aiLL 10/5/23 - Democratizing Creativity and Connection with AI

Description

Chapters

Transcript

0:24 [Music]
0:27 we're going live is anybody here is any
0:30 anybody joining hello Tik Tok you going
0:32 to make us live
0:34 hello looks like some people just joined
0:37 steo what's
0:40 happening quick on the trigger tonight
0:45 dude I just got back from a little
0:47 networking event what
0:50 shaking I was uh it was this these CEO
0:55 dinners in Denver called thunderview
0:58 so I was thunderview
1:08 [Music]
1:10 doing zero
1:13 likes the hell is going on people got 11
1:17 people in here I haven't said a word why
1:18 would I not have any likes come
1:20 on techy Becky what's
1:23 happening Tyron truth hello hello hello
1:27 hello hello there's there's the likes
1:30 let's get this like game going I feel
1:33 like all the Irregulars have been
1:34 slacking off on their like game
1:35 competition we got to fix that [ __ ] cuz
1:38 otherwise we can't you can make money
1:42 with can't make money without the likes
1:45 although from all I can gather the likes
1:48 there's no there's no money associated
1:49 with the likes I think they're just I
1:52 don't know I think they're just a game
1:55 this whole platform is just a gamified
1:57 mess isn't
1:59 it
2:00 I
2:01 know here's six other ways to pump their
2:07 dopamine oh man welcome everybody
2:11 getting a little bit of a late start
2:12 tonight I won't be on too long but we'll
2:15 have a good time my name's Kyle Shannon
2:18 this the AI learning lab there's Jim
2:19 Ross AI
2:24 Superstar doing some cool stuff getting
2:27 articles written about him you know just
2:30 k killing it killing it how'd he do that
2:32 he just paid
2:34 attention well he's got other skills too
2:38 he's like applying it but uh but man the
2:42 opportunity right now you just got to
2:44 pay
2:45 attention you just got to pay attention
2:48 so what we do here a little bit about my
2:50 background I'm an entrepreneur many time
2:53 entrepreneur officially unhirable at
2:55 this point I would say
2:58 and and uh I'm the co-founder and CEO of
3:02 storyvine and I'm the uh founder of a
3:05 community uh here in Denver and online
3:07 called the AI salon and then I host this
3:11 what else do I do I don't know do a
3:13 bunch of [ __ ] in the mid90s uh I started
3:16 an one of the early digital agencies and
3:18 I say that because in the in uh in 1994
3:22 I had a big Epiphany about the worldwide
3:24 web about holy crap this changes
3:25 everything and I didn't think there'd be
3:27 another time in my life where that
3:29 happened again and it did with this chat
3:31 GPT stuff with the the generative Ai and
3:34 so all of this this Channel and my salon
3:37 and you know have been around pretty
3:39 much since the beginning of
3:41 that and that's what we talk about so if
3:43 you have questions about generative AI
3:45 pop them in the comments below um you're
3:48 free to troll me if you're creative
3:50 about it uh if if you're not imaginative
3:53 just shut your mouth the Irregulars will
3:56 they will take you down those are the
3:59 people that are in here a lot
4:00 uh if you got a good troll troll me up
4:03 uh if uh if you have questions ask those
4:06 um trying to think what else actually
4:09 you know something that's been
4:10 interesting in the past I don't know two
4:13 three weeks is there's been a lot more
4:15 people in here instead of asking
4:16 questions they've been sharing
4:19 experiences like hey I started talking
4:21 about this at work and all of a sudden I
4:22 got invited into a meeting and now I'm
4:24 going to go meet with the big boss
4:26 things like that so that's uh that's
4:28 pretty exciting that's pretty cool stuff
4:31 so um let's let's let's get started this
4:34 way as well um if you are brand spanking
4:38 new to this generative AI stuff and
4:40 you're annoyed by people like me and
4:44 others that are like have you tried chat
4:46 gbt it's amazing you should try chat gbt
4:50 if you're really annoyed with those
4:51 people I agree we're [ __ ] annoying
4:54 I'm pathologically annoying at this
4:56 point like I don't know I honestly don't
4:59 know how people in my life can stand me
5:01 but I'm not going to [ __ ] stop
5:03 talking about it so if you're one of
5:06 those people hasn't tried it yet go
5:08 there right now chat. open.com just go
5:10 there just go there and do it just do it
5:11 do it do it do it hey zip zip it do it I
5:16 don't care if it's evil I don't care if
5:18 the robots are going to kill us I don't
5:19 care if this thing steals from artists
5:21 all that sort of stuff's true yep great
5:23 perfect anyway go there anyway go there
5:26 anyway go there anyway well what I do
5:28 there good just go it's just a little
5:31 box just like a Google box ask it
5:33 anything can I input whole books yes you
5:36 can in uh Claude
5:38 doai cloud. A has 100,000 token context
5:42 window that's that's that third one down
5:45 there oh uh Tark alien thank you very
5:49 much I appreciate it wow digging it got
5:52 some music notes got a little what is
5:54 that a paw I can't see what that is I
5:56 think it's a paw and a hat
5:58 GG man oh I'm getting all the goodies
6:01 this is awesome thank you very much um
6:04 yeah so you can input whole books so
6:06 claw. has 100,000 token context window
6:09 which means uh about 75,000 words it's
6:13 amazing you're amazing and so is what
6:15 you're doing to benefit others thank you
6:17 well thank you very much I really
6:18 appreciate your generosity and I
6:20 appreciate the kind words um sometimes
6:23 as you can imagine it feels like I'm
6:25 sitting in my home office talking into a
6:28 phone
6:31 cuz that's what I'm doing but there are
6:34 people on the other side of it and and
6:36 you know we're actually hanging out I'll
6:37 talk about the salon here in a second it
6:39 was funny I was at this networking event
6:40 tonight here in Denver and one of the
6:43 guys that hosts it he's a guy that owns
6:44 the Mercury Lounge here in Denver and uh
6:48 I said something about doing a bunch in
6:50 AI on AI and he goes he goes you're on
6:53 Tik talk a
6:55 lot I said yeah I know I've been going
6:57 live every night he goes I know I watch
6:59 you
7:00 so you never [ __ ] know man you never
7:04 know um but I appreciate the kind words
7:07 okay so these four are what I call the
7:10 adults table these are the four to
7:11 really pay attention to they're not
7:13 going away anytime soon they're going to
7:15 get more and more and more and more
7:17 sophisticated um they are all quite
7:20 powerful right now they all have their
7:22 quirks about them um uh you're supposed
7:27 to talk into phones
7:30 know it's like it's like phones are
7:33 anything but phones these days aren't
7:35 they it's like I I saw that uh Apple
7:39 just added a mode where you can
7:41 basically just turn your phone into a
7:43 phone again where it basically just
7:45 blocks all the apps and you put like
7:47 four icons on it it's like phone text
7:50 you know whatever it was photos and
7:52 music or something like
7:55 that so go to chat. open.com if you have
7:59 haven't played with it um B that's
8:01 Google's entry that's going to get
8:03 dramatically better in the next month or
8:05 so cuz they're going to release this new
8:07 thing called Gemini which is the engine
8:09 that sits underneath it Claude um is
8:13 already remarkable and a and a real
8:16 competitor to chat
8:17 GPT um and it's got 100,000 token
8:20 context window so again that means
8:22 75,000 words so you could take um a
8:26 50,000-word novel literally paste it
8:28 into the prompt and start asking
8:29 questions about it I love your knowledge
8:32 and inspiration but really appreciate
8:34 your entertainment value thank you Tech
8:36 Becky I try to be a little entertaining
8:38 I try to try to make it not so [ __ ]
8:41 boring this stuff is you know what's
8:43 funny about this Tech stuff
8:45 is
8:47 technologists get seduced by the
8:50 tech and it's really easy to get seduced
8:53 by the tech but in the end this is
8:57 about people
9:01 and artists and humans self-expressing
9:04 and these tools are amplifiers of that
9:07 and and I think it's one of the big
9:09 misconceptions about you know it's it's
9:12 it's really what this channel is about
9:13 that this channel I mean listen we we
9:17 get Pate comes in here and he works at
9:19 Google on tpus and and Sam Horn who's in
9:21 here is a a master [ __ ] geek if
9:23 you're a Super Geek and you enjoy this
9:25 [ __ ] great but the you don't have to be
9:28 that and you don't have to be that with
9:30 chat GPT which is the best I'll get to
9:32 that in a second
9:35 um it's a large language model what
9:38 what's what's happening the shift that
9:40 we're in the middle of right now is all
9:44 of the Decades of work that these
9:46 scientists and researchers and Engineers
9:48 have
9:50 done with large language models with
9:53 chat GPT and Bard and Claude and Bing
9:55 and all of
9:57 them we get to just use our words
10:01 just use our words just have our ideas
10:03 we share our words and and ideas and
10:05 these things amplify that it's it's
10:09 [ __ ] magic Arthur C Clark one of my
10:12 favorite quotes from Arthur C Clark is
10:14 any sufficiently advanced technology is
10:17 indistinguishable from Magic these
10:19 things are [ __ ] magic Miss some
10:22 episodes what's new hey corporate
10:23 digital um I'll show you I'll show you
10:26 what's new we'll we'll go look at chat
10:28 GPT with do inside it I got access to
10:31 DOI inside chat GPT holy [ __ ] is it
10:34 amazing and it you know
10:36 it's it makes images well but what's
10:39 amazing about it is the interaction like
10:42 how you interact with it is just magic
10:45 so anyway I'll show you that so go with
10:47 these the question about which one's
10:49 best if if you were just going to use
10:51 one the the only one I would recommend
10:54 right now is chat GPT chat.
10:57 open.com and I personally would pay the
11:00 20 bucks I pay the 20 bucks for their
11:02 plus
11:03 model because the 20 bucks gives you
11:05 access to
11:06 gp4 and even though Bing is built on top
11:10 of gp4 it is not the same as open AI
11:14 chat GPT so you get gp4 you get browse
11:18 with Bing so it's re they reconnected it
11:20 to the Internet so it's now connected to
11:21 the internet again you get Advanced Data
11:24 analysis which is [ __ ] remarkable you
11:27 get plugins which are you know hit or
11:29 but there's some really cool ones uh and
11:32 you and you also are going to get you
11:34 may or may not have it now do which is
11:37 the image generation within this thing
11:39 it'll also have the ability to read
11:41 images and then at some point here
11:43 you'll be able to talk to it and listen
11:44 to it I
11:45 mean this thing is
11:47 getting really really
11:49 powerful Claude is apparent or Bard is
11:52 apparently going to do that um Claude
11:54 just got $4 billion from Amazon and if
11:57 you want to do long documents or
12:00 document analysis claude's really good
12:03 um Bing is on this list because they
12:05 have Dolly 3 but the their uh safety
12:08 guard rails are real pain in the ass um
12:12 hello every time I put like a celebrity
12:14 reference for AI art it doesn't look
12:16 like them at all yeah I don't know I I
12:19 mean the the getting one of the one of
12:22 the biggest things about these all these
12:26 generative image things right now is
12:29 getting predictable results is still way
12:32 more art than science so just keep
12:34 [ __ ] around with it you'll you'll get
12:36 it um that little thing at the bottom
12:39 there prompt. chat that is a just a
12:41 primer to to teach you how to talk to
12:43 these things and give you some copy and
12:45 paste prompts to start with um the kids
12:48 table as I like to call it um there's
12:50 these three things there's there's
12:52 perplexity doai in the middle there's Pi
12:54 at the top and po at the bottom so Pi is
12:58 designed to be Aver conversational bot
13:01 conversational you know chat
13:03 interface it is horrible if you want to
13:06 get work done it is [ __ ] brilliant if
13:10 you want something to talk to that feels
13:13 like a human like like a good human who
13:16 listens and encourages you and talks
13:20 with nuance and can explain things and
13:22 is infinitely patient and infinitely
13:25 knowledgeable go to the I go get the IOS
13:27 app of Pi search for pi personal
13:31 assistant and and there's a little green
13:33 telephone icon in the search bar just
13:35 hit the telephone icon and start talking
13:37 to it you'll you'll see it's just it's
13:40 just [ __ ] weird it's good
13:43 weird um perplexity lets you play with
13:46 meta's llama 2 models Winston llama 2
13:50 Winston paging Winston paging Winston
13:52 llama llama llama
13:58 l and I know I don't need to actually
14:01 call you out on that because you're like
14:02 a flipping machine with that thing
14:06 um Can Claude AI do
14:11 sop oh what's sop again I forget what
14:15 sop
14:17 is what's sop
14:20 again uh pop it in there I can look it
14:22 up I don't I don't want to look it up
14:24 um I don't know uh seriously Pi pie AI
14:29 is becoming a friend I know what did I
14:33 do tonight I met there was a a CEO
14:36 tonight she runs a um like an end of
14:39 life technology thing for people that
14:41 are in hospice and things like
14:44 that and she was talking about how oh
14:47 people people are like sharing their
14:48 stories and things like that I'm like oh
14:50 you got to try pie and so I I popped it
14:52 in pie I mean I popped open pie and I
14:56 said um what did I say said I said
14:59 something about I want help preparing
15:01 because my mother's dying my mother's
15:03 actually dead but um but I was just
15:05 saying that the pie is an example and
15:07 Pie came back and went no no no no no no
15:10 no is that true is that really true it
15:13 like it totally got like like worked up
15:16 that I said my mother was dying it was
15:19 just
15:20 crazy um yeah so it's pies really
15:23 interesting standing standard operating
15:25 procedure oh
15:27 um well I don't know what you mean can
15:29 it do standard operating procedure like
15:32 like like can it teach you about it does
15:34 it know
15:35 it or or can it be can it be really good
15:39 at doing doing standard operating
15:41 procedure across a series of
15:44 tasks
15:46 um all of these models still hallucinate
15:49 so it may or may not be able to um but
15:52 you can certainly take sop kind of
15:55 documentations and pop it in there you
15:57 could take a big old you know sop
16:00 document pop it in there and then start
16:04 um interacting with it so um it's it's
16:06 worth experimenting with that's that's a
16:09 a level deeper than I have any knowledge
16:11 on okay here's another one that's worth
16:13 screenshotting this is for the pretty
16:15 pictures um Dolly 3 is is in bing right
16:20 now so if you go to Bing the the chat if
16:23 you go to regular Bing and say make me
16:24 an image if you go to Bing SLC create if
16:26 you use the Bing app those are all
16:28 dollar 3 and then it's also in chat GPT
16:32 and I'll show that in a second here um
16:35 Mid journey is if you want high quality
16:37 art mid Journey still the best for for
16:40 the art crowd ideogram I don't know that
16:43 I'm probably going to use again but
16:45 because of because I have access to G or
16:47 Dolly 3 and chat GPT it's just it's just
16:51 it just hits different um but ideogram
16:54 is a really simple site it does it does
16:57 uh text in images so all of the um those
17:01 are all do but these are
17:04 all
17:05 ideogram so ideogram is doing some
17:07 pretty cool stuff right but it like you
17:10 know like all of them they they don't
17:11 spell really good learning with two
17:15 ends but it but it occasionally gets it
17:18 right and then these are some of the
17:19 ones that came out of Dolly 3 out of
17:21 chat
17:24 GPT
17:25 so how did you get Dolly 3 um retro Punk
17:29 one night after the live said don't go
17:31 to
17:33 bed and I guess he was on um he was on
17:38 open ai's Discord and one of the
17:41 developers of of Dolly 3 um said they
17:45 were looking for Alpha testers and put a
17:47 Google form up and so he said it was a
17:50 couple of hours old so I thought we
17:51 missed it but we uh we both filled out
17:55 the form and within 10 minutes had
17:58 access to Dolly 3 and then they shut the
18:00 form down
18:01 so I guess they got their Alpha
18:05 testers
18:07 um oh where are we let's see welcome to
18:10 Chad
18:11 add um ideogram and then Firefly so
18:14 listen if you work in a big company or
18:17 if you work for a company that cares
18:19 about osay
18:21 lawsuits um adobe's image models were
18:26 not trained on copywritten material so
18:29 from a from a intellectual property
18:32 standpoint it's a safe choice because
18:35 it's not been trained on copy materials
18:37 it's not as good so the the the quality
18:40 of the image generation within Firefly
18:43 and within Photoshop generative fill is
18:46 generally weak it's generally kind of
18:47 like Dolly 2 level it's a little better
18:50 than that I I I suppose it's between
18:52 dolly2 and you know stable diffusion
18:56 1.5 but it's getting better and it's
18:59 Adobe they're going to they're going to
19:00 do it right and uh the Photoshop
19:03 generator fill is quite good Pi must be
19:05 better on iOS
19:07 because won't Converse on web or SMS
19:11 it's so funny I got it I got it into
19:15 conversation mode once and I haven't
19:17 been able to do it since I'm wondering
19:18 if they killed it on the web pi. let me
19:22 see if I
19:23 can figure out how to put it in that
19:26 mode uh uh uh uh
19:29 let me close
19:32 that
19:36 what I don't want
19:38 onboarding here we
19:43 go oh yeah okay so in the upper right
19:47 hand corner on the website
19:54 is oh wait I'm not logged in I wonder if
19:58 that's part of it so here's these
20:02 voices
20:03 hey hey there great to meet you I'm P
20:06 your personal AI my goal is to be useful
20:09 friendly and fun ask me for advice for
20:11 answers or let's talk about whatever's
20:12 on your mind how's your day
20:15 going blah blah blah can I you can't
20:18 hear me
20:19 wait hello can you hear me now no hey
20:23 there great to meet you I'm P your
20:25 personal
20:26 AI hey there great to meet you I'm P
20:29 your personal AI my goal is to be useful
20:32 friendly and fun ask me for advice for
20:34 answers or let's talk about whatever's
20:35 on your mind how's your day going yeah
20:38 this used to this used to put it into
20:40 conversation mode yeah I guess it's only
20:41 on iOS all right well it's only on iOS
20:45 then okay um what else we got here um so
20:48 th so those are those things and then
20:49 prompt book that that thing at the
20:51 bottom there open art. a/ promptbook
20:53 that's really good uh it's like a 996
20:56 page document that will explain to you
20:59 all of the [ __ ] that these things know
21:01 these things like these image generation
21:03 tools similar to the um to the word
21:07 tools like chat GPT they just know a lot
21:09 they've been trained on a on a
21:11 ridiculous amount of
21:12 data um so so those are good um these
21:17 are the video ones so video or or other
21:20 sorts of things um
21:22 runwayml will do um text to video image
21:27 to video hey gen does synthesized Bots
21:31 like with synthesized voice and
21:33 animation they also have this remarkable
21:36 thing called video translate that will
21:38 translate your voice into I think it's a
21:41 dozen other languages and resync your
21:43 lips and everything really quite
21:46 remarkable excuse me um by the way we we
21:50 um we have a we have a couple of farma
21:52 clients that want to do something their
21:55 their basically legal agreements on
21:57 privacy are
21:59 non-existent they they are a very not
22:01 together company so they got to get
22:02 their [ __ ] together um cuz they're going
22:05 to get eaten alive by any any big
22:07 companies that want to use
22:11 them um blockade Labs uh is really
22:15 interesting text to 3D worlds and then
22:17 D- ID is also animated um
22:22 animated
22:24 uh like talking faces basically Sher
22:28 thanks you said Runway ml question mark
22:32 what's the question what is
22:35 it I can show you you want me to show
22:39 you how did you get dolly3 um just uh
22:43 retr Punk got us a
22:45 form I'll show it to you in a second
22:47 here it it's really quite remarkable one
22:50 of their one of their developers at
22:51 openai was looking for Alpha testers and
22:54 posted it to their Discord so you should
22:56 join the open AI Discord if you're
22:57 interested in getting [ __ ] early
23:01 um the one oh the one that has no legal
23:04 stance right now hey Jen hey Jen is the
23:06 one that they they they just they just
23:08 don't have their [ __ ] together I think
23:10 Runway ml seems better funded I honest
23:12 hey Jen this video translate thing I
23:14 don't think they have a clue how big a
23:17 product that is I don't think they have
23:19 a clue how big a product that is they
23:21 think that they little talking avatars
23:23 are the thing nope it's that it's that
23:26 tool so anyway they'll get it together
23:29 but like with all this stuff it's a bit
23:30 of a bit of a [ __ ] show right now uh
23:33 what was the links for prompts again it
23:35 is prompt.
23:37 chat down at the bottom
23:40 there blue G blue
23:47 lady um let's see what else um
23:53 so these are just some resources so the
23:55 first two things here futurepedia doio
23:58 and there's an AI for that.com are both
24:01 directories of AI tools and then Ethan
24:04 mullik David Shapiro and Paul rer are
24:07 all I I strongly recommend following
24:11 them if you're not geeky you're probably
24:13 not going to enjoy David Shapiro but I
24:15 think what he talks about is actually
24:19 really important he can go super geeky
24:21 but he also does these sort of really
24:24 high philosophical like what's the
24:27 future of work look like what what is
24:29 the fourth industrial re Revolution his
24:32 his um his name there for ir. David
24:36 Shapiro that stands for fourth
24:37 Industrial
24:39 Revolution Wicked smart guy if if you
24:42 are interested in
24:44 in the philosophy and and the
24:48 implications of this [ __ ] um the the
24:52 first two dudes Ethan mullik is just a
24:56 baller Wharton business school professor
24:58 just putting out killer content and
25:00 David Shapiro and then Paul ratzer the
25:03 marketing AI Institute they have a
25:04 weekly podcast that if you just want to
25:06 keep up on what's going on in this world
25:09 um they cover it really really well and
25:11 he's just he's just like a pragmatic
25:13 thinker like I I tend to resonate with
25:16 what he says because he's he's really
25:19 excited about this stuff he's really
25:20 bullish on it and then he's like Yep
25:22 this is a problem or yep that thing's a
25:24 piece of [ __ ] like he's he's got a
25:26 similar sort of you know he looks at the
25:29 reality of what's going on he's not too
25:31 far bought in or not too far you know
25:34 demonizing
25:35 it um this is uh this is a workshop that
25:40 started Wednesday there's two sessions
25:42 left I think you can buy individual
25:44 sessions the next session is uh about
25:47 applying AI to work but it kind of
25:49 Builds on the first session so I don't
25:51 know if it would be interesting but if
25:52 it is you can go check it out Ronda
25:54 voice.com I'm doing that on another
25:56 creators uh platform with with her
25:58 community um so that kicked off that was
26:01 that was exciting and then the final
26:02 thing um in this little opening banter
26:06 here the salon. this is a um a community
26:11 that I formed back in December we have
26:14 uh meetings every two weeks and we have
26:16 a Discord Community where we can hang
26:18 out and so a lot of the people that are
26:20 in here regularly have a little Channel
26:23 over there and uh it's a really good
26:25 group you can ask questions you can make
26:27 mid-journey images you can share stuff
26:29 and uh some good conversations going on
26:32 so what we're going to do here people
26:35 what we're going to
26:37 do what we're going to do what are we
26:39 going to do do do eat the food eat the
26:43 [Music]
26:44 food
26:49 [Music]
26:51 lucky is I'm going to go start looking
26:54 at questions we're going to start
26:55 answering questions oh I think people
26:57 wanted to look at Dolly 3 in chat GPT
27:01 will Apple buy pie to replace Siri oh I
27:03 [ __ ] hope so um are they going to buy
27:12 pie it's probably that's that's cuno
27:16 that's a that's an interesting I hadn't
27:18 really thought about
27:20 that but inflection the company behind
27:23 Pi they want to make the world's best
27:25 personal assistant right they want it to
27:27 take action and they just raised $1.3
27:29 billion and they're putting like 950
27:32 million of that into a 22,000 GPU server
27:36 Farm to build their own foundational
27:38 model we are the
27:40 Irregulars
27:42 um so they're they're a real serious
27:48 player and I'd be curious to know what
27:51 they have up their sleeves change shirts
27:53 all right yeah I I got to do that well I
27:56 can just put on my jacket hang on
28:00 I just I just took it off cuz I was
28:02 getting a little I was getting a little
28:04 warm cuz I was a little worked up I was
28:06 a little I got the adrenaline flowing in
28:09 here in the AI learning lab it's so
28:11 exciting in here you know what I mean
28:12 it's so exciting look at that
28:16 fancy fancy now you can read [ __ ] hang
28:19 on let me open my I'll open my window so
28:22 so I'm not quite so hot all right shirt
28:27 changed
28:31 um so Pie and
28:35 Apple if they've got some sneaky [ __ ] up
28:37 their sleeves like if if Pi is is really
28:40 much better than what we're seeing right
28:42 now cuz cuz PI right now is um fairly
28:46 useless if you want to do things like
28:48 work but it's really it's really really
28:50 good at the conversation stuff if they
28:54 if they make that autonomous in any way
28:56 where it's it's able to take action on
28:58 your
28:59 behalf um I think that's what Apple's
29:02 going to be doing anyway
29:04 so what was Apple's research budget this
29:08 year was $
29:09 22.4 billion and a lot of the increase
29:12 in their research budget was for
29:14 generative AI so they're probably
29:17 building the [ __ ] that inflections got
29:19 going but oh apple pie is too cute
29:24 lunatick good Lord someone else you know
29:27 I I think it was Nancy someone else in
29:29 here said you know what you should do
29:31 Kyle
29:33 is make a wait wait for
29:39 it make a yellow lab it's the AI
29:43 learning lab that was really good and
29:46 apple pie is in a similar
29:51 Place Apple should buy pie and they
29:54 should they should they should retire
29:56 Siri and they should they should name it
30:00 apple pie they really should it's really
30:02 good okay Irregulars get your ass over
30:06 to Dolly 3 to Bing or to uh wherever we
30:10 we got to start a whole new apple pie
30:12 meme apple pie and you can't spell it
30:15 with the E people you show a pie but but
30:19 you don't put the e in there it's apple
30:21 pie CU that's that's and then you put TM
30:26 lunatick
30:27 [Laughter]
30:30 Tim Cook get on this yeah exactly I
30:32 think I think there's something there I
30:33 think that's a really really good
30:37 call all
30:49 [Music]
30:53 right do you YouTube uh I do not YouTube
30:58 well I've got a YouTube channel so there
31:01 is a YouTube channel oh you can't see it
31:03 for
31:06 [ __ ] right up there at the top tippy top
31:09 it's it's AI learninglab dtt for Tik
31:13 Tock get it TT Tick Tock huh huh always
31:17 thinking like a steel
31:20 trap um and right now what's there are
31:24 archives of these lives um with
31:27 description and chapters and things like
31:29 that created by Claude so I'm taking
31:31 like the three-hour transcripts of these
31:33 things popping po popping them into
31:35 Claud and then I've got kind of Master
31:37 prompt that generates all that [ __ ]
31:39 that's been an interesting experience
31:41 getting that stuff to actually work CU
31:44 this this AI stuff demos amazingly and
31:48 then when you actually go to do
31:49 something you're like
31:52 H it's like 80%
31:56 good
31:59 oh man but why do you ask if I YouTube I
32:03 I don't know
32:05 I I barely [ __ ] Tik Tock I mean I do
32:08 it a lot but I'm not very good at it I'm
32:11 getting better I have moderators now
32:12 thank you irregular who who moderate
32:14 here that helps a lot um so I've got
32:18 I've got a little bit of
32:21 support but um I'm not a media operation
32:24 I'm I'm running a company right I'm a
32:26 CEO of a company
32:28 so you know um AI can interact with apps
32:32 or programs by telling them what they
32:34 are and where to go and and what to do
32:37 oh kenet are you you're asking right now
32:41 they cannot there there are projects
32:44 like Lang chain and like Auto GPT and
32:48 baby
32:49 AGI and I don't know there there's
32:51 another one called open interpreter
32:54 there's there's a a handful of projects
32:57 out there right now that can um take
33:01 action on your
33:02 behalf uh my late dog was named
33:05 Archimedes because he was born on March
33:07 14th and because I'm a huge nerd that's
33:12 awesome Archimedes screw is one of the
33:14 coolest inventions in all of
33:19 history water goes uphill so
33:26 cool
33:29 apple pie is up nice I like it apple pie
33:33 oh the one with the pie on it really
33:35 good I like it very
33:39 solid very
33:45 solid oh
33:48 um corporate digital you wanted me to
33:51 show you
33:54 um what's it called chat
33:58 jpt so we'll show you a couple of things
34:00 in Old chat
34:04 jpt flip over to NOP I'm going to try
34:09 one other thing settings and beta beta
34:11 features nothing new there okay all
34:14 right so here we are at chat GPT so I've
34:16 got chat GPT plus that's the 20 Buck a
34:19 month you you get that little yellow
34:21 button if you pay $20 but what you also
34:24 get is all this [ __ ] so look at this
34:26 look how long that list is getting
34:27 people how exciting is that so you got
34:29 default gp4 which is cool they have
34:33 brought back browse with Bing so let's
34:35 just do a quick browse with Bing and
34:36 I'll do
34:39 a what are some local sports activities
34:43 going on in Denver this
34:47 weekend and it says
34:56 browsing
34:59 finish browsing various
35:03 Sports there's snow shape winter Fitness
35:06 event at Red Rocks
35:08 Amphitheater I'm good thanks Broncos at
35:12 the Jets right so it's connected to the
35:14 internet now so all of that thing where
35:16 everyone's like it you know it it only
35:19 knows up to September 2022 with browse
35:21 with Bing it doesn't you could also sort
35:23 of hack this with with plugins before
35:25 but browse with Bing is back so that's
35:27 good
35:29 um you've also got Advanced Data
35:32 analysis and plugins but Dolly 3 this is
35:34 the one we all want to play with so this
35:36 this one's really cool so you go into
35:37 Dolly 3 mode and it's just like normal
35:40 chat
35:41 GPT um but I'll I'll do I'll do
35:44 something I did this this after I did I
35:47 did a Tik Tok video on this this is the
35:49 the latest video on my channel is is
35:50 some version of
35:53 this you're a world famous tattoo artist
35:56 and graphic
35:59 designer and I want you to help me
36:02 design some exciting new t-shirt lines
36:09 period I don't have any
36:12 ideas other than wanting to start making
36:17 Graphics period can you give me five
36:23 Concepts that would make great
36:26 t-shirts
36:30 I'll just leave it
36:31 there we'll see what it comes back
36:34 with of course here are five t-shirt
36:37 Design Concepts one celestial bodies and
36:47 mythology it's slow tonight I'm going to
36:50 stop it
36:52 too let's see those are okay but they're
36:56 a little cliche period I want them to be
36:58 super original and I also want you to
37:01 put
37:07 a text saying on each shirt as well as
37:11 describe the
37:15 style all right let's see what this does
37:18 this should be
37:19 better all right let's get creative
37:22 bonsai cityscape a miniature City
37:24 cityscape intricately growing out of a
37:26 bonsai
37:27 love that that's great so there's the
37:30 design style detailed ink illustration
37:34 text Nature's Metropolis right so so the
37:38 first thing the first thing that strikes
37:40 me
37:42 about Dolly 3 inside of chat
37:45 GPT is that you don't have to have an
37:51 idea you can you can lean on chat GPT to
37:55 help you brainstorm and then if and when
37:58 it comes up with something interesting
38:01 then you
38:02 can um make it into images so let's see
38:05 Soundwave stories Quantum playground my
38:08 God this is slow tonight it's slow
38:11 people it's slow you know what it is all
38:13 you freaks are like I want to go on chat
38:15 jpt and I want to make all the pretty
38:17 pictures let me make
38:19 pictures um by the way if you don't have
38:21 access to Dolly 3 within um chat gbt
38:26 some people do that didn't even fill out
38:27 the form so it's they're just kind of
38:29 rolling it out intermittently it seems
38:31 like but but
38:33 Bing and bing.com create are also dolly3
38:39 um it's it's not quite as good as as
38:41 this experience because you can do
38:42 multiple aspect ratios here and Bing
38:45 seems to still be um locked down to
38:47 square
38:48 images but it's doly 3 and it's actually
38:51 really good okay would you like visual
38:53 representations of these Concepts let's
38:55 just say Yes um I'm going to say
39:00 um pick the four that you think are the
39:03 best and give me an incredible t-shirt
39:08 design for
39:10 each now watch what this
39:14 does so it says so it pops it open like
39:17 a plugin and then it starts writing a
39:20 prompt photo design of a detailed ink
39:23 illustration featuring a miniature
39:24 cityscape intricately growing out of
39:27 bonsai tree skyscrapers and Tow houses
39:29 so it writes the first prompt now it's
39:32 writing the second prompt then it'll
39:34 write the third and the fourth this is
39:35 normally faster than this so I don't
39:37 know how long it's going to take to
39:38 render these images um but the little
39:40 spinny wheel there that's that's it off
39:43 generating the
39:46 images and so it goes and does it and
39:50 then you can tell it to make them
39:51 portrait landscape you know super skinny
39:55 um
39:57 you can say I like the contents but I
39:59 don't like the style or vice versa keep
40:01 the style but change the contents it
40:04 does text um it does text I when it gets
40:09 it right it's better than ideogram and I
40:12 would say it gets it right about as much
40:14 as ideogram maybe a little bit more I
40:17 think I think dolly3 is maybe
40:20 maybe 50 nah it's maybe 35% of the time
40:26 you'll get mostly what you want with
40:28 with Dolly 3 and maybe 15 or 20% of the
40:32 time with with ideogram and maybe you
40:35 know 1% of the time with mid
40:40 Journey um has it done it did we get all
40:43 four oh we did so Isn't that cool so you
40:46 get your little prompt that it did here
40:48 so and then there's your
40:51 image that's absolutely Nature's
40:55 Metropolis that's absolutely a t-shirt
40:58 like you could just upscale that in one
41:01 of those upscaling programs that would
41:02 be a kick-ass
41:04 t-shirt um that's kind of cool kind of
41:07 like the early 2000s combined with the
41:10 70s beyond the microscopic slightly
41:13 misspelled but pretty close close enough
41:15 that you can certainly Photoshop it
41:17 right that's
41:19 cute every shadow has a tail huh I have
41:24 no [ __ ] idea what that means but kind
41:26 of cool
41:28 um so let's say let's say
41:34 um I really like number one
41:41 period it's surreal and reminds me a bit
41:45 of MC
41:46 eer can you create four different
41:50 subject matters but keep the style of
41:54 that design
41:55 period
41:57 and
41:58 so if you've ever worked with an art
42:02 director or a
42:03 designer this is the
42:06 experience right I don't have to know
42:09 what are the commands I need to type
42:10 into mid journey I don't have to know--
42:13 AR 16 colon9 and if you if you do 16-9
42:18 or just 169 it it doesn't work and it
42:21 doesn't tell you why like like all
42:25 of you know the fact that this is these
42:27 things it's writing its own prompts
42:30 based on my creative
42:32 Direction This is where these tools go
42:35 right this is where copy generation goes
42:39 this is where picture generation goes
42:41 this is where video generation goes this
42:43 is where music generation goes this is
42:45 where animation 3D model worlds games
42:49 computer programs business
42:52 strategies project plans all of
42:55 it goes to this
42:59 place and
43:01 so what does that mean what are we
43:03 supposed to
43:05 learn you're going to need to use your
43:08 [ __ ] brain as opposed to oh those are
43:10 [ __ ] cool as opposed to I mean God
43:16 damn I
43:19 mean I remember being a kid I I remember
43:22 being at reoboth Beach Delaware and and
43:25 the T-shirt make were all you know up
43:28 and down the
43:29 boardwalk and I remember seeing some of
43:32 the art on those t-shirts and and just
43:35 marveling and I still Marvel at
43:37 illustrators that can do
43:40 this
43:45 and that specialty is is is being
43:51 democratized it's amazing look at that
43:54 [ __ ] thing that's like a combination
43:56 of Falling Waters Frank Lloyd Wright's
43:58 house and the M mcer you know infinite
44:01 staircase thing that's [ __ ]
44:04 remarkable fluid
44:08 foundations
44:10 amazing Nature's Tim keeper very
44:15 cool I don't know what that
44:18 is
44:22 huh amazing right so
44:28 what do we do with this what do we do
44:29 with this new found knowledge with these
44:33 infinite the the infinite ability
44:37 to personalize our lives and and express
44:41 ourselves you know I was thinking about
44:43 that tonight I my company Story vine
44:46 like it exists because I believe that
44:48 everyone's got a story to tell and
44:50 telling your story well is actually
44:51 really hard cuz it's kind of a lost art
44:55 and and you know social media has made
44:57 that worse worse in a lot of ways even
44:59 though people are putting themselves out
45:01 there they're not necessarily telling
45:02 their
45:03 stories and I was thinking about it a
45:07 lot tonight I'm like you know with all
45:09 this AI stuff just is telling your story
45:12 you know still the the kind of mission
45:15 that I want to be
45:17 on I mean I still believe passionately
45:20 about it but what hit me tonight is I
45:22 think that everyone wants to
45:28 ex be fully expressed like everyone
45:30 wants to feel confident
45:33 enough to express who they are and what
45:36 they
45:37 want and we're all going to have access
45:40 to these tools in in you know a
45:43 relatively short amount of time like
45:45 over the next 5 years these things are
45:47 going to get ubiquitous and they're
45:49 going to get incredibly powerful and
45:51 every single person no matter where they
45:54 are what they're interested in are going
45:56 to have tools to express themselves in
45:59 ways we can't even imagine right now
46:02 like I that to me feels
46:05 like I don't know that that that to me
46:08 feels like just the deepest most
46:12 profound
46:15 thing yeah I mean I I I think that's why
46:18 I'm here whoa my whole thing is the
46:21 importance of Storytelling and why I'm a
46:23 writer yeah I mean yeah Story vine
46:27 that's the the whole point of Story vine
46:29 is is you
46:31 know sharing story we we should make it
46:34 easy for people to share stories and
46:36 it's not easy to do it well and I think
46:39 that's shifting I think with these
46:42 tools it's not just our story it's it's
46:45 what we want to put in the world it's
46:46 what we want to invent it's what we want
46:52 to like not everyone is an amazing
46:55 writer not every Everyone's an amazing
46:57 illustrator not everyone's an amazing
47:01 coder but we're going to be everyone is
47:04 going to have access to all the tools
47:07 the democratization of
47:09 specialization that's what's
47:12 happening so if you're a specialist
47:15 that's why this feels so [ __ ]
47:16 existential because it
47:21 is if you've spent 30 years doing one
47:25 thing but boom
47:28 law creative writing illustrating this
47:31 particular style um this particular kind
47:35 of research this particular kind of data
47:37 analysis and you're the [ __ ] best at
47:41 it all of a sudden that
47:44 specialty is going to be
47:47 given to a lot more people who haven't
47:51 put in 30 years haven't put in 10 years
47:54 maybe haven't put in 10 days
47:57 and they're going to be able to do do it
47:59 to some percentage of what you can do it
48:02 now will it be as good as yours like up
48:04 to 100% probably not but if you're one
48:09 person that can do your thing and all of
48:11 a sudden there's a thousand people that
48:13 can do it 80% as well what does that do
48:15 to the
48:16 world I mean I know it [ __ ] with you as
48:19 an
48:20 individual but you're leveling up the
48:22 whole world to do all that stuff I had
48:25 chat GPT qu
48:26 questions to have my 80-year-old mom
48:29 answer to document her life story yeah
48:32 that one's [ __ ] brilliant and and you
48:35 know I don't know if Pi well so so we
48:38 can do this with with I don't have
48:40 access to the Talkie Chat GPT yet the
48:42 chat GPT app doesn't talk for me yet but
48:46 a combination of
48:49 um a combination
48:52 of um what's it called the system
48:54 instructions the the custom instructions
48:57 in chat gbt and the talkie mode you
49:00 could write a set of custom instructions
49:02 Jim where you put it into talkie mode
49:05 and it interviews your
49:07 mother and every one of those chats is
49:10 captured and and you could just say you
49:12 know I want you to ask her about you
49:15 know a single story in her life and once
49:18 a day for the next you know 60
49:21 days and so every day there's a
49:23 conversation with her and this chatbot
49:26 where the chatbot is just interviewing
49:27 her pulling this story
49:30 out people have been asking to see
49:32 custom instructions please okay custom
49:35 instructions are are really quite
49:38 powerful they're they're a little hard
49:40 to get your head around but everyone has
49:41 access to them so I'll even do it in GPT
49:44 3.5 let me flip over
49:47 here
49:49 um so we're in GPT 3.5 I'm doing this
49:52 partially because GPT 3.5 is fast and
49:55 and GPT 4 is really slow tonight okay so
50:00 when you're in chat GPT so if I'm just
50:02 in here and I say let's see what am I
50:04 going to ask about um I got one what is
50:08 the average speed of an unladen
50:12 swallow gen xers will get that you
50:15 younger folks will
50:20 not all right so it just answers you all
50:22 right okay what you can do with with
50:26 custom instructions is you can tell chat
50:28 GPT to act a certain way to act as a
50:31 certain kind of person and respond in a
50:33 certain kind of way when people talk to
50:36 you so you go to your settings or or you
50:40 go to your name you click on your name
50:41 and that's where settings are but then
50:43 there's this thing called custom
50:44 instructions so we go to custom
50:46 instructions and then right now I've got
50:49 them turned off but I'm going to turn
50:50 them on and I'll read them to you
50:51 because I I did these I don't know last
50:54 night or the night before and I I just
50:56 like what they do because they do
50:57 something visual
50:59 so I'm a creative director that has a
51:02 passion for structure and Clarity I can
51:05 synthesize big Concepts into very simple
51:07 metaphors that people can understand
51:09 right so so that's who chat GPT is going
51:12 to start acting like now you can get
51:14 ridiculous with this I I I had a joke
51:16 one in here before about um You're A
51:18 pompous ass that uses words so big that
51:21 even you don't know what they understand
51:23 right and I had it respond in these
51:25 really comp kind of way so so you can
51:27 sort of set it up as a personality and
51:29 it and it really does work and then you
51:32 can say um how do you want chat GPT to
51:35 respond so what I wrote here was when
51:37 someone asks a question respond
51:39 following these rules mirror back what
51:41 you think you heard answer with a non
51:45 obvious innovate Innovative idea to help
51:48 clarify uh to help clearly answer the
51:50 question when you answer include a
51:53 metaphor that is Visual and will make
51:55 your IDE idea easy to remember follow
51:58 your idea with five to seven or five
52:00 bullet points of 7 to 10 words to
52:03 support your idea and finally create a
52:05 table formatted in markdown I'll explain
52:08 that in a second that compares your idea
52:11 with the traditional way of thinking
52:13 okay so every time we ask this thing
52:16 something it should give us some kind of
52:18 cool answer and some sort of innovative
52:20 thing and a metaphor and a table
52:22 comparing [ __ ] okay that's what it
52:24 should do so we'll save this we're going
52:25 to create a new
52:27 chat and then let me go copy that exact
52:30 prompt what is the average speed watch
52:32 is the average speed of an unladen
52:35 swallow African or
52:37 [Applause]
52:39 european and so I'm going to pop this in
52:44 here there's your bullet points there's
52:47 your table so so so so imagine this I
52:51 mean you could put in those custom
52:53 instructions here's you know I'm an
52:55 executive at at this company my company
52:58 does X Y and Z um most of what I'm going
53:02 to use you for is to write responses to
53:04 emails so so when someone gives you an
53:07 email you know respond using this
53:10 example and you could paste an example
53:12 of your writing in there and then every
53:14 time you use chat GPT to process an
53:17 email it will sound like you now one of
53:20 the drags about custom instructions is
53:22 there's only one right so if if you if
53:26 you want to save multiple of them like
53:29 dozens of them like I think would make
53:31 sense you could have silly ones and
53:33 serious ones and work ones and play ones
53:35 you're going to have to do that just
53:37 like your prompt Library you're going to
53:38 have to do that outside of chat
53:41 gbt um that's because all these tools
53:43 are janky pieces of [ __ ] as we all know
53:46 and what we talk about in here um but
53:49 isn't that wild so let's see mirror
53:51 you're asking about the average speed of
53:53 an unladen swallow Innovative idea with
53:55 a metaphor calculating the average speed
53:57 of an unladen swallow is like trying to
53:59 measure the speed of a leaf in the wind
54:01 instead of pursuing an impossible task
54:03 let's focus on finding a practical
54:06 solution supporting points unladen
54:09 swallows have varying speeds measuring
54:10 their speed accurately is challenging we
54:13 can explore related bird migration
54:14 patterns consider analyzing the birds
54:17 right and then here's the table
54:20 feasibility Innovative approach focus on
54:22 practical Solutions traditional approach
54:24 attempting to measure the unladen
54:26 follows speed they're not necessarily
54:28 good this is GPT
54:30 3.5 but but it it followed our
54:32 instructions so that's custom
54:34 instructions pretty crazy and it it'll
54:36 do that in all in all uh Whata callit so
54:40 let's go to G let's go to dolli 3
54:43 mode for in GPT
54:46 4 and I'll say
54:54 um I want want to come up with
54:56 non-traditional graphics for
54:59 t-shirts wait what's going
55:01 on I want to come up with
55:03 non-traditional graphics for
55:08 t-shirts and don't know where to
55:13 start what should I
55:20 do what I heard you're seeking
55:23 innovative ideas for non-traditional
55:25 designs considering evolving art
55:29 t-shirts consider evolving art T-shirts
55:33 instead of static designs imagine
55:35 t-shirts that tell a story through a
55:37 sequence of Graphics each shirt
55:40 represents a chapter or phase and when
55:42 lined up they narrate an evolving tale
55:45 that's actually pretty [ __ ]
55:47 cool think of it like a comic strip each
55:50 t-shirt is a panel when Warden
55:52 sequentially they unfold a captivating
55:55 story line live it's actually a really
55:56 good idea and so you know what we're
55:58 going to do now don't
56:01 you all right here's our
56:06 table evolving t-shirts traditional
56:09 t-shirt
56:13 graphics design philosophy here you go
56:16 lunatick story driven
56:19 evolving story driven t-shirts that's
56:22 actually really cool God look how slow
56:23 it is tonight it's just it's just you
56:26 know what's funny this reminds me of in
56:28 the in the early days of when I was
56:30 first learning personal computers we had
56:33 dialup modems and my friend Scott Bart
56:35 had a 300 bod modem and it used to be
56:38 this slow like this is this is what it
56:40 looked like when you called another
56:41 computer it would type about this fast
56:43 and then when you got a 1200 bot it
56:45 would go fast enough you you couldn't
56:46 quite keep up with
56:48 it all right so now I'm going to
56:50 say I love that idea please come up with
56:54 five brief
56:57 concepts for four panel
57:02 Story
57:11 series oh I don't want the whole thing
57:14 all right no let's stop let's screw
57:19 that all right just write a story and
57:22 show
57:24 me
57:25 the four t-shirt
57:27 designs for a series of
57:32 four now this should fire off Dolly
57:38 3 yep yay it worked I've had trouble
57:42 moving tables or charts from chat GPT to
57:44 word okay I'll show you how to do that
57:46 cam katkin cam katkin trying to figure
57:49 out how to get the old uh the old uh
57:52 what you call it tables tables are pain
57:55 in the ass there's there's a a
57:58 non-obvious way to do it and I will
58:00 teach you
58:02 that look I'm all storyvine branded got
58:05 the dirty storyvine hat The Story vine
58:07 jacket
58:09 huh all right let's see if this makes us
58:11 pretty pitches pitches pitches pitches
58:14 all right let's let's read what it
58:16 said photo t-shirt design showing the
58:19 first stage of a Seedling sprouting from
58:22 the earth with Morning Dew surrounding
58:24 it
58:26 showcasing a young plant reaching toward
58:29 the sun symbolizing growth oh there we
58:32 go look well that's actually quite
58:35 that's quite remarkable I kind of missed
58:38 it on that last one but actually the the
58:40 last two are not quite right but so
58:43 there's the little baby now it's growing
58:46 now it's flourishing now it's
58:48 dying I mean the trick would be do it
58:50 all in the same you know make it the
58:52 same plant but
58:53 still it came up up with that you know
58:56 it's it's funny like like the thing
58:58 about people like w but chbt can't be
59:00 creative well it just [ __ ] was I
59:03 didn't have an idea to come up with a a
59:06 storytelling series of
59:09 t-shirts I said I don't want a boring
59:11 t-shirt company and it said here's a
59:14 concept and now with this cat GPT Dolly
59:21 integration what was that 2 minutes of
59:24 in ual
59:28 input and what is this at
59:33 least eight hours of traditional like
59:37 you know if you wanted to create these
59:38 images it would be more than eight hours
59:40 if you wanted to go find images like
59:42 this on stock you know that's an8 hour
59:45 job finding it photoshopping it
59:47 licensing it buying yet another
59:50 subscription to a stock photo place
59:54 right all right let's show let's show
59:56 show cam katkin what she needs to know
59:59 about tables so
1:00:03 um so the the thing about tables is you
1:00:07 basically just need to copy them
1:00:10 within within
1:00:13 um what's it called within here what is
1:00:17 this within chat gbt you just kind of
1:00:19 highlight it like this and then there's
1:00:21 two ways that you can paste it um so let
1:00:24 me copy
1:00:25 and let me go to
1:00:32 numbers so I'm going to pop open a blank
1:00:35 spreadsheet so if I just go in the first
1:00:38 uh column I think I did it down here um
1:00:43 wait what's my table here yeah so we'll
1:00:48 go here and I'm going to hit
1:00:50 paste so it pastes it all in one
1:00:54 cell but if I oh wait hang
1:00:59 on oh that didn't do it too bad well
1:01:01 anyway it it worked there but you can
1:01:03 also do a thing where you hit command
1:01:06 option shift
1:01:07 paste and then it it it
1:01:11 um inherits the the um stylesheet of the
1:01:16 document you're pasting
1:01:17 into so if you just hit paste it kind of
1:01:20 does whatever the [ __ ] it wants to do
1:01:23 but if you hit command shift shift paste
1:01:26 command option shift paste on a Mac
1:01:29 anyway then it then it matches the
1:01:32 design the the uh whatever the style is
1:01:34 that you're pasting into so hopefully
1:01:36 that
1:01:38 helps the other the other thing that I
1:01:40 do a lot on chat GPT is um I'll just
1:01:45 [ __ ] screenshot it so like if I want
1:01:48 that table somewhere like if I don't
1:01:50 need to edit it I'll just do command
1:01:52 shift 5 and you know throw a little
1:01:55 little you know throw a little lasso
1:01:59 around
1:02:00 it and go
1:02:02 bang and then I'll go to do I have
1:02:05 anything I can paste into yeah I can do
1:02:08 this oops Yeah and I
1:02:10 do command let's see command new and
1:02:15 there's my image
1:02:18 right and if I zoom in on it it's just
1:02:22 you know text it's just pixels
1:02:27 so those are those are the two ways that
1:02:29 I get [ __ ] out of there but that's is
1:02:32 it's a pain in the ass to to get stuff
1:02:34 out of there so hopefully that
1:02:37 helps the smell of t-shirt press brings
1:02:40 me back to M days exactly right right
1:02:43 Sherry
1:02:44 Banks when I was when I've been doing
1:02:46 these t-shirts I'm I'm uh I'm Consulting
1:02:49 with someone right now who's who's
1:02:50 putting together some print on demand
1:02:52 stuff and um it's it's it's totally
1:02:56 bringing me back to all those smells of
1:02:57 those those t-shirt shops in the 70s and
1:03:00 80s man that was like when they didn't
1:03:02 have any regulations about chemicals
1:03:04 that that was some nasty [ __ ]
1:03:06 chemicals they would melt into your
1:03:12 t-shirts all right within your prompt
1:03:15 you mentioned an artist as reference
1:03:17 language that is
1:03:19 informed about art
1:03:23 styles language that is in informed
1:03:26 about art Styles um I don't quite know
1:03:29 what you're saying there Barbara b60 but
1:03:31 yeah so so there's a couple of things
1:03:35 um a number of my my artist friends um
1:03:40 are trying to figure out where their
1:03:41 boundaries of Ethics
1:03:44 are and one of my one of my good buddies
1:03:46 from the AI Salon a guy named Kenny
1:03:48 Freedman really brilliant creative
1:03:50 director if you need a good creative
1:03:51 director let me know I'll hook you up um
1:03:55 and he does he does really amazing work
1:03:58 and really amazing um AI art and he's
1:04:03 kind of drawn a line in the sand for
1:04:05 himself where he won't use the name of
1:04:08 other
1:04:10 artists he what he tries to do is
1:04:13 describe the prompt in such a way so
1:04:14 that he can get the look and feel that
1:04:16 he wants if it's of a particular artist
1:04:18 but he doesn't use the artist's name um
1:04:21 I think it's an
1:04:22 interesting it's an interesting way to
1:04:24 deal with the with the Dilemma of hey
1:04:27 these things were trained on lots of
1:04:28 people's work but if you're using
1:04:31 someone like mcer I know mcer has an
1:04:34 estate but he also has you know he he's
1:04:37 got so much work out in the world that
1:04:39 lots of his images are in these models
1:04:42 so it's a shortcut to getting the kind
1:04:44 of work you want
1:04:47 but even with that the way that chat GPT
1:04:52 is working now where it's writing its
1:04:54 own own prompts it may call for
1:04:56 someone's I I think they've got this the
1:04:59 safety guard rails on chat GPT right now
1:05:02 or such that I don't think it uses
1:05:04 specific artist names I think I think
1:05:06 they've done that on purpose but that's
1:05:07 a it's a really interesting point and
1:05:09 it's something that you can choose to do
1:05:10 or not not
1:05:16 do that's the AI learning
1:05:20 lab blah blah blah blah blah blah
1:05:26 can you do prompt variations on an image
1:05:29 you like yes you can in fact let's go do
1:05:33 that oh
1:05:37 like yeah I don't know how to do that
1:05:39 let's let's go let's go learn that one
1:05:44 [Music]
1:05:46 together so if I click here okay so I
1:05:49 can I can go over here and I can get the
1:05:53 prompt
1:05:59 so here's what we're going to do so so
1:06:04 since so it did these in different
1:06:06 styles for one thing and then this
1:06:11 started as like a plant went to a plant
1:06:13 then went to just this bizarre flower
1:06:15 that seems like it doesn't resemble that
1:06:19 one so I think what would what what the
1:06:22 the way I would modify this PR as I
1:06:25 would say let's let's just go with uh
1:06:27 let's say we like the style of number
1:06:30 four but we want all four of these to be
1:06:34 the same kind of plant so
1:06:38 so while I like these images and they do
1:06:41 tell a story
1:06:45 comma I want the four of them to be the
1:06:47 same kind of plant period I really like
1:06:51 the plant and style of number four can
1:06:54 you
1:06:56 redo the four
1:06:59 Images so that they're all the same
1:07:01 plant in the same style at different
1:07:10 [Music]
1:07:11 stages so so I think you could is it
1:07:14 doing it I think you could go copy and
1:07:17 paste The Prompt out of that you know
1:07:19 out of the interface and just paste it
1:07:22 into your prompt and modify it but I
1:07:24 think you could also just give it art
1:07:25 Direction like I just did let's let's
1:07:28 see let's see how well it follows this
1:07:29 art Direction I mean the thing about
1:07:32 this is as remarkable this as this is
1:07:34 this is still Alpha like they're still
1:07:36 figuring out like it's gotten it's
1:07:38 gotten a bit more predictable since I've
1:07:40 been using it I've only had this for two
1:07:42 days so it's it's evolving as I'm using
1:07:45 it um is it still going yeah it's still
1:07:50 going
1:07:52 uh so you are drawing with words but it
1:07:56 goes nowhere without intellect that's
1:07:59 correct Barbara b60 that's that's what
1:08:01 I'm saying um the most
1:08:05 important ingredient in all of these
1:08:08 things is is human intention what are
1:08:11 you trying to do how are you trying to
1:08:14 express
1:08:15 yourself did it do it oh it did it sort
1:08:20 of well no it didn't did
1:08:23 it
1:08:25 it kept everything the way it
1:08:29 was okay I'm going to I'm going to tell
1:08:34 it either I didn't explain myself well
1:08:37 or you didn't follow my instructions
1:08:39 period but I like the
1:08:42 dandelion kind of flower of the last
1:08:45 t-shirt
1:08:47 period and each of your four t-shirts
1:08:49 have a completely different
1:08:51 plant can you have all four be the same
1:08:55 dandelion plant but at different stages
1:08:57 of life and please follow the style
1:09:02 of the dandelion
1:09:05 shirt right now the other shirts are
1:09:08 completely different
1:09:13 styles so so again if you're working
1:09:15 with an art
1:09:17 director sometimes you give them what
1:09:20 you think are really clear instructions
1:09:21 and they come back and completely
1:09:23 ignored you this is exactly like
1:09:27 life but yeah Barbara critical thinking
1:09:31 and human intention are everything and
1:09:34 and I think go up in value moving
1:09:38 forward the Tactical execution is just
1:09:41 going to happen and what's going to
1:09:43 set talented people apart from
1:09:46 non-talented people is their
1:09:49 capacity to
1:09:52 imagine and ask for or what these things
1:09:55 can do in a way that serves their
1:09:58 purpose or serves other people or right
1:10:00 creates a better spreadsheet creates a
1:10:02 better formula creates a better computer
1:10:04 application creates a better image
1:10:07 creates a better
1:10:09 song how do they do that and then once
1:10:11 it creates it then what's their role
1:10:13 their role might be to curate and refine
1:10:16 and maybe they're using AI for a lot of
1:10:18 that but maybe they're using traditional
1:10:20 tools for a lot of
1:10:22 that so we're ultimately
1:10:24 we as humans are ultimately the ones
1:10:27 that get to say this goes out to the
1:10:29 world that
1:10:31 doesn't that's the art the art is not
1:10:36 in the pushing of the
1:10:39 pixel CU these tools are going to do
1:10:41 that the art is in all right which of
1:10:44 these acceptable images is the one I
1:10:47 like well I like this
1:10:50 one you know I really don't like that
1:10:52 one someone else probably loves that one
1:10:55 you know this looks somewhere between
1:10:56 photographic and not photographic to me
1:10:58 I'm like make up your mind but it's cool
1:11:01 someone else may love that one that's
1:11:03 the human
1:11:05 piece would it be good to put the prompt
1:11:08 you used in the alt image tag I'm pretty
1:11:11 sure that
1:11:13 these
1:11:15 images keep that
1:11:18 um metadata let me save image
1:11:23 as
1:11:25 oh no they
1:11:27 don't they've got their hashtag they've
1:11:30 got they've got a a unique hash of that
1:11:34 original image but they don't have like
1:11:36 stable diffusion and mid Journey Keep
1:11:39 some of the metadata in the image itself
1:11:42 um yeah that it it would be a good idea
1:11:45 to do
1:11:50 that so it didn't quite understand the
1:11:54 exercise but it got these two right well
1:11:56 no it got this right like that's the
1:11:59 right number
1:12:02 three but this is not the baby dandelion
1:12:05 and this is these are both old dandelion
1:12:07 right so so it's it just got this wrong
1:12:10 so you know what I would probably do at
1:12:12 this point is okay it gave me the
1:12:14 concept let me just go Um single single
1:12:17 design by single design right have it
1:12:20 give me the broad Concepts and like
1:12:22 there's a point of diminishing return
1:12:23 where you just just can't if it if it's
1:12:25 just not getting it you could spend the
1:12:27 next hour just trying to prompt it
1:12:29 that's what I I I swear to God I
1:12:32 burned 20 hours 30 hours trying to get
1:12:36 clawed to to make my Youtube
1:12:38 descriptions for the for this channel
1:12:40 from from the big long transcripts it's
1:12:43 it's a
1:12:44 mess um sometimes you just have to you
1:12:47 know fake
1:12:52 it chat GPT just tell it to leave off
1:12:56 the numbers and then make a
1:12:58 table yeah are you pasting the table
1:13:01 into Excel I was pasting it
1:13:02 [Music]
1:13:04 into um
1:13:09 sheets that's exactly the same problem I
1:13:12 had with Turtle Army images yep yeah no
1:13:14 wonder you are we are special sniffing
1:13:19 Pence uh I screenshot too but then the
1:13:22 images need Alt imag tags yeah I know
1:13:25 listen I so I mean this the deal cam
1:13:29 katkin is like you know we go back to
1:13:31 our our our little inside inside joke
1:13:34 here of you know all these tools are
1:13:37 janky pieces of [ __ ] they're all janky
1:13:39 pieces of [ __ ] right now they'll get
1:13:41 better they they'll have export options
1:13:44 and oh you know what you can also do cam
1:13:47 is
1:13:49 um if you use code interpreter if you if
1:13:52 you use advanced data
1:13:55 analysis it can generate Excel files you
1:13:59 just tell it to generate a downloadable
1:14:01 Excel file and it will so pop your data
1:14:04 in there or have it generate a
1:14:06 table and then say you know generate
1:14:09 this table as an Excel file and it will
1:14:11 generate it and you just click on the
1:14:13 link and it'll download it to your
1:14:14 computer and it's an Excel file I did
1:14:17 that I I created like
1:14:19 a seven worksheet Excel file out of data
1:14:24 from a single CSV in code interpreter
1:14:26 it's pretty it's pretty slick also Excel
1:14:29 can import data from an image oh I
1:14:32 didn't know that is that true Rich 14370
1:14:35 that's pretty
1:14:42 cool I was saying that all aart will
1:14:45 only be as good as your
1:14:47 intellect yeah
1:14:52 although
1:14:56 that's an interesting
1:14:57 one I don't think we know yet
1:15:00 Barbara the the speed with which these
1:15:03 tools are are
1:15:05 improving
1:15:07 um I think they're going to improve
1:15:10 beyond our capabilities pretty quickly
1:15:13 like
1:15:15 within 5
1:15:18 years um and I think I think on things
1:15:21 subjective things like art and
1:15:22 creativity and things like
1:15:24 that I think there's going to be systems
1:15:27 where you can just give it a topic and
1:15:30 it will just do everything it'll create
1:15:32 a write a creative brief it'll write
1:15:35 marketing strategies it'll write copy
1:15:38 it'll generate images it'll generate
1:15:41 audio branding it'll generate
1:15:43 logos right I think we're going to see
1:15:45 systems that just just do that M smells
1:15:50 should be series of scratch and sniffs
1:15:52 that's a great idea
1:15:54 scratch and sniffs were the best except
1:15:56 there was like wasn't it wasn't Pizza
1:15:59 the one that always just made you almost
1:16:01 vomit smell a pizza
1:16:05 [Laughter]
1:16:14 one all right what time is it I think
1:16:17 that's getting late for me a little more
1:16:18 few more
1:16:20 questions a few more questions is this
1:16:24 helping tonight is this good is
1:16:25 everything good is everybody good
1:16:27 tonight tell me what's happening how
1:16:28 y'all doing how's everybody doing you
1:16:31 janky pieces of [ __ ] today was a weird
1:16:34 day for me it was a it was a weird it's
1:16:36 been it's been a weird couple of
1:16:38 days how's everybody doing let me scroll
1:16:40 to the bottom and see how people are
1:16:41 doing what are you what are y'all doing
1:16:43 yeah Pizza one was gross a thank you for
1:16:46 the dog butt that loves me a good dog
1:16:48 butt dog buts is
1:16:50 cute um M smells
1:16:57 um I had a call with CPS today not a
1:17:00 great
1:17:00 day but I think it was a false report oh
1:17:04 that
1:17:05 sucks that sucks
1:17:08 bad sorry about that felonius um who
1:17:11 will be our
1:17:13 heroes well I think we become Our Heroes
1:17:16 I just don't know that it's going to
1:17:17 require our intellect to do good
1:17:20 work I think what what it's going to
1:17:22 require from us
1:17:26 is why are we doing the
1:17:29 work and who are we doing the work for
1:17:32 and what do we want the work to
1:17:34 do so I guess that's intellect I I I I
1:17:38 think that's right maybe maybe maybe the
1:17:40 word rather than intellect is
1:17:44 intentionality that what these what
1:17:46 these tools are going to require from us
1:17:49 is our
1:17:51 intentions because I don't think I don't
1:17:54 think they will have intentions they
1:17:55 might have imperatives where we say we
1:17:57 want you to get stronger or better or
1:18:00 faster but it's our intentions that that
1:18:03 are steering these things and I think
1:18:05 it's going to be our intentions for
1:18:06 quite a while madcap you are the [ __ ]
1:18:09 Kyle thank you so much for sharing with
1:18:12 us every night oh thank you very much
1:18:13 madcap I appreciate that that's really
1:18:16 nice that's really really nice I had the
1:18:19 best day ever so much kindness love
1:18:22 gratitude learning thanks all oh that's
1:18:25 really sweet hair growth geek
1:18:27 congratulations those are rare days
1:18:29 that's that's really that's really
1:18:32 amazing good for you see I like that
1:18:35 that's the the other thing
1:18:39 here like let's not be [ __ ] to each
1:18:42 other as a as a wild
1:18:46 concept let's just be nice recognize
1:18:49 people are going through [ __ ] sometimes
1:18:52 if people act a certain way it's because
1:18:55 of fear it's because they had a shitty
1:18:57 day they had a shitty
1:19:00 life they just stubbed their toe we
1:19:02 never know you
1:19:04 know did you talk about Google AI phones
1:19:08 I didn't um I I didn't pay super close
1:19:11 attention to the thing but the new
1:19:12 Google 8 Pro has all sorts of AI fix it
1:19:18 things like the coolest one that I saw
1:19:21 is you can take a series of group fot
1:19:23 photos so like like's say let's say you
1:19:25 got four people in a group photo um it
1:19:28 will isolate all the faces and let you
1:19:31 choose the best face of each of the
1:19:33 people and then recompose the image with
1:19:36 the best
1:19:37 faces of for each person so you don't
1:19:40 have someone with their eyes closed I
1:19:42 thought that was pretty
1:19:44 brilliant um you can grab objects and
1:19:48 remove them or move them you can like if
1:19:50 you've got like a photo of a tent and a
1:19:52 photo of people and there's too much
1:19:54 space between them you can just you can
1:19:56 just grab the tent and slide it over um
1:19:59 it's going to get [ __ ] trippy I saw a
1:20:03 client use sign language to communicate
1:20:06 can't wait for AI to help William that's
1:20:08 going to be so here I mean chat GPT is
1:20:12 about to get Vision right and then once
1:20:15 it's got Vision right now it's vision of
1:20:17 an image but pretty soon it's going to
1:20:19 be vision of a video feed and pretty
1:20:22 soon that's going to be in your chat GPT
1:20:25 app and so you're communicating with a
1:20:28 deaf person
1:20:30 and they're going to sign it's going to
1:20:33 translate for you you're going to talk
1:20:34 back to it and then I suppose it'll make
1:20:37 words on the screen you know that they
1:20:40 can
1:20:41 see it's going to be
1:20:46 amazing crazy or the worst fakes yeah
1:20:50 it's going to be that too 2024 is is
1:20:53 going to be insane for The Fakes The
1:20:54 Fakes are already get they're already
1:20:56 starting right have you
1:20:59 noticed there was like the porn one and
1:21:02 then there was Tom Hanks they Tom Hanks
1:21:05 is making a commercial for a bank and he
1:21:07 had to put out a statement going that's
1:21:09 not me I didn't put out a thing and then
1:21:12 we're going to have politicians going
1:21:14 that photo of me French kissing that dog
1:21:17 behind the alley wasn't really me French
1:21:19 kissing a dog behind an alley but it was
1:21:22 but now because of AI I can deny it I
1:21:25 it's just G to be [ __ ] crazy it gonna
1:21:28 be crazy people GNA
1:21:29 be czy you own those children the state
1:21:33 has no authority over
1:21:36 that oh man
1:21:41 alive I took a picture of a menu today
1:21:43 and asked chat GPT to identify the
1:21:45 healthy options it did a pretty good job
1:21:48 that's a great idea cool I can mix all
1:21:51 the faces of my ex-husbands
1:21:56 no I only have one ex
1:21:58 ex-husband yeah you can just grab his
1:22:00 face and drag it off
1:22:04 screen actually yeah this will the the
1:22:07 Google pixel will be the the new version
1:22:09 of sort of ripping someone out of the
1:22:10 photo as you just grab them slide them
1:22:13 off screen that's
1:22:17 great Slide the new boyfriend
1:22:19 [Music]
1:22:21 on that's great
1:22:23 um how are you guys uploading pictures
1:22:25 to chat GPT I'm not right now I don't
1:22:28 have access to that tool um the the
1:22:31 images that I showed there are being
1:22:33 generated within chat GPT with the do
1:22:36 function but not everyone has that yet
1:22:39 just be just be patient mcgarrett it it
1:22:42 what I'm guessing is that so November
1:22:45 6th is open AI big developer so so the
1:22:50 first part of November mark your
1:22:52 calendar ERS kids if you want to geek
1:22:55 out
1:22:57 um November
1:22:59 1st Microsoft is launching co-pilot so
1:23:03 chat GPT within all of Office 365 is
1:23:06 launching November 1st November 6th is
1:23:10 the open
1:23:11 AI um developer conference so I think
1:23:15 open AI is probably trying to roll these
1:23:17 features out in anticipation of that
1:23:20 November 6th deadline so I would assume
1:23:23 that most people are going to get most
1:23:26 Plus members will get most of this
1:23:28 functionality over the course of October
1:23:31 who knows what they're going to announce
1:23:33 in November and then sometime in that
1:23:36 time I'm guessing sometime right before
1:23:39 November 1st or right in the November
1:23:41 1st area Google's going to release
1:23:44 Gemini um and Gemini is supposed to be
1:23:47 better than chat gp4 so so all of these
1:23:50 guys are kind of vying for attention
1:23:52 right now
1:23:54 um so the next month just just be
1:23:57 patient and we're going to get some cool
1:23:58 [ __ ] in our in our
1:24:01 toys so Bing has a tool I think I don't
1:24:05 know that what that's in reference to
1:24:08 not enough what did that
1:24:17 say not enough blood and Sacrifice by
1:24:20 Kyle to the ultimate altar
1:24:25 listen Sam's going to [ __ ] up here at
1:24:27 some point I mean listen the fact that
1:24:30 they that they pulled browse with Bing
1:24:33 for four months or three months or
1:24:34 whatever it was that was a pretty decent
1:24:36 [ __ ]
1:24:37 up
1:24:40 um the fact that um Google cut a $100
1:24:44 million deal with the New York Times
1:24:46 while open AI was negotiating with them
1:24:49 and I heard that negotiation wasn't
1:24:51 going well so Sam Alman he he ain't
1:24:54 walking on
1:24:56 water and if listen pate if if uh if
1:24:59 Gemini is even as good as gp4 that's
1:25:05 going to be a major [ __ ] headache for
1:25:08 uh for open
1:25:09 AI I mean the thing about open AI is
1:25:12 they've got Microsoft incorporating it
1:25:14 into everything so they're not I don't
1:25:15 think they're going
1:25:17 anywhere
1:25:20 um but we'll see I mean I just
1:25:23 fascinating times I mean I I listen I'm
1:25:26 impressed with with what Alman has done
1:25:28 at this point I'm impressed with his
1:25:32 self-awareness and his awareness
1:25:36 of his company's responsibility for this
1:25:40 stuff whether or not he can cash that
1:25:42 check is a totally different thing you
1:25:45 know and is that all PR spin I don't
1:25:47 sense that I sense that he's authentic
1:25:50 but we'll see met is working with
1:25:52 Microsoft now yeah the the
1:25:55 uh one of the things they just released
1:25:58 you could play with on Microsoft or vice
1:26:00 versa so so yeah it's that's going to be
1:26:04 interesting Microsoft's kind of hedging
1:26:06 their bets on open AI so I don't know
1:26:08 we'll
1:26:09 see we will say good deal I've seen and
1:26:13 heard some horror stories I don't know
1:26:16 on what but oh I think you guys are
1:26:19 talking to each other I hear Alman and
1:26:22 for a second confusement him with
1:26:24 bankman oh Jesus Sam bankman freed
1:26:27 that's terrifying I don't get that sense
1:26:29 from Alman oddly enough Sam bank mfried
1:26:33 I couldn't listen to the guy talk it's
1:26:34 like the the CEO of mid Journey skes me
1:26:37 out skes me the [ __ ] out he reminds me
1:26:39 of bankman Sam mman doesn't I mean Sam
1:26:42 malman definitely seems like a San
1:26:44 Francisco a product of San Francisco's
1:26:47 you know Silicon Valley scene for
1:26:51 sure
1:26:54 you know unapologetically over
1:26:56 intelligent it's
1:26:58 like but but he's but I don't know he
1:27:02 seems way more self- Weare than most of
1:27:03 those other numb nuts from that part of
1:27:06 the
1:27:08 country um I was hooked on Claude for a
1:27:11 while but ever since chat GPT brought
1:27:12 the code interpreter analysis out for
1:27:14 everyone I've been stuck there yeah it's
1:27:16 it's funny they're kind of different
1:27:17 things for me like Claude for me is oh I
1:27:21 got a big long document or just like
1:27:23 some big long thing I don't want to read
1:27:25 I just immediately go to Claude cuz I
1:27:26 know it won't choke on it and then if
1:27:29 it's actual data analysis I go to
1:27:31 Advanced code interpreter and then if
1:27:33 that fails or's too much of a pain in
1:27:35 the ass then I'll go to Claude to see if
1:27:38 it can do something different and I've
1:27:39 had Claude succeed there and I've had it
1:27:42 fail there so it's you know again all
1:27:45 the to all the tools right now it's all
1:27:47 about experimenting so all right
1:27:50 everybody listen thank you very much
1:27:52 much for hanging out with me I know it
1:27:53 was a bit of a short session tonight but
1:27:55 but I got to go let me do a little
1:27:57 Romper Room Roundup here we got who we
1:28:00 got we got cuno we got rich 14370 thanks
1:28:04 for contributing Thelonius I hope things
1:28:06 work out Pate as always thank you I
1:28:08 appreciate it Joker you're always joking
1:28:10 Joker you know why don't you be serious
1:28:12 for once and stop joking all the time
1:28:15 it's like what is your name
1:28:16 Joker what's that well one of the
1:28:19 comedians got pulled
1:28:21 over was telling the story of getting
1:28:23 pulled over and the cop walked up what
1:28:25 are you a comedian he was like how do I
1:28:27 answer
1:28:29 this oh man fantastic digital Gods thank
1:28:33 you very much uh who else is in here
1:28:35 Silver Fox good to see you Dr J jrc got
1:28:39 the Irregulars in here
1:28:42 Firefly Arizona Mama You've been in here
1:28:45 before welcome cam katkin cam katkin
1:28:48 from
1:28:50 Cleveland all right have a good rest of
1:28:54 the evening everybody tomorrow's Friday
1:28:56 hey you know what we're going to do
1:28:57 Friday night date night no no we're
1:29:00 coming to the AI learning
1:29:02 lab dear internet thank you for all your
1:29:05 efforts you are quite welcome dear
1:29:07 internet I appreciate it all right
1:29:08 everybody peace out have a good night
1:29:11 I'll talk to you tomorrow