aiLL 09/21/23 - The Future of AI: From Janky to Sophisticated

Transcript

1. 0:01foreign [Music]
2. 0:21[Music] getting the glasses cleaned off so I can see can I see now can I see can you see me where are you people where are you
3. 0:37people why are you late why are you late AI learning lab we come here every night you're not supposed to be late why are you late all right
4. 0:53it's a little warm it's okay I've got my iced coffee oh yeah my my voice was having a rest yeah that was that was pretty pretty
5. 1:07nasty last night I think it's okay now we'll see we'll see when I get talking we've been hanging around here for for an hour
6. 1:17hey Ted looks at life what's happening Omar [Laughter] welcome everybody the beginning is near I don't know if you knew that but it is
7. 1:38I was about to give up don't give up Stevo I had a uh it was Denver startup week and tonight was the uh
8. 1:47the last night bash I tend not to go to Denver startup weeks because I'm busy running a startup so I went to the party it was fine I saw
9. 1:59a couple of people but I am not a fan of uh those Mega networking events they just don't I went to two two uh sessions
10. 2:09today one was okay the other one was a copy waste of time oh oh so anyway Robert Rossi what's happening
11. 2:22Ted looks at life waiting for the Great Pumpkin hey that's a fat joke isn't it I resemble that remark I resemble that remark see yeah yeah
12. 2:33his head was the size of a giant pumpkin he looked like Rosie O'Donnell Tom Arnold maybe yeah see yeah you palooka
13. 2:46pouring water down my face a terrible drinking problem so welcome everyone my name is Kyle Shannon this is the AI learning lab
14. 3:02you know a lot of people out there in this world they're saying the robots are gonna kill us the end is near
15. 3:09no I disagree the beginning is here we're at the beginning of this thing if you're feeling behind if you're feeling
16. 3:18left out or you're feeling like everyone knows more about AI than I do they don't nobody knows more than anyone right now
17. 3:25everyone's trying to figure this [ __ ] out and everyone that thinks they haven't figured out they're going to get
18. 3:30their ass handed to them like say mid-journey because because Dolly three was announced and it can do text and it can
19. 3:39do coherence meaning you can identify a character at the beginning of a a series of prompts and it'll maintain that
20. 3:47character throughout that series of prompts and uh it can do really complicated uh object integration into the scene and and then today
21. 4:01Elon Musk because he's Elon Musk and he doesn't want old Sam Altman to to get too big in his britches tweet it out um
22. 4:10mid-journey's coming up with something pretty cool soon too so anyway there's there's this is this is gonna be a crazy fall if you think
23. 4:21all of these tools we we are we are moving out of so one of the things I say on a channel here because I'm just a
24. 4:29smart ass I'm a smart aleck I'm a smart alec and one of the things I say here I say it pretty much every night I say
25. 4:37all these AI tools are just Yankee pieces of [ __ ] I do I say it I say it just like that I say it so much somebody
26. 4:45made me a graphic for it with a little poop it's poop on top of the word janky we are entering like in the next three months
27. 4:59I may have to retire the janky pieces of [ __ ] graphic I think these tools are going to get pretty sophisticated pretty quick
28. 5:08so anyway listen if you're new here welcome everyone got a lot of folks in here um if you see all the hearts going
29. 5:14that's people tapping the screen like mad now you're thinking like what he hasn't said anything yet why are they
30. 5:20what what is there's nothing to like here exactly those are the Irregulars there's people that show up here night after night
31. 5:28after night including me and they're just not right they're just not right they're irregular and so so they have little games they play
32. 5:39don't we have a we have a t-shirt for them we made like a little t-shirt graphic see they're not weird they're Irregulars
33. 5:46so they're trying they're trying to beat their record from last night what are you up to you're up to one point six
34. 5:52thousand so you're doing good um here's what we do here if you're new to this generative AI stuff if you've not played with chat GPT go to
35. 6:04chat.openai.com and just start playing while you listen to me ramble you'll you'll see what goes on in the channel
36. 6:11um I'll answer any questions you have I'll demo stuff if you want to see it um and the the uh the non-profit uh fundraising person in here
37. 6:22um dm'd me before the things that you wanted me to talk about the future work I don't know if you're in here and but if you are
38. 6:28just sort of pop in kind of what you want me to talk about I'm happy to talk about that but go to chat.openai.com
39. 6:34start playing with it and when you're like but Kyle I don't know what that means I'm not very technical I don't you
40. 6:40say play with these tools you can't play with these tools this is technology this is artificial intelligence you have to
41. 6:46be a mathematician to do that no you don't you don't chat CPT is just like the Google box you know the Google box where you put [ __ ]
42. 6:56in and you get a bunch of ads this you put [ __ ] in and you get back remarkable stuff so go play with it if you don't know
43. 7:05what to say there go to prompts.chat that thing will teach you a little bit about how these things work and then I'll talk about those in a
44. 7:14second but anyway um pop any comments you have down below a bunch of stuff going on oh cool little cool little star thank you digital Gods
45. 7:22appreciate that what do you think about Bard using it to search Gmail yeah so so I did finally get it to work today so
46. 7:29here's here's the skinny on Bard so Bard my top three right now are chat.openai uh that's the the chat GPT
47. 7:38I think it is worth paying the 20 bucks for that oh by the way I want to go see something there's a guy on Tick Tock named Abram apron
48. 7:52a Brawn like LeBron a brawn who's an open AI Ambassador and he did a video today on Dolly three and he said dolly three is going to show
49. 8:04up as a plug-in not as a mode like code interpreter but I'm gonna go look for something I'm gonna go check and see if it's there it's not
50. 8:16but I am going to go look in plugins for a second hold on please let's let's see if we got a little action snuck into our
51. 8:26chat GPT we could we could play with this stuff for free d-a-l-l Dash e nope nothing all right it's worth a shot okay anyway
52. 8:43um yeah so go play with these tools um the question about bard so if you don't know bard yesterday released a new version of bard
53. 8:56which is good they improve the interface it still seems to be decently functional and they added a thing called extensions
54. 9:05and extensions is pretty wild they're kind of like chat CPT plug-ins except what does Google have well Google has
55. 9:14all of your Gmail and all of your Google Drive and travel and all that sort of happy horseshit so so you can now um
56. 9:29search across your Google Drive now here's the deal from my limited experience with it it ain't great yet but the fact that you
57. 9:41can essentially GPT across all of your Google Drive and and your emails you know your gmails from history it's pretty wild
58. 9:51let me get that let me get logged into that because I oh oh here's the here's the other thing about the Google thing
59. 9:57so all day yesterday I couldn't get it to work I even went on the the salon Discord and I I said can anyone help me
60. 10:04make this work like the old man I am hey Marge how do you make the tick tock work is this is this thing on how do I know
61. 10:13if it's on I see myself I'm right here um so I was doing that a little bit yesterday but um if you have a Google workspace account
62. 10:27so basically if you pay for Google you don't get these new services which is stupid now I understand why it's like that
63. 10:37because I would imagine that the permissions for the workspace you know if you're you know gpting across Google Drive for multiple
64. 10:47users is probably a [ __ ] nightmare so I kind of get why it's not there but they didn't they don't give you any indication that
65. 10:56like there's no dialog box that pops up and says this is a Google workspace account go [ __ ] yourself it's just like
66. 11:01just doesn't work so let me see if I can get logged in here so you go to bard.google.com and I gotta switch over to
67. 11:19my other account accept my problem is LastPass has the wrong password in it so I think
68. 11:33open the Gmail app continue okay so uh let's go look at this so here we are at Bard so this is Bard experiment
69. 11:48just so you just so you know this is not a real thing this is just an experiment they reserve the right to steal all your
70. 11:54data and turn this thing off just when you get addicted to it um so it's just like that now if you go bard.com so so if you're logged into a
71. 12:07single account not a workspace account you go to bard.com bard.google.com extensions extensions it should then take us
72. 12:22huh it didn't this is what was happening oh wait it re-logged me into the other account no I want the other account come
73. 12:32on you weirdo okay here we go good Lord you know is it so hard to make [ __ ] that doesn't suck apparently apparently it is okay
74. 12:47so uh uh one conversation many extensions so they have Google flights Google hotels Google Maps Google workspace so this is
75. 12:57all of your drive Gmail and Docs and then YouTube oh that's let's play with YouTube so so the way you call these things is you
76. 13:09call like at YouTube at GMail at drive at docs right so these are I've got these all turned on okay so go back to Bard Bard experiment
77. 13:20and then I'll go at uh at YouTube and that was kind of cool so it says this this extension's enabled so it's recognizing it as an extension
78. 13:37so let's say at YouTube um what is the last video I uploaded
79. 13:55um query unsuccessful well to be fair I might not like this might be a dead uh YouTube account hang on so let's let's ignore that one uh all
80. 14:08right let's see um we'll do at gmail what
81. 14:24um what kind of product is being sold in most of my spam email
82. 14:39let's see if it can do this by the way like I I this is not prepared at all I'm just I'm just playing here kind of in
83. 14:46real time with you so this is kind of the stupid [ __ ] I cannot determine specific products sold out of most of
84. 14:51your spam emails because the emails you have shared do not contain that information that's not true thank you for your order PlayStation
85. 15:02so it gives me some some links to emails I think it takes me over to Gmail yeah it takes me into that that's interesting all right
86. 15:17this lovely Ford Flex what what did I just close I'll just close the window oh good God all right whatever anyway
87. 15:28so what I think what what do I think of Bard I think that um the the tools are weak right now the the janky piece is a
88. 15:44[ __ ] they're they're weak and the large language model that's powering them is not great but Gemini's
89. 15:51coming out so I think they're laying the foundation for some pretty amazing [ __ ] coming up I hope I hope
90. 15:58um okay so how this channel works is ask any questions you have wait tell it that's not true yeah well I
91. 16:07would have but I closed the window so I just bailed I just gave up I'm done I'm done with I'm done with the Bard I don't
92. 16:13know what I'm doing I don't I don't like doing live kind of like I don't mind doing live [ __ ] but like I
93. 16:20haven't tested it at all so I haven't figured out like how to how to use it and I haven't like what what I need to
94. 16:25do because that Gmail account that I have to use to be able to see the extensions I haven't used it in like eight years so it's like
95. 16:35a completely dormant account so I don't know what's in it I don't know what to search for I don't know what would yield
96. 16:39an interesting result all my good stuff's in my workspace account that's the one I want to play with but you
97. 16:44can't you just can't do it they're not gonna let you do it you know what I'm saying I don't know if you know what I'm saying yeah all right
98. 16:57good Lord good Lord all right fantastic Bob let's put a nice doggie up let's put the AI learning lab up this is
99. 17:07Ali the AI learning lab ain't she sweet she sounds white she's so sweet she's a hunting lab she likes to chase ducks she's duck chaser okay
100. 17:19so I'm gonna go start answering some questions this shirt is really hard to read against I might need to go flip on
101. 17:26a uh um what's it called t-shirt a dark t-shirt because this is this is hurting my eyes my eyes are old and tired people they're
102. 17:37old and tired they're old and tired chat GPT helped me land a big client today by creating killer marketing campaign Jim Ross Bang
103. 17:49Jim welcome good to see you again that's awesome so Jim Jim I met Jim I do uh live office hours on LinkedIn every Friday afternoon
104. 18:03and I met Jim last week or the week before and he runs a bunch of uh storage businesses and so he's been
105. 18:14using he's been using I said I said well how much you use an AI I'll just use it a little bit here and there I said well
106. 18:20how long you've been using it he goes since last November I'm like dude you've been using it since day one
107. 18:26you you're Advanced and so now he's like out there Landing business with it boom we'll wait wait A Change Kyle okay fine I'll go change
108. 19:10all right I got my little crescent moon shirt over the high tide the waves of of your Silver Fox many music notes many music
109. 19:25notes [Music] I don't know he put on his shirt he seemed to he tucked at the neck a lot I
110. 19:40don't know what was going on there oh I wonder if that's really an office or a backdrop oh wouldn't you like to know
111. 19:54you've seen this before if this was a backdrop could I grab a screwdriver off a shelf nope could I bag it grab a Ziploc baggie
112. 20:03full of change nope [Laughter] Kyle is your office a big pile of trash shut up shut up don't pick on me all right
113. 20:19do anyone have any insights into a 3D printing generative AI not yet um there's ways you can do it there's some there's some
114. 20:28AI models in the open source space on GitHub that are there's papers on it there's some early models that'll take an image
115. 20:39and convert it into 3D objects there's other ones that will do text to 3D objects they're they're a little too early right now um
116. 20:50but it won't be long it won't be long so um yeah I would say just be patient just give it another six months or so and
117. 21:01then you I mean like like we we're within a year of ask for it and it'll start printing you know if you ever if
118. 21:08you've ever owned a 3D printer they take for [ __ ] ever so so it's not like you ask for it and it shows up like a
119. 21:14mid-journey image it's like you ask for it and then you go to bed and you wake up and maybe the print hasn't failed
120. 21:20and you get something that looks pretty cool Home Collection huh why is that pinned
121. 21:34why is there a multi-carat color area rug pinned oh let's see the CW said the other shirt reminded him of rugs that's hilarious
122. 21:47yeah it's that sure gets a lot of compliments in in person it's a nice shirt it's got all this embroidery on it
123. 21:53but it's pretty much impossible to read here um dude you know you can press the yellow heart at the top oh and send Kyle a free heart oh
124. 22:03didn't know that can I sell myself a free heart right hey nope I got 88 free Hearts that's pretty cool
125. 22:12thanks chair bear I didn't know that was possible um classic Google everything's in beta for a decade yeah exactly exactly experiment
126. 22:25but yeah Gmail was an experiment for 15 years uh at one point A buddy of mine Ben Jones got hired at Google he still works there and uh
127. 22:40I was having some weird issue with Gmail and I reached out to him like dude man's Gmail sucks he's like how much you're
128. 22:46paying for it I'm like oh you're already a douche you worked for Google for like three weeks and no you're already a douche yep [Music]
129. 23:07crypto joke Joe just joined Emilio's wife what's happening Ahoy sir Ahoy irregular cello Ron how are you
130. 23:16what's the best place to start it seems like a lot of public-facing products are I hope this says pieces of [ __ ] are wait wait
131. 23:27kneecapped oh yeah so okay so it depends what you want to start with but if you're just trying to like figure
132. 23:40out this AI [ __ ] just get your head around it there's only one yeah there's only one you need go to chat.openai.com
133. 23:48and you don't need to pay the 20 bucks for it I personally would because you're going to get access to Dolly three when
134. 23:55it comes out and then you won't have to pay for mid-journey um you also get access to code interpreter or Advanced Data analysis
135. 24:02you also get access to plugins and you get access to gpt4 if you pay for the the 20 bucks a month um it's it's the first one it's the one
136. 24:14that started all this nonsense if you're wondering why everyone's talking about AI it's because of chat gbt so they've
137. 24:20got like a six month head start on the industry which with the speed this industry is moving six months is a lot
138. 24:28right so we're gonna see Gemini come out which is going to kick GPT for his ass and then November 6th or 8th open AI is
139. 24:37having their developer conference I have a feeling they're going to announce some things so we're going to keep seeing these
140. 24:43these things do this but GPT right now is the one just go start playing with it go to prompts.chat that'll teach you how
141. 24:50to talk to large language models which is what all of these things are um start with parlor tricks like write me a
142. 24:58poem write me a country song write me a marketing plan just just tell it just give it you know compare two obscure uh
143. 25:06philosophical treatises it doesn't matter like just just throw language pasta at the wall with it have it write you a Python program or a
144. 25:19c-sharp program have a do coding for you write some HTML rewrite an XML schema depending on what you're uh where where you are on the
145. 25:27geeky scale and just start playing with it and then once you once you kind of say okay this is I think this is a little bit more than Google
146. 25:37then start to think about what you do and start to query can it do what I do and then at some point in there you will
147. 25:50have your Kevin McAllister moment and then then the adventure begins and and by the way this is a red pill blue pill moment
148. 26:01if you don't have the Kevin McAllister moment it's your blue pill you can just sit on the sidelines and say I'll just
149. 26:06wait for this AI stuff to show up for real or you dig in and you have that moment where you realize holy [ __ ]
150. 26:17these things are something different this is something different um language pasta that's a t-shirt a language pasta sort of the language
151. 26:30pasta at the wall I didn't even know I said that all right let me add that to the let me add that to the Trello board that's a good one
152. 26:43oh that's really funny 90 tabs let me restore my 90 tabs just to give those of you who actually close your tabs anxiety [Laughter]
153. 26:59where is my Trello where's Trello where are you Trello I guess I'm gonna have to type those letters t-r-e all right that's all
154. 27:10by the way is an autocomplete the best that's that's you know we've been using AI for decades right you do know that
155. 27:18right like all the autocomplete all the spell check the grammar [ __ ] all that stuff say I um so where am I going here t-shirt sayings uh
156. 27:31language pasta I don't know what that means next time I come look at this cue the Jeopardy music I guess that was when I changed my shirt
157. 27:43I wonder if that's really an officer backdrop oh that's what that was in reference to he shows his stuff from the shelves
158. 27:50anyone have insights into the 3D printing generative AI just wait uh listen so so Ron the the Insight I have is go just go to Reddit
159. 28:02and search for um 3D printer AI or you know AI 3D prints or whatever just I guarantee you there's some subreddits in there where
160. 28:13there's some Geeks that have figured out that if you take this tool and you've Stitch it together with that tool and
161. 28:19you do this it'll generate of this which you can bring into Cura slicer and you can slice it and you can make your 3D print
162. 28:26I can't imagine that at this point you know sort of you know I what we're probably like two years into
163. 28:33gpt3 being around two and a half years um someone will have solved some version of it it's just not gonna be super easy
164. 28:41right now so if you want to geek out just go to Reddit I guarantee you there's something there someone's gotta
165. 28:45probably got a Google collab notebook cutting up tonight um Tech Becky techie Becky oh techie Becky 10 [ __ ] out of 10 on the name techie
166. 28:58Becky oh that's [ __ ] gold that you know what that deserves yeah okay sorry what's your comment just saw that there's a class action
167. 29:13lawsuit from The Writer's Guild against open Ai and Sarah Silverman and lots and lots and lots of them did you also see
168. 29:21that Microsoft said that if anyone gets sued for copyright infringement they're gonna pay the lawyer bills
169. 29:29um yeah I don't think these lawsuits really stand much of a chance um but I think they're gonna keep coming
170. 29:37for for a couple of reasons one is copyright holders are rightfully concerned that their livelihood could be [ __ ] hosed like like like it
171. 29:55at a minimum things are going to change dramatically if you understand actually how large language models work and how embeddings work
172. 30:06what what Chachi PT is not is it's not a copying and pasting plagiarism machine that's how it's characterized by everyone but that's not
173. 30:18what it is like the original documents that were used to train jet chat GPT they do not exist in the chat GPT um Vector space
174. 30:30what lives in there is individual words that are clustered semantically all the all the like words kind of hang out together
175. 30:41and they're they're disassociated from their Source documents so depending on how you prompt them you can get them to give you things
176. 30:52maybe even exact copies of stuff that was copywritten but I I just I think there's a technical argument here that you know this is the
177. 31:02equivalent of having books in a library and people going in and learning and then being able to recite something you know
178. 31:09I can sing all of the words of Gilda radners let's talk dirty to the animals but it you know it doesn't mean
179. 31:18I'm a copyright infringer now if I record it and try to sell it then I am I gotta the reason that one's on my mind
180. 31:24is I got a one of my tick tocks one of these lives I played the Gilda Radner let's talk dirty to the animals thing I
181. 31:31played the whole song I got a copyright strike on that video on on YouTube um is AI out of control and can't turn
182. 31:39back due to lack of policies um no not yet but we can't turn back um for a couple of reasons one is so so when Chachi PT launched
183. 31:55um nobody including open AI thought it was going to be a big deal nobody and then it got to a million users in
184. 32:04five days and it got to a hundred million users in six weeks and that shooketh the Earth for these tech companies and so everyone kind of
185. 32:15panicked and everyone who had AI projects that they were kind of slow rolling and we don't want to get it out until it's
186. 32:22right all of a sudden they're like holy [ __ ] we gotta we gotta get out there including Google and Google invented the
187. 32:29core technology that chat CPT is built on this thing called the Transformer um so so there's there's just too much
188. 32:38money at stake you know Microsoft invested 10 billion dollars in open AI 10 billion and um they're rolling it into all of their
189. 32:48applications all of Office 365 all of azure all of uh it's in GitHub co-pilot it's in slack it's in teams it's gonna be in everything
190. 33:02so they're making a big big bet they're not going to let it die because they're they're you know they're sort of betting their future on it
191. 33:09um Google is now responding pretty aggressively meta this is the big reason why it's not going to go away
192. 33:17meta has open sourced some very powerful large language models I don't think they've open sourced their most powerful
193. 33:23ones but they open sourced a 70 billion parameter model and they open source this thing called code code llama which is fine-tuned on coding
194. 33:32um it's a 34 billion parameter model and those are out there for commercial use freely opened no safety guard rails so
195. 33:41could nefarious actors use the Llama 2 models to do bad [ __ ] yes go look at worm GPT 100 bucks a month
196. 33:56product design for hackers to use GPT to go do bad [ __ ] so yeah they can do some bad [ __ ] with it the good guys can do bad [ __ ] to the
197. 34:07bad people too so so so that's a bit of a Saving Grace the models that are out there right now are probably not powerful enough
198. 34:18to do major damage but the ones that are coming probably are and so if if you're curious in into what
199. 34:26those companies are doing look up let me see let me I want to get you the right um whatever you call it the name of the
200. 34:36document the front here look up Frontier AI article
201. 34:58[Music] open AI Google Microsoft and anthropic form a body to overseer oversee safe frontier AI development so the big companies
202. 35:10trying to think I'm trying to see where the actual link is to it uh is it um
203. 35:25I don't know anyway you can go find it it's basically the the frontier AI group or commission or commitment or whatever the [ __ ] it is um
204. 35:34but the ones that have enough money and funding and gpus to be able to do the next level of these things like gpt5 and
205. 35:44Gemini that's coming out and whatever meta is working on that's going to be amazing um and and Elon Musk all of the
206. 35:52companies that have have the resources to do things that could truly do some damage if they if they got out of
207. 35:58control they've agreed to cooperate together to do everything they can to not let them get out of control Sam Altman has
208. 36:06committed 20 of open ai's compute moving forward for what he's calling super alignment if you don't know what
209. 36:14alignment is alignment is basically this is so [ __ ] trippy are these things these large language models the AI the robots
210. 36:27are they aligned are there imperatives aligned with these things imperatives because if they're not when these things get to the next level
211. 36:41and when they can start executing code which they already sort of can and when they can start taking actions
212. 36:48like go do a Google search go book me an airline go you know reach out and communicate with someone when they can start doing that which
213. 36:57they kind of can right now it's a little janky but when they get really sophisticated at that if if the alignment of them and
214. 37:06the the alignment of us isn't right and these things at some point decide well these pesky humans their imperatives
215. 37:14don't align with mine so let's just find a way to slow them down a little bit it seems like they're they're pretty
216. 37:22reliant on Logistics to eat at least in America huh well since I have access to some systems why don't I just uh confuse the logistics
217. 37:35and then groceries store shelves would be bare for a while right so so that's the fear um you know I don't think the tools are
218. 37:46are strong enough that we have that to fear but these things are improving very very quickly and because of that competitive pressure
219. 37:56of chat GPT getting to 100 million users thank you sherbert um they're going to keep they're they're going to keep
220. 38:07innovating it's not going to get uninvented it's not going to go away no single country can shut it down so anyway
221. 38:15I don't think that even answered your your uh question but hopefully it did I honestly recommend Claude AI to start
222. 38:24free is full model just limited usage that's actually not a bad I that's that's not a bad thought Sam
223. 38:31so the thing about Claude AI that that's that Sam says that's that's where he recommends people start it's not a bad thought so Claude AI is
224. 38:40uh the company's called anthropic it's a bunch of X open AI people the people that helped build the gpt3 model are the
225. 38:48ones that started anthropic and they built Claude what Claude has that chat GPT doesn't and this is actually a big
226. 38:54deal and and you know what I I think you know what now that you say that Sam I might I might switch those two maybe I won't switch them
227. 39:04but I think you're right I think I think you start with Claude because here's the deal it's got a hundred thousand token context window
228. 39:11which if if you don't know anything about these things you're like Kyle I don't know what a token is what's a
229. 39:15context window why are you talking so technical I didn't realize you were some sort of Brainiac or something you seemed
230. 39:22you seem like normal people and then you go busting out hundred thousand token context what is that so so what that is
231. 39:33is how much it can deal with in a single chat session BT can deal with about 3 000 words so if you put in like a two thousand word
232. 39:43prompt and it responds with a thousand words that's kind of its whole context window and then anything you do after
233. 39:48that it sort of forgets some of the [ __ ] at the beginning of the conversation or if you put in like a 5 000 word
234. 39:55um document or or you copy and paste five thousand words into it it'll just go gee I can't do that I'm not interested in your
235. 40:04your words there's too many words there I can't I'm a little sleepy chat TBT gets a little sleepy sometimes it's just like Claude
236. 40:16can do about 75 000 words into and if you're like I don't know how many words are in a newspaper article what does that mean
237. 40:25the what is it the Great Gatsby yeah The Great Gatsby I think that's the one fifty thousand word novel so with Claude
238. 40:32you could take the entire text of The Great Gatsby paste it into Claude and just start interacting with it like
239. 40:38it'll it'll have read the whole thing and you can have it write a little book report for you there you go first any little high
240. 40:46schoolers in here and you got a Great Gatsby book report coming up tell them that the AI learning lab told you to use
241. 40:52chat or to use Claude to cheat and then when they call you into the principal's office tell the principal
242. 41:02you'll teach them how to use chat GPT if you if he lets you off the hook because they're gonna need to figure that [ __ ] out too
243. 41:11uh that's that's a really good thought Sam I think I'm gonna make a little shift there language pasta that's a
244. 41:16t-shirt that's been added I'm using it to learn new ml package in go long oh go Lang which is a language I really don't know
245. 41:25well that's super cool yeah you can use these things to teach you how to program did an apple to Apples between GPT and
246. 41:34club today not impressed with Claude what am I missing um well the thing that I just talked about uh Source Camp Kelly camp
247. 41:42is the the big one is the the context window there's another thing about Claude that so so here's the deal
248. 41:51if you're a writer if you're a lover of words gpt4 is the best chat gpt4 at that address chat.openai.com
249. 42:01is really good the Bing version of it not so good because it's got too many safety guard rules but that is really
250. 42:09really good so if you want words that don't suck use that Claude is closer to GPT 3.5 in terms of its capability and its scale
251. 42:21um but because of that large context window it might be a little less frustrating it might not be a sophisticated output but it might be a
252. 42:28little less frustrating here's the thing about anthropic the company behind Claude I'm a huge fan of so the way they
253. 42:35train these things the way open AI trains their models is what's called human reinforcement learning where as
254. 42:42they're training these models they have the model spit out and answer and they have humans go thumbs up thumbs down and
255. 42:48as they do that the model gets smarter and smarter and smarter and that's how they do it couple of things about that approach
256. 42:56humans are expensive and slow and have biases so so every time a human does this or this they're introducing their bias into the system
257. 43:06um and you know they kind of famously got dragged drug over the coals of like well you all your people are in San
258. 43:13Francisco so it's you know it's it's you know it's very left-leaning um Claude anthropic uses what's called a
259. 43:22constitutional learning model which is they write a constitution a set of imperatives a set of rules that when Claude answers a thing it's
260. 43:31compared against those rules and if it doesn't pass the rules it goes back and iterates on the answer um and so that approach I think moving
261. 43:40forward is going to be an important approach and anthropic's whole mission in life is not to create the best large
262. 43:47language model it's it's to dramatically improve large language model safety for the entire industry they just have to
263. 43:57make their own large language model to be able to build out the tech right so there's a reason to support them and I think I
264. 44:05think that you know I think all of these things are going to get much much better in the next I don't know six months or a year so
265. 44:15anyway yes chat GPT is still better um there's a lot of people who don't understand AI so they fear it exactly wait where did that go
266. 44:27but it may come across as anger yeah well that are [ __ ] angry about it right so the The Writer's Guild and sag are angry about it
267. 44:42now I personally think um this is coming I have a degree in acting and I've written seven screenplays so I
268. 44:52know the world this is not this is not coming out of nowhere um I think what they're striking for and
269. 45:00and what they're trying to get in their contracts I absolutely support but the position that they've taken of AI is
270. 45:07evil and if if that's true that the wga is suing open AI they're taking a you know scorched Earth we will never use AI for anything kind
271. 45:21of approach to this and I just think that's not the wisest thing to do because this [ __ ] isn't getting
272. 45:28uninvented it's common whether we want it or not Microsoft announcement today Windows toolbar next week Office 365 co-pilot
273. 45:38goes live wait what did that say when hang on let me go find this um let's see uh Microsoft co-pilot release date
274. 45:55[Music] um [Music] 12 hours Windows 11 co-pilot just got
275. 46:11okay Windows 11 co-pilot just got a release date is that what you said yeah yeah so that's the so so they're
276. 46:17building so into Windows 11 you're gonna have a toolbar that's basically a chat thing where you can just say put my computer into dark mode
277. 46:27um I don't want my computer to go to sleep after 10 seconds I want it to wait 15 minutes whatever and it'll just go
278. 46:34find the [ __ ] and do it that's really cool and then Office 365 goes live November 1st I told you it's
279. 46:42gonna be it's gonna be a wild wild wild wild wild wild wild fall so um duet is live in Google right now but it's kind of buried November 1st
280. 46:58GPT is going to be live in Windows and office 365. Dolly 3 will probably be out maybe a new mid-journey Google Gemini
281. 47:09which is supposed to be better than chat gpt4 um those are the biggies I can think of right now there might be something else
282. 47:18I missed but there's some big stuff coming what's up talk Dylan haha Kyle and Irregulars what's happening to you um
283. 47:35I went through the whole exercise with Bing today they failed Claude Hit it first try oh I went through a-hole exercise
284. 47:50with Bing today they failed Claude hit it the first try yeah yep uh Bing so so what I did what I did to Bing not not
285. 48:02that Microsoft cares that the AI learning lab has done this but so you can barely see the word Bing there underneath my fingerprints that's
286. 48:10because I made it dark gray I I dimmed it I turned out Bing's light because I'm pissed off at it it just it shouldn't suck this bad
287. 48:24and it doesn't suck because the technology is bad it sucks because wait never mind Kyle I was on the quad train there
288. 48:32oh that's right I don't mind that I don't listen here's here's here's the deal about about this channel
289. 48:40I don't give a [ __ ] if people like try this AI stuff and just say I [ __ ] hate it it's awful or if they try it and
290. 48:49they say I love Bing or I love Claude or I love that like I don't give a [ __ ] the thing I'm most interested in is getting
291. 48:55people off the sidelines so that they have an opinion that is informed right if you're just sitting on the sideline
292. 49:05come on I heard chat DBT lies it can't even it it it stops it uh September 2021 it doesn't even know anything in the future
293. 49:14or in the present whatever but if you're sitting on the sidelines not actually playing with it not actually understanding how it works what
294. 49:22it does what it makes possible what it's good at what it's not good at then you're just a whiner don't be a whiner be engaged
295. 49:34if you're gonna [ __ ] all over AI [ __ ] all over it from a place of of knowledge and make a cogent argument for why you
296. 49:41think the robots are gonna kill us why they are going to take all the jobs you know I go the other way but like that kind of that kind of
297. 49:54um opinioneering is great I love that stuff it's the ones that have clearly never used it that have all sorts of
298. 50:03pontificating opinions about it it's like shut the [ __ ] up have you used it and it's clear that they haven't that drives me [ __ ] crazy
299. 50:15have you heard anything about Google using AI to sort through users docs and steal info oh well Laura um I I don't so
300. 50:27probably I mean but here's the deal Google has been okay no let's not put this on Google Google created create created a thing
301. 50:42called the terms of service that over the years just got worse and worse and worse and we all us stupid human beings that were like oh free
302. 50:51email oh a free search engine willingly gave them all of our data so they've already got all of our data Google Docs has been free for
303. 51:04I don't know ages so they don't need AI to mine that but I guarantee you that they've been mining that with AI for many years
304. 51:18it's not just in the past year and it's not just this generative AI [ __ ] right the data is already gone our data is already gone
305. 51:28they have so triangulated when you talk to people that really know what the [ __ ] is going on with data between Facebook
306. 51:36and Google and all of these services they they triangulate everything they know where you are they know when you're
307. 51:46there they know what you're doing what you're thinking what you bought what you didn't they know it all and they know it all in real time
308. 51:53and that's before any of this [ __ ] came out that what I think what makes this stuff scary is we now have access to
309. 52:00tools where we can look across all of our Google Docs and go oh I wonder if Google can do that well yeah
310. 52:10they've been doing it for [ __ ] ever so so you know that ship has already sailed that ship is is is you know that ship has sailed and sunk
311. 52:23the do they have access to our data yeah and they've sold it it's good it's long gone long gone um we being where I work oh cool very
312. 52:34nice so Claude is best for free um can you use AI for advertising you can yeah it's it's it's really good at
313. 52:44you know what Sam Quinn said here um gpt4 in particular is really good at writing now you have to get good at
314. 52:51prompting so it doesn't just give you um pablum [ __ ] if you just say write me an ad it'll write you an ad and it'll
315. 52:59suck but if you say hey here's my creative brief and here's who I am and you know I want you to act like a master
316. 53:08copywriter for a world-class Ad Agency and I want you to come up with you know a whole campaign inspired by Lee Klaus work with Apple
317. 53:19it'll know what all that means and it'll go write you some cool [ __ ] and and tell it to write you 30 you know
318. 53:2730 headlines you know tag lines for a campaign and then out of the 30 taglines it writes five of them won't suck and then you can
319. 53:37say hey I like these five give me 50 more that are more like that than the other 25 and it'll so yes really good for advertising and
320. 53:50also for marketing strategies and I'm a writer and actually enjoy prompting so maybe that's the difference for me so so
321. 53:57Source Camp Kelly camp I like your name it's a lot there's a lot going on with the source and the Kelly but I like it um
322. 54:10this this is the thing that frustrates me about the The Writer's Guild and the SAG thing and sag's a bit a bit more
323. 54:17complicated but the writer's Guild in particular these in my humble opinion in my my very humble opinion
324. 54:28are the most profoundly powerful writing tools in the history of humanity and you've got a professional organization that represents writers
325. 54:38going [ __ ] that we'll never use it really when it's built into Microsoft Word and it's built into your screenwriting
326. 54:48software and it's built into your note-taking applications and it's built into Google Docs and it's built into [ __ ] everything windows 11.
327. 54:56you're not going to use it really if you can improve the quality of your outline of your screenplay by having this thing get you to 80
328. 55:10percent in 10 minutes rather than three days and then you spend a day refining that outline without chat GPT why would you not use that
329. 55:22and I think the fact of the matter is a lot of writers will but the Union's taking the stance like AI is evil well no what okay
330. 55:32you can judge it all you want it's not going away it is the most profoundly powerful writing tool in the history of mankind
331. 55:40and you're like we don't want to use computers we like the sound of typewriters we we like the letter hitting the paper chick
332. 55:48it's not a screenplay unless a metal letter hits a ribbon onto a piece of paper storing screenplays on magnetic media
333. 55:58how dare you that's not real writing I mean that's that's literally what's happening it just baffles my mind it doesn't
334. 56:10because it is [ __ ] scary like I get why it's scary I get why they're saying hey producers don't [ __ ] use these things
335. 56:21but the producers are going to use them anyway and if they can say listen these tools are inevitable we will like
336. 56:30to likely incorporate them into our workflow but even if something is AI enhanced you know from the production company if I'm
337. 56:41the original writer on it here's the rights that I want like they can still fight for for rights and sag can do the same thing without
338. 56:50denying that these things are coming they can say hey they're coming and given that they're coming here are our demands
339. 56:58that's cool like I don't know if it's cool that makes sense to me but this you know we're gonna Sue open
340. 57:06AI to get rid of it it's not going to slow it down I mean llama is open sourced so let's say let's say they win they shut down open AI
341. 57:19they even shut down Microsoft right the writer's Guild Microsoft can no longer use chat GPT what about all these open
342. 57:27source projects built on top of llama 2. what happens to them does the wga go after meta and say put the horse back in
343. 57:36the barn and Matt is like uh it's not one horse now it's like 10 million horses and nothing we can do about it
344. 57:44Fair Point I've been hearing Rumblings from writer talk about Google taking story ideas oh that's interesting
345. 57:51Google taking story ideas from writers but but is Google like they don't have a Netflix arm do they I guess YouTube might have a
346. 58:03production arm but I don't think YouTube's doing like like full-on full-on shows I I I find that that sounds to me like just
347. 58:12paranoid [ __ ] like what would be in it for Google to steal story ideas because story ideas are [ __ ] dime a
348. 58:20dozen and now with chat GPT I mean honest to God I've I've done some experiments I did an experiment where I said I took gpt4 and I told it
349. 58:34to give me the original Star Wars plot it did and then I said I said okay now write me an outline that matches you know scene
350. 58:45for scene matches Star Wars except instead of being in space I want it to be um animals on the Galapagos Island at
351. 58:56the time of Charles Darwin and it came up with character names that were [ __ ] genius it came up with a plot that was [ __ ] amazing
352. 59:07and so so I I just I I just like it just feels like conspiracy [ __ ] to me I don't what's in it for Google to steal [ __ ] story ideas they
353. 59:21just it just doesn't make sense to me I mean Google's got much bigger fish to fry like like predicting consumer
354. 59:28behavior of every person on the planet there's big money in that you know if they get story credit for some
355. 59:38Film Production that might make them 150 million dollars or a billion if they really knock it out of the park
356. 59:44and meanwhile they're making trillions with their data it just doesn't make sense to me again I understand fear
357. 59:54like like what that probably was was like that Air Force thing where some some writers said it hypothetically
358. 1:00:00because Google's got access to Google Docs they could steal our stories and then someone else heard that and said
359. 1:00:05Google's stealing stories that's probably what happened there good use of chat GPT input customer personas and use
360. 1:00:12it as a focus group ask questions it's really good for that how would they respond it's really good for that it's really good for that
361. 1:00:20I did some I did some Persona stuff where I I actually brought one of the personas to life um using did and the voice synthesis
362. 1:00:29stuff it's pretty good I wonder if I have that somewhere I do you wanna see it yeah this is it
363. 1:00:47view play in a window play Ed oops [Music]
364. 1:01:07February and this was something I wanted to do I wanted to experiment for my company storyline so we have a video storytelling platform
365. 1:01:19and we do a lot of work in healthcare and Pharma and we wanted to see if like a lot of times our Pharma clients
366. 1:01:27can't find patients like they don't have the right indication or whatever they all the stuff they all their fancy s words and so
367. 1:01:39I wanted to see you know if we needed to generate a patient if we needed to generate a patient's story or generate a patient
368. 1:01:48how might we do that well you know they always want them to be you know diverse and represent different groups and things like that so
369. 1:01:56what I'm about to show you from the from the minute I went into chat GPT and said huh wonder if I could do that
370. 1:02:06to the video I'm about to show you took less than 30 minutes and this was the first time I tried this this was last February
371. 1:02:13so shit's gotten better since since this so the first thing I did was I had chat GPT come up with um a Persona framework and I said you
372. 1:02:23know how do you create Persona Frameworks and it educated me on that and it gave me a bunch of different
373. 1:02:27options and so I selected some of the things and I said okay these look good and then I said now write me 12 personas
374. 1:02:34that that you know have different backgrounds in all these different areas bang out they came so I took one of them this is Amina and
375. 1:02:43the way storyvine software works is we have an app that guides an end user it prompts it prompts a patient in this case
376. 1:02:53to answer a series of questions about their disease when were you diagnosed and what do you enjoy doing and you know
377. 1:03:01how have you been treating it whatever whatever the prompting questions are in a person answers them so I then had chat
378. 1:03:06GPT have Amina tell her story using our prompts so so basically she answered our prompts just like she was a real person and then
379. 1:03:17um Leah fasten who co-hosts the salon with me she's a photographer and she had this really cool style that was super
380. 1:03:23authentic so so she gave me a whole bunch of images to start with so I picked this one for Amina and then I took this story and I took a
381. 1:03:34um South Asian accent in um it was at 11 labs yeah uh yes no it wasn't it was in did so in did you
382. 1:03:50can pick different languages and so I just copied in these and then I made video clips and then I put those
383. 1:03:56video clips in my company software which automatically assembles them into a fully edited video hang on
384. 1:04:07and this is the video that was created so again this was from start to finish less than 30 minutes [Music]
385. 1:04:23I was diagnosed with illness about four years ago what brings me joy is spending time with my family especially my parents and siblings
386. 1:04:34we enjoy cooking and eating traditional South Asian dishes together moving beyond my diagnosis was not easy
387. 1:04:41but I found comfort in my faith and my family's support I also started to focus on what I could still do instead of so anyway you get
388. 1:04:51the basic idea so yeah so not only can it do personas you can bring those personas to life in ways that that you just couldn't otherwise and
389. 1:05:01you know again this still sounds a bit monotone I think I think with the accent it it sounds you know decently realistic but
390. 1:05:09it's still fairly flat and monotone but within a year it's not going to sound flat and monotone it's gonna have
391. 1:05:17it's gonna have you know Cadence and and Dynamics and things like that and it's going to seem more increasingly
392. 1:05:24real and so you know the like the amount of use cases here like like just just dealing with um things like um
393. 1:05:35you know there's patients who want to tell their story that can't tell their story well now they're going to be able to
394. 1:05:42um you know so all sorts of stuff ideogram site any tips to using it um a little bit a human being so um be patient with it
395. 1:05:55um put your text in quotes use the buttons at the bottom like the typography ones a good one like if you
396. 1:06:02do typography and poster it tends to do pretty good typography AI soapbox preach it Ted look at life that's cool look look at what Ted
397. 1:06:12putting put in the in the uh in the Discord that's pretty funny that's awesome um so so that's the first thing is um just
398. 1:06:27uh experiment with a different thing the the different buttons will just give you different results the other thing is go
399. 1:06:36find a video that um or I mean go find an image an ideogram that looks close to what you want and just copy the post because in copying
400. 1:06:46the post it'll have what buttons they pushed um so that's a way then you can remix it another way is to remix it where you you
401. 1:06:55take one of their an image that already exists and just hit the remix button and just keep spinning it so so what I tend to do on it is just
402. 1:07:05experiment experiment experiment until I get something that's close like AI learning lab with this dog I would get
403. 1:07:11like AI leaning lab or or learning lab or AI lab it like it was always all over the place but if you get it close like
404. 1:07:21the one that said AI leaning lab I took that one and I remixed it and I just kept remixing it until it spelled it right
405. 1:07:31because because once once it it sort of you know understands the sort of generals you know amount of words and
406. 1:07:39letters in in the neighborhood it it tends to get it right at some point in two or three iterations what was the video platform you just
407. 1:07:49showed Kyle so the the thing that assembled that with all the graphics is my company's platform called storyvine
408. 1:07:56storyvine.com but the thing that generated Amina talking is D Dash ID so Dash id.com I think let me just make sure that's right d
409. 1:08:09Ash id.com yep the digital people platform here I'll show it to you
410. 1:08:32[Music] so like here's all the videos that I've made over the over the you know months and then let's see if if I want to
411. 1:08:46create a new video um I've got here are the presenters some of these I made in
412. 1:09:01mid Journey some of these I made in stable diffusion um here's ones that they give you they've got high quality ones that I
413. 1:09:09guess those are 1080p ones or instead of 720p or whatever um we've got close-ups these are all ones so you basically just pick a pick a
414. 1:09:18thing and then you can um foreign you can either type your text or you can upload where do I do it oh yeah so right here
415. 1:09:33you can either type text in right so you type in a script here like hi um hi my name is Bob I like
416. 1:09:48Turtles and then you can choose different accents um we'll just leave the United States and
417. 1:10:04then certain ones like the United States they have lots of different characters so we'll do that let's go let's get a let's get a Bob here
418. 1:10:14where's Bob there's Bob that's Bob there's Bob and then you say generate video it's gonna cost me one credit let's go
419. 1:10:32it should go really quick it's funny every time I demo this they've slightly changed the uh the interface that because that's the thing
420. 1:10:47you know all these tools being janky pieces of [ __ ] they they get less and less janky every month so you know you
421. 1:10:53go back in to use these things and you're like oh they actually made this usable now hi my name is Bob I like touchels [Laughter]
422. 1:11:10okay [Laughter] so that's D Dash ID so that's that's the tech I use to make that um that South Asian woman telling her
423. 1:11:26patient story that woman Amina hello I'm Amina Ahmed I was diagnosed with Alice about four
424. 1:11:43years ago cuckoo cuckoo and then and then you know then you've got things like um okay I'll show you we've got 100 people
425. 1:11:55in here so there's people that haven't seen this yet so so that's her so that's just sort of inventing a Persona like literally out
426. 1:12:03of nothing in in like 25 to 30 minutes and having her tell her story and then um and then you have a tool like hey Jen
427. 1:12:13which just sort of sneak sneakily launches a feature they don't even realize is a big deal and it does this
428. 1:12:24this is [ __ ] this is so stupid I mean so stupid like crazy like holy [ __ ] we live in this reality right now this
429. 1:12:34exists right now so this is uh look more Kyle this is a video I filmed last uh February this was just a clip in
430. 1:12:45a in a video I made you know for marketing or this was like a video I sent out to my advisors or something
431. 1:12:51like that so I just I just took some clips from this I uploaded it to this hey Jen video translate thing and watch this
432. 1:13:01so we've heard from a number of clients that modular content is important to them getting more efficient with how we
433. 1:13:07create content and how we leverage content is behind me you see these little green boxes um that's how we think about
434. 1:13:20stories at storyline we break stories down into little atoms right little modules
435. 1:13:36right so like that exists today you just upload files 10 or 15 minutes later you get back something and I think it's up to eight
436. 1:13:45languages right now the co-founder I watched it live with the co-founder last week he said they're going to be up to 20 languages pretty
437. 1:13:54quickly and then a lot more after that and and he said there's nothing this is astounding to me he said there's
438. 1:14:03not much present preventing them from being able to do that in real time other than enough compute power
439. 1:14:11so as algorithms get better and gpus get better and you know more server Farms are are you know serving up more GPU Cycles
440. 1:14:26um we're gonna have a literal Babel Fish um kind of situation where you're going to be on a zoom call with people in
441. 1:14:36multiple countries where everyone's speaking their native language and everyone's hearing all the other participants speaking their language
442. 1:14:44fluently [ __ ] Wild crazy all right time sensitive bedtime reminder your
443. 1:14:58bedtime is approaching time to wind down and prepare for a good night's sleep thanks Apple I know you've got my best in mind
444. 1:15:08today you know what I need to do no Kyle what I need to charge my little watch hang on I'll be with you you people just
445. 1:15:18calm down just calm down I'll be back just hold your horses hold your horses I know I was late tonight everyone was
446. 1:15:29all like where's Kyle where's Kyle class didn't start yet we've been here for an hour where are you where are you
447. 1:15:38Sam corn Kyle oh have you seen Google's new generative Image Dynamics yet um I don't know is this a new thing I
448. 1:15:48saw their style drop thing are you talking about style drop or is this a new thing because if it's new no
449. 1:15:55oh and by the way for those of you that are here and you're like um I'm so far behind if you pay attention to this [ __ ]
450. 1:16:04every single day and like you do everything in your human power to keep up with it you're not even [ __ ] close
451. 1:16:16not even [ __ ] close so let's see if Google did something new with Image Dynamics Google image I'm telling you Google's gonna come out
452. 1:16:25[ __ ] swinging Dynamics I didn't have faith in them after their Paris debacle back in February that was [ __ ] embarrassing
453. 1:16:41and then they did another one in June that was [ __ ] embarrassing and then just all of a sudden in July Bard started sucking less
454. 1:16:49in this latest set of updates and Bard is actually pretty good uh
455. 1:17:03um generative Image Dynamics from oh September 18th okay so this is three days ago generative Image Dynamics from Google
456. 1:17:16brain research is this it sorry I might I might have some dust in my throat or something maybe my throat's going again um
457. 1:17:33let's watch it together Google just dropped a new paper with the name generative Image Dynamics which is
458. 1:17:41the proof we are living in a science fiction World they created a diffusion based model which takes it still image
459. 1:17:47as input and outputs Dynamic looping video okay I can't understand that um from static to stunning Google Images
460. 1:18:00generative Image Dynamics so this is a paper okay this is a paper yeah September 14th was this is this the one from Google
461. 1:18:30is this the one yeah that's the one what's it do um oh you want to see something fun yes Kyle that's why we come here show us
462. 1:18:39something fun don't ask us questions like that silly you know we want to see something fun my goodness he's such a silly bird all right
463. 1:18:50so we're gonna go to chat gpt4 I don't know if you ever heard of chat jpt but uh [Laughter] so chachifi teased this thing it's going
464. 1:19:04to change everything I don't know if you heard about it gonna totally [ __ ] up uh I didn't talk about the future of work I
465. 1:19:10don't know if uh if and fundraising is in here but if you're in here raise your hand and I'll I'll wax
466. 1:19:17philosophic about the future of work but anyway we're gonna go to plugins and then I'm using a plug-in called Web
467. 1:19:26pilot and web pilot can look at websites and PDFs so I'm going to say based on this PDF and this is by the way so one of the
468. 1:19:40nicknames for the channel as you might imagine is chat add which is I think it's my favorite nickname of all the stuff uh I I have 80d and
469. 1:19:52while I like technical [ __ ] the thought of reading a long technical paper and actually parsing all of it
470. 1:20:01makes me insane and and with these technical papers even reading the abstract is like you know they kind of
471. 1:20:08assume you know what you know certain abstract mathematical concepts are so when you read the abstract you can't
472. 1:20:14even understand what the [ __ ] they're talking about so that's that's no longer an issue so based on this PDF um please summarize this
473. 1:20:31um at a fifth grade reading level I'm going to talk it knowing that
474. 1:20:46I am an artist and trying to understand what is significant about this paper in the context of digital art
475. 1:21:02and so now it's going off it says using web pilot so it's using that that plug-in it's going off and reading the paper for us
476. 1:21:11and then it's going to come back and it's going to do a little writing it's going to do a little creative writing
477. 1:21:18all right this paper is about making pictures move summary the scientists again I did this for a fifth grade
478. 1:21:25reading level the scientists in this paper have created a way to make pictures look like they are moving like
479. 1:21:31trees swaying or flowers Dancing In The Wind all from just one single image they use something called neural stochastic motion textures
480. 1:21:42see you don't you don't know what those words [ __ ] mean like they're it's four words neural stochastic motion
481. 1:21:51textures I don't even know what some of the individual words mean much less what that grouping of forwards means
482. 1:21:59to predict how each part of the picture should move to make it look real this means they can turn a still image in
483. 1:22:07into a video that Loops over and over and over or even let people interact with the picture like dragging and releasing a point on an object
484. 1:22:16of the picture and see it bounce back so you could grab a flower and drag it and it'll snap back that's amazing
485. 1:22:22this is really cool for artists because it can bring your artwork to life why it's cool for digital art bringing art
486. 1:22:29to life interactive art creating Dynamic scenes questions to think about how can I use this you can use this technology so anyway like
487. 1:22:38even though that's kind of a silly little way I did that you know what I'm gonna do I'm gonna say hang on neural stochastic motion
488. 1:22:47textures copy I'm going to say watch this this will be fun I want you to pretend like you're a marketing master
489. 1:23:01and your goal is to help these scientists repackage the concept of and I'm gonna and then I paste it in neural stochastic motion textures
490. 1:23:19um come up with a way to say that that will make sense to non-scientists and let's use that as the tagline
491. 1:23:35and then I want you to come up with 10 creative names for what this technology might be called if we were going to Market it
492. 1:23:48absolutely tagline bring your art to life with magical motion textures perfect creative creative names art mover Magic
493. 1:23:58picture pixel pulse Creations a live brush technology motion craft Studios these are all shitty Dynamic canvas Innovations interactive wait
494. 1:24:09interactive imagination flow interesting wondrous wave Artistry no alive and art Solutions okay um let's see the names are too slick
495. 1:24:23can you keep them simple and compelling but have them sound less like you're trying to sell me something
496. 1:24:33I don't even know if it'll understand that but Art In Motion living pictures that's pretty good living pictures simple motion art
497. 1:24:43breathe art picture pulse quiet motion gentle move art but anyway you get the idea like one of the one of the things that
498. 1:24:52researchers often face is they'll create something remarkable and they they have a tough time communicating like the
499. 1:24:59value of it because they're they're interested in like how to describe what the thing actually is but if they want
500. 1:25:05to get the thing out there you could use something like this to to you know create a marketing version of of your
501. 1:25:14paper where one goes to Archive for the scientists and another one goes to to Twitter to understand what the [ __ ] this thing is
502. 1:25:23that's really cool thanks for sharing that I think that was Sam or no is that joker Sam or Joker shared that thank you for sharing that
503. 1:25:33meth lab AI learning meth lab yeah all right um yesterday I saw a video that said dolly is now embedded in chat GPT can
504. 1:25:46you show it's not it's uh I can I can show you the site it's not in chatbt yet I'll go look now the thing that was
505. 1:26:00announced yesterday is it was announced yesterday and it sounds like from a Braun um this guy I know that it's one of the
506. 1:26:08open AI ambassadors it's going to show up as a plug-in um I'm gonna go just make sure that they didn't sneak it in while we've been
507. 1:26:22sitting here this Happened One Night with uh code interpreter I was like when am I gonna get access the code interpreter
508. 1:26:30they never gave me eyes as the code interpreter I talk about oh here it is that's pretty funny um and then we'll go look and see if
509. 1:26:38they snuck it into plugins because that's where he said it's going to be so in plugins we're going to go to the plug-in stowa downtown
510. 1:26:52dolly no dog Dash e nope not there um but I'll go show it to you
511. 1:27:06well that lost its CSS connection didn't it there you go all right so here's Dolly three this is the website so if
512. 1:27:15you want to see this go to openai.com d-a-l-l-e-3 because why make it easy to type in [Laughter]
513. 1:27:30here's here's some of the images the this this one right here um there's a couple of things going on here
514. 1:27:40so the prompt for that image where is it is let me open this up all right The Prompt for this image is an
515. 1:27:51illustration of an avocado sitting in a therapist's chair check saying I just feel so empty inside check
516. 1:28:03and if you know these tools they can't do that well with a pit sized hole in its Center so that's not a pit it's a hole a pit sized hole
517. 1:28:15the therapist a spoon wait a spoon check is taking notes check so it can take all of the context it's also built in natively in the gpt4
518. 1:28:33um gpt4 is going to help be able to help you prompt and it's also going to be in context of a larger conversation so
519. 1:28:43let's say you describe the plot of a graphic novel and then you want to start creating concept art for it it
520. 1:28:51understands the context of the graphic novel that you're that you're writing about um so that's that guy um
521. 1:29:00here's another one where the the oops what did I do oh old man trying to use computers old man trying to use computers yeah
522. 1:29:13um so this is cool this is a lychee and inspired spherical chair with a bumpy white exterior and plush interior set
523. 1:29:23against tropical wallpaper yep that pretty much hits it out of the park so it does that um it also does this is a thing that
524. 1:29:35again just it's like a really complicated prompt so they broke the prompt up and just showed all the different things that uh it captured
525. 1:29:45um here's dolly two with this same prompt and here's Dolly three with the same prompt so Dolly three is just way
526. 1:29:54better and then I'll play this video here because this video is crazy [Music]
527. 1:30:18oh my daughter says its name is Larry can I see more like this so it now it it now remembers that that character
528. 1:30:33is Larry and so now he just goes on and prompts it to put Larry in a bunch of other scenes which if you've ever tried to do a
529. 1:30:43children's book this is one of the most frustrating things because you get a cute thing like Larry and then you try
530. 1:30:49to get Larry to be in other scenes and you just [ __ ] can't it's like it all of a sudden the Hedgehog is a platypus
531. 1:30:55and then it's a squirrel and then you're like where the [ __ ] is Larry Larry's gone not anymore [Music]
532. 1:31:33look at that you tell the whole story then you then you put Larry to bed and he goes to bed this it's like
533. 1:31:41Emilio's wife it's like Emilio's wife Larry was made for you look at that bed look what we have for Emilio's wife
534. 1:31:49she's very concerned about being late and she's just got lovely Emoji game and so so fans made her an am I late sign and it's
535. 1:32:07just like little Larry look look how sweet you see how long this steps in her related we're all part of the same Oneness you
536. 1:32:22know that right so we shouldn't be mean to each other anymore how about that what if we use AI as an excuse to stop
537. 1:32:30being [ __ ] to each other how's that for a radical idea you in Microsoft debut debuts AI powered
538. 1:32:45co-pilot platform new Surface Hardware interesting Microsoft to sell AI assistant software the biggest clients starting November 1st yep
539. 1:32:56Kyle there's a demo of the moving images online oh cool where is it let me go see if I can find it um what was that called um
540. 1:33:08generative Image Dynamics generative image Dynam
541. 1:33:23X demo generative image dynamics.github.io paper demo I met Sam Sam's the best
542. 1:33:37Sam has has has brought me to the other side of if someone says where should I start I think they should start with
543. 1:33:45Claude I think you're right Sam and now Sam's like but Kyle showed demo don't just read the paper with your stupid chat JP hey
544. 1:34:01all right seamless looping video it's just wiggling interactive Dynamics it's wobbling the picture that's super cool
545. 1:34:07oh yeah look at these things oh this is so cool you know what just I mean think about this for just um just for marketing for for marketing
546. 1:34:23if nothing else you generate a cool image or you take a cool photograph you pop it into a tool like this and it just brings it to life
547. 1:34:31like this it just so imagine the web and and social media is just gonna start having like more emotion in it because it will be just as
548. 1:34:45easy to make an image that does this is to just make an image that's pretty slick can we grab any try it yourself yes oh look
549. 1:35:00by the way I'm at generative Dash Dynamics slash GitHub dot IO wow wow wow can I grab the background
550. 1:35:16nope so it understands that this is in the foreground so it's the thing in Focus so it separates it from the background wow that's wild
551. 1:35:25let's see let's grab another one a little white one I mean it's a little boring but it is it's it's a lot of cool it's like what's
552. 1:35:35nice about it is how fluid it is and how it's it's instant oh that's cool oh wow it's even doing some parallax
553. 1:35:50you can't really see it it's doing some kind of Parallax stuff on like that little Branch right there is moving slower than the one that's
554. 1:35:59closer so like it's it's understanding some depth stuff in here there's some artifacts that are weird uh
555. 1:36:09see see these blurry lines here there's some weird artifacts on that one yeah yeah it's it's sort of [ __ ] in the bed there but this right
556. 1:36:21here is perfect huh but it understands that this is closer so it's trying to move it more and so it's just not interpolating that
557. 1:36:36movement right but the the one where it's not moving it a lot it's really cool motion magnificent oh that's so cool we can mimify top or
558. 1:36:48magnify bottom Minify oh zoom out by adjusting the amplitude of the most motion textures so this is motion textures
559. 1:37:04Wild motion related work that's super cool all right see please repeat the name again yes the the paper is called generative Image
560. 1:37:16Dynamics from Google research and the URL is generative Dash Dynamics dot GitHub dot IO and in fact
561. 1:37:34um [Music] uh-oh what happened no stop oh Bing whoa
562. 1:37:48look what Bing did this is like Siri taking over I was just trying to show you the URL of that thing um and Bing Bing heard me and said
563. 1:38:00generative Image Dynamics demo and made an image to kind of go with it that's kind of cool um anyway there's the there's the URL
564. 1:38:12gender generative Dash Dynamics dot GitHub dot IO and then there's a link to the uh to the paper to the archive and to the demo
565. 1:38:27so that's pretty Swale all right what time is it I got Listen People I got I have a 7 A.M meeting it's 10 53 it's
566. 1:38:36very late people are ruining my life you're ruining my life um oh you're not You're Swell um a couple of things let's see I'm
567. 1:38:50totally up for that idea I'm gonna I'm gonna scroll to the bottom here just so the Takeover begins it starts with our web browsers
568. 1:38:58uh off to work I go later Thelonious peace now Bing is jealous exactly what the hell is going on here the dot IO
569. 1:39:05extension is assigned to the British Indian Ocean territory that's really funny um let me show you I want to hit a
570. 1:39:16couple of things before I go um first thanks for hanging out this is cool um I gotta go to bed though uh uh uh uh
571. 1:39:24where is my there it is okay um if you want to join a community of people exploring Ai and trying to figure
572. 1:39:36out what the [ __ ] all this stuff is um this is this is a group I started back in December called the AI salon we meet every other Tuesday
573. 1:39:47um in person here in Denver and then online and then we have an online Discord and a lot of the people from the
574. 1:39:53AI learning lab have infiltrated the salon and are kind of like you know playing around in there it's really cool
575. 1:39:59there's a lot of new energy a lot of new activities so if you want to you know be part of a community that's trying to
576. 1:40:06figure this stuff out Larry's gone keep the digital Gods just made it funny uh people are making funnies what's fun
577. 1:40:18is people are using ideogram with the with the text generation as I say stupid [ __ ] they make stupid images and put
578. 1:40:24them in there to remind me of the stupid [ __ ] I said which I think is it's kind of the most flattering kind of trolling
579. 1:40:30I've ever experienced I like it a lot it makes me it makes me smile inside um so anyway go do that um uh Story vines my company uh but the
580. 1:40:41the other thing is I'm putting on a workshop um for creatives so I've been invited by another Creator who's a creative coach
581. 1:40:50and a voiceover artist and we've put together a three-day Workshop it's it's two hours once a week for three weeks
582. 1:40:57starting October 1st so if that's something you'd be interested in just go there um and I'm trying to think what else
583. 1:41:04just I'll go back to this this is kind of the most important thing um if you haven't played with these tools yet
584. 1:41:14start with Claude as Sam told you um or Chachi PT chat.openai.com or Bard any one of these three and just start playing with them just
585. 1:41:26start seeing what they do and how they do it and and it will serve you well also I've got um the recordings of these lives I'm hosting on
586. 1:41:39a YouTube channel AI learning lab Dash TT um and if you go there you'll look at the impressive descriptions of each
587. 1:41:47episode that was a that's all Claude um I figured it took me God it took me it probably took me 50 attempts
588. 1:41:59at that prompt to get it even close to giving me what I wanted because I'm trying to figure out how to
589. 1:42:06chapterize these these big long things that are just kind of random but anyway um hope that's good listen I appreciate you all
590. 1:42:16let's get up a pretty little image here to go to go night night with we started with that one let's see let's see which one we got here
591. 1:42:26that one's good the little the little cranky the cranky bartenders with the AI salon the AI Salon booze um
592. 1:42:37all right so we're gonna do a little Romper Room here Nancy good night micro thanks for coming Sam corn thanks for
593. 1:42:44your contributions tonight that was awesome um forever hooked welcome thank you micro Thelonius Emilio's good to see you uh Nancy jrc
594. 1:42:56um thank you all peace out digital Gods thank you so much I appreciate the lemonade lemonade there that tasty lemonade
595. 1:43:06can't wait till Ben gives you the survey share bear thank you oh cool Shivers signed up for the workshop that's awesome
596. 1:43:15um is Claude still exclusive to Slack it really is [ __ ] days of our AI learning lab models man I appreciate these lives thank you
597. 1:43:29Paulie Z I appreciate that and thanks for whoever pinned that that's really cool all right everybody nostalgic fix
598. 1:43:34thank you Joker good to see you micro peace out everybody I hope you have a good night and I will see you tomorrow's
599. 1:43:43Friday so I'll see you Friday night what are you doing Friday night I'm watching some old guy on Tick Tock what are you
600. 1:43:48doing I'm doing the same thing thank you I don't know why you would do that but I'll see you there