomorrow I can't miss this
2:33:12 live yeah you don't need to watch it
2:33:14 right now just go watch it at some point
2:33:16 know David Shapiro is one of those he's
2:33:19 one of those dudes like I know I've I've
2:33:21 gotten the the semi-roll semi compliment
2:33:24 that I fall asleep to your lives um
2:33:26 David Shapiro is definitely one of those
2:33:28 dudes that I fall asleep to because he's
2:33:30 got he's got a very sort of monotone
2:33:32 kind of delivery is just very matter of
2:33:34 fact
2:33:35 and you know here's what's going to
2:33:37 happen when the robots do this and
2:33:39 here's what's going to happen with this
2:33:40 and he sort of ties everything back to
2:33:42 history so he's very good to fall asleep
2:33:44 to but
2:33:47 you know in those times when I haven't
2:33:50 fallen asleep to him he's it's some
2:33:53 pretty eye opening stuff it's just it's
2:33:55 just very well thought out his his
2:33:57 work's very well thought out as as
2:33:59 opposed to this channel welcome to Chad
2:34:04 add oh man if we don't get it it's
2:34:08 probably be not because it isn't there
2:34:11 it's because the world isn't ready yet
2:34:13 yeah you know it's funny uh JPC guy
2:34:17 um one of the things that's hit
2:34:23 me is that so I experienced this with a
2:34:26 client I had a client hire me and
2:34:28 another guy to
2:34:29 do an experiment I
2:34:32 said I said to
2:34:35 her I think that a 10p person agency is
2:34:39 going
2:34:40 to kick the [ __ ] out of a 100 or 200
2:34:43 person agency like in the next year
2:34:47 I said because if you get 10 smart
2:34:49 people that have kind of different
2:34:51 disciplines that all go all in on AI and
2:34:55 you and you build an agency from the
2:34:57 ground up using AI
2:35:01 tools you're going to be able to out
2:35:04 execute
2:35:06 underprice and like overd deliver from a
2:35:09 timing standpoint big agencies just cuz
2:35:12 it's a it's just a fundamentally
2:35:14 different model and so she she hired us
2:35:17 and she said well why don't we do an
2:35:19 experiment so it was the two of us and
2:35:20 we met with her on a Friday he and I
2:35:23 spent an hour and a half on Saturday
2:35:24 morning with chat
2:35:26 GPT came up with four strong Concepts we
2:35:31 each spent about two or three hours on
2:35:33 that Saturday took Sunday
2:35:37 off delivered her
2:35:40 four fully fleshed Out Concepts like
2:35:44 with creative on Monday she picked one
2:35:47 of the campaigns and we delivered it on
2:35:52 Tuesday like he and I maybe put in in in
2:35:56 just in terms of man
2:36:01 hours maybe a day and a half maybe two
2:36:04 days let's let's say two days 16 hours
2:36:07 we maybe put in 16 hours but I don't
2:36:10 even think it was that but let's say it
2:36:11 was
2:36:12 that and it was work like it it was work
2:36:17 that you would have taken at least two
2:36:19 weeks to do and probably a month to do
2:36:22 with with with an agency with you know a
2:36:24 small
2:36:27 agency she wasn't ready for it like we
2:36:30 delivered it faster than she was
2:36:32 prepared to accept it so I think I think
2:36:35 there's there there is a weird
2:36:38 thing where the Cadence of business
2:36:42 there's going to be some people
2:36:44 operating at this hypers speed
2:36:47 but the Cadence of business is expecting
2:36:49 things at their Cadence and and there's
2:36:53 going to be a whole other class of
2:36:54 people that are
2:36:57 like and I think that's going to take a
2:36:59 while for these these two things to
2:37:03 normalize because it like
2:37:07 like I assume this goes down to like
2:37:09 some deep reptilian [ __ ] like circadian
2:37:12 rhythm kind of stuff that we sort of
2:37:14 expect if I ask someone to do something
2:37:17 it's going to be at least a week until
2:37:19 they get back to me and then it's going
2:37:20 to be at least a week for them to do
2:37:22 this part and at least a week to for
2:37:24 them to do that part if I now have that
2:37:26 meeting where I'm expecting that 3- week
2:37:29 Cadence and I get back the answer the
2:37:31 next day I'm like oh [ __ ] like it's
2:37:35 jarring so I think
2:37:39 I and like does that mean
2:37:44 that there's this separate economy that
2:37:46 starts to
2:37:48 happen there's a separate economy that's
2:37:50 operating at
2:37:52 this 2hour turnaround speed and this
2:37:56 economy is operating on a two- we
2:37:57 turnaround speed boom
2:38:01 boom boom
2:38:05 boom and how do those two things come
2:38:08 together I don't that I don't know mgpt
2:38:10 any comments yes my comment is I haven't
2:38:12 heard of
2:38:14 it let's go look
2:38:18 mgpt is this the memory
2:38:20 thing teaching llms memory okay so this
2:38:23 is like Lang chain it's a system to
2:38:26 intelligently manages memory tiers and
2:38:28 llms in order to effectively provide ex
2:38:30 extended context um yeah is this how new
2:38:33 is
2:38:36 this is this built on top of Lang
2:38:42 chain event
2:38:45 mgpt
2:38:47 main context external context
2:38:50 function looks great I mean like like
2:38:53 the drawing looks I mean the the concept
2:38:55 looks
2:38:57 great what's the date on this is there a
2:38:59 date on this
2:39:00 paper no UC
2:39:03 Berkeley paper let's go look at the
2:39:07 paper uh
2:39:09 uh October 12th
2:39:11 2023 you're learning it here first so it
2:39:14 seems like a a Slicker maybe more
2:39:17 automated simplified version of Lang
2:39:21 chain yeah I think I I
2:39:25 think piie right now has a little bit of
2:39:28 memory in it which is interesting like
2:39:31 the other day I I was talking to Pi and
2:39:34 Pi
2:39:35 said like good idea Kyle like she used
2:39:39 my name I was like oh [ __ ] right she
2:39:42 remembers [ __ ] um if recursively self-
2:39:46 improving AI is achieved it could
2:39:48 potentially do 10 10,000 years advance
2:39:51 in two days yeah Sher bear thank you so
2:39:53 much I appreciate that I appreciate the
2:39:56 gifts when you when you guys do the
2:39:57 gifts it helps a lot I appreciate it
2:40:00 um yeah Sam Sam Alman Herod talked about
2:40:05 um he feels confident that in the next
2:40:09 few
2:40:11 years we're going to have a single
2:40:14 year of scientific
2:40:17 advancement that equals the last 500
2:40:20 years of scientific
2:40:21 advancement and if you want to do a
2:40:23 [ __ ] thought
2:40:25 experiment
2:40:27 go
2:40:30 um have chat GPT create a table for you
2:40:34 of the past the the major scientific
2:40:37 advant advancements over the past 500
2:40:41 years and just look at it just read it
2:40:45 and then imagine
2:40:48 if the equivalent of those advances
2:40:51 happen in a single
2:40:56 year like that is
2:40:58 within the realm of our very near future
2:41:02 like it's within the realm of
2:41:04 possibility of our very near
2:41:06 future well what's that mean well it
2:41:10 could mean Fusion it could mean solution
2:41:12 to the climate [ __ ] up
2:41:17 it could mean the cure for cancer it
2:41:19 could mean the cure for genetic diseases
2:41:20 it could mean the cure for paralysis it
2:41:22 could mean the cure for people trapped
2:41:25 in neurological
2:41:36 hell what I I mean what I find
2:41:38 fascinating about that one
2:41:42 is like what do that break we we have
2:41:46 have we our our entire economy in a lot
2:41:48 of ways is based on these over
2:41:52 complicated systems that are there you
2:41:56 know to to you know profit the companies
2:41:58 and things like that like
2:42:01 like curing cancer is not necessarily
2:42:05 something that Pharma companies want
2:42:07 right it's not very profitable to cure
2:42:11 it so they're going to slow roll that
2:42:14 well what if that's no longer in their
2:42:17 hands to solve what if because you're
2:42:20 democratizing
2:42:23 expertise that gets solved independent
2:42:26 of the economic
2:42:28 interests and it gets solved in such a
2:42:32 transparent and ubiquitous way that they
2:42:35 can't stop
2:42:38 it what's that
2:42:40 do the cure for aging yeah exactly
2:42:44 Gordon what what's your name Gordon
2:42:47 espan one Gord wait Gordon o desan
2:42:52 one
2:42:54 yeah what if we cure aging is is that
2:42:57 necessarily a good thing right what if
2:43:00 all of a sudden average lifespan goes
2:43:02 from you know whatever it is 85 or 89
2:43:05 years to 150 or 200 oh [ __ ] what do we
2:43:10 do what do we do with all those people
2:43:12 aging is in a disease why a cure well
2:43:16 aging isn't a disease but living longer
2:43:18 is a desire
2:43:25 right and and and whether or
2:43:29 not you classify it as a disease or
2:43:34 not and whether or not you think
2:43:36 extending lifespan is a good
2:43:40 thing it's likely there's going to be
2:43:42 advancements made whether you want them
2:43:44 or not right just like this AI stuff
2:43:47 you're an inspiration Kyle awesome
2:43:48 thinker thank you very
2:43:50 much yeah I feel like I had to kick it
2:43:52 up a notch when when when we got trolled
2:43:54 they said is all you're making pretty
2:43:56 pictures is this just a pretty picture
2:43:58 Channel I'm like all right [ __ ] it let's
2:44:00 go to
2:44:03 AGI we can we can you you want to turn
2:44:06 this up a notch but I still
2:44:10 think that llamas farting rainbows is
2:44:13 the way to go like not for nothing oh
2:44:17 thank you very much
2:44:19 Zeb first gift that's really awesome
2:44:21 thank
2:44:22 you
2:44:27 uh oh yeah this is his thing that's
2:44:29 David Shapiro's AGI within 12
2:44:32 months this is where this is where he
2:44:35 upgrades from 18 months to 12 he he
2:44:38 thinks we're there
2:44:41 sooner oh this is another good Ted image
2:44:45 sorry we got to get back to the images
2:44:47 the images are just too good what if Ai
2:44:49 and Manufacturing means iPhones can be
2:44:51 made in the USA exactly exactly exactly
2:44:58 exactly what if you solve
2:45:09 energy what if what if I I mean just
2:45:14 it's like what does it do if if every
2:45:17 house could have its own little box that
2:45:19 runs all the that gives that house all
2:45:22 the power it needs ever Kyle I disagree
2:45:25 all this takes action and all llms can
2:45:27 do is generate strings of symbols or
2:45:31 images
2:45:34 um I so Brian Pullman
2:45:38 pulum um
2:45:42 understood go look at um some of the
2:45:46 Boston Dynamics things where they're
2:45:47 putting chat GPT into the robots go look
2:45:50 at have you seen the
2:45:52 um it's not so so technically what
2:45:55 you're saying is correct large language
2:45:58 models are nothing more than a
2:46:00 probability engine from a from a from a
2:46:03 what they do technically perspective
2:46:05 you're absolutely correct they don't
2:46:07 have sensient they don't have
2:46:09 Consciousness they don't have empathy
2:46:11 they don't have
2:46:12 compassion they're probability engines
2:46:14 absolutely agree
2:46:16 with that technical definition but if
2:46:20 you look at what they actually
2:46:23 do when chat gbt actually writes a
2:46:27 python
2:46:29 application the first thing I did with
2:46:31 chat GPT that that made it click for me
2:46:34 that this is
2:46:36 different this
2:46:39 is this is not Computing as we've ever
2:46:41 known it before was in 90 minutes
2:46:46 just [ __ ] around while my wife was
2:46:48 watching some shitty TV show I wrote a
2:46:51 fully functional python application that
2:46:54 made five different calls to open ai's
2:46:57 API and was a social media generating
2:47:01 machine I wrote it 90 minutes I've never
2:47:05 written a line of python in my life ever
2:47:08 and still
2:47:10 haven't that's not just putting symbols
2:47:13 out that's taking action
2:47:16 you now combine that with
2:47:19 robotics let me let me let me find this
2:47:22 this
2:47:23 um this thing I saw on on Twitter
2:47:29 today
2:47:33 x.com let's see
2:47:37 robot pen
2:47:40 spinning do you see this thing
2:47:43 today yeah this
2:47:46 can cat gbt teach a robot to do pen
2:47:49 spinning tricks better than you
2:47:51 do I'm excited to announce Eureka an
2:47:54 open-ended agent that designs reward
2:47:57 functions for robot dexterity at
2:47:59 superhuman
2:48:01 level they train the robot to do pen
2:48:04 spinning in 2 minutes with chat
2:48:11 GPT that Hardware exists it's being
2:48:15 built by by five or six different
2:48:17 companies oh thank you very much Ron it
2:48:20 is [ __ ]
2:48:23 magic so so while technically large
2:48:25 language model are are exactly as you
2:48:28 describe them what these things are
2:48:30 actually
2:48:32 doing the images that I'm showing here
2:48:36 are just people in
2:48:39 seconds taking a brain fart of an idea
2:48:43 putting it into cat GPT and it emerges
2:48:47 it can write code it's going to be able
2:48:49 to reason it's going to have memory go
2:48:50 look at the cognitive architecture um
2:48:53 that that David Shapiro has open
2:48:56 sourced the components are actually
2:48:58 quite simple that he's putting together
2:49:00 but when it starts having memory and the
2:49:02 the the mem GPT that someone just you
2:49:05 know shared with
2:49:06 me when these things start having memory
2:49:09 and can start training Hardware like
2:49:11 this and then you connect it to the
2:49:12 hardware this is significantly
2:49:16 Beyond this is just generating symbols
2:49:19 so you you might disagree with it so I
2:49:22 think I think I agree with you as a
2:49:26 technicality but I massively disagree
2:49:28 with you as a
2:49:30 reality because these things are
2:49:33 actually doing something significantly
2:49:36 Beyond how you just categorized
2:49:43 it Sherry d
2:49:45 [Laughter]
2:49:51 this is this is the new Kevin mallister
2:49:55 moment that's [ __ ]
2:49:58 awesome oh
2:50:02 man all
2:50:06 right good Lord hey Cowboy wish I could
2:50:09 quit
2:50:10 [Laughter]
2:50:14 you
2:50:20 I just wish I knew what to invest in
2:50:22 yeah I know uh Nvidia although you know
2:50:26 you know what's funny so over the past
2:50:29 few months like I've sort of been kind
2:50:31 of going on about I can't believe that
2:50:34 the entire computer industry has
2:50:36 effectively
2:50:37 seeded the the hardware production to
2:50:41 Nvidia that that like everything's gone
2:50:43 to gpus and Nvidia is the the only one
2:50:45 doing it now meta open
2:50:49 AI
2:50:51 um someone else there have been like
2:50:53 three or four announcements that people
2:50:55 like [ __ ] it we're making our own gpus
2:50:57 now like like all of a sudden people are
2:50:59 like no we can't no Invidia doesn't get
2:51:01 all the business so so all the big
2:51:04 companies now are like okay we're going
2:51:05 to make our own ships now so it's like
2:51:09 damn it's like it's amazing it's amazing
2:51:13 like and and even that I mean think
2:51:14 about that so so right now you know gpus
2:51:18 do what they do and tpus do what they do
2:51:20 P's in here he's working on TPU stuff
2:51:24 so
2:51:26 so we're figuring out which of these
2:51:29 architectures are good for AI now that
2:51:32 you've got three or four major
2:51:33 manufacturers going [ __ ] it we're going
2:51:34 to make our own chips that really
2:51:36 specialize in this they're likely going
2:51:38 to come up with even more advanced
2:51:41 architectures that that do [ __ ] at some
2:51:44 other level above so the hardware is
2:51:46 going to continue to
2:51:49 get better and better and innovate more
2:51:51 and more I took a pick of a rest
2:51:53 restaurant buser's cart and asked what
2:51:56 business I could start interesting it
2:51:59 came up with
2:52:01 five yeah exactly hey Kyle I missed
2:52:04 those names you were mentioning
2:52:06 recommending the the name Ethan mullik
2:52:09 he's Wharton business school
2:52:11 Professor one useful thing. org is his
2:52:16 uh blog but he also follow him on
2:52:20 Twitter on X David Shiro we just talked
2:52:23 about a lot and then Paul rer from the
2:52:25 marketing AI
2:52:27 Institute and then I would also follow
2:52:29 on Tik Tok I would follow Rachel woods
2:52:32 from the AI
2:52:34 exchange um she she's she's
2:52:38 good she kind of does generative AI
2:52:41 applied to business and and she's a lot
2:52:44 about AI oper operations right now like
2:52:45 how do you actually take this stuff and
2:52:47 start to build tools and build
2:52:49 automations and things like that she's
2:52:51 got some workshops and things like
2:52:54 that so Ron thanks for everything sir
2:52:58 you are welcome sir I am getting crispy
2:53:02 I think I
2:53:03 gotta I gotta I got to start whining
2:53:07 this bad boy down that that got a little
2:53:09 peppy that got a little peppy there my
2:53:12 hair my hair is looking like an 80s uh
2:53:16 like Thompson Twins kind of
2:53:19 situation it's it's like I put hairspray
2:53:22 on it in the 1980s and haven't changed
2:53:24 it
2:53:28 since AI learning lab
2:53:33 M why can I not make that bigger oh
2:53:38 well I've been trying Kyle but
2:53:41 famously known robot models almost all
2:53:44 of gpts
2:53:48 modalities that's good every live oh
2:53:52 yeah
2:53:54 uh I think nvidia's up over 4:30ish per
2:53:57 share I know like Nvidia Nvidia they're
2:54:02 like it's it's just amazing it was just
2:54:05 like everyone was like all CPUs and they
2:54:08 were like [ __ ] gpus and then like well
2:54:10 it's just for
2:54:12 gamers and then gaming like
2:54:15 dwarfed Hollywood like gaming became
2:54:17 this massive thing and then it was
2:54:20 Bitcoin and everyone's doing Bitcoin and
2:54:22 then they sold all those gpus for that
2:54:24 and now it's AI it's just like everyone
2:54:27 else drifted away in Nvidia was just
2:54:29 like [ __ ] it we'll just we'll we'll
2:54:31 stick to the GPU game so now now they're
2:54:35 going to get some competition but it's
2:54:36 what's it going to take it's going to
2:54:38 take at least a year probably two or
2:54:41 three to spin up GPU factories from the
2:54:43 big companies hopefully AI can solve the
2:54:46 problem of large hyron collider and get
2:54:48 us to the right Universe I know I do I
2:54:51 do feel like we've been shifted a a
2:54:53 reality or two to the left it would be
2:54:56 nice to it'd be nice to be able to like
2:54:58 can we just kind of flip some of those
2:55:00 switches and get us back back to
2:55:02 something resembling normal wind down
2:55:06 you know an anniversity Marathon to
2:55:08 train for I know I do it's true I got to
2:55:11 figure out how I'm going to do that if
2:55:13 I'm going to do that it's not a foregone
2:55:15 conclusion it's just an idea right now
2:55:17 but you know me I'm batshit crazy so
2:55:20 hopefully AI can solve wait was that
2:55:23 it's
2:55:26 been it's been 15 minutes since my last
2:55:30 dol3 hi
2:55:33 Kyle I'm making one right now by the way
2:55:36 if you haven't played with um Dolly
2:55:40 image generation on the iPhone app the
2:55:42 chat GPT app it's like
2:55:47 I have in my head that anything you
2:55:49 generate on a phone is kind of of lower
2:55:51 quality and lower
2:55:53 whatever and then you create these
2:55:55 dolly3 images on the phone and you look
2:55:57 at them on the computer and they're like
2:55:58 these really big things signing off Kyle
2:56:00 explore the Gen Gen X connection here
2:56:02 it's been real and we have a big role
2:56:04 here absolutely Brian Brian I listen I
2:56:06 really appreciate your point of view and
2:56:08 your contributions so please keep coming
2:56:10 back um
2:56:12 because like all of this [ __ ] is not a
2:56:15 foregone conclusion right all of this is
2:56:17 happening in real time this channel is
2:56:19 just about trying to keep up with it in
2:56:21 real time and explore it so please come
2:56:23 back I really appreciate your
2:56:24 contribution dog food is done we can go
2:56:26 to bed
2:56:27 now all right everybody listen I'm going
2:56:30 to wind it down um Vicky thank you Joker
2:56:34 Silver Fox great seeing you Amelio's
2:56:36 wife as always you're late um Joker
2:56:40 seril Simpson has a country song called
2:56:42 Turtles all the way down good we need to
2:56:44 make that
2:56:46 image Holly Allison thank you so much I
2:56:49 appreciate that um who else is in here
2:56:52 Ted looks at life killing it with the
2:56:53 images Brian thank you um Kevin Vasquez
2:56:57 very nice to see you cockal thank you
2:56:59 Vicky as always good to see you Clint
2:57:02 Copperhead Clint are you ready you going
2:57:04 to there we go everyone's your regulars
2:57:06 and mods present Turtles llamas
2:57:09 hedgehogs or your favorite animal thank
2:57:11 you for the donut there Ted looks at
2:57:13 life I appreciate get
2:57:15 it all right everybody listen happy
2:57:19 Saturday night it's Saturday Night Live
2:57:21 don't forget you can make
2:57:25 money I'll see you tomorrow everybody I
2:57:28 hope this has been fun we made some cool
2:57:29 images um follow the channel follow the
2:57:32 channel follow the channel tell people
2:57:33 about
2:57:34 it I will see you tomorrow night
2:57:38 peace