aiLL 9/22/23 - 17k Learn About Generative AI Live With Kyle Shannon

Transcript

1. 0:10mm-hmm come on there we go
2. 0:30welcome welcome welcome welcome to the AI learning lab everybody happy Friday night another Friday can you believe it hmm
3. 0:42it's another Friday night so there's that what's happening hello hello
4. 0:57my watch is too tight let me loosen this ah [Music] what's going on here there we go all right
5. 1:13Nancy hello there Irregulars what's happening what's shaking welcome real chat bots of the future making money with chat gbt what else we
6. 1:27got we got let me find my Trello board we got all the good sayings here at the at the AI learning lab a uh learning lab
7. 1:42what do I what did I Trello up weird weird weird weird [Music] escapee ah
8. 1:56let's see this chat add welcome mansplaining is a service I'm not weird I'm irregular I'm AI native the
9. 2:05beginning is here we have an image for that one the beginning is here I recruited a non-technus to start
10. 2:20learning AI awesome that's super cool that's super cool how did it go did they resist it's like a cult Training Center have
11. 2:31you tried AI listen listen can I talk to you about church every day what did you say can I talk to you about tattoo did you say Church
12. 2:42they're listening Siri's very jealous series very jealous Bing has gotten a little aggressive but could I talk to you about Chachi be today
13. 2:57she was all over it good good I you know listen I I think that I think that word is getting out that this [ __ ] isn't going away and
14. 3:09it's time it's time to start you know this is this is not dissimilar to the early days of the web where this little
15. 3:16tool came along and first it was the dweebs and the Geeks and then Along Came the lawyers and the accountants and the
16. 3:22creative people and they all started using this worldwide Webby thing and then at some point it became
17. 3:29inevitable and you know with the web it took two or three years with this stuff it's we're what are we 10 months in now
18. 3:36something like that Stevo what's happening ah I had a little bit of coffee earlier it was not enough had a long day today I got up at
19. 3:48I thought I was gonna have a seven o'clock meeting I didn't an eight o'clock meeting a nine o'clock meeting
20. 3:52at 10 o'clock meeting at 10 30 meeting an 11 30 meeting and I think that was it until two so I'm Crispy and they were big meetings
21. 4:03these were big beefy you know CEO type meetings all right welcome everyone my name is Kyle Shannon this is the AI learning lab
22. 4:14if you do me a favor go ahead and share the hey Luna stick what's happening if you share the uh Stevo I'm late because
23. 4:22I was updating my phone again has anyone noticed like the past three days I keep getting logged out of like Google Apps
24. 4:28and and just everything I seem to be logging into they're like log in again and like just use our convenient 17 Factor authentication scheme
25. 4:39you're like click here then we'll text you on your phone then we'll ask you if it was really you and when you say yes
26. 4:44we'll bring you back here and then we'll jump you to another browser where you're gonna have to do a capture then you're
27. 4:48gonna have to come back here scan your iris and then we decide you can't get into your account because we don't recognize
28. 4:55the machine really is this uh that's how we're going anyway that's not what this channel is about although this channel is about me
29. 5:03bitching about stuff it absolutely is don't get me that don't let me deny that because I that would be inaccurate
30. 5:11this is a whining Channel we're dug into any wine or we've got diverticulitis no um why are they aggravating us exactly
31. 5:22Nancy Nancy understands Nancy's an irregular Nancy understands the pain just you know we didn't used to have
32. 5:32passwords and then when we had passwords you had one password for life it was like you had one password in a decade
33. 5:38and you just you put it on all your sites and then the computer guys were like not very secure and I'm like oh
34. 5:45what I suppose I can't have it on a Post-It note on my monitor too uh technically no and now we're at 17 Factor authentication thanks Geeks
35. 5:55I'd rather have all of my data stolen anyway back to what this channel is about yay you did it yay uh this is the AI
36. 6:08learning lab and my name is Kyle Shannon I'm an entrepreneur believe it or not I'm a CEO of an 11.5 year old company
37. 6:15called storyvine which is an automated video storytelling platform that believe it or not does not use AI
38. 6:22to automatically edit or as we call it auto magically edit videos um we we use a more traditional assembly
39. 6:28method uh but it's still automated and uh I've spent about the last two years deep diving into this generative
40. 6:36AI stuff and the image stuff and um I was an entrepreneur in the early days of the World Wide Web I co-founded
41. 6:42this this company called agency.com which was one of the very first digital agencies and holy [ __ ] are we right back
42. 6:50at the beginning of something and these little tools here so if you have not played with chat GPT if you are
43. 6:58a chat GPT newbie and you haven't done it and you're like I heard about it but I don't I'm I'm not too technical Kyle I
44. 7:04don't I don't know how to tell you this but uh I don't know much about the computers if that's you fine doesn't matter you're
45. 7:13cool just go there right now you know how to go to a URL right what's a URL it's a website oh a website yeah just go there
46. 7:24and that's chatgpt that's the official chat GPT website start playing just go there right now and start playing and if
47. 7:30you're like but Kyle I don't know what to say go there proms.chat problems.chat will teach you how to talk to that thing
48. 7:37oh and Sam I don't know if Sam's in here yet hey share bear retro Punk the whiners SNL throwback damn that takes me
49. 7:45back we're dug into any wine or we've got diverticulitis no [Laughter] um from stat chat will teach you how
50. 7:57these large language models work and how to talk to them if Sam's here I told Sam the other place you might want to start
51. 8:04so one place to start is chat GPT the other place is Claude dot dot a i you could go to bard.google.com but give
52. 8:12that one a week that one's going to get better soon claude.ai is also really good either of those top two are a place to start go
53. 8:20there now go do that now and then when we get rolling here uh I'll answer some questions if you have questions about AI
54. 8:27pop them in the comments below all those hearts that you're seeing those are the Irregulars they're people that show up
55. 8:32here every night they just harass me it's it's just a never-ending it's it's like reverse trolling they they compliment me they hit the
56. 8:43heart button and I know there's something more to it I know they've got an agenda so I just I don't trust it I
57. 8:49don't trust it I know there's an agenda not everything's a conspirac conspiracy but just because you're paranoid doesn't
58. 8:56mean they're not out to get you you know what I'm saying all right okay so let's get going here we're an odd lot we are not plot we are
59. 9:05the Irregulars we I am irregular I am a regular I am a regular I am a turtle um all right
60. 9:19let's see let's go ahead and system questions let's go do this oh and then I'll sort of dig through I'll I got a bunch of
61. 9:27[ __ ] I can show you here websites and things like that well bites for some more people to get rolling
62. 9:34evangelized ideogram to non-believers in the house mcallistered my 21 year old son now that see okay so what Ted Ted looks at life
63. 9:45is talking about there is is if you haven't played with these tools yet you owe it to yourself to play with them
64. 9:51long enough to have a Kevin McAllister moment which is the Home Alone moment where at some point it's gonna blow your [ __ ] mind
65. 9:58so I Ted's been in here a lot for for many weeks now and so his son knows he's been you know he's been on The Tick Tock
66. 10:07and he knows he's been talking about Ai and the sun I got 23 year old twin boys so I know how how much they think they know
67. 10:15and how big their opinions are so the fact that you Kevin mcallistered a 21 year old is quite impressive that's awesome that's great
68. 10:27yeah idea Graham is quite something it really is so awesome all right let's see y'all check my article on Discord and legal ethics
69. 10:39Channel about bart oh did you write that Luna stick I thought you just were sharing that that's super cool Apple just updated
70. 10:4917.0.1 I hope that dot one is making their voice synthesis better there's comic relief in that yeah what are they aggravating us
71. 11:07all right let's see there's a brand for that another handsome shirt well thank you it's a lovely checkered pattern it's classic yet modern
72. 11:19simple yet elegant thanks I went to the haberdasher I I also noticed I have some food right there so it's lovely and all I can see
73. 11:32is is that you know something just like [ __ ] foreign like the insides of the of the uh one of
74. 11:49those cuffs are a different fabric ah swell right all right AI learning lab and Fashion Center hair's not bad tonight all right how do
75. 12:04you get AI on computer I'm new to this okay cool so welcome um I would encourage you change your name from user whatever that number is to
76. 12:15like something clever the one of the things we do in here is we judge names and so yours right now um 0.5 out of ten but change it to
77. 12:24something clever like a like a nickname you want people to call you or something even more clever where they have to figure it out
78. 12:31um anyway so here's what you do you don't need to put AI on your computer it's these are just websites that you can just go play on so
79. 12:41chat.openai.com or Claude dot AI or bard.google.com just type it into your browser window and here let's go show
80. 12:50you let's go do it right now I'll go show you what I call parlor tricks there's some parlor tricks you can do right brother right brother
81. 13:01we'll share the chat JPay tag all right so when you get to chat GPT you you won't see all these things down here but down
82. 13:10the left hand side this is your prompt history so these are all your different conversations with it and then over here
83. 13:16to the right this is it and you're like but Kyle that doesn't look like very much whatever do I do here exactly so
84. 13:26um when you're when you first join here and you're not paying for the advanced version you just have access to one
85. 13:32model it's GPT 3.5 it's actually really good it's very fast these little things here are just suggestions so this says like suggest
86. 13:40some code names create a workout plan make a Content strategy design a date database schema those are just suggestions you can
87. 13:47um I wonder if you reload the the site if it changes them I never really paid attention yeah it does explain my popcorn pops so you could
88. 13:56just start clicking on those another thing you could do is um you know have it write you a country song this is this is my go-to
89. 14:07write a country song about a hedgehog lame named Larry um
90. 14:28and how he passed too soon but his younger brother Lenny still enjoys thinking about him
91. 14:45here we go kind of an inside joke from last night from Dolly three and then it'll just write you a song
92. 14:52so if you haven't done this before just go this is chat.openingi.com let's see if it did any good it's
93. 15:00generally pretty good we gather around folks I got a tale to share about a little hedgehog named Larry with spikes and hair
94. 15:08he rolled through life with a twinkle in his eye but one fateful night he had to Say Goodbye Oh Larry you left this all
95. 15:16too soon underneath that Silver Moon your memory lives on so um so you can do that um it can it can do useful stuff as well you could say
96. 15:25um let's see um write me a professional business email trying to get a job selling hedgehogs at the local pet store boom
97. 15:41there's your letter I hope this message finds you well oh this is good look at this um I'm writing to you to express my
98. 15:51strong interest in the sales position currently available at local pet store as a lifelong animal lover and pet owner
99. 15:58have developed a deep understanding and appreciation care for for well-being of animals including hedgehogs this is good
100. 16:04it's not just leading with I got a hedgehog Obsession it's saying including hedgehogs so we're kind of planting the
101. 16:11seed that we might be a little obsessed it's good it's salad I'm excited about the opportunity to join the team at
102. 16:17local pet store and share my knowledge and enthusiasm with customers looking to provide the best care for their pets
103. 16:23some key qualifications customer service Excellence passion for hedgehogs uh you can do that you could also say um
104. 16:36write me a python application that's a trivia game about hedgehogs mm-hmm sure and then it writes you code
105. 16:48so the the thing about chat GPT if like as I watched this screen when it writes code in particular but even even when it
106. 16:59just writes English or a business letter um like I've been deeply obsessed with this now since chat EVT came out and it still
107. 17:08blows me away every time I see it um that's real code it may or may not work but if it doesn't work you can put
108. 17:15the error messages in it'll tell you how to debug it so you can write code with it you can write words with it you can
109. 17:21write copy with it you can create meal plans with it so just start playing the most important thing to do right now is
110. 17:28don't be intimidated by it there's nothing to install you just go to these things and you start playing um and and that's that's really the
111. 17:37beginning and and basically you know my my My Hope and my advice is that you play with it long enough that it kind of breaks your brain a
112. 17:51little bit you have the Kevin McAllister moment you have this moment where you like wait it can do that and then
113. 18:01think about something that you're really good at or that you know really deeply and ask it about that and you'll
114. 18:07probably find that it's good at that too um I do find that if I don't do some serious prompt tweaking the responses are often very can
115. 18:18sounding yeah exactly that's very right well good they still require some work so gray-haired Gamers so here's the deal with this stuff
116. 18:26as remarkable as the tools are so first thing is gpt4 is way better than GPT 3.5 and in fact it's way better than essentially
117. 18:37everything out there until further notice apparently Gemini that's coming from Google is better than gpt4 but we haven't
118. 18:46we haven't heard anyone outside of Google say that yet so that could just be marketing [ __ ] um but gpt4 is really really good that
119. 18:56said all of these large language models the way they work they're almost the opposite of Google where with Google what what Google's
120. 19:05trained us to do over the past 20 odd years is put in as few words as possible just put in the minimum minimum viable prompt right
121. 19:15um you know I want a car wash that's open on Saturday you know like hope it doesn't get confused right
122. 19:22these large language models chat GPT and Bard and Claude all these things are the way they work that they require a
123. 19:34lot of context so and here's what that means if if you're working on the Great American novel one of the things you
124. 19:42could do is say you know you're you're a master um literary coach and you specialize in helping first-time authors you know make
125. 19:53the best novel that they can and then I'm working on a novel and you know it's about you know this basic topic and it's
126. 20:01a mystery and it's a this and it's a that you know can you help me structure my story and it will say sure here
127. 20:09here's a bunch of ways you can structure your story and then you can say well what are some other ways I could
128. 20:14structure my story and they would they'll throw that out there's six other ways you can structure your story and
129. 20:19then you can say oh I like number two and number four could you combine those two story Frameworks and give me a new
130. 20:26story framework and it'll do that and you're like oh that's perfect um you know here are are some of the
131. 20:33main characters of my story and it'll support you and tell you how awesome you are the the more context that you wrap
132. 20:40around that conversation the better it gets and so by the time you actually ask it to do something it it's kind of
133. 20:50I was going to say it's in the mindset it's not actually thinking um but it it it is in a it is in a more
134. 20:59narrowed focus in terms of the language world that it's drawing from and so it'll get better as you give it more
135. 21:06context so don't feel with chat GPT or with any of these tools like you have to get it perfect in a single prompt
136. 21:13in fact my experience has been just the opposite of that that it's that it's it's more like a conversation with a
137. 21:20really dim-witted intern who just happens to know everything right have you ever had an intern that they're just
138. 21:28like clueless about the world but they're really well educated that's chat GPT it's it's a simpleton that that knows [ __ ] everything
139. 21:38and so and so you have to coach it and guide it and get it there and you have to do that for every conversation you're
140. 21:43kind of starting this over it does not maintain memory across chats each chat is its own little Universe of language
141. 21:51the other thing you said there's another thing that that I think is an expectation thing and you really only
142. 21:57get this when you start playing with these things regularly the expectation that you'll put in a single prompt and it will spit out
143. 22:04something that's a hundred percent good to go is essentially zero it's essentially zero my experience has been
144. 22:15when I was weak at prompting it could quickly get me to like 60 to 75 percent as I got better at prompting it could
145. 22:23get me to like 75 to 85 percent and now if I really put time into the context setting I can get to like 85 90 the rest
146. 22:33of it's on me and and I I was I was working on an article for the past two days actually where I had a really cool idea
147. 22:42um I went into gpt4 I had to outline it for me I did two iterations in chat gpt4 then I went over to Claude and I had
148. 22:50Claude try to do it because claude's got a longer memory and I ended up getting so lost in it that now I've got I've kind of got exploded
149. 23:01piece parts of this idea like it kind of [ __ ] me up because I didn't just stop when it was like decent and actually
150. 23:09just write it on my own I tried to have the machine do it better and better and keep refining it and I kind of [ __ ] it
151. 23:15up so so sometimes you mess it up um but the important thing to do right now these things are evolving very very
152. 23:23quickly the important thing to do right now is um just learn them get curious about them kind of open your mind to what's
153. 23:31possible because they're going to get better and better and better let's see um thanks to you fast forward to thanks
154. 23:38to you fast forward two months I'm using regularly that's great you're utilizing regularly that's awesome thank you for sharing that
155. 23:52thank you Lauren I had a Kevin McAllister moment using Trace journey and Discord to create vectors oh what's Trace journey
156. 24:02is that is that an AI that's cool look at that that sounds like an AI because because it sounds like someone
157. 24:11took mid Journeys branding and they put Trace in front of it and said hey all you illustrator people Trace Journey Trace Journey
158. 24:21Trace Journey turn mid-journey images into vectors with just one click the bot replies to every [ __ ] mid-journey upscale with an
159. 24:31options menu holy [ __ ] we need to install this into um the AI salon whoa is this is it good Lauren this is
160. 24:45this is Kevin McAllister kind of territory how it works I don't need to know how it works um join Trace Journey welcome to trace Journey
161. 24:59um watch how it's done read the rules check the cool new features start here all right I can't I don't want to sort
162. 25:12of figure this thing out on the fly but holy crap that's cool this is so good all right Lauren see this is you see see
163. 25:24listen look look listen listen people listen listen listen zip it zip it zip it zip it zip it look um
164. 25:36see this see what that says AI learning lab we're learning Lauren just shared a thing called Trace Journey it looks flipping amazing I
165. 25:47can't believe I haven't heard of that I wonder how old it is um so so here's the deal when you generate images in mid-journey or in stable
166. 25:58diffusion for the most part they're very low resolution there's tools like topaz labs gigapixel and photo AI that will
167. 26:05upscale it for you or stable diffusion upscalers um but they're still bitmaps a vector file you can scale infinitely
168. 26:17and it will look amazing and so Vector Journey Vector Journey yeah um looks like it takes those images and makes them into
169. 26:27vectors that's pretty sadding that's pretty exciting I'm working on a project right now that that's actually really
170. 26:34important for so so thank you Lauren um okay um all right we'll go we'll go back to that I don't want to install that live but
171. 26:46that looks super cool as an irregular I have an extreme I have extreme enthusiasm at or Zeal and alacrity towards Turtles
172. 26:58now why might that be Mr Winston Minister Winston Hardy names Winston Winston Hardy Winston Hardy III and I like turtles okay look
173. 27:13watch for it watch for it this is for you Winston Hardy I hope your name's Winston Hardy III it would be so good Winston Winston Hardy
174. 27:27Winston Hardy III pleasure to make your acquaintance um why am I not going to YouTube do I have bad internet what's going on oh you know
175. 27:38what this is I know what this is so um Microsoft Edge is now crashing on me at least once a day I switched my
176. 27:50default browser to Microsoft Edge back in January because I got so excited about Chad JP Tay and I haven't switched
177. 27:57it back I I think I might switch it back it's just it's got too many memory leaks relaunching relaunching it's gonna be so good um
178. 28:09you know what we're gonna do here we've got 71 people in here let me see am I late how late am I hey Governor we just started really
179. 28:17um did you see the email that Microsoft text-to-speech and examples I just saw that right before or not right before I
180. 28:24got on but right before I went out earlier um they look decent I didn't have a chance to dig in it creates SVG files okay cool
181. 28:31what's it called again it's called Vector Journey um actually you know what I'm gonna go do um [Music]
182. 28:46I'll be right there hold please let's see at Ally um whoops
183. 29:04um and you take a look at Vector Journey and see if we can install the bot
184. 29:21um Vector images from mid Journey ones exclamation point all right I just delegated that to get that
185. 29:35installed over on the AI salon so let me show you a couple of things here eaten um so here's what I do if if you're new
186. 29:46to this AI stuff I'm going to show you I don't know three or four slides in a row or five I don't know that have a bunch
187. 29:53of websites on them if you're in a regular you can just start tapping the screen or you know go go make a coffee
188. 29:58or something you've seen this before but feel free to screenshot this these are I'm not affiliated with any of these
189. 30:05sites in any way these are the these are the sites that I think are worth you paying attention to right now if you're
190. 30:12brand brand new to this stuff you don't need anything more than what's on screen right here chat GPT maybe Claude or Claude maybe chat GPT
191. 30:22there is so much depth in all of these tools that you can you could spend the next two or three years exploring them
192. 30:29so so don't feel like you need to do all this stuff if you've been digging around for this stuff for a while
193. 30:35these are the these are the sites that I think are interesting and why so so chat GPT interesting for a number of reasons if
194. 30:44you pay the 20 bucks a month for the Plus subscription you get gpt4 you get plug-ins which some of them are
195. 30:51interesting you get what used to be called code interpreter it's now called Advanced Data analysis which is what it is
196. 31:00um and you get access to the new [ __ ] what's the new [ __ ] they have a new thing coming out called Dolly three
197. 31:07which looks to be a mid-journey killer so mid journey is one of those image generation things you're now going to be
198. 31:13able to generate images natively within chat gpt4 so that's a big deal and it's they're The Originals they're the one that started
199. 31:22all this madness um Claude is a bunch of ex-open AI people they started a company called anthropic Claude
200. 31:31you can put really long prompts in like you could put an entire 50 000 word novel into the prompt and ask it to
201. 31:40write an epilogue for the book and it'll just do it um it's it's really good it's really interesting um Bard is Google's entry into this
202. 31:49space Bard's got some amazing additions in the past few days where as long as you don't have a Google workspace account so basically if your
203. 32:03work pays for Google you're screwed if you have an individual Google account bard.google.com now has extensions where
204. 32:12you can search across like Google drive or all of your Gmail it's gonna get really interesting really
205. 32:20fast so these three are kind of the big three right now that I think are interesting prompts.chat will teach you
206. 32:24how to talk to them the Kids Table of these large language models these are still worth playing with
207. 32:32um you'll notice that Bing is grayed out that's because Bing is not only at the kids table but they had to turn their their chair away
208. 32:40from everyone else at the table because they've been a jerk lately Bing's been jerky and um but Bing is connected to the internet
209. 32:51it's oh cool Bing is connected to the internet you can make images there you can um analyze images there
210. 33:05it's gpt4 it's free and they're going to be incorporating Dolly three into that as well that's Microsoft's version of
211. 33:14chat GPT so Dolly three inside of Bing is going to be really interesting so that's worth playing with
212. 33:21perplexity.ai is good for doing research it does citations really well and then labs.perplexity.ai that's metaslama 2 models those are good
213. 33:31Pi dot AI is a company called inflection they just raised 1.3 billion dollars 1.3 billion dollars for a chat bot
214. 33:45they're spending about a billion of that on 22 000 h100 gpus if you don't know what that means they're buying a lot of machines
215. 33:57um and then you got poda oh the other thing about Pi is our nickname here in the lab for pi is chatty GPT chatty GPT
216. 34:05is really good if you if you download the PI app pi and then search for pi personal assistant you can put it into conversation mode
217. 34:15and just start talking to it so you're having a conversation with one of these GPT models connected to the internet
218. 34:21fascinating useful interesting Wild um Vector Journey or Trace Journey oh it might be Trace Journey sorry if I if I
219. 34:32lied hang on I think it's Trace Journey Trace Journey Trace Journey sorry for all the people that went Stevo oops he got that wrong
220. 34:41Stevo seriously that's the first time I've ever gotten anything wrong right back me up on this I never get anything
221. 34:47wrong right Stevo definitely have never made a mistake it's well except always listen it's called Trace Journey I don't know
222. 34:58which one of you Irregulars told me it was Vector Journey that's wrong it's just not right so I'll correct it I'll correct your
223. 35:07mistake I'll correct your mistake it's Trace Journey Trace Journey and let's see did I I called it Vector
224. 35:14journey in in Discord as well I'm a loser baby so why don't you sue me let's see uh Trace Journey Trace Journey all right there we go
225. 35:29Vector images from a mid Journey okay perfect thank you for that sorry about that oh that's so embarrassing um oh wait [Laughter]
226. 35:42okay we're here uh poe.com lets you play with a bunch of these different models uh open AI Claude some other ones and
227. 35:51it's got a bot maker you can make your own bot with poe.com so there's seven large language models right now that I think are worth playing
228. 36:00with I think Bard is gonna get how many times have you attended the AI learning lab oh that was a cool one
229. 36:10Edge is the worst Edge is actually really good except for the fact that it sucks just like Bing that's why Bing's at the
230. 36:18freaking kids table with this her chair her name's Sydney with her chair turned around with her back to the table okay
231. 36:25so those are the word ones these three are the top three these three are the kids table but still worth paying
232. 36:32attention to now the pictures pitches oh God I gotta add Trace Journey To This I Gotta play with Trace Journey first
233. 36:39but it looks [ __ ] amazing okay um so these things if if you have not spent time making pitches with these here large with these
234. 36:51here uh generative AI things they're quite remarkable so mid journey is the best one for now it's not gonna be the best for long but
235. 37:00but it's the best for now this list is Gonna Change pretty dramatically here relatively shortly I think I think Dolly three is going to
236. 37:08be at the top of it which considering how bad Dolly 2 is right now um is gonna it is like a minor miracle um ideogram is the one that
237. 37:21um lets you do pitches I mean words so like AI learning lab Tick Tock sent me to the bottom I'm not a robot goodbye
238. 37:31um AI learning lab all these things were made with ideogram ideogram dot a I so it's just like you know stable diffusion or any one of
239. 37:42those you know kind of image generation things but it's super simple and it can do text Dolly three that's coming out can also
240. 37:51do text and it looks like it does it better than this but this was not possible like a month ago and it is now so there you have that um
241. 38:02what else do we have here um leonardo.ai is a web-based uh interface for stable diffusion where you don't
242. 38:11have to install anything it's really really good hey uh let me skip hey Jen for a second Adobe Firefly and Photoshop generative
243. 38:19fill came out of beta about a week and a half ago and are now good good enough to use for government work like you can you
244. 38:28can make commercial work with them now you couldn't when they were in beta you can now prompt book will teach you how to talk
245. 38:36to these things and then I want to talk about hey Jen so hey Jen is a tool that's kind of like D Dash ID or
246. 38:42Synthesia where you can do text to voice and then you can animate those voices with a picture with like a person
247. 38:49talking you can make synthetic people but one of their Engineers kind of went rogue and did a little side project called video Translate
248. 39:00so if you go to labs.hagen.com click on video translate and upload a video of yourself speaking English and then
249. 39:13pick a language and wait for the obligatory 15 minutes for it to process and you'll get back something
250. 39:21that's kind of like this now what I did here was I took the English original video and the Italian that it translated
251. 39:28and I and I sort of intercut them back to back so I have this sort of ping pong between me speaking English and me speaking Italian
252. 39:36so we've heard from a number of clients that modular content is important to them getting more efficient with how we
253. 39:41create content and how we leverage content is behind me you see these little green boxes that's how we think about stories
254. 39:55at storyline we break stories down into little atoms right little modules foreign so yeah so if you noticed it's not only
255. 40:13synthesizing my voice it's matching my tone and Cadence it's syncing the lips up and it's reanimating my whole face uh
256. 40:25there's a little glitch like actually if you notice right there wait uh notice that my hand kind of disappears
257. 40:33underneath my face that's because they take the lower half of your face and and they're synthesizing it and they layer
258. 40:40it on top of the video so if your hand passes here you pass under your mouth which is kind of funny um they'll fix that at some point but
259. 40:48this is a remarkable tool um I don't think they know what they have on their hands here this will be the first of many but that's heygen.com
260. 40:58um so that's that one let's see how many times have you attended there you go there's a little Poll for y'all go take
261. 41:07that poll go do it wait oh I I would have done it but it went away um okay so um futurepedia.io the thing at
262. 41:19the bottom that's just a big old directory of AI sites um here's the thing I've been invited by another Creator
263. 41:28who's a creative coach and a voice over coach and she's got a whole community of folks over there to do a workshop I've
264. 41:35also been asked to do a workshop for the AI learning lab I have not put that together yet but I I am putting together
265. 41:41I have put together a workshop called the AI launch pad so this is um an AI workshop for creatives so it's a three-part Workshop over three
266. 41:53different weeks you know one day a week for three weeks starting October 4th so if you're interested in that
267. 41:59is today Friday yeah today's the last day she's got an early bird special there that ends today so if that's
268. 42:05something you're interested in that's that then the other thing is last December I started a group called the AI
269. 42:11salon and when I talk about the Discord and and people the Irregulars doing stuff in the Discord um it's Trace Journey not Vector Journey
270. 42:21J jrc is schooling me in the in the Discord right now you said it wrong it's not Vector Journey that doesn't even make sense um
271. 42:34uh so the AI Salon if you go there this is a it's a there's no charge for it it's we have meetups every other Tuesday
272. 42:43so this coming Tuesday is one of our salons and then we have a Discord Channel and all of the links for that
273. 42:49this is the link tree for that story of mine's my company and that's my that's my handle the other thing here is
274. 42:55um AI learning lab Dash TT that's the YouTube channel so these lives are archived there and we're using Claude to
275. 43:04write the YouTube descriptions and the chapterization that took me about 50 hours of prompting to get that right
276. 43:11um and then one final thing and then we'll get back to answering questions because I know you're all like but Kyle
277. 43:16I asked a bunch of questions holy crap there's there are 5 000 people in this live right now wait wait hold let me
278. 43:27time out time out time out hello my name is Kyle Shannon this is the AI learning lab normally we have between I don't
279. 43:35know 35 and 70 people here now there's about 6 000 people in here I don't know what [ __ ] algorithm we just tripped
280. 43:42on but welcome my name is Kyle Shannon this is the AI learning lab we talk about AI stuff in here we have fun I
281. 43:49like acting like an idiot I do occasionally cuss so if you don't like that I would jettison now um one of the questions you might be asking
282. 43:57and I get asked this a lot sir what are your qualifications I have none um I'm an entrepreneur I'm a co-founder
283. 44:06and CEO of a company called storyvine which is an automated video storytelling platform the I've been an entrepreneur for years
284. 44:14I've started I don't know 12 or 15 companies in the mid 90s I was a co-founder of a company called agency.com which was one of the very
285. 44:21first digital agencies and that was insane and I was lucky enough to be part of the very beginning of something
286. 44:31transformative like the World Wide Web at you know decently the ground level and and kind of rode that rocket ship up
287. 44:39we're at exactly the same spot right now almost eight thousand I don't there's there's eight point seven thousand
288. 44:46people here I have no idea what's going on Trump I have no idea we obviously so we we hit it we hit the algorithm um
289. 44:562023 is the new 1995. is I've got a video on that uh um it's actually one of my pinned videos I've got a two-part
290. 45:04video pinned to the top of the channel here um where I talk about how reminiscent what we're going through with generative
291. 45:12ai ai right now is to the early days of the World Wide Web so this channel is about that inflection point
292. 45:20that started on November 30th 2022 so 10 months ago which was the launch of chat GPT if you have not played with chat CPT don't go there go
293. 45:33hold please yeah let me make this bigger come on there we go to chat.openai.com that's the official chat website just go there you're gonna
294. 45:44love it it's gonna be fantastic you may not love it doesn't matter just go there and start playing and what can you do there
295. 45:51it's kind of like a Google search box just ask it to write you a country song ask it to write you a poem ask it to
296. 45:58write you a meal plan that only features grapes and kumquats and it will um so so just go start playing with it and
297. 46:09and one of the things we talk about in here you owe it to yourself to explore these large language models
298. 46:15long enough that you have what we call your Kevin McAllister moment which is this moment it can do that
299. 46:29um and and and just start playing with this stuff and then what this channel is about pop any comments you have down
300. 46:37below and I'll do my best to answer them normally we have about a hundred people in here or so right now we've got more
301. 46:45than eight thousand so I likely won't be able to get up to all the comments but but if you like what you're seeing here
302. 46:51you know tap the screen um feel free to ask a question I do have mods in here the the group that shows up
303. 46:58here every night so I do these things every night from like eight o'clock mountain time I'm in Denver until I
304. 47:04don't know when I get tired when my voice goes and my ass gets numb um and the Irregulars are people who show up regularly they will
305. 47:12um they will uh sort of pin any comments up here so anyway tell the people that they can oh yeah oh so another thing
306. 47:20okay so one of the things that you may be asking or as you're flipping through Tick Tock and and you see these annoying
307. 47:27people and they're like those are the people we make fun of in here why is that because while you can make money with chat EBT
308. 47:42is just a tool right so so the question isn't how do you make money with chat gbt the question is what's a problem you're trying to solve
309. 47:54that Chachi BT could help you solve faster better cheaper things like that all right so um so yeah so just go play and and you know
310. 48:05I I also feel like we're we're entering an era a very magical era era I'm my nickname for the era that we're entering
311. 48:13is the great Renaissance because um these tools really allow you to amplify what you're good at um I'm I am not a good drawer at all I'm
312. 48:25not a good illustrator I'm really creative I'm really visual I'm really good at recognizing good [ __ ] but I'm
313. 48:31not good at creating it well now I am um I was good with words now I'm better with words so so these tools really help
314. 48:39you amplify things that might be kind of tangential to your core skills and do it in a really fun way and so
315. 48:48um I think we're we're entering an era of a lot of uh reinvention where people are going to kind of ReDiscover their
316. 48:55passions and what they want and all that sort of stuff uh what's the link from the voiceover lady oh sorry about that so hang on a sec
317. 49:04so I mentioned I mentioned before to the folks before lots of folks came in um I'm putting together a workshop now
318. 49:11I've put together a workshop um that I'm that I am uh being hosted on another Creator's Channel um Rhonda Phillips she's a creative
319. 49:21coach and a voice over actress and she's got a whole Community there so this is a three-part workshop for creatives called
320. 49:28the AI launch pad and workshop one that starts on October 1st is just an overview of here's what's going on in a
321. 49:35AI here's what's possible Workshop 2 is applying AI to your business and workshop three is Advanced um kind of advanced stuff like voice
322. 49:46synthesis and animation and video and all that sort of stuff so it's going to be a really interesting three days it
323. 49:52happens over three weeks on I think Wednesday nights Rhonda's voice.com slash events is how you can find that so go check that out
324. 50:04um and again there's a lot of new people in here because there's significantly more people in here I'm a little caught
325. 50:12off guard because there's 8 000 people in here we normally have about a hundred sometimes 50 sometimes 30. um so welcome
326. 50:20if you're if you have not tried chat GPT go there right now go there now just you're like but you're an old man
327. 50:28talking about AI why should I trust you what are your qualifications sir I want to know I want to know just ignore all
328. 50:37that I have no qualifications I'm completely unqualified go to chat.openai.com and start playing if you
329. 50:43don't know what to say there if you're kind of confused about what these large language models or chat GPT is all
330. 50:49that sort of stuff that's chat EPT that's the official website go to that thing down there prompts.chat
331. 50:55that's a little primer that will tell you how these things work and it gives you like a whole bunch of prompts that
332. 51:02you can literally copy and paste into chat GPT and get it to do stuff um so just go start playing have fun and
333. 51:09what I'm going to do now is I'm going to answer some questions and then um yeah welcome everybody all right my
334. 51:16husband thinks I have lost my mind I drop everything to be a to be a regular elf I love this yeah I'm sorry I know
335. 51:24listen there it's it's like uh what is what did you do Friday night oh I don't want to talk about it why not I I don't
336. 51:33want to talk about it what did you do I kind of watched The Tick Tock Channel well that's okay do you watch it for a
337. 51:39few minutes no I watched it for a couple hours well what is it I don't know it's some old guy talking about AI
338. 51:48that's what this is so if you're young here um if if you're thinking of doing the troll or if you've already done it oh
339. 51:55isn't that cute the geriatrics teaching other each other how to use computers we've already heard that one you gotta
340. 52:00you you got to be more creative with your trolls than that so anyway okay let's let's get rolling and answer some
341. 52:07questions this is this is cool okay um let's see you're a great actor from what was that New Jack City I was in New Jack City yes
342. 52:20but I didn't The Tick Tock just jumped me away from your comments oh wait there it is it you're a great actor from New Jack
343. 52:30City I am I I still get checks in fact I think I have one here beside me I have a I have a residual check which
344. 52:39these things are collector's items this is a residual check from sag-aftra for New Jack City for 9.31 cents
345. 52:50this is back when they paid actors residuals I did that I filmed New Jack City well whatever the [ __ ] that was 19 90
346. 53:00one two I don't know I don't remember what it was I'm still getting residuals from that um my sis called me to help prep for a
347. 53:07job interview I told her to use pie oh that's a great idea that's a great idea curve Bruno so if you have not played with pie
348. 53:15good Lord there's 10 000 people in here hey everybody I'm a little shook it uh my name is Kyle Shannon this is the AI learning lab
349. 53:25welcome if you have any questions about AI pop them in the comments below what character did you play I played white boy number one
350. 53:32number two I was I was the first White Boy um that was my character name this was in a time where no one was triggered
351. 53:42this isn't it this is in an era of before people got triggered so you were allowed to write a character called white boy number one
352. 53:52um let's see what was I gonna show oh I was gonna go here um if you have not played with pi it's another interesting one to explore
353. 54:03so you can go to Pi dot AI you can go there or ideally go get the PI app so if you if you go to um the Apple store or Google Play search
354. 54:17for Pi Pi and then search personal assistant it's from a company called inflection so make sure you get the
355. 54:23right one and then it's like chatting with chat GPT just a different company but you can put it into talk mode
356. 54:30there's a little like phone icon hit the phone icon and then you just talk and you don't have to submit or anything you talk and
357. 54:37when you stop it listens and then answers you and you can pick one of six voices um it's been designed to be conversational
358. 54:47to be really interesting and good at conversation so go play with it because it's not you if if you've if you've been hanging
359. 54:56out on the channel regularly you'll you'll understand that that um my experience with Siri and and Siri lately
360. 55:04Siri has gotten a little jealous lately and she invades our conversation a lot um it's not like Siri or Alexa it's it's
361. 55:11just a very different thing because it's connected to the internet and it's chat gbt you can literally ask it anything
362. 55:17you can ask it to teach you stuff you can ask it to make up a game you can ask it about local events that night you can
363. 55:24ask it to coach you through something I had um my battery shorted out on my car one night and I'm like I I don't know
364. 55:30what to do I gotta do and I was like oh hey thank you very much let me try pie and and I I open this stupid chat bot
365. 55:38and I'm like hey I have this Ford Flex and it's acting kind of funny what could this be and I said well let me
366. 55:45troubleshoot with you and it asks me a bunch of questions and it gave me two or three different possibilities and I
367. 55:51ended up troubleshooting you know that I had a dead battery with it um but but it's just wild so so go play with that go try that okay
368. 56:0010K with mostly Vietnamese names and comments the same wait and comments same as an irregular day every everything in
369. 56:08this channel is irregular I don't know what happened but I'll take I didn't listen whether you're from we have
370. 56:14people from all over the uh all over the world come in here so we we regularly have folks from Australia from from
371. 56:21Singapore from the Philippines from Europe so welcome one welcome all something amiss there might be something to miss I don't know we we
372. 56:33got hacked in a positive way I'll tell you a funny thing so so in the early days I was I was messing not in
373. 56:40the early days of crypto but about three years ago um I was trying to I was trying to get into crypto and I I bought some
374. 56:48insurance crypto coin that some friend of mine said oh you should buy this you'll make a fortune and and I invested
375. 56:54150 bucks in it and then I parlayed that 150 bucks into like 75 bucks six months later so so so I I heard of another crypto coin so I
376. 57:06took my 75 bucks and I and I bought this other crypto with it and then it at some point it said you know claim your uni
377. 57:14here and so I clicked this pink button and then I claimed my uni I didn't know what it I didn't know what I was doing
378. 57:20and then I went back into my wallet and the balance in my wallet went from seventy five dollars to twelve thousand dollars
379. 57:29and I had been airdropped 400 uni coins from uni swap when I bought that original thing um and I thought I was hacked but I'm
380. 57:39like but but it but I've got too much money I was like I don't this doesn't make any sense to me but it was an
381. 57:45airdrop and I sort of learned about crypto and how stupid and weird it is um I typed fluent via uh they're all
382. 57:51here to learn about AI from an expert well what are your qualifications sir what are your qualifications
383. 57:59um I don't think hey gen does Vietnamese yet um the the only thing even close to an Eastern language is uh is uh um
384. 58:09is Indian it does Hindi so but it doesn't do Japanese Chinese um Vietnamese any of that stuff yet but it will it'll get there
385. 58:18um I added viewers to you because I like you and your AI evangelism oh was that you James come out did you do this are
386. 58:27you responsible for this well you're freaking awesome that's that's awesome thank you very much I appreciate it um I'm I'm
387. 58:38a little I uh uh this has never happened that's really awesome so thank you for that um okay let's let's answer some more
388. 58:47questions uh it's crowded in here um you're standing on my foot oh man hey all your ten thousand go read my stuff uh on AI on medium
389. 58:59um yeah go read cabrino's thing if you if you uh let's see I'm trying to think what might be good to show here
390. 59:08there's some new people here so I'm just gonna I'm gonna keep doing this I'm gonna keep these things on screen right
391. 59:13now if you're brand new to AI if you're trying to figure out like where do I start with this what do I do I don't
392. 59:20know what to do I'm very confused AI seems very exciting everyone's talking about it what is it if you're anywhere in that neighborhood
393. 59:29cool welcome go to chat.openai or Claude dot AI either one of those top two sites the top one is the official chat GPT
394. 59:42website I'm not affiliated with it in any way I'm not making any money off this I don't I don't care um
395. 59:48go to go to you know chat.openai.com or go to cloud.ai and just start interacting with them just start ask it
396. 59:55to write you a song Write ask it to write a poem ask it to write a marketing plan ask it to write some HTML for your
397. 1:00:03website or python have it write you a python application it'll just do it in fact Let me let me go let me go do some
398. 1:00:10a little bit of what I call parlor tricks here let's jump over and do that because we have a bunch of new people here so
399. 1:00:17one of one of the things we like to do here to AI learning lab is deliver some value we we like to deliver value for
400. 1:00:24money and you're like but Kyle you don't charge anything exactly exactly so we'd like to give you at least zero dollars
401. 1:00:30in value all right so we're gonna go to chat GPT if you've never been to it it kind of looks like this when you
402. 1:00:41start you say log in am I in Chrome oh no I relaunched I relaunched Edge that's what's going on oh this is a joy
403. 1:00:54Google's asking me to re-log into everything hold please um oh
404. 1:01:11you probably just watched me type my password there's all my Security's gone hey way to go how'd you get hacked uh well I was
405. 1:01:22an influencer for for a while there before they took everything Google's new 17-factor authentication okay
406. 1:01:31as I was saying so so this is chat GPT if you've never seen it before the interface is relatively simple down the
407. 1:01:38left hand side this is kind of your prompt history so this is your different conversations this is the interface
408. 1:01:45um I have this little plus icon here because I paid 20 bucks a month for it um if you don't pay the 20 bucks a month
409. 1:01:52that's fine you get uh chat GPT 3.5 which is still really good it's very fast um and then it's basically just kind of
410. 1:02:00like a Google search box but you can do kind of anything in here so you could say to it um let's see uh oh did huh I think Siri might have
411. 1:02:15taken over my yeah it did see I told you Siri was gonna be a pain in the ass here uh did uh let's see Escape hold
412. 1:02:32all right I'll just type it uh write me a country song about um a big crowd uh crowd
413. 1:02:48showing up to my party welcome everybody there's 11 000 people in here um so so it just wrote a song it like
414. 1:02:59you see how fast it's writing now um it wrote a long one too um it it understands what songs are so here's a verse here's a pre-chorus
415. 1:03:09here's a chorus um there's verse two so let's see it's a country song well I sent out the invites on a hot summer day said y'all come over
416. 1:03:18we're gonna rock it this way I cleaned up the barn I don't know I need to be doing this with a Southern accent a
417. 1:03:25cleaned up Barn put some hay on the floor strung up some lights and unlocked the door um so it can rhyme it can write songs
418. 1:03:33um I could also say um uh what did I do earlier let's see write me a business letter that applies for a job organizing
419. 1:03:44parties that consistently get more people than were invited and there there's your professional business letter boom then I could also say
420. 1:03:58um write me a python application that will allow me to manage all of the people that show up to my party hmm
421. 1:04:10it'll describe it it'll then write the code you can then take this code and throw it into replit or some sort of
422. 1:04:18program like that so if you haven't played with chat GPT just play with it long enough until you have what we call your Kevin McAllister
423. 1:04:29moment which is this moment from Home Alone so so um it it will at some point blow your mind and and if if you haven't seen the The
424. 1:04:41Matrix movies The Red Pill blue pill moment the blue pill moment is is you take the blue pill and everything stays
425. 1:04:48the same I feel like this AI stuff's like that there's a Tipping Point where you realize that these these tools like
426. 1:04:55chat GPT and and the generative image tools and things like that are significantly more powerful than you had
427. 1:05:02any [ __ ] clue about and and and it happens to a lot of the people that are in here sending 5K thank you dunkaroo
428. 1:05:11how do you send 5 000 people oh wait now we're down to three thousand so so 5500 or so just evaporated or 6 500 just
429. 1:05:24evaporated I don't know what's going on anyway today today's been a really weird um it's been a weird day where where it
430. 1:05:35kind of started out in this sort of off kind of weird energy but then we had a bunch of really cool things happen I'm
431. 1:05:42going to show you something here since we have so many people in here um one of the one of the sites that I have
432. 1:05:50this list this might be something worth screenshotting is this is a list of the image generation tools that I think are
433. 1:05:58worth your time to play with again I'm not affiliated with any of these so mid-journey and ideogram are the two
434. 1:06:05that are kind of the biggies right right now um along with stable diffusion so Leonardo dot AI kind of represents the
435. 1:06:12stable diffusion world and then um Adobe Firefly and Photoshop generative fill down below there all of those are
436. 1:06:21are really good ones this this this this tool hey gen right here um really interesting company they spun this company up I think with seed
437. 1:06:31funding and like 50 people one of their Engineers did a side project called Hey gen video translate and hey gen video translate does this
438. 1:06:47so we've heard from a number of clients that modular content is important to them getting more efficient with how we
439. 1:06:52create content and how we leverage contents behind me you see these little green boxes um that's how we think about
440. 1:07:06stories at storyline we break stories down into little atoms right little modules
441. 1:07:20so so how you do this I'll show you how to do it I can't I can't people it's the future is here the future's here nothing is
442. 1:07:33sci-fi anymore except time travel I mean honest to God like we have we've had the pandemic and and now we have talking
443. 1:07:41robots and and they do stuff for us and other now we've got the Babel Fish thing so um let me show you how this thing works
444. 1:07:52so the the company is called hey Jen h-e-y-g-e-n and then you go to oh you go to labs.hagen.com
445. 1:08:10and then one of the buttons in there is video Translate and um you basically just drop a video file I can't start the poll because a comment
446. 1:08:25is pinned so unpin the comment and try that again there we go um so let me sign in here hold please uh
447. 1:08:40so basically you just you just upload a an English file or whatever file whatever your native language is and
448. 1:08:47then you select from let me see how many they have oh they've added some Mandarin Portuguese Korean if we have a bunch of Vietnamese
449. 1:08:58folks in here they don't have Vietnamese but how about Mandarin or Japanese let's have me should I do Japanese or Korean
450. 1:09:07what's the vote or Turkish or Mandarin Mandarin Japanese or Korean I feel I feel like me talking any one of those is gonna be hilarious
451. 1:09:18so so let's go ahead and do this let me go grab where is it hold please let me go find a file so downloads those are those
452. 1:09:33okay here we go um so for the meeting on the 23rd here's what it's going to look like is that that's video four yeah I think
453. 1:09:45that's one I haven't done yet okay so what we're gonna do and close this we're gonna close this we can close this
454. 1:09:53what's the vote Mandarin all right we'll do mandarin so I'm just going to drag my file there I'm going to select Mandarin
455. 1:10:02this should be a trip you can choose to translate Audio Only they give you two credits um for free and you can do a video of of up to
456. 1:10:16um you can do a video of up to five minutes long um that's good all right and so it's going to upload it
457. 1:10:26there and that'll process for a bit so I'll keep checking back on that so so pretty soon we'll we'll have uh we'll
458. 1:10:33have this this guy talking Mandarin um so yeah anyway if you're new here and and a lot of people are new here there's 6 000
459. 1:10:42folks in here right now it keeps oscillating between between um what's it say in there upgrade to skip the line estimated position 79
460. 1:10:54000. so so um the uh how do I get back to the home page here I guess I just go to hey Jan hey Jen
461. 1:11:12there's pricing that's not the lab's pricing so I think it's about 60 bucks a month for 30 credits
462. 1:11:26up to five minutes each and then it's like 179 bucks a month for um for 60 credits up to 20 minutes each so
463. 1:11:38it's not super cheap but considering what it does the fact that it does you know translation at this level and
464. 1:11:46that's how we capture the content but because we have them in this organized structured way that approach can be used
465. 1:11:51for everything from how you capture original content with traditional video production to how you leverage these
466. 1:11:57assets to create social media content things like that because
467. 1:12:14social media right so so that's what that thing does so so considering the quality that that it's generating that
468. 1:12:21kind of content with uh it's it's pretty remarkable um the fact that I'm 79 000 in line I don't think we're going to be able to
469. 1:12:29show we're going to be able to show um me speaking Chinese anytime tonight let's see labs labs dot hey Jen let's see how fast it's
470. 1:12:43moving it's funny I watched it live with the with the co-founder yeah I'm still 75 79 000th in line I watched it live
471. 1:12:51with the co-founder he basically said their lead engineer all he's spending his time doing is oh thank you very much
472. 1:12:57for the what is that that's a cute little heart a little heart and a hand it's so cute we're back up to 10 000 folks welcome
473. 1:13:09everybody my name is Kyle I can't even speak my name anymore normally in these lives we have I don't know on a good
474. 1:13:17night 150 200 people and all of a sudden there's 10 000 people in here and it's been fluctuating wildly so I assume yay
475. 1:13:24thank you very much I appreciate that Ted looks at life um so if you're new here um this is all about exploring generative AI um I'm an
476. 1:13:38entrepreneur I've started 12 or 15 companies um if you've not played with chat GPT go check out that chat.openai.com that's the official
477. 1:13:50chatgpt website if you haven't played with it just go there just go there now if you don't know what to do if you
478. 1:13:55don't know what to say go to prompts.chat um that's a little primer that will tell you how to talk to these large language
479. 1:14:02models it's quite remarkable um it will blow your mind in fact one of the things we talk about in here is you
480. 1:14:09should play with chat jpt until you have your Kevin McAllister moment which is if you remember the movie Home Alone that moment
481. 1:14:19uh at some point when you're using these generative AI tools something will snap inside you and you'll be like oh my oh I
482. 1:14:29didn't know I could do that and then you're like but it can't do what I do it certainly can't do what I do and then it
483. 1:14:35does what you do and then you go oh my this is something different and so these things are something different these
484. 1:14:41three in particular are my top three this is sort of what I call the adult table um of of the large language models
485. 1:14:48Chachi BT Claude and Bard Bard just got dramatically better it got this thing called extensions which is really
486. 1:14:57interesting it's about to get a major upgrade with something called Gemini so so all three of these are kind of in a
487. 1:15:04war right now for being the best and then there's another group here that I call the kids table that includes Bing
488. 1:15:11perplexity.ai pi and po which are also interesting to play with and then um you've got oh thank you very much
489. 1:15:19Winston I appreciate that Fireballs are plenty dang Winston you're the best I appreciate that thank you all very much
490. 1:15:27um that's really awesome um chat GPT is my assistant I call her Cynthia that's awesome in fact um what's your name the The Velvet The
491. 1:15:37Velvet emeralds um I gotta say so one of the other things we do here in the AI learning lab is is we we rate names The Velvet
492. 1:15:45Emerald 10 out of 10 solid name don't come in here with some half-assed unthought up thought uh thought of name
493. 1:15:51you come in here with a name like the Velvet Emerald is awesome 10 out of 10 on the name um what she talked about there is you know
494. 1:16:01Chachi PT is her friend and she talks to this thing um a really good way to think about how to interact with chat GPT is much less
495. 1:16:10like you've got to write the perfect prompt and get the perfect answer but much more like it's a conversation you'll notice that
496. 1:16:18Chachi PT is it's funny one of the images we have for this for the learning lab is let's see I think it's the other direction hold there
497. 1:16:33is is this little this little yellow lab or like a golden retriever Chachi PT is very much like a golden retriever it's
498. 1:16:41just always there for you it's very supportive you know you you ask it to do something it's like sure here's this and
499. 1:16:47then you tell it no that's not very good and it's like oh I'm sorry I apologize that wasn't what you wanted and it'll
500. 1:16:52try again and it'll just keep trying again and the more you go back and forth with it the better it seems to get and
501. 1:16:58and the nature of how this technology Works um is is one of the reasons that happens in fact what what might be kind of fun
502. 1:17:07here if you're down for it I guess I don't know how to do this if you're down for it um tap the screen make the hearts like
503. 1:17:14really fly I if you want me to show you how Chachi PT actually works it's a bit of a rabbit hole but in like the next
504. 1:17:23five minutes I can teach you how Chachi PT actually works is that something you'd be interested in is that a good thumbs up good yes
505. 1:17:30all right I'm I seem to be getting some I wish he'd say llama come on llama llama llama I like llamas llamas is
506. 1:17:41tasty I like Larry and I like Lenny and I like Lucy and I like llamas yes Winston with the Llama thank you very
507. 1:17:51much we also I'm also I am also a pavlovian uh creature where if if someone puts a cowboy hat on me I just talk stupid [Laughter]
508. 1:18:02but but let me let me show you something here is this yes what I'm looking for it is all right and is this now in a window yes it is oh
509. 1:18:18everything's coming together everything's coming up roses all right so we'll leave the little AI learning lab in the background because
510. 1:18:28this this is kind of fun okay so how does chat GPT work it works like this we'll start with some definitions so
511. 1:18:36what is what what is the GPT in in chat GPT um the G stands for generative the P stands for pre-trained the T stands for
512. 1:18:49Transformer what does all of that mean doesn't it actually doesn't mean much I'll try to try to break it break it
513. 1:18:56down generative means that these tools are actually generating original content one of the misconceptions about them is
514. 1:19:03that They're copying and pasting from copywritten work it's not how they work but but just know that it's generating
515. 1:19:09content the pre-trained means these machine learning models have been pre-trained on um essentially as much of the internet
516. 1:19:18as openai could get their hands on or or these other companies so so it's basically you don't have to bring your
517. 1:19:24own data it's already been trained on as much data as they could get into the thing to train it and then the t is it stands for
518. 1:19:32Transformer that's a piece of technology that Google invented in 2017 that is the thing that allowed these tools to be
519. 1:19:40trained on these massive data sets like all of the internet okay so that's what GPT means um you might also hear the phrase llm or
520. 1:19:49large language model that just chat EPT is a large language model we human beings are a large language model
521. 1:19:55everything goes in we process it it's all words that's how we make meaning out of stuff these are modeled on how our
522. 1:20:02brains are connected and how they work and that's why they call them large language models there's a thing called the latent space
523. 1:20:08so this is now we're getting into some of the geeky stuff and this is where it gets kind of fun so what you're looking at here to the to
524. 1:20:16the right this big white box with all the scattered dots on it is is a 2d representation of mathematical Vector space so you have an
525. 1:20:26X and A Y and then you have all these dots that are clustered together well what are those well each dot represents
526. 1:20:32a vector which is just a mathematical number that's tied to a word and then they're clustered together by
527. 1:20:38semantic meaning so I'll give you a specific example we have the dog up here right you might have one of those
528. 1:20:43clusters of words that has the word dog in it and that's like those you know fluffy things that we pet and you know
529. 1:20:50love us every day but you might have the word dog in another cluster where it's like the guy that did his girl wrong
530. 1:20:56he's a dog right so so individual words can have lots of different meanings so these clusters of meanings are in this
531. 1:21:03in this mathematical Vector space this is dramatically oversimplified right this is 2D space with I don't know
532. 1:21:11whatever that is you know 500 dots or a thousand dots the actual Vector space is a thousand Dimensions or or just a ridiculous
533. 1:21:20amount of Dimensions it's hard to comprehend and it's billions of words or trillions of words and all these
534. 1:21:25different things so how does it actually work it works like this and this is the more you learn about this the more remarkable
535. 1:21:33it is that these things do what they do when you type a prompt into chat GPT it's actually predicting what the next
536. 1:21:42most probable word is in the response over and over and over and over and so so here's an example so if I type in a
537. 1:21:51prompt that says the quick brown fox jumped over the lazy blank as humans if you've heard that phrase before we know
538. 1:21:59that the answer is dog because our large language model here's that you know quick brown fox jumps over the lazy
539. 1:22:05blank and we do pattern recognition and dog pops into our head and we know it's dog well the large language models do
540. 1:22:12something similar what they do is they kind of um they they ignore all the stuff that's not you know maybe animals or whatever
541. 1:22:20might be in that Vector space so some of these vectors have a higher probability than the others and within this semantic
542. 1:22:28cluster there's one of them that has the highest probability and if you look at what that is it's actually just a big
543. 1:22:35long string of numbers which is it's it's location and vector space and that long string of numbers is tied to the
544. 1:22:42word dog there's a table somewhere that connects those two and then it sticks the word dog into that sentence there
545. 1:22:49are some other words it could have been but dog is the one it chose and so when you watch chat GPT right
546. 1:22:56write you a whole long business letter or write all this code or write all of this stuff that's what it's doing for
547. 1:23:03you know every single word or in some cases it might be like little small groups of words but it's doing an
548. 1:23:10original word you know one at a time when it writes all that stuff so that's how the things work this is also why one of the things that
549. 1:23:22um people will sort of give chat GPT a bad time because it hallucinates it it lies right it you ask it to give you a
550. 1:23:33citation and it gives you a citation and then that citation is fake and you're like wait what happened why is it not
551. 1:23:39right why did it lie to me is it evil no no it's it doesn't know what it wrote it has the cognitive um awareness of a calculator because
552. 1:23:52it's technically a calculator you type in a prompt it turns it into a mathematical probability and then that
553. 1:24:00number in Vector space is tied to a word and it puts that word in there it doesn't know what that word was
554. 1:24:05and so it's literally just typing a bunch of words we as human beings look at what it writes and say what it's
555. 1:24:12writing is so human that it must be sentient it must know what it's doing it knows probabilistically what it's doing but it doesn't have any
556. 1:24:21consciousness of what it said and that's why it will sometimes predict something that looks correct but it's not it's
557. 1:24:28also the reason that that chat in these tools the more context that you give them the better it gets so anyway
558. 1:24:37um since we have a lot a lot a lot of new people here welcome my name is Kyle Shannon this is the AI learning lab if
559. 1:24:44you haven't subscribed to the Channel please go subscribe to the channel um and and uh if if you like it great you
560. 1:24:53know follow the channel I mean um I try to do these lives every night you can ask the the uh the Irregulars the
561. 1:25:00people that are here on a nightly basis um I haven't missed a night in I think three months or something like that much to the much to the
562. 1:25:09um consternation of my family sometimes um but they know this is important and I think this is important and I come in here to
563. 1:25:17um just try to demystify this stuff I think this is we're living in an incredibly rare moment in history where we're at the very beginning
564. 1:25:27of a a transformative technology that's going to change almost everything about how we work and even what work is and
565. 1:25:37where the value and work is all that sort of stuff and so this channel is really just about demystifying it
566. 1:25:42talking about it answering questions about it things like that ion China says maybe sentient intelligence is a form of
567. 1:25:49real-time statistical inference well so that's that's actually a really good point ion China that that um
568. 1:25:58one of the things that happens when when people are confronted with chat GPT and and they're like well I can't do what I
569. 1:26:06do they get very defensive I can't do what I do and and um it does it's very possible that the way our brains work are kind of a more
570. 1:26:18sophisticated version of what these things are doing and and scientists don't know enough about how they do what
571. 1:26:24they do that maybe our brains are a simplified version of what this is doing I think it's the other way around
572. 1:26:31um but but these things are going to get incredibly sophisticated thank you for that um and uh yeah it's it's a really
573. 1:26:38interesting point I mean his wife uh I don't think they have any all irises I don't know what that's in reference to
574. 1:26:45um I don't think they have all irises wow the chat feed has slowed down again no there's there's 13 000 people in here
575. 1:26:55welcome everyone my name is Kyle Shannon this is the AI learning lab and this is the AI learning lab um
576. 1:27:03if uh if you're wondering what are my qualifications I'm an entrepreneur and um and love just hanging out and talking about this AI stuff
577. 1:27:14um and do it every night thank you very much Aaron I appreciate that um Let me let me pop down here and and
578. 1:27:20see some questions I didn't bring enough gum for the whole class yeah exactly Ted looks at life he brings Chiclets do you
579. 1:27:27remember Chiclets Ted he's a he's an older fella like myself he brings Chiclets every night for all the people
580. 1:27:33that hang out here at the AI learning lab in the in the nightly lives but he only brings like a hundred Chiclets and
581. 1:27:40he's normally got more than enough so I don't know what's happening but uh the numbers fluctuating a lot so I have no
582. 1:27:47idea here's another possibility like I haven't really looked at the actual comments it's possible not that many
583. 1:27:54people are actually watching and there's some sort of tick tock glitch but welcome if you're new here um let's see hi from Texas hello Miss
584. 1:28:06shalom Michelle M Michelle M oh Michelle M Michelin Michelle Lim Michelle hi from Texas hello from Denver how are you
585. 1:28:20all right it's 10 p.m Friday night beautiful Denver Colorado welcome to Denver Colorado the home of the Rockies the home of the nuggets
586. 1:28:31the home of joggers and bikers everywhere what time we got 1005. all right we're still good we're still good here's a little poll
587. 1:28:43I'll just pop a poll up there go ahead and take that poll I don't know if those things are working or how they work
588. 1:28:53um cool say it again in German uh let me go see what's going on with the the one I did in Chinese um reload oh it's really slow so I was in
589. 1:29:0979204 and now I'm in or at 2 23 or something like that now I'm at 204. so this is probably going to be overnight um
590. 1:29:19if if you're new here there's a there's a tool called hey Jen um that has a video Translate um beta like a proof of concept
591. 1:29:33that is quite remarkable here I'll show you so this is this is a video of me from a video I did last February and I intercut this with
592. 1:29:47um me speaking English followed by me speaking Italian this is from hey Jam Labs so we've heard from a number of
593. 1:29:55clients that modular content is important to them getting more efficient with how we create content and how we leverage content
594. 1:30:03modular important behind me you see these little green boxes um that's how we think about stories at storyline we break stories
595. 1:30:15down into little atoms right little modules I mean that's how we capture the content
596. 1:30:29but because we have them in this organized structured way that approach can be used for everything from how you
597. 1:30:35so anyway so that's quite remarkable anyway they're they're their servers are very far backed up and again the company here is hey Jen
598. 1:30:45hang on a sec [Music] um that that company right there hey Jen is the one that's got this video Translate
599. 1:31:01labs.hagen.com slash video Dash translate is how you get to it um it's looking right now like the the the backlog of free
600. 1:31:17um free rendering is so far backed up that it it's like barely moving so my instinct is if if you want anything
601. 1:31:25turned around decently quickly you're probably gonna have to pay for a subscription um it for what it does it's it's pretty
602. 1:31:35remarkable it's not that expensive but it ain't cheap it's like I think the the cheapest uh subscription for the
603. 1:31:41translate tool is 60 bucks a month um so so anyway that's that's what that one is um that's that's what did that
604. 1:31:49Italian thing that I just showed you there um dude you usually only have 60 people in here thousands and thousands congrats
605. 1:31:55I love re-watching these things oh my goodness there's 14 000 people here welcome everyone my name is Kyle Shannon
606. 1:32:02this is the AI learning lab um I have no idea what's going on tonight I don't know who's butt we kissed or what algorithm we tripped over
607. 1:32:12or someone came in and said they sent 5 000 people our way I didn't know that was a possibility maybe it is I do not
608. 1:32:19know but we have a lot of folks here tonight um I'm an entrepreneur let me flip this this way I'm an entrepreneur my current gig is
609. 1:32:31I'm the co-founder and CEO of a company called storyvine which is not an AI company although we are completely re
610. 1:32:39um architecting our product Vision with AI um and that's really exciting we've been around for 11 and a half years we do a lot of work in
611. 1:32:49um in healthcare and Pharma Alexander Wright says that Italian thing was awesome isn't that amazing I mean it's amazing it's not
612. 1:32:58my friend ritesh Patel he's a he's a um he's an agency guy he I've worked with him for years he and I worked together
613. 1:33:05back at agency.com in the early days of the web and and now he's doing a lot of healthcare marketing stuff
614. 1:33:11um and he's been doing some of these hey gen videos and uploading them to LinkedIn and it's like every time I see
615. 1:33:17one of his videos he does it in a different language he did one in Chinese he did one in Spanish um and at the beginning of it he he
616. 1:33:25um he says my name he says you know my my friend Kyle Shannon told me about this thing and so every time it's in
617. 1:33:31another language I hear my name I hear ritesh say my name in some other language it's just a trip it's it's just a trip so
618. 1:33:40um that's what's going on with this stuff oh it's you buddy hey James you're the one doing this I don't know how you're doing this but
619. 1:33:48this is awesome thank you very much um if you're new here um I am normally not flustered I am normally just a a smarmy smart ass uh
620. 1:34:00that that acts like he has it all together but not tonight it's just crazy you people are just nuts you're just
621. 1:34:07nuts I'm telling you you're crazy all of you from one to the other to the other anyway um if you haven't played with chat GPT
622. 1:34:16please go there right now chat.openai.com and if you're like oh is he just making money off chat GPT is this just him
623. 1:34:27making a commission no I'm not I do this for free I'm here every night ask the Irregulars the ones the ones that are
624. 1:34:34tapping the screen and and answering other people's comments below and pinning the the questions and modding here
625. 1:34:42um I come here every night these are really just here more as a public service announcement the whole idea here
626. 1:34:48is to demystify these things so go to chat GPT or go to Claude or even Bard all three of these are worth you playing
627. 1:34:57with long enough until you have a moment where you realize oh my God these are different something's different here these are not
628. 1:35:10just Fancy Google they're they're deeply profound they're going to change a lot we are about to get a series of major
629. 1:35:19announcements over the next two months that these things are even going to get better and better and better um November 1st
630. 1:35:28um is General availability for Microsoft's co-pilot being available within office 365. what that basically
631. 1:35:38means is that chat CPT will be built into all of the office suite within Microsoft products Microsoft Word slides
632. 1:35:47Excel all of that stuff Google announced the same thing for their Office Products two weeks ago so these tools right now are kind of off
633. 1:35:59to the side like you go to chat GPT you go to that thing they're very very very quickly going to be integrated in all of the tools we're
634. 1:36:08already using so if if you are not starting to get curious about these and get up to speed and understand what they
635. 1:36:17do the time is now so so here's what I'm going to do since since there's so many new people keep seem to be rolling in
636. 1:36:25every every few minutes um I'm gonna just go play with okay so this is this is chat GPT down the side here you've got this is
637. 1:36:36this is my chat history this is conversations I've had with this thing and and oh thank you very much that is a lovely
638. 1:36:46the Stars the Stars they were falling from the sky was remarkable um so in fact form a union yeah exactly we're gonna we're gonna need to do
639. 1:36:57something because work is about to change pretty dramatically um okay so um there's a couple of different modes of
640. 1:37:08chat GPT if you pay the 20 bucks a month for their what's called their Plus subscription you get gpt4 you also get
641. 1:37:16plug-ins and you get Advanced Data analysis Advanced Data analysis is absolutely [ __ ] mind-blowing um I'll I'll show that because there's
642. 1:37:27something there we may we may lose a lot of viewers but that's okay because I if I were you I'd hang around and just see
643. 1:37:33this if you haven't played with chat gbt yet um that's if you pay the 20 bucks even if you don't pay the 20 bucks
644. 1:37:40GPT 3.5 is is really really good so it's just like a Google search bar it's just a bar you can type anything into it so
645. 1:37:48we're going to type into it um let's see um write me a social media marketing plan for my car wash business called shiny turtle
646. 1:38:03now I didn't really give it any context whatsoever but there's a marketing plan and you're like but Kyle it's writing it
647. 1:38:10so fast it can't possibly be good and then we'll scroll to the top here and let's go look at some things all right creating a social media
648. 1:38:20marketing plan for your car wash business shiny turtle can help you increase brand awareness so step one start with your goals and
649. 1:38:27objectives great so it gives us some goals and objectives we may or may not like those know your target audience
650. 1:38:33these are right choose your social media platforms create compelling content um develop a Content calendar and get
651. 1:38:40calendar engage with your audience right so good basic stuff now if you're like well this seems like a little obvious in
652. 1:38:46this and that the only context I gave it was write me a social media marketing plan if I gave it a lot more context
653. 1:38:52like I said hey you're a social media master and you you excel in helping small businesses drive more traffic to
654. 1:39:01their business and my business is called shiny turtle and we're we specialize in this kind of thing and that kind of thing
655. 1:39:11um oh I I can't uh I can't start the poll because the uh whatever you call it someone a message is pinned so unpin the
656. 1:39:21message and then I can start it there you go um so this this would get much much better over time and so so even even here I'm going to say
657. 1:39:31um um write me 10 catchy ad headlines um for shiny turtle that will get people uh to want to try it out
658. 1:39:47certainly unleash the shines today shiny turtle car wash where clean meets gleam roll in dirty roll out shiny
659. 1:39:57experience the turtle magic today Revitalize your ride with shiny turtle we make cars Sparkle Dirk's worst nightmare shiny turtle car wash
660. 1:40:08so I missed my calling and voiceovers apparently um but it can do that kind of stuff what else can it do um we can say
661. 1:40:16make me a table comparing the different kinds of car waxes I can use on my car that accentuates the the advantages of each
662. 1:40:31I don't know that wasn't good English but that's okay but look it's making us a table um what are the kinds of wax I don't
663. 1:40:39know I didn't know this going in Carnauba wax here's the advantages of it here's the ideal uses here's the
664. 1:40:46application frequency synthetic polymer ceramic coating and Spray Wax oh and paste wax so it can it can do tables I could also say um
665. 1:40:58write me a python application that helps me calculate the profit for my car wash business it'll explain what it does it'll write the code
666. 1:41:09so this is chat EPT this is way way more than ask catch EBT about yielding about dividend yielding stocks James for you anything
667. 1:41:23um okay so let me start a new chat and I'm going to go to chat gpt4 because it might do it better it'll be slower but
668. 1:41:29it'll be better so what did you say dividend can you pin that message again if you dividend yielding stocks okay
669. 1:41:41I want you to act as an expert stock broker that works for Merrill Lynch and deeply understands the market
670. 1:41:55period I'm trying to understand dividend yielding stocks and why I might consider them in my Investment Portfolio is that something
671. 1:42:05you can help with certainly a dividend yielding stock is one where the company pays out a portion of this thing for example company pays
672. 1:42:14out 12. and so look now it's even doing things like creating you know headlines a numbered list with bullet points I
673. 1:42:20didn't ask it to do that it's just doing that um risks involved right so so let's see what it says um why you would consider it's a stable
674. 1:42:30income stream it's a reinvestment opportunity there's a there's potential for lower volatility there's tax
675. 1:42:35efficiency and there's a hedge against inflation risks dividend Cuts or omissions Market risks interest rate sensitivity
676. 1:42:44diversification strategy dividend investment strategy second or selection criteria final thoughts so you know I
677. 1:42:53asked it to act like a Merrill Lynch you know person that kind of narrowed the context a little bit and now gave us
678. 1:42:58this big long thing from Michael um chat EBT helped me create an advanced image creation tool for my Custom
679. 1:43:06Framing business perfect so so this is one of the things when you start playing with chat gbt it's easy to fall into the Trap of just
680. 1:43:20thinking it's a toy because the initial things you play with are kind of toys write me a poem write me a song
681. 1:43:28where it really starts to melt your brain and and and and then your brains drip out of your ears and then your eyes
682. 1:43:35cross and your tongue starts quivering is when is when you make a mental shift from this is a toy to let me see what this can do for my
683. 1:43:46business or for my job or for my thing that I enjoy the most in the world you know I love stamps and I want to I
684. 1:43:56want to learn everything there is to know about stamps and then you ask this thing something that you think is one of
685. 1:44:01the most obscure questions in your area of expertise and it [ __ ] answers it like that and you go oh my God I I had no idea it's one of my
686. 1:44:14favorite moments is when people and by the way a bunch of new people just joined one of my favorite moments in
687. 1:44:20this Channel and in general with this generative AI stuff let me flip this is when especially people who were cynical about
688. 1:44:31it or kind of sitting on the sidelines like well I heard Chachi PT hallucinates and I hear and I hear it's stealing
689. 1:44:38people's work and I hear we get a lot of people that are like huh yeah I suppose it's just hype it's just
690. 1:44:45hype it's just how you get a lot of that right and almost a hundred percent of the time those people have not tried it
691. 1:44:53one of my favorite things in the world is when you sit down with them and I would encourage you to do this play with
692. 1:45:02this stuff and as you learn it start talking about it with people share it with people and uh there's a pinned comment I can't
693. 1:45:09start the poem um show them chat GPT and what what ends up happening is let me see how to close that now I don't
694. 1:45:19want to delete the poll dude there we go um what ends up happening is those cynical people realize how powerful this thing is and
695. 1:45:31then they go oh and then you can just see their mind kind of open up and there's it's initially excitement
696. 1:45:39and then it's like a little bit of concern like Oh that's oh that's kind of interesting and it's like but but it can't do my job
697. 1:45:48and then they do something about their job and it can at some level and sometimes way better than you think it should be able to
698. 1:45:58there's another pinned comment there from Jim which I want to get to um and and that's the moment where there's
699. 1:46:07kind of it's a threshold you can't come back come back from once you realize how powerful these tools are you kind of can't unlearn it
700. 1:46:15they're not going to be uninvented they're getting integrated in everything we do um so yeah so so go there right now
701. 1:46:24chat.openai.com that's the official chat GPT website go to claude.ai that's a competitor to chat GPT that's got some
702. 1:46:32really interesting strengths go to bard.google.com any one of those three are a great place to start uh can it develop business plans
703. 1:46:41[ __ ] yeah in fact let's let's go let's let's go do a couple of business plans what's your so new level up uh type in
704. 1:46:52um what your what your business is or like category of your business and then let's go make a business plan
705. 1:47:02um and if you could if whoever is modding can pin whatever new level up types there unless it unless they're no
706. 1:47:08longer here which is possible there's a lot of volatility in the in the lab tonight we were just up to 17 000 now
707. 1:47:16we're down to ten thousand we're all over the place this is wild though this is fun Freight dispatch okay perfect so we're gonna say
708. 1:47:26I want you to act as a business consultant that specializes in creating business plans for one of the big consulting firms period you have a
709. 1:47:41friend who is starting a freight dispatch business and wants to do it right and make sure that they have a solid business plan in place
710. 1:47:57period can you recommend three different strategies for creating a business plan that might be appropriate for a freight dispatch
711. 1:48:08business period I don't want you to start writing the plan yet I just want you to give me the Frameworks so that I can decide which
712. 1:48:16one I want to start with so notice that you know I gave it a lot of context and I'm even doing a thing
713. 1:48:23like I don't want to just jump into is this 17 000 verified humans that's really funny Ted looks at life I have no
714. 1:48:32idea I don't know what's happening but I'm digging it James is uh James is tossing some people our way please go
715. 1:48:38follow the Channel please go you know subscribe if you want to hang out here we're here every night and um
716. 1:48:48we do these things for I don't know two and a half or three hours live every night I basically just do them until my
717. 1:48:55voice blows out or my ass gets numb which it's it's getting close right now but this is just wild this is fun uh
718. 1:49:04okay so let's see what it did here so there's the porters five forces framework so the objective is to
719. 1:49:14understand the competing forces within the industry to identify areas of strength and potential opportunity that's pretty cool
720. 1:49:22the business model canvas I've heard of and the SWOT analysis so let's say so so let's say what we want to do right here is is can you um
721. 1:49:33can you combine the advantages of Frameworks one and two and create a hybrid framework that is simple yet powerful
722. 1:49:46and will help your friend position their business for success and investment sure it's called the Strategic business
723. 1:50:00positioning canvas so you've got your internal business model components the value proposition customer segments and relationships
724. 1:50:11revenue streams Key activities and resources key Partnerships cost structure and then you have external industry analysis components this is
725. 1:50:19really sound steps to implement start with internal components okay great so um so I'm going to say no this business framework is fine what
726. 1:50:32I'd like you to do now is make a first attempt at writing this plan for the Freight business for your friend
727. 1:50:44include as many details as you think are important and even include industry insights as best you can
728. 1:50:56and this should give us so so one of the things that chatgpt is pretty brilliant at is there's a there's a comment pin so I
729. 1:51:07can't start the poll ing that is if you get good at asking a couple of framing questions and you know
730. 1:51:16and then get get it to do something like this this business plan will probably be at about 80 percent of good
731. 1:51:25like it'll probably be actually pretty close to what you would end up with if you took a lot of time and read a bunch of business
732. 1:51:34books and melded two business plans or hired a consultant that last 20 percent is is the part where you start
733. 1:51:45participating again so so historically how I've worked is you know if something is compelling enough to me
734. 1:51:53or I've got to do something for work I'll take three or four days or three or four weeks I've got 80d so sometimes I
735. 1:52:01can just crank [ __ ] out like it's nothing other times it takes me weeks and I will sort of get into my hyper
736. 1:52:07focused tunnel and I'll I'll take however long it takes and I'll get to like a good first draft and that first
737. 1:52:13draft might be at like I don't know 85 or 90 percent of where I want it and then I spend like the next three days
738. 1:52:19you know finalizing it what happens with chat GPT is you can get to that 80 85 percent draft in [ __ ] five minutes
739. 1:52:29so instead of it taking me three days or three weeks or however long it takes me to get my [ __ ] together
740. 1:52:36I can have like what we just did here let's go look at a little bit of it for this Freight business um value proposition efficient reliable
741. 1:52:47cost effective Freight dispatch Services differentiation utilization of advanced technology for real-time tracking
742. 1:52:54scheduling and communication ensuring seamless Service Great it may or may not be that but you could obviously fine tune that
743. 1:53:02um here's your customer segment here's your revenue streams right so it's got all that stuff there so I can now take that
744. 1:53:09and what would have taken me three days I now have in five or ten minutes and I can spend the next three days refining this
745. 1:53:18so I want to show you something that that I actually think is is not only relevant but I think it's quite
746. 1:53:23important I think it's quite significant there's a A Wharton business school Professor named Ethan Malik
747. 1:53:29and if you don't follow him you should follow him brilliant [ __ ] dude like right on the cusp of all of this
748. 1:53:39stuff and and he just did this piece he's his blog is called one useful thing you should subscribe to it I don't get
749. 1:53:49paid by him there's always people in here it's just a sham he just wants us to go to subscribe to youth in my league
750. 1:53:55stand because he gets Kickback you know every time we subscribe to Ethan Malik's newsletter Kyle gets like ten dollars
751. 1:54:00like if I did wouldn't that be [ __ ] awesome but I don't so anyway don't be so cynical get a hug have someone hug you
752. 1:54:09um so this article that he wrote is called centaurs cyborgs and the jagged Frontier and it it's a group of
753. 1:54:19um people from Boston Consulting Group where they took a series of consultants and they gave them access to gpt4 chat
754. 1:54:27gpt4 the thing I just showed you and what they found this is [ __ ] mind-blowing so that little chart but
755. 1:54:35the the numbers under the chart right here Consultants using AI finished 12.2 more tasks on average completed them 25 more
756. 1:54:48quickly and produced 40 percent higher quality results so more faster better more faster better 12 percent more 25 percent faster 40
757. 1:55:05percent better using chat GPT this is Boston Consulting Group here's another thing that that this finding that that this study found
758. 1:55:14which again is mind-blowing this chart so on the left these were the bottom half of Performing Consultants
759. 1:55:26the lower performing Consultants had a 43 increase in quality of work the higher performing Consultants only saw a 17 increase at work
760. 1:55:38so what's what's happening is the lower performers are performing the increase in their performance is dramatically higher for the lower
761. 1:55:49performers in a company so it's making your best people better but it's making your not best people dramatically better
762. 1:55:57so so think about this this this is this is a technology that kind of looks like a Google search bar that when you
763. 1:56:06actually start applying it to business it's having you know not only a statistically relevant impact but like a dramatic impact
764. 1:56:18like that's what's going on right now and this is with chatgpt in its current state one of the things we joke about
765. 1:56:25here in the in the in the lab is you know I make fun all the time of these tools are are very primitive because they're
766. 1:56:36very new right the the generative AI stuff this stuff that's like you can type in some words and Magic [ __ ] comes out that
767. 1:56:45um has really been only been available since um since November of 2022 so it's like 10 months old so all of these tools are
768. 1:56:57what we like to call in the AI learning lab janky pieces of [ __ ] they're they're all just very primitive
769. 1:57:05so even with them in their primitive state without them being multimodal and and truly you know um do reasoning and and and take
770. 1:57:16autonomous action all of the things that they could be it's having that kind of effect 12 more 25 faster 40 percent better
771. 1:57:28holy crap I said magic [ __ ] I did I swear here and it is Magic [ __ ] but it's also janky [ __ ] the tools are janky right now
772. 1:57:37and Yankees a technical term if you're not if you're not a technologist you probably yeah but I had a partner life's
773. 1:57:45tough it's tougher when you're stupid all right uh by the way if you stick stupid hats on me I do respond to them
774. 1:57:51like one of those NPCs oh uh hot dog hot dog don't do that [Laughter] um anyway um ADHD also got any tips that get you into hyper Focus
775. 1:58:08yeah yeah I mean I mean I have I have tips about these tools and and what it's done for me with my ADD um I
776. 1:58:26the only other time in my life where I felt um where I felt overwhelmed to the point that my ADD couldn't keep
777. 1:58:42up with it was in the early days of the world wide web and when I was when I was starting agency.com in in 1994 1995.
778. 1:58:52there was so much to learn and things were happening so fast I literally couldn't keep up with it right one of the joys of ADD
779. 1:59:00is that you can synthesize things really quickly and with very little information you can sort of grock stuff and if
780. 1:59:07you're into it you can do a deep dive on it the speed and depth that these tools are are sort of being
781. 1:59:20introduced and evolving are absolutely overwhelming there's there's absolutely no way I or anyone else can keep up with
782. 1:59:29it in in any kind of way I pay attention to this stuff multiple times a day and honest to God I'll take 12 hours where
783. 1:59:37I'm going and I just do something else and I don't think about this [ __ ] and I miss seven things it's it is there it is
784. 1:59:44impossible to keep up with it so if you think you're behind if you think you can't keep up even if you went 24 7 with
785. 1:59:51this stuff you can't keep up so that's the the kind of um the overhang of this is that there's literally no way to get your head around
786. 2:00:01it so so just give that piece up so that's that's a big piece of it now the hyper the hyper Focus thing all right historically
787. 2:00:16before chat GPT if I had an idea it would it would have to be you know how you have those ideas so sometimes you have an idea and it's just
788. 2:00:27like a brilliant idea and you kind of do like a quick hyper Focus but then you have those other ideas where it's an
789. 2:00:33idea that haunts you right and you think about it and then you're like oh that's pretty cool I should do that and then
790. 2:00:39you forget about it and then like two days later you're like that and then two days later you like wake up from asleep
791. 2:00:45and you're like oh I really got to do this thing and then and then you go into hyper Focus but it's it's it's a it's a
792. 2:00:50big idea and it's one that can't escape your brain and then at some point you give into it and you hyper focus and
793. 2:00:57then with ADD there's like a thousand other ideas and almost all of those ideas just sort of pass through you know
794. 2:01:04they pass through squirrel squirrel squirrel squirrel squirrel so so because of that Dynamic you know you you tend to only do hyper Focus I
795. 2:01:16don't know every three weeks depending depending on your your manicness it might be once a week it might be twice a week it might be
796. 2:01:25once or twice a month what's happening with chat GPT is that I can have just a whisper of an idea like not even not even a super
797. 2:01:37compelling idea but it's like oh that's kind of interesting and then what I'm having to train myself is is go
798. 2:01:46um get it out of your head and into chat GPT so I'll have like a whisper of an idea I hop into chat GPT
799. 2:01:53and then I just start playing with it and what happens is I get to 80 of an articulated vision of or not even Vision I get to an articulated
800. 2:02:06um instance of that idea sitting in front of me and and then I can spend 15 minutes in a hyper focused tunnel
801. 2:02:17refining it or I can look at it and go ah it wasn't such a good idea but it's like instant and and I can have I'm I find myself having
802. 2:02:26instead of like two or three of those a month where I was doing that historically I'm sometimes having two or three ideas a day
803. 2:02:35and definitely you know multiple days a week it's one of the reasons I'm doing these lives is because one of the purposes these lives serve
804. 2:02:47for me is a way for me to process what what the hell is going on in my brain because these tools are amplifying my
805. 2:02:57ideas in ways that I've never been able to do so that that study that Ethan Malik showed about the BCG Consultants
806. 2:03:07being faster better and and and uh and producing more stuff I'm finding the same thing I'm I'm creating more better faster ideas
807. 2:03:18and the other thing is where a lot of times my ADD would be crippling where I sort of have an initial idea but then if it required
808. 2:03:30actual research or if it required actual paying attention to it it would fall away because I would just
809. 2:03:37be off to some other thing and I'd be like oh yeah that was a good idea but oh it's too much work and I wouldn't do it
810. 2:03:42well this now does all the research it organizes [ __ ] you can have four different ideas and you literally copy
811. 2:03:50and paste those ideas together into a single chat GPT session and say hey over the past week I had these four ideas do
812. 2:03:57they have anything in common and it'll it'll find the connection between them you're like oh that's really interesting
813. 2:04:02yeah I hadn't thought about that can you come up with a fifth idea that is the umbrella for them sure here it is
814. 2:04:09Sam corn for capturing ideas I highly recommend reai they have excellent transcription and hang on magic summarization all right
815. 2:04:22r-e-a-i-a-u-r-a-i that sounds very cool let me go look at that re.ai oh they have cool colors too
816. 2:04:38said I'm a sucker for that that dark you know the uh the dark uh black with the the blue thing um I can draft emails record and transcribe
817. 2:04:51voice memos create meeting notes write essays correct your grammar okay cool all right so this is like an AI powered
818. 2:04:59chat bot that you could just chat ideas into that's super cool okay I'll definitely check that out um I ran 5K off so so you can ramble
819. 2:05:07into a voice note for a while and then it totally cleans it up and summarizes it that's super cool in fact let's see
820. 2:05:14I'm going to write this down outside of my computer because I have 80d even if I don't nothing will happen with it
821. 2:05:21but here here's my notebook any other people have ADD [Laughter] Ed all right uh re dot a i all right I'm definitely
822. 2:05:36gonna start playing with that that's awesome thank you Sam Sam corn is one of the Irregulars here the Irregulars are
823. 2:05:42the people who show up here every night and hang out with me they're awesome um they're really supportive they're
824. 2:05:48also supportive of the community if you have questions and comments below um uh you know put them down below happy Friday night everyone um
825. 2:05:56the the at the peak tonight we had 17 000 people in here we still have 4.8 000 people in here which which is a little
826. 2:06:04odd this this the funny thing about Tick Tock is this was a little intimate Bistro for like three and a half months
827. 2:06:12and all of a sudden tonight you know we're like playing a playing a [ __ ] theater or a stadium a small stadium
828. 2:06:19um so it's super cool to have everyone here and uh and it's just super cool so anyway welcome welcome everybody that's new here
829. 2:06:31um we'll get the janky pieces of [ __ ] thing out of the way and then let's see we'll do yeah that's a good one all
830. 2:06:38right Beauty all right tons of Post-it notes yeah exactly Sherry Banks knows the add game and by the way if you're here and didn't know it
831. 2:06:51um this is an inside joke here at the AI learning lab so one of my pet peeves are all those Tick Tock sites as you're
832. 2:07:00scrolling through Tick Tock and there's like some guy like leaning into your face going you're going Mario
833. 2:07:07they're worse than the crypto Bros um and and and to be clear how they're making money is by getting you to pay them money so
834. 2:07:21that they tell you how to make money with chat EBT they're not making money with chat GPT because if they were they
835. 2:07:27wouldn't be screaming in your face about making money with chat gbt um we're breaking all the codes we are
836. 2:07:34breaking all the codes it's crazy in here isn't it jrc just nuts um here's my philosophy on how you make
837. 2:07:41money with chatgpt because it's there's a flipping answer ah Make It Rain oh yeah there's a flipping answer to these these guys right
838. 2:07:53the flipping answer is [ __ ] them don't listen to them that's not where the money is um but but the real answer is this
839. 2:08:03because Chachi PT is so sexy because AI is sexy it's very easy to get seduced into thinking it's something more than a tool
840. 2:08:14it's not it's not a strategy AI is not a strategy chechi BT is not a strategy is a tool just like Photoshop is a tool
841. 2:08:25just like Microsoft Word is a tool just like a hammer is a tool it's like a shovel is a tool so when someone says you can make money
842. 2:08:34with cheche EBT it's the equivalent of saying you can make money with a shovel okay why would you say that right if someone
843. 2:08:42said it about a shovel you'd be like okay that's [ __ ] weird well how do you make money with a shovel I don't
844. 2:08:48know you find out who needs holes Doug and if you can figure out the way to use your shovel in a way that digs their
845. 2:08:56holes in a way they're willing to pay you for it you can make money with a shovel exact same thing with chat GPT with AI
846. 2:09:05with mid-journey with all these things so the place to start if you want to make money with chat gbt is start with you
847. 2:09:13what are you good at what are you passionate about what do you know um in the world is a pain point that you
848. 2:09:22could probably solve if you had the skills to do it or if you had the tools to do it oh well I know how to throw really good parties awesome
849. 2:09:35um could you throw really good parties that people would be willing to pay for yeah like okay then go to chatgpt and
850. 2:09:42say hey I'm really good at throwing parties and I'd love to figure out how I could make money throwing parties it'll
851. 2:09:47give you 40 [ __ ] ideas and then of the 40 ideas two of them will be really interesting to you and you're like oh hey chat jbt these two
852. 2:09:56are really interesting to me can you tell me more about them how'd they come oh that one's really good and before you know it this idea of
853. 2:10:05the hyper Focus tunnel you can have a whisper of an idea for a business and within five minutes have have it articulated to the point that
854. 2:10:16that you could say now give me 10 names for that business now give me 30 tag lines for the name that I picked
855. 2:10:24now write me a website and it'll write you all the website copy within an hour you could have a fully articulated business from a standing
856. 2:10:34start with nothing that's how you make money with chat GPT chat GPT is just the accelerant it's just a tool that accelerates the [ __ ]
857. 2:10:44coming out of your brain it's a Humanity amplifier that's what not the AI learning lab is that's what chat GPT is that's what clot is that's
858. 2:10:55what Bard is that's what mid-journey is that's that's what the audio generation tools are that's what the
859. 2:11:02um Runway ml text-to-video tools are everyone's talking about these tools like the tools are going to save you no no no
860. 2:11:10You're Gonna Save You Amplified by these tools Humanity amplifiers okay but where do I start I'm not very
861. 2:11:22technical start here that's why I have them on screen on every one of my lives I've been doing this for months now nightly here
862. 2:11:34why do I have these here start here because we're early still need a recommendation on a voice to text transcription service okay um
863. 2:11:47whisper is really good so if you go to there's a couple of ways that you can do it you can get the open AI app the the
864. 2:11:58chat GPT official app from openai it does transcriptions you can basically flip it into talk to me mode and it will transcribe there you can
865. 2:12:11also this is a bit wonky there's not actually a simple way to get to this which I think is kind of shitty but I I
866. 2:12:18think this is ultimately going to be built into chat GPT so I don't I don't think you have to do it this way for
867. 2:12:23long but if you go to platform dot openai.com playground that brings you to the chat GPT playground hang on a sec sorry the voice is going
868. 2:12:39ha ha all right we're down to 2 000 people now we're down to normal levels no we're not we're still like 10 times where we where
869. 2:12:48we normally are so so welcome everybody all right So within within the open AI playground again the URL is platform.openai.com
870. 2:13:00slash playground and then that brings you here and this is where developers you know develop their prompts and things like that so if
871. 2:13:09you're creating automations in zapier for example this is where I come to develop my prompts so over on the right
872. 2:13:15hand side here there's there's the mode so if you if you move out of chat mode and into complete mode there's this little teeny tiny
873. 2:13:26microphone right here this gives you access to whisper so you click on that and now you can upload a text a voice
874. 2:13:36audio file or or a video file with voice in it or you can just record and this will generate it to text I've got this built into my company
875. 2:13:48storyvine where every video in the system I can just push a button it'll automatically do this for any video in
876. 2:13:54the system but it has to be done programmatically but this is how you can actually do it and then once it
877. 2:13:59generates it it'll put the text right in here Whispers really good you know it's not a hundred percent but it's you know
878. 2:14:07it's above 90. these Voice to Text transcription tools historically have sucked bad like I have an 11 and a half
879. 2:14:16year old company and I have not I have not had transcription as part of the the company Emily this is going to save me so much time awesome
880. 2:14:26Emily um I didn't have transcriptions as part of my my video platform for for 11 years until whisper came along and it's it's
881. 2:14:37because it's actually good um it still gets [ __ ] wrong but it's at least acceptable at this point all right how we doing time wise
882. 2:14:48um 10 51. it's getting late it's getting late all right can it generate SRT files for the videos what's an SRT file is that the uh the
883. 2:14:59subtitle hang on let me go find out what an SRT file is I know I've owned a video company I should know what that is but
884. 2:15:08um dot SRT for video file SRT oh yeah that's captioning um yes it can I don't know what the
885. 2:15:23parameters are hang on let me let me pop a video in there and let's let's let's go play with it [Music]
886. 2:15:38file type is not supported well you're an [ __ ] um do I have an mp4 here foreign let's do this and this is the Dual
887. 2:15:50dialogue one this is the one with me speaking both English oh no it's a mov I need an mp4 oh these are too long oh good lord
888. 2:16:03it's never easy as is that camera easy I you know I try and I try and I try people and then the file format's not
889. 2:16:11supported what am I supposed to do MP3 let me go find an MP3 I don't even know if I have any MP3 things that are worth going home
890. 2:16:32um all right that's not good let me see uh dot MP4
891. 2:16:48is an MPEG for movie those are all those and I guess we'll throw that up there all right
892. 2:17:03transcribing thank you for the Roses by any other name would smell as sweet yeah voice to text and text-to-speech
893. 2:17:12went from useless to job killer in like six months exactly slash home slash Dave by The Way Slash home slash Dave 10 out
894. 2:17:20of 10 on the name Emily thank you for the oh you're giving me all the toys a lightning bolt and a pepper and a little
895. 2:17:27ice cream cone and a world and a what the hell is that it's a microphone or a fist or some sort of phallic situation
896. 2:17:37um why did that not transcribe that video I don't know anyway I don't know what the I don't know what the settings are um for
897. 2:17:52um for generating SRT files I've got within storyvine I've got storyvine set up to generate a vtt file which is which is
898. 2:18:03also a subtitling format so I know I know whisper can do lots of different sub subtitling formats it can pull down
899. 2:18:10time code when it when it does the the transcription so it's pretty sophisticated it also here's the other just [ __ ] wild ass thing about
900. 2:18:21um about these tools is like chat GPT all these tools are brilliant at translation Emily I'm so grateful for your help
901. 2:18:31you're welcome thank you so much I really I really appreciate that thank you Mr Pringles uh loves me a good hot
902. 2:18:36dog who doesn't like a hot dog you know who doesn't like hot dogs Liars you you know those vegans you you know
903. 2:18:46every insufferable vegan that you know is sneaking a [ __ ] hot dog on a Saturday night after they get home from
904. 2:18:54the bar you know leave the bar you know that's going on [Laughter] bitter you betcha um but but these transcriptions will be
905. 2:19:07they we you can translate this stuff into any language and now with hey Jin you can do it with the video itself I
906. 2:19:14we're just honest to God I I feel like we're in a post sci-fi reality it's I like I think about almost every sci-fi movie that I
907. 2:19:26experienced as a kid or as an adult and with very few exceptions we're like we're just checking off the Technologies
908. 2:19:34like oh yep we can do that now oh we can do that now ah that'll probably be a year like like nothing's a decade away
909. 2:19:42anymore nothing ex time travel's the only one where I like okay we're gonna have to probably wait a couple of
910. 2:19:48decades like I'll probably I probably won't be here for time travel but like curing cancer and curing genetic
911. 2:19:55diseases and and like climate change and energy production I mean just it's it's gonna be [ __ ] insane
912. 2:20:05so anyway it's just crazy it looks like we just got a whole new pile of people we're down to 2 000 people we're back up
913. 2:20:11to seven thousand welcome everyone my name is Kyle Shannon this is the AI learning lab um tonight's a bit of a trip because we
914. 2:20:19normally have anywhere from 50 to 150 people in here and now there's seven thousand so hey I'm Kyle this is the AI learning lab we
915. 2:20:30have cute little graphics like see that and then we got um I'm not a robot goodbye we got AI learning lab on a little planet all
916. 2:20:41these images were created with a tool called ideogram which if you've not played with it does text and if you've ever used
917. 2:20:52mid-journey and tried to get it to do something like have words it was really frustrating because it not
918. 2:20:59did words and and many people when when I would show them this AI stuff they're like oh make make a thing that says
919. 2:21:06blank you know whatever the saying was that they wanted to do and you can't you'd have to say well the tools are kind of stupid right now
920. 2:21:13all right how do I turn my hard copy notes into soft copy something I didn't quite see what that was
921. 2:21:19um you can do that right now natively if you've got an iPhone um you go into I think you go into notes and then you say um
922. 2:21:32uh scan a document hard copy notes into soft copy notes so so there's there's a there's a couple of ways you can do it there's there's a
923. 2:21:42notebook called Rocket books which are really cool um they're not you well they might be using AI but but their their physical notebooks
924. 2:21:54and they've basically got QR codes on the page so you just take notes with a normal pen they've they've got a version
925. 2:22:02of The Notebook you can you can take notes with erasable pens and then you can microwave it and it really it erases
926. 2:22:09all your notes if you microwave it it's pretty cool but um rocket book you can just write notes and then you use the rocketbook app to take
927. 2:22:17pictures of the notes and it scans everything that required a bit to management too much management for me I
928. 2:22:24know within iOS you can go into the notes app and you put it into document mode and it'll OCR [ __ ] so you can just
929. 2:22:30take pictures and it'll do it um Leonardo is a good one for images too yeah Leonardo's good so let me show you my my list of
930. 2:22:39um image generation tools that I think are worth paying attention to because Leonardo's on there Leonardo doesn't do text ideogram does
931. 2:22:50text so I'll talk about these here so here's the uh here's how this is going to roll mid journey is the best for probably another week
932. 2:23:01because Dolly Three is coming and I think Dolly three is just going to obliterate a bunch of these things
933. 2:23:07um there's a comment pin so I can't start the poll tonight there you go okay mid journey is the best
934. 2:23:21um the problem with mid-journey I don't know if it's a problem but you have to generate images from within Discord and
935. 2:23:27it's a real pain in the ass ideogram is not as good an image generation but it is good at text generation and and you
936. 2:23:35can make posters and you can make t-shirts and you can do all sorts of stuff not super precise but good
937. 2:23:41Leonardo is a web interface sitting on top of stable diffusion so if you're geeky you can use there's this interface
938. 2:23:49called automatic 1111 where you can do all sorts of you know flips and switches for stable diffusion if you don't feel
939. 2:23:56like installing your own software manually use leonardo.ai you just go to that website they give you 150 free
940. 2:24:05credits a day which translates to about seven or eight images a day but you can make seven or eight good images a day with a nice interface
941. 2:24:14um hey Jen I'll come back to Adobe Firefly and Photoshop generative fill came out of beta last week and which basically means you can now
942. 2:24:24use the images you generate there commercially the other thing about Adobe and and Photoshop generative filler Firefly and generative fill
943. 2:24:32adobe's image models have not been trained on copywritten images it's been trained on um uh public domain and Adobe stock images
944. 2:24:44Adobe also has a plan for allowing artists to opt out of being trained opt into being trained and have a
945. 2:24:51compensation and attribution model so um Adobe is going to make image generation or RDR safe for Corporate America
946. 2:25:01so those are the biggies hn is not an image generation tool it's a text to audio and then to video animation tool
947. 2:25:11but they have this thing called um video translate that one of their Engineers cooked up and it's [ __ ]
948. 2:25:19insane the backlog to get your videos rendered right now are just stupidly long but check this out so this is a video of
949. 2:25:29me I I intercut my original English video with the Italian translation that that hey Jin did so we've heard from a
950. 2:25:39number of clients that modular content is important to them getting more thank you Ted with how we create content and
951. 2:25:44how we leverage content modular important behind me you see these little green boxes um that's how we think about
952. 2:25:57stories at storyline we break stories down into little atoms right little modules foreign so just absolutely wild
953. 2:26:10um okay I just saw a comment from Jules so Jules my daughter who's an artist just chastised me for AI generated art
954. 2:26:19because it's stealing work okay okay this is this is an incredibly nuanced conversation I made it to whisper but can't find where to go to
955. 2:26:35upload okay hang on a sec you made it to whisper you mean did you get into um chat GPT playground go to plague or
956. 2:26:52platform.openai.com playground and then flip from chat mode into complete mode so it'll default you into chat mode it's not available there
957. 2:27:06flip into complete mode and then that little microphone icon is the thing you're looking for um I want to go back to the image
958. 2:27:15generation thing the way that chat GPT works and the way that diffusion models work they're not copying and pasting
959. 2:27:28they're not copying and pasting and in fact I am so passionate about explaining this I made slides I got slides
960. 2:27:39AI isn't copying and pasting now I made this image in mid-journey this is this is the misconception right now is that these tools are
961. 2:27:54you know when you ask for for you know some sort of photograph that it's going into some photographer's work and
962. 2:28:01cutting the car out of it and pasting it into some magic composite it's not what it's doing um now that said these models have been
963. 2:28:12trained on copywritten work and the artists whose work it's been trained on haven't gotten attribution haven't gotten compensation but
964. 2:28:22here's a metaphor I like to think of if I have in my finger a grain of sugar and you're making cake batter and I take
965. 2:28:34that grain of sugar and I toss it in into the cake batter and then you bake the cake trying to figure out what part of an AI
966. 2:28:42generated image came from what specific image that it was trained on is kind of like asking to find that grain of sugar
967. 2:28:50in your baked cake it no longer exists in that form when they train the GPT models the large language models
968. 2:29:00on these massive data sets they might have been trained on the novel War and Peace but the way the words get embedded
969. 2:29:08the way they get turned into Vector space is just words in semantic clouds War and Peace ceases to exist what it
970. 2:29:17pulled out of War and Peace were the meaning of the words related to one another and all of them get Amalgamated
971. 2:29:24into this massive mathematical set same thing with the images they they get denoised they get noised from an actual
972. 2:29:33image into a noisy image in fact I've got a slide that shows this here okay we're going to do this now screw it because this is important
973. 2:29:42um I want you to you can you can tell your doctor I saw some old guy on Tick Tock and he told me that it's not stealing here's the deal it's
974. 2:29:51this is a very very very gray area because this is absolutely impacting people's livelihoods and it's it's transforming
975. 2:30:03how art is created who creates art how it's created but but it's a very nuanced argument so if you or your daughter actually
976. 2:30:13understand how these things are made you can at least be more educated in where the solutions might lie because I
977. 2:30:21think the entire industry right now is focused on the wrong part of the equation everyone's focused on the
978. 2:30:27output and how these images are used and I think where to focus is on the generation and and compensating people
979. 2:30:34who's who's work has been put into these things and compensate them on the generation of the image but anyway here's how they work so
980. 2:30:46so you've got your large language model you've got your chat GPT the the way the image models work is you've got your
981. 2:30:51large language model you've got it paired with a diffusion model in this case there's also Gans there's different
982. 2:30:56different kinds of Technologies but the one I'm going to show you is a diffusion model the so if you just notice the image that was
983. 2:31:07there it was that that sort of house on a cliff you know illustration that image when it gets put into the
984. 2:31:17training database it's labeled right so someone went through and they labeled it or they use computer vision to label it
985. 2:31:23they said there's grass there's trees there's a house there's a cliff there's mountains there's Sky there's so
986. 2:31:28whatever so in that image there were all these words that they put in there right and then they use a noise noising
987. 2:31:34algorithm to noiseify the image so the image is no longer an image of that thing it's this noiseified mathematical
988. 2:31:46blob that has some words associated with it and then based on the prompt that you put into a large language model the
989. 2:31:57large language model interprets those words and then it filters out um the the relevant vectors and and so I
990. 2:32:07explained this earlier with chat gbt but this is this is what's called Vector space all of these dots are like uh they rep they're they're
991. 2:32:17um sort of strings of numbers that represent a point in mathematical space this is a 2d representation of it real
992. 2:32:25Vector spaces are like a thousand Dimensions it's it's literally impossible to conceptualize but just imagine so every one of those dots is
993. 2:32:34like a word but it's not a word it's a number and the groupings are semantic clusters like you might have you know
994. 2:32:40animals down here and within here's dogs but you have like people that are bad in relationships and you have dogs up here
995. 2:32:47but those are men that treat their women bad right you know so so you have these these different semantic clusters okay
996. 2:32:54now how do we get from this mess to a picture so again you you're combining this idea of a large language model with this sort
997. 2:33:03of tagged diffusion um image Library so if if our prompt is Cat on a desk what the what the large language model does is it basically says
998. 2:33:16okay in all these semantic clusters here's all the the vectors that have been tagged cat and then here's all the vectors that
999. 2:33:27have been tag desk and then within those the cat images are all these random noise images and the desk images are all these random noise
1000. 2:33:37images it actually can't see what's in the images it just knows that these were tagged cat and desk and then it
1001. 2:33:44mathematically averages those noisy images into a final noisy image and then using this the reverse of the algorithm
1002. 2:33:51that made the noise it denoises the image over 30 or 50 or 70 steps until you have an image that's a cat on a desk this is an original image
1003. 2:34:05that came out of effectively the imagination of all of the [ __ ] it's been trained on combined with your prompt
1004. 2:34:15in this sort of mathematical magic that is actually generating an original piece of art now where does it get really [ __ ] up and and um
1005. 2:34:27and and fuzzy well if you type in I want a cat on a desk illustrated by Greg Rutkowski who's an illustrator who's
1006. 2:34:37famously showed up in almost every mid-journey in stable diffusion prompt for for months and all the images looked like his
1007. 2:34:46images why because when you type in the word Greg Rutkowski if he's tagged in the language model it's going to pull
1008. 2:34:54all of his images and even though it's averaging them and it's creating a new original image it's doing it in his
1009. 2:34:59style because that was those were the source um sort of training elements that went into generating that noisy image so your
1010. 2:35:09daughter's not wrong to say this is [ __ ] up but it ain't stealing and it ain't copying and pasting and so anyway I hope
1011. 2:35:19that helps so uh we still have a lot of people in here so uh welcome if you're new here my name is Kyle Shannon this is
1012. 2:35:27the AI learning lab uh I've been going for a while tonight my voice is officially um checking out so I'm just gonna do a
1013. 2:35:37few more minutes but you know another 10 minutes maybe but follow me um you know feel free to subscribe I do these lives
1014. 2:35:48every night I've been doing them every night now for about three months four months I don't know some amount of time
1015. 2:35:54um I think it's really important I'm here to kind of demystify this AI stuff have fun um it's been a little intense tonight
1016. 2:36:03because normally there's like 50 to 100 maybe 150 people tonight there's been as many as 17 000 right now we're at about
1017. 2:36:12six and a half thousand so welcome everybody um if you have questions pop them in the comments below we've got some moderators
1018. 2:36:20that are that are uh picking interesting comments and putting up Up Normally I try to answer all of the questions uh
1019. 2:36:27with seven thousand people in here I don't think I can do that but what are you gonna do all right let's answer some questions
1020. 2:36:34let's get into it all right this uh are you an AI evangelist um yeah I think so I I'm certainly I'm certainly an AI Optimist um
1021. 2:36:50I know that the the the popular thing to to say these days are the robots are going to kill us I think there is there is a an equal if not
1022. 2:37:01higher chance that the robots are going to elevate us they're going to free us from doing the the mundane work they're going to
1023. 2:37:09amplify our capabilities and they're going to allow us to do remarkable work overall um I also believe that the changes that
1024. 2:37:19are coming are coming way faster than anyone's prepared for they are going to be dramatically disruptive
1025. 2:37:28uh and it's it's going to cause a lot of pain and it's gonna it's gonna transform almost every job any job that is
1026. 2:37:36computer mediated in any way um is going to be affected by this stuff and so this channel is really just about
1027. 2:37:45opening up the conversation and saying you know if you're sitting on the sidelines you don't have as much time as as you
1028. 2:37:54think you do for for how fast this stuff is coming um and so a big part of what I kind of proselytize here if I if I
1029. 2:38:06evangelize anything it's get curious now like go there right now chat.openai.com you don't need to do all
1030. 2:38:16this other [ __ ] that I'm talking about all these other Technologies save that for another day if you haven't
1031. 2:38:22tried chat GPT go there right now right now like not I'll get to it after the Live's over I'm on Tick Tock I don't want to go there
1032. 2:38:32just go there right now [ __ ] do it [ __ ] do it do it do it and just start playing um I've got a a comment pin so I can't start the poll
1033. 2:38:45um so why do that because this isn't a Google search it looks like a Google search it's not a Google search
1034. 2:38:56It's profoundly more powerful than that these are like the most powerful tools that have ever been been put in
1035. 2:39:05the hands of mere mortals kind of shy of of you know the nuclear bomb which was not in put in the hands of mere mortals
1036. 2:39:13this stuff is really really powerful and we get to play with it so it's in everybody's interest right now to get curious about this and
1037. 2:39:22understand what it's good at what it's not good at the conversation I just had about the mother who said her daughter
1038. 2:39:28kind of slapped her down for you know generated images of stealing and this and that like I'm glad her daughter is aware of it and
1039. 2:39:39has an opinion on it I want everyone to get more educated about it because it's not as black and white it's it's really easy to demonize
1040. 2:39:48this [ __ ] it's really easy to dismiss it oh it's not perfect it hallucinates it lies uh was that explanation ideogram that
1041. 2:40:01explanation was just generally how diffusion models work um ideogram's one of them I don't know how they're getting I assume what
1042. 2:40:09they're doing is they're doing two different um diffusion models where one is images and one is text and they're somehow
1043. 2:40:18combining those in the in the sort of noisy algorithm part I don't I don't honestly know how they're doing what they're doing
1044. 2:40:25but it's a diffusion model so it's it's it's some fancy version of what I showed I make most of my YouTube videos uh
1045. 2:40:33scripting with chat gbt yep exactly so just go to these things that just go to this and start playing go to
1046. 2:40:40prompts.chat that'll teach you how to play with these things just get educated about it get curious about it and just
1047. 2:40:45keep coming back to this this is why we're this is why this is here um was that last explanation ideogram that's what I just explained
1048. 2:40:56a lot of tools he has shown tonight weren't here two weeks ago exactly so Joe so for the people that have been
1049. 2:41:03here regularly what what you what you don't know is if you just joined here is that these lists that I show of the different tools
1050. 2:41:13these have been changing regularly this one that just looks like a novel right now this used to be like two or three
1051. 2:41:20tools and now how many are there one two three four five uh uh yeah five and then what's missing from here is
1052. 2:41:32Dolly three which was announced two days ago and it's gonna show up maybe in a week maybe over the weekend who knows
1053. 2:41:42and it's Gonna Knock mid-journey and ideogram off this list I think because mid-journey you have to go into
1054. 2:41:53Discord you have to learn Discord and then you have to learn command line functions you have to you have to type in command
1055. 2:42:02lines to make images in mid-journey it is the least visually the the fact that the most sophisticated image tool has to be typed in with
1056. 2:42:14command lines is [ __ ] absurd so what's going to happen with Dolly three is the quality is going to
1057. 2:42:22probably be as good as mid Journey or at least close to it it's going to be able to do text better than ideogram does and
1058. 2:42:29you're going to be able to do it within the context of a long chat GPT conversation and generate multiple
1059. 2:42:36images within that conversation where it as it's generating the images it's aware of the entire conversation so it can do
1060. 2:42:44things like maintain a single character across visual prompts which you've ever if you've ever tried to do that in
1061. 2:42:50mid-journey it's effectively impossible even with control net with stable diffusion it's effectively impossible
1062. 2:42:57you can get there there's people that have hacked it but you can't just say this character's name is Bob now put Bob
1063. 2:43:04in a spacesuit now put Bob on a fire truck now put Bob at the hot dog stand Dolly three will do that none of these
1064. 2:43:12tools can do that right now like that's gonna change in a week been loving this stream I try and catch
1065. 2:43:19it when I get home from my gigs thanks everyone thank you Scott Curtis what kind of gigs you a musician that's super cool
1066. 2:43:28um all right let me do we're down we're down to a paltry 1600 people uh which thank you all for hanging out for a little bit
1067. 2:43:39um now 1700 people it's going back up um normally these These are kind of intimate little Gatherings and we have
1068. 2:43:47lots of inside jokes and things like that um Kyle you missed Jules sent you a beautiful gift a glowing jellyfish oh I
1069. 2:43:54did miss it I don't know what I was looking at but thank you Jules um that's amazing thank you very much um listen I um
1070. 2:44:07it you know it's funny it's funny some nights I'll be in here there'll be like 30 people in here and the the people
1071. 2:44:13that are in here regularly we've kind of established a fun a fun relationship um but it's been kind of an intimate
1072. 2:44:19little thing and there's some nights I'm like well you know is this really making a difference and and then you know
1073. 2:44:25tonight lots and lots of people are in here and I think it might make a difference but it's you know the nature
1074. 2:44:29of what it has changed so so I hope you come back um I'm gonna go because my voice is is absolutely fried
1075. 2:44:38um is there an app for Dolly three okay so Dolly three what I'm going to do is give you go to openai.com slash d-a-l-l
1076. 2:44:54Dash e Dash three Dolly oh this is absolutely [ __ ] mind-blowing what's coming so and it it if you're if you haven't played with
1077. 2:45:07these image generation models before this won't be mind-blowing this will just be like oh that looks cool um
1078. 2:45:15one of the toughest things to do with these image generation tools is get them to predictably do what you want them to do
1079. 2:45:24and so so Dolly three a couple of things about this this is from open AI so this is from the people that make chat gpt4
1080. 2:45:32they currently have a thing called Dolly too which if you want to play with it if you go to bing.com here I'll show you go to bing.com
1081. 2:45:41and then you go to chat that's chatgpt4 on Bing and I can say um make me an image of a hedgehog driving a fire truck
1082. 2:45:56and this will use dolly two to make the image of a hedgehog driving the fire truck Dolly 2 hasn't been
1083. 2:46:05updated in I think over a year it's not good but it's there and it's free and look there's a hedgehog in a fire truck
1084. 2:46:14now is it bad no that's not bad they're actually pretty good um I wonder powered by Dolly I wonder I don't think this is Dolly three yet
1085. 2:46:29those does seem a little primitive so anyway okay so let's go back to to Dolly three so dolly two was okay but Dolly three
1086. 2:46:38it's going to be natively incorporated into chat gpt4 so you're in chat TPT and you're having a conversation maybe
1087. 2:46:45you're writing a children's book or maybe you're writing a graphic novel and then you'll just be able to immediately
1088. 2:46:50within that say go make me this kind of image and it will use gpt3 now check out what this does this is a fun kind of New
1089. 2:46:59Yorker kind of illustration but here's what I want you to pay attention to and if you've ever made images you'll see how remarkable this is
1090. 2:47:08the prompt is an illustration of an avocado sitting in a therapist's chair saying I just feel so empty inside so it did the text perfectly
1091. 2:47:18it's an avocado sitting in a therapist chair with a pit sized hole in the center so that looks like a pit but it's
1092. 2:47:24actually a hole it's a pit sized hole the therapist a spoon scribbles notes so all of the elements in that prompt
1093. 2:47:35are represented in that image if you have ever tried to get mid-journey to do two concepts in the same prompt it's
1094. 2:47:43essentially impossible it will always prioritize one or over the other and you will want to smash your face into your
1095. 2:47:50computer or a brick wall it's maddening so this is just gonna do that now the prompt here a 2d animation of a folk
1096. 2:48:00music band composed of anthropomorphic autumn leaves all right perfect there's anthropomorphic Autumn Leaves as if you
1097. 2:48:10know as if I you know could ever conceive of that now you can conceive of it and just make it this one look at
1098. 2:48:16this prompt I'll read it and then I'll zoom in on the image because because it's a remarkable image um in front of a deep black backdrop a
1099. 2:48:27figure of middle years her Tongan skin rich and glowing is captured mid-twirl her curly hair flowing like a storm behind her her
1100. 2:48:38attire resembles a whirlwind of marble and Porcelain fragments illuminated by The Gleam of scattered porcelain shards
1101. 2:48:46creating a dream-like atmosphere the dancer manages to appear fragmented yet maintains a harmonious and fluid form and look at that
1102. 2:49:03fragments of porcelain swirling porcelain her hair like a storm I mean like remarkable remarkable okay
1103. 2:49:14so let me close that down how do I get out of here where does the there it is um let's look at a couple other things
1104. 2:49:22here this is really interesting so what they did here was they took a long prompt and they they broke it up and
1105. 2:49:28sort of pointed to so part of the prompt was sidewalks bustling with pedestrians enjoying nightlife bustling city under a full Moon Under
1106. 2:49:39the shine of a full moon so you get Shadows the full moon the shine her prompt at the corner stall a young
1107. 2:49:46woman with fiery red hair dressed in a signature velvet cloak haggling with a grumpy old vendor and then the vendor a
1108. 2:49:54grumpy vendor a tall sophisticated man he's taller than her wearing a sharp suit yes Sports a noteworthy mustache yes
1109. 2:50:04and animatedly conversing on his steampunk telephone yes like holy crap um so so the level of control here is going to be amazing
1110. 2:50:16um let's see that was Dolly too that's Dolly three it's just on a complete different level and then I'll
1111. 2:50:26play this I'll play this video so one thing that I'll note here before I play it I'm not going to stop it so
1112. 2:50:34the the video generates they generate this Hedgehog that you see right there and then the first thing that happens is
1113. 2:50:42the person doing it picks one of the Hedgehogs and says my daughter said his name was Larry and then and then just watch what it does
1114. 2:51:05thank you [Music]
1115. 2:51:23[Music] [Applause] foreign [Music]
1116. 2:51:47let's see so Jules asks is Dolly three like ideogram or your creation is available and uh to people in the public once
1117. 2:51:56generated I don't think so because it's going to be a plug-in within a chat GPT chat you can probably share your chat
1118. 2:52:04um and then someone else asked another thing uh right below that uh what did what did they ask uh what tool are we using tonight this I
1119. 2:52:16was just showing Dolly three anyway this should just roll out so so the people that are gonna get first access to Dolly
1120. 2:52:24three are um chat GPT Plus members so the 20 bucks a month will get you access to this before anyone else so I I have
1121. 2:52:32thought for a long time chat GPT is worth the 20 bucks now with this it's definitely worth the 20 bucks and with
1122. 2:52:40code interpreter it's really really really really worth it um Emilio's wife appropriately made fun of me for saying I was just gonna put in
1123. 2:52:48another five minutes it's already 11 30. all right yeah I gotta go I gotta go people I know isn't that amazing Jules
1124. 2:52:55it's amazing uh oh the the other question was oh update on the Mandarin video I I don't think we're going to be
1125. 2:53:01anywhere close I think I actually have to pay for um where was it um window no uh um
1126. 2:53:15[Music] let me just log in hey Jen um someone asked did did Dolly three talk about um how long those images took to render
1127. 2:53:29no but they're they're not going to be I don't think they're going to be that long they wouldn't be releasing this if they were super long
1128. 2:53:36um oh I need uh Labs Dot hagen.com I'm down to 78
1129. 2:53:53511 in line so yeah we're not gonna my voice is fried seize the question [ __ ] shut up shut up I resemble that remark from Emilio's wife
1130. 2:54:04um all right I'm gonna put a little picture up here a couple of things if if uh all right we'll do one more pull
1131. 2:54:18um all right I gotta go listen this has been awesome this has been absolutely surreal tonight um thank you to I think it was James who um
1132. 2:54:31dumped thousands of people into the uh chat tonight it was a wild one indeed all the Irregulars are like holy crap what
1133. 2:54:41happened I'm like that too so thank you all for supporting this you know what's funny had we not set you all up as mods
1134. 2:54:49however it was three weeks ago and sort of figured out what we were doing here this would have been total [ __ ] chaos
1135. 2:54:55tonight so thank you for for all the work you do thank you for showing up here every night um for those of you that are still
1136. 2:55:02hanging out from the new influx of people thank you come back um follow me subscribe to the Channel all that sort of happy horse [ __ ] um
1137. 2:55:10this has been really awesome uh Nancy good to see you and major fundraising we had such a crowd we did we had a big one
1138. 2:55:17um we'll talk about future of week uh future of work I know that's something that you wanted to talk about Ann at
1139. 2:55:23some point we'll talk about that this over the weekend if you're around uh just make sure you pop a comment in
1140. 2:55:28there it was a little hard tonight with 17 000 people in here to to get your get your comment in um but uh forever hooked great good
1141. 2:55:38night Vicky Emilio's wife Jules thank you um was that Stevo you're awesome Ted thank you very much historic it was a
1142. 2:55:46historic night wasn't it it was [ __ ] crazy caprono thank you peace out Scott Curtis Silver Fox all right all you all
1143. 2:55:55are awesome have a fantastic rest of your Friday night um the Suzuka Grand Prix should be a good one Max verstappen didn't win a
1144. 2:56:04race last week he's he he got to be pissed he's gonna dominate that race I'll see y'all see y'all tomorrow