aiLL 09/23/23 - Exploring the State of AI Tools Like ChatGPT, Midjourney, and More

Transcript

1. 0:02mm-hmm [Music]
2. 0:24all right let's see
3. 0:51welcome everybody how can I use AI to find clients well I don't know it depends what's your business what are the clients
4. 1:06what do you mean by find clients um you know they're like it might give you strategies for for looking on things like um
5. 1:19LinkedIn groups or Facebook groups or strategies for for Twitter um
6. 1:33yeah I'm not sure companies need help Staffing oh companies that need help Staffing I don't know I would I mean I I would I
7. 1:46would use General sort of chat GPT for that to um to give you strategies and then you might experiment with Bing or Bard that are
8. 2:00connected to the internet and see if it can find you companies that have posted something that indicate that they're you know hiring
9. 2:14but you could probably you could probably do some some paid stuff with LinkedIn I'm not I'm not anything less than 15 000 people a dramatic
10. 2:24disappointment if you weren't here last night we were going along fine and it was 50 and then it was 75 and then it was like I don't
11. 2:34know 100 we broke a hundred and then thanks Stevo and then all of a sudden hate Tobias um all of a sudden I look up and it's
12. 2:43like 4.6 K and then it was like 7K and it was like 10K we hit a peak of 17 700 viewers so so one of our uh one of one of the
13. 2:56Irregulars from South Africa decided that that the AI learning lab needed a boost last night so I was just like holy [ __ ]
14. 3:08um what was amazing was was afterwards I looked at the stats and the stats the the engagement of those 17 000 people
15. 3:14was very very low like the average view from them was like uh four seconds or something like that so I don't it must
16. 3:22be it must be some sort of service where they're they're literally just clicking into lives clicking out clicking into
17. 3:26lives clicking out I don't know who knows anyway that's kind of fun um oh right uh looks like Silver Fox
18. 3:34shared this life so yeah if you'd be willing to share the live um let people know that you're a weirdo you're not weird you're in a regular
19. 3:43that that you're curious about this AI [ __ ] and if they're like oh AI isn't that stupid aren't you weird yeah let
20. 3:51them let them think you're weird let them think you're weird go ahead and do it do it see if I care see if I care
21. 3:57Paulie Z what's happening thanks for the input last night great info awesome appreciate it they know I'm weird already see this is
22. 4:06so you have to at some point you just have to embrace the uh the oddness that is you and and you know quite frankly with with all the stuff
23. 4:15that's coming with this AI stuff they're gonna be a lot of weirdness there's gonna be a lot of weirdness
24. 4:21so I think I think we're all just trying to figure it out so why not figure it out and have a little fun that's what
25. 4:27this channel is all about uh my name is Kyle Shannon this is the AI learning lab and that's Ali the AI learning lab get
26. 4:33it it's it's a play on words a lab is like a laboratory like like this well you might think this just looks like a
27. 4:43home office but no it's a laboratory and then also a labrador you see like it's just I don't because I just I'm
28. 4:53clever I'm clever a lot of times people come in here and they're like sir what are your qualifications watch all your own qualifications huh
29. 5:01well I can do word play huh lab and lab it wasn't my idea I think it was actually Nancy's idea she's one of the
30. 5:11Irregulars yeah she came in one night and she said uh make a graphic with a yellow dog on it you'll thank me I was
31. 5:18like oh a lab the lab oh I see so anyway that's why we have the Irregulars here by the way they're the ones that are
32. 5:26here every night with me they're odd Tobias I'm so sorry that 17K freaked me out I bailed on you I figured I figured some hack was happening
33. 5:38that's really funny you're like I'm going down they're going to take all the money out of my bank account through my
34. 5:45Tick Tock that's [ __ ] awesome was it thirteen thousand nine hundred thousand Bots it was 17 700 at the peak that I saw um
35. 5:59I don't know that it was Bots I it was because a lot of the a lot of the Irregulars were saying it was a bunch of
36. 6:05people from Vietnam I don't know how they knew that but um it was probably you know engagement farms in Vietnam is my guess but yeah
37. 6:14but it was it was it was just crazy it really was crazy I I wasn't you know as funny as I was okay we're up to 114 people now so we're
38. 6:25good so yeah last night we had a crazy influx I didn't really talk about the salon and the Discord because like I had
39. 6:33no idea who the people were so I didn't want to necessarily invading invading you know that sort of ongoing space you
40. 6:41know [ __ ] up alive is fine I'm I'm exploring using it for I don't wait for identifying learning outcomes for student readings
41. 6:51oh that's really interesting for identifying learning outcomes I don't what does that mean brother 52
42. 6:59I don't know what that means identifying learning outcomes like like analyzing their work or or analyzing what they
43. 7:06learn and then how do you do that is that a is it like a a form Oh I thought I was it was just all Librarians I told you about finally
44. 7:15tuning in yes yeah it was they put us up on the uh the Nash the big screen at the at the stadium event for librarians in Detroit
45. 7:25everybody logged in with their phone could have been that it could have been a bunch of librarians I don't know
46. 7:33oh that's really funny uh welcome everybody my name is Kyle Shannon this is the AI learning lab and uh a little bit about my background
47. 7:41I'm an entrepreneur um I've been I've been uh digging deep into this AI stuff for about two years this channel is about kind of that that
48. 7:51inflection point that started on November 30th 2022 when Chachi BT was launched uh you did great and and not
49. 7:58flustered I would have froze I was someone noticed that that I blushed I was like what do I do with this and like
50. 8:05every time I looked up there was another 5 000 people so I just kept restarting like you know my little my little starting Spiel that I do
51. 8:13so but thank you for the kind words um anyway I've been digging into this uh this this generative AI stuff for a while and and
52. 8:22that moment of when chat GPT launched and then went from zero to 100 million users in six weeks changed everything
53. 8:31and it's this inflection point that prior to that inflection point machine learning and artificial intelligence was
54. 8:37reason really the purview of researchers and scientists and engineers and things like that after that day it's the rest of us
55. 8:46the rest of us get to play with these tools so so we get to we get to we're lucky enough to be living at a time
56. 8:55when we can just go to this thing called Chachi PT chat.openai.com or this thing called Claude or this thing called Bard and we
57. 9:04can just ask it questions and it answers them and it's and it's not a Google search it it may seem on the surface
58. 9:11like a Google search but if you have not played with Chachi BT go there right now go to chat.openai.com start playing with it
59. 9:18and you'll see it's there's there is a there's a there's a something you gotta play to win you know what I'm saying you just go
60. 9:28there you start playing you gotta play to win I actually wrote an article today on LinkedIn um about lateral play and and what I mean
61. 9:39by lateral play is like you know I've got my skills that are you know I've spent you know my my five plus decades honing
62. 9:48and lateral play is like playing outside of that like I was never good at drawing or art so me going to mid-journey is is
63. 9:57lateral play it's it's off to the side of what I do but I really enjoy creativity and I enjoy visuals and I'm a
64. 10:04photographer like there's a lot of things I do that if I had the skill to draw I would do good stuff with it
65. 10:13and now with mid-journey I can I can take my wordy brain and make it a picture brain or I can go over and play
66. 10:21with code where you know I've got add and I never really learned code Beyond you know scripting in HTML and [ __ ] like
67. 10:28that and now I can go write python right so lateral play is this idea that we can use chat GPT to improve our productivity at work
68. 10:41but all of that lateral play feels like what I wrote in the article is is that lateral play kind of felt like I was
69. 10:48cheating on my business like like this is too much fun like I'm having fun over here that's not really work though but
70. 10:55what I learned is it was work and it was contributing and it was contributing to the quality and the productivity and in
71. 11:01ideation and all sorts of stuff like that so so a big part of what this channel is about is just [ __ ] play
72. 11:07just go there right now don't you don't have to be technical I have a degree in acting all right you know I spent years as a Tom Arnold
73. 11:20impersonator look at this it's [ __ ] uncanny he and I are like we're best I've never met him but that's not the point okay
74. 11:30I made as much as ten dollars an hour impersonating Tom at malls at mall openings but they there aren't malls are
75. 11:38all closed now but it was a good gig for the week that I did it it was I anyway back to AI he seems like kind of a dum-dum I I
76. 11:51don't know I went on this place it's called AI learning lab and I thought oh it's going to be a scientist and I was
77. 11:56expecting him to wear you know like a like a white coat like a lab coat and then and it was gonna be like there's
78. 12:02gonna be like computer screens behind him and it was just like a basket and some magazines and they seemed like kind
79. 12:08of a dumb dumb and then and then he kind of made fun of people asking what his qualifications were and I was sitting
80. 12:13there thinking well what are your qualifications what are you qualificating I don't even know how long did you watch it I watched it
81. 12:20for I I don't know I watched it for like three hours or so you spent three hours on Tick Tock with some idiot well yeah
82. 12:26yeah did you learn anything I kind of did I kind of I kind of I think they say I think it's going to be big
83. 12:33it could be bigger than the internet that's what he said he said it could be bigger than the internet and I and I
84. 12:38think the internet's big I got it installed on my phone and it's am I I'm all out of I'm all out of room on the
85. 12:44phone now because I think I put all the internet on it you know what I mean it's just [ __ ] this is crazy it's just crazy anyway
86. 12:52um do you want some corn pie I made a lovely corn pie today would you like a slice okay let me go get that for
87. 12:58you all right and scene all right trying to learn how to advertise with AI where's the best play to go to learn
88. 13:06place to go to learn so so what I would do he surmise that I'm frustrating to a lot of people um my answer to almost every question
89. 13:18right now and probably for the next 18 months is there's no place to go to learn how to do advertising with AI
90. 13:29because all of this stuff is changing so fast now now that doesn't mean you can't learn with AI about advertising
91. 13:38but what I would do is and I why don't we why don't we go do it right now let's let's go play um
92. 13:54and I sing Hawaii Five-O songs yeah all right so we are at chat GPT this is chat GPT if you've never seen it
93. 14:06this is it it's you're like that's it why is everyone talking about it doesn't look very exciting it's not it's a janky
94. 14:14piece of [ __ ] right now so down the left hand side here are your little uh your your history these are little conversations I've had
95. 14:23with chat GPT these things here are just uh suggestions like if you literally can't think of anything you just click
96. 14:30it just go click and it'll make a prompt for you um there there are two different kind of modes of chat EBT there's the free mode
97. 14:39which you get access to chat CPT 3.5 and then there's paid mode which is called plus and you get access to gpt4 and to
98. 14:47plugins and to I don't know a bunch of stuff so so let me show you one other thing because I I'm kind I I'm not copping out
99. 14:58on the answer wait why oh no I think Microsoft Edge just ate my memory because I can't go you know and quit
100. 15:09Microsoft Edge hang on we'll come back to this okay I'm gonna quit leave
101. 15:25Microsoft Edge is full of memory leaks yeah if you've got too many tabs open because you have the add Edge will crash on you
102. 15:38every three hours or three and it looks like it hung my machine yay you did it yay yay Microsoft you're a winner keynote not responding
103. 15:55uh good Lord force quit I know hold please well he you know someone asked him his qualifications and then he said he kind
104. 16:09of made fun of it like well he doesn't need to have qualifications and then he couldn't get his computer to work and
105. 16:14it's like I just felt like I was watching a geriatric trying to make a computer work and he kept asking about how does tick tock work
106. 16:22he I think his wife's name is Marge or something he goes Marge how's this thing work anyway it it was sort of funny but
107. 16:29then I just wondered well what are his qualifications what what were his qualifications and I I still don't know I still don't know
108. 16:37but you watched it for three hours well that's not it's not really really I had a Coco I had a little cocoa and I was I
109. 16:45was making the corn pie all right I gotta start I gotta restart my machine people it's it's just not in a happy place yeah
110. 16:55QuickTime Player interrupted it try again do I want to save that why is that distorted there's just weirdness going on
111. 17:08I'm gonna say don't save there uh I'm gonna say don't save there I will say oh that's recovered hang on sorry about this people
112. 17:21um did you find out why so many people were in the in the live last night um yeah it was we're down we're down to
113. 17:2750. we're down to a manageable 55 tonight the uh the peak last night was 17 700. um um what's his name James John uh wait hang on
114. 17:42one of the one of the Irregulars the dude from South Africa whose name James kamua he's from he's from South Africa he's
115. 17:53from uh nehru Nairobi County Kenya according to Firefly Firefly did some some sleuthing um no I don't want to force quit
116. 18:07um I'm gonna have to yeah boy yeah this this machine is not happy ignore okay hang on hang on people it's gonna be so
117. 18:23worth it [Laughter] oh man so I think so James said last night that it was him that that he liked he liked the
118. 18:37live and so he just sent like 5 000 people and I I like I don't know how you um get like thousands of people to view
119. 18:47alive I assume there are farms like there are engagement farms in China a lot of the people last night said it was
120. 18:54uh Vietnamese a lot of Vietnamese people so it might have been some engagement Farm in uh in Vietnam that that he used
121. 19:04um but it was it was really exciting I mean it was just [ __ ] insane we went from like 100 people to like 17
122. 19:12000. and I was just like huh what do I do um what was wild though is that it um boy this thing is just this is bad I
123. 19:22think I'm just gonna do a hard restart on this thing um [Music] um am I doing
124. 19:37yeah I went and I looked at the um I looked at the engagement after the live was over I went for like an
125. 19:45extra hour because I didn't know what was happening I didn't know if it was tictographers James or what it actually
126. 19:51meant and I looked at the engagement for the um you know for the live and it it wasn't dramatically different than you know
127. 20:00when we just have like 50 or 75 people in here so so that's why I think it was engagement farming which is just basically like P
128. 20:10and the the average view of non-followers of the channel was four seconds and the average view of followers was like 18 minutes so so
129. 20:19there was definitely a difference between people who knew about the the lab and those that didn't um okay there's Poe
130. 20:29all right so it restarted all that stuff let me restart Keynote I'll be with you let's go seriously this is gonna be really good what's your
131. 20:37opinion on the famous authors taking opening out of court for copyright um I understand why they're doing it I
132. 20:43don't think they understand how chat GPT works or how large language models work here here's this is my real opinion on it
133. 20:56if mid-journey or stability AI or open AI or Google or any of them went to all copyright holders first and
134. 21:07said hey we're thinking about training up a model on all of the internet which includes some of your [ __ ] do we have
135. 21:12your permission everyone would have said no they would have been stuck in legal battles for decades so so
136. 21:20we are in a situation where they all basically had to break some eggs and ask for forgiveness rather than permission
137. 21:29and that's that's where we are right now I equate it if you understand how embeddings work and the fact that the
138. 21:36original documents are not being stored in some big database they're they're turned into vectors they're turned into
139. 21:42mathematical numbers in a thousand dimensional Vector space where all these words are represented in semantic clusters now those semantic
140. 21:52clusters are derived from all that copywritten work but it's not like when you search Chat GPT it's like oh let me
141. 22:00go into silver Sarah Silverman's book and copy out a passage and paste that into your answer that's not how they
142. 22:06work it's more it's more equivalent to someone goes to college for four years and they study the literary works of
143. 22:17stand-up comics and they come out after four years and they know all these different references from all these
144. 22:22books from the papers they've written and what they've studied and all the stand up they watched and this and that
145. 22:26and you ask them to write something and they quote something from Sarah Silverman's book that they love like
146. 22:32it's more like that it's more like recall from some nebulous mathematical brain of all these Concepts and semantic clusterings of words than
147. 22:45it is copywriting but it was trained on some of that work and mid-journey was trained on copywritten images and but it's not stealing them
148. 22:56it's not how the technology works so I think so it's so it's a very gray area I understand why they're doing it um and uh
149. 23:06I don't fault them for it but I I don't think they're going to hold up in court what's likely going to happen is open AI is going to cut
150. 23:18um you know 100 million dollar deals with big Publishers like the New York Times and time Inc and they already did
151. 23:26one with the AP the Associated Press and so all of these all of the big companies are going to cut deals with
152. 23:32big companies and then basically below the 100 million dollar deal threshold it's going to be the legal equivalent of
153. 23:40go [ __ ] yourself is is what I think now does that mean that people aren't going to keep suing them oh no people are
154. 23:47going to keep suing them again and again and again and again and again and at some point they'll probably counter Sue
155. 23:55um Microsoft put up put out a statement last week or the week before that they're going to pay the legal
156. 24:04bills of anyone who gets sued because they used chat GPT to generate content basically so that says to me that their customers
157. 24:13are worried about it right and their customers are saying well we can't use this if there's any risk we're going to
158. 24:17get sued so Microsoft said we'll pay for your lawyer bills um that says to me if I were a lawyer I'd be like let's go sounds like class
159. 24:28action it sounds like class action it sounds like class action let's go yeah all right see you Denver okay
160. 24:43all right now I've got keynote back to where it was let me get edge up and running thank you to Microsoft for absolutely smashing
161. 24:55the memory in my Mac to the point that it was non-recoverable that's an awesome feat it's actually hard to get Max to crash that hard
162. 25:05very accomplished if if opening I lose this case though what would it mean would they have to retrain on copying so here's the thing
163. 25:14they can't you can't so so like I said it's not like there's a giant database with all this data in it what they do is they take all the
164. 25:25data they found and they put it into um into into a form that that it can be embedded right it can be turned from
165. 25:33words into mathematical vectors and and they do this across billions if not trillions of documents right at least trillions of words
166. 25:46probably billions of documents trillions of words and what goes into the vector database are all these numbers that are
167. 25:53tied to all these words and there there's literally no way to go well you know let me go into the Sarah
168. 26:02Silverman section and pull that data out if she if she won that lawsuit so the only option is they shut the [ __ ] down
169. 26:11and start over and I just I don't think that's gonna happen I like it's I I don't think there's a a judge or like I
170. 26:19don't know how how it would get adjudicated but I I just the the stifling effect that would have on on the industry it just I I just
171. 26:30can't imagine it happening right I think I think all of those lawsuits are going to settle um outside of it all right let me see
172. 26:42what was that gonna show oh I was gonna show let's go to uh what's it called chat.openai.com so if you're new here
173. 26:52welcome got a few new people in here my name is Kyle Shannon this is the AI learning lab if you've not played with chat GPT go right there
174. 27:03chat.openai.com that's what I'm doing right now I'm going to chat.openai.com a while ago before where before I found
175. 27:11out my computer was hard crashed um someone asked me about what AI would I use to um to learn advertising and I said well
176. 27:24there's not really an AI that you can go to to learn advertising there might be that URL on the very bottom futurepedia.io
177. 27:34is like thousands I don't know four thousand forty five hundred AI tools so you can go there and you can search
178. 27:41for advertising but I do not recommend it my stock answer to any very very specific request like how do I learn
179. 27:51this how do I learn that how do I you know how do I get AI to do you know this specific thing for me there is probably
180. 27:59some sort of tool at futurepedia.io most of those tools are just tools that someone figured out a prompt that generated a cool output in
181. 28:10chatgpt and then they turned it into an app they're called wrapper apps they're basically just doing calls to open AI
182. 28:17with with a clever prompt or a hundred and then they charge you 39.95 a month and it's all stuff that you could do anyway so
183. 28:27so what I wanted to do let's go show everyone um how I would attack the problem so here we are at chat gbt this is my chat history
184. 28:38um there's Chachi PT 3.5 and 4. you have to be paying the 20 bucks a month to get to gpt4 I think it's worth it but you
185. 28:45don't have to so we'll just do 3.5 for now these things are just you know uh if you literally can't think of anything
186. 28:51you click on those and it'll write a prompt for you um so what I what I'm going to do here is I'm going to say oh come on
187. 29:07why is that not working now uh it's serious messing with me again Yeah Siri I thought so
188. 29:24stop Siri go away Siri shut down uh I don't know how to turn her off
189. 29:40I'll just return Syria this is really annoying
190. 29:55um [Music] oh good God okay I suppose I could type but I don't want to type I want to use my microphone
191. 30:19oh something happened hello ah this is this is maddening this is just Madness you know it's got to be
192. 30:29worse than what I'm going through is you watching this there we go okay now it's working I want you to act like David Ogilvy
193. 30:45and help me understand the world of advertising period I'm just getting started and want to learn advertising and I want you to
194. 30:58serve as my mentor period do you understand what I'm asking for and what do you think is the best way for us to begin
195. 31:12so and and so now it's it's just popping up an answer now why did I do that I could have just said teach me about
196. 31:19advertising the way the large language models work is if you just don't give it any context it basically pulls from everything so
197. 31:30there are idiots that talk about advertising and there are ignorant people that talk about advertising and
198. 31:35there are brilliant people that talk about advertising and there are clients that talk about advertising so if you
199. 31:41just say tell me about advertising it's kind of pulling from all these different data sets at the same time and they just
200. 31:47kind of get mushed together and what you end up getting is very very generic [ __ ] so by me starting out saying I want you
201. 31:54to act like David Ogilvy who's you know one of the most famous admin in all of the land um and say that I want David to Mentor me
202. 32:05and ask things like do you understand what I'm asking you it's providing what's called In The Biz context right
203. 32:12so it's kind of narrowing the window where chat GPT is going to focus when we're having our conversation so the
204. 32:19answers tend to get much much better so let's see what it said of course I understand your request I'd be delighted to guide you through the
205. 32:28world of advertising much like David Ogilvy would David Ogilvy was a true advertising Legend known for his principles
206. 32:34to begin your our mentorship in advertising we should start with some fundamental concepts understanding your
207. 32:41audience the big idea research creative execution headlines and copywriting testing and measurement media planning
208. 32:48brand image and continuous learning um so then I'd say um well I just want to make some social media ads should I use Facebook or Twitter
209. 33:02now this should slap me back a little bit and say well you should start with a strategy it should do that it shouldn't
210. 33:10just let me go right to a tactic but let's see yeah choosing between Facebook and Twitter for your social media depends on
211. 33:19your target audience goals and the nature of your product or service both platforms have their strengths and
212. 33:24weaknesses so let's consider some factors to help you decide audience demographic demographics Facebook
213. 33:30Facebook has a broader and more diverse user base Twitter attracts a younger audience with a significant presence of
214. 33:3718 to 34 year olds although I don't know that that's accurate these days but it might be um so anyway so so this is what I would do is if if
215. 33:51if I were trying to figure out essentially any problem right now I would start with having a conversation
216. 33:58with chat gbt now or or Claude or Bard it doesn't matter which tool you use um start by having a conversation with
217. 34:06one of these things because it will educate you really quickly and then you can start to go explore specific areas
218. 34:13and things like that so anyway have you seen chat Dev yet no I haven't I haven't even heard of it let's go see what chat dev's all about
219. 34:22I assume it's a coding bot is it built on top of code llama chat Dev GitHub chat Dev create stands uh
220. 34:37the primary objective of Chet chat that was creating easy to use highly customizable extensible framework which
221. 34:43is based on large language models oh I see so this is this is a oh llama I got myself a llama from Winston
222. 34:52thank you Winston hello Silver Fox I am llama you guys are awesome um yeah it looks like this is a thing to
223. 35:01to allow you this is this is probably not just similar to uh what the hell is it called GPT for all there's a project called GPT for all
224. 35:10where you can install um an application on your Mac or Linux or Windows that you can then install large language models in and create your
225. 35:20create your own chat Bots you can also do that in zapier they've got this thing called interfaces now um so
226. 35:30let me look over on the Irregulars Channel okay everything's good there turn Siri off by going into settings
227. 35:38scroll down to Siri and search top three buttons thanks Nancy you know I think I'm going to keep it on because the
228. 35:45comedic gold that Siri gave us when she got jealous of Bing I don't want to lose that but boy is it annoying when she
229. 35:52takes over my microphone uh oh I don't have my photos thing up hang on let me get photos up so I can get all that pretty pictures back like
230. 36:05like Ali the AI learning lab there she is she's so cute all right okay so welcome everyone uh if you're new
231. 36:19here pop any questions you have about AI in the comments below I'm happy to talk about them I'm happy to talk about my
232. 36:25background I'm an entrepreneur I was an entrepreneur in the early days of the web um the parallels between the early days
233. 36:33of the web and the early early days of this AI stuff are uncanny and wild and exciting and um I feel lucky to be alive
234. 36:46at the beginning of one of these things for the second time in my life um I just missed the PC one but um but I got the World Wide Web one
235. 36:56and I'm here for this one so anyway it's it's exciting times do [Music]
236. 37:11all right so we don't have a ton of questions maybe did I miss a bunch I may have it might have jumped me to the bottom
237. 37:17Tick Tock jump into the bottom you talk Camelot oh yeah I missed a lot of questions okay if you have questions pop them in the
238. 37:31comments below I also do have some mods in here they are they are formed from the Irregulars so the Irregulars are
239. 37:37people who show up here regularly um if you're interested in modding and you show up here a lot just let me know
240. 37:44DM me and I'll find a way to get you hooked up uh yeah there's turn Siri off I saw that earlier I'm I'm Gonna Leave Syrian just
241. 37:53for the comedic value [Laughter] um Kyle seems so excited finally the
242. 38:07crowd's showing up for the play yeah well I I mean I was excited because I thought it might have been like Tick Tock just blasting the
243. 38:17algorithm but it wasn't it it was engagement farming and it you know when it went from like a hundred to forty six hundred and then
244. 38:29from that to 10 000 and then from that to 17 000 that's that's not you know I could see if Tick Tock did it it would
245. 38:36probably go to like 100 then 200 then you know 500 maybe but it's not going to jump you know 10 you know thousands at a
246. 38:44time I'm hosting my first AI seminar in two weeks friends signed up six people then told me um wait why is this not working
247. 38:55told me what she'd done oh that's funny I'm I'm doing a uh I'm doing a workshop too in two weeks starting on October 1st
248. 39:03or 4th October 4th so here let me let me let me pull up the info for it and I'll go through I'll actually I'll
249. 39:10I'll start with the the workshop that I'm doing so I got invited by a woman named Rhonda Phillips who's got a uh
250. 39:18she's a creative coach and a voiceover actress and she's got a whole community of creative people so we're doing this
251. 39:25thing called the AI launch pad um where we're doing uh three two-hour sessions a week apart so October 4th 11th and 18th if you go to
252. 39:35rhondasvoice.com events you can see that there um and yeah it's a similar sort of thing where it doing workshops right now I've
253. 39:44I've real mixed feelings about it I think it's a good thing but everything is changing so fast that I'm really
254. 39:51reticent to put too much down in writing to say this is a good thing so so what I'm going to try to focus on as much as possible is
255. 40:00the craft of of playing with AI as opposed to here are six prompts that are perfect and and things like that so
256. 40:09anyway it should be really good um and yeah so that should be fun um okay so if you're new here let me let me
257. 40:20show you a couple of things these are these are my kind of recommended things to go play with if you're brand new to
258. 40:27AI if you haven't played with chat gbt before that's it it's not chatgbt.com it's chat.openai.com go there right now and
259. 40:35start playing or go to Claude or go to Bard either one of these all of these are have a free option Claude is
260. 40:44really powerful because you can put in giant prompts and it doesn't choke so you can copy and paste really big
261. 40:49documents into it and and just start interacting with it in fact it's a it's most of the way that I use Claude is if
262. 40:55I've got a document that's bigger than a page or two because I know chat CPT will choke on it Claude won't
263. 41:03so just go play with those if you don't know what to say there at all if you're just like I don't know what to do Kyle
264. 41:09it's just a box what do I put in the box go to prompts.chat go to that thing down there and that'll tell you what to put in the box
265. 41:20and you you all you always want to make sure you put good stuff in the Box you can put good stuff in the Box you get
266. 41:26good stuff out what is corn pie you ain't never had corn pie oh sweetie you ain't live till you had some of my
267. 41:36corn pie now you got to get to Del Monte you get to Del Monte whole sweet corn don't get the store brand don't get the
268. 41:45store brand you want the Del Monte okay and you got to get the sweet one because the one they got the one in
269. 41:53Spring Water you don't it's it's flavorless it's like eating it's it's like eating low beads of Styrofoam no
270. 42:00you want the sweet whole corn Del Monte that's this that's the secret that's the whole secret and I wanna I'm not even
271. 42:08gonna get into my pie crust okay all right um anyway we do skits here too it's like Saturday Night Live if Saturday Night Live we're
272. 42:21even less funny [Laughter] uh okay um so those are the three this is what I call the adult table this is the this is
273. 42:31the adult table these are these are the big three that I use pretty regularly this is the Kids Table you'll notice
274. 42:37Bing is barely readable because because I dimmed I dimmed Bing because Bing pissed me off because the safety guard
275. 42:45rails on Bing are so egregious right now that it makes it essentially unusable but it's still really cool conceptually
276. 42:52it's connected to the internet it can generate images it's going to have Dolly three in it really soon which is gonna be super amazing
277. 42:59um it can analyze images it's it's just there's there's things about it that are right and then it but it's it's in
278. 43:06timeout right now so that's why it's there um perplexity.ai pi and Poe are all fun to play with okay so those are those now
279. 43:14for the image things if you if you're like but Kyle I don't care about words I like images I am a purveyor of fine
280. 43:23images and I would like to understand how I make images with um with these tools these are the these are the ones mid Journey midjourney.com
281. 43:34will take you to a website and when you say join the beta it will take you to a whole nother application called Discord
282. 43:40and if you've never been on Discord it will confuse the ever living [ __ ] out of you wait wasn't I just in a website yes
283. 43:48you were and now I'm in a thing that looks like slack but it's kind of a shitty version of slack yes it is and
284. 43:55and now I have to go into slack and I have to make an account there yes you do and then you have to go into these weird
285. 44:01Channels with these these young people and they're saying things and making slash commands yes and then I have to type command line
286. 44:12interfaces to make images yes but it's the best so if you want the really good images you go to Mid Journey
287. 44:22um ideogram not as good images but it can do text so that AI learning lab there that was done with ideogram ideogram ideogram idea
288. 44:32Grant idea Graham janky pieces of [ __ ] all these tools right now these are all I'm not weird I'm irregular laughs
289. 44:44uh these are all ID gram okay um leonardo.ai is a is a website that you can make images at it's really good
290. 44:53it's built on top of stable diffusion hey Jen I'll come back to Adobe Firefly and Photoshop generative fill are out of
291. 45:00beta now which means you can use those images commercially those are if you're in a company that's probably your best
292. 45:07option because it hasn't been trained on copywritten stuff and then openart.ai promptbook will teach you about how to talk to these
293. 45:16these things I'm going to show you one example of something that hey gen does
294. 45:31that will blow your mind hey look more me [Laughter] okay I'm not really Tom Arnold this is Tom Arnold okay
295. 45:47um hey Jen if you go to labs.hagen.com and then click on the video translate button and if you upload a video of yourself
296. 45:58speaking I assume English if you're listening to this in English and you can pick one of I think it's 12 or 14 languages now that you can
297. 46:06translate it into watch this so I I actually cut together some of me speaking English with some of me speaking Italian check this out
298. 46:16where is it um so we've heard from a number of clients that modular content is important to them getting more efficient with how we
299. 46:25create content and how we leverage contents behind me you see these little green boxes um that's how we think about
300. 46:39stories at storyline we break stories down into little atoms right little modules I don't say storyline like that I say
301. 46:54Story vine and and this makes me say esta Divine uh it's amazing it's sinking the lips it's syncing my voice it's
302. 47:04matching the Cadence and the inflection of the English perfectly the timing to all of my many hand gestures because I'm a hand talker
303. 47:13um are all there that tool is labs.hagen.com and it's their video translate tool um they've got a backlog right now that
304. 47:22is gargantuan so if you want anything um done quickly you're gonna have to pay for it um but that that's a game changer in my
305. 47:31opinion and the uh the uh what's his name one of the co-founders um was talking about they're not that far from that thing being able to be
306. 47:41essentially real time so so we are probably within a year or 18 months of true real time translation
307. 47:56in video so imagine being on a zoom call with someone in another language in another country that doesn't speak your language
308. 48:04but hearing you in your voice in your Cadence speaking their language fluently and you hear them speaking yours fluently
309. 48:14we are essentially a year out from that odd so anyway that's cool um the other thing that that might be of
310. 48:25interest if if you're into this AI stuff um I think it's an adventure what this channel is about is about kind of
311. 48:32leaning into the adventure of whatever the hell this stuff's gonna be I created a community called the AI Salon we meet
312. 48:39every other week we've got a meeting coming up Tuesday night we've got a Discord server um we've got a a special little section
313. 48:46in there for the Irregulars to hang out so if you start hanging out here you can hang out there as well and talk you can
314. 48:53make mid-journey images there you can ask questions there you can share news it's it's a it's a good community so
315. 49:01um so that's what that's about storybinds my company that's my social media tag so anyway that's it that's that's some of that stuff
316. 49:14um by the way Irregulars you're you're up to 20 21 000 likes you you're like you're out like machines tonight
317. 49:21have a hot Saturday Saturday night date with Kyle just going click click click click click click click what'd you do
318. 49:27Saturday night well I don't want to talk about it it's a little embarrassing how do people create the AI videos I've
319. 49:39seen okay Kim Let's go do that since I have the the thing down there let's go make some videos I'll show you two different ways to make
320. 49:49them let's see all right we'll start we'll start with Runway ml there there are a couple of things there's one now called Pika I
321. 49:58think it's called but there's also Runway ml so we're gonna start with Runway ml because that's kind of the the first one to get
322. 50:04it get it kind of right and um it's pretty slick so it's it's uh app.runwayml.com so they so what they
323. 50:20say here is bring your image to life with Gen 2 and then you can start with an image or you can start with text
324. 50:27and if you start with text you can literally just describe something well let's we'll do both of these so start with text um
325. 50:50a man and a woman holding hands walking through a field of wild flowers on a breezy day I don't know that's good
326. 51:05generate four seconds so um I haven't spent a lot of time doing text to video but for the most part it
327. 51:15will give you what you ask for the thing I have done is the next thing I'll show which is image to video so so one of the
328. 51:23nice things about image to video is you can go to Mid journey and make all sorts of cool images and then you come here and you just
329. 51:30upload the image and basically it takes that that first frame is the start frame of the video and then it makes a four
330. 51:36second video after that and you can extend that four different times you can make up to an 18 second video clip from
331. 51:44a single start image or from a single um text prompt like this so what the people are doing that are doing this is
332. 51:53um you know they're they're probably generating you know 25 videos for every every single clip that they put in one of those hand edited
333. 52:04um sequences um but so this looks like it gave us the woman but that not the band but let's see if the man mysteriously emerges no
334. 52:19but that's actually not bad I mean you know with the exception of the fact I wanted a man and a woman um it looks like a woman walking you
335. 52:29know down a trail uh you know about to have a hay fever attack so this this could this could be the the good first scene from a uh from
336. 52:39an allergy medicine commercial okay so that's so that's starting with text and and now I'm going to start with
337. 52:48an image and so let's see I've got here upload a file select from assets I've got a bunch of assets already uploaded where are they
338. 53:03why are they not showing up there we go because they were loading
339. 53:19oh that's cool all right so I got I've got I'm a photographer so this is a like a sunset Cloud thing I I took a picture of so
340. 53:31one of the things you can do here is let's see uh [Music] is it this thing
341. 53:47image plus description okay motion so I can I can change my motion how much motion I want so I'm going to put that
342. 53:54kind of high so I'm going to move that up to eight and then that's duration I thought there was some oh Advanced
343. 54:03Camera controls okay I I thought there was something here so there's Advanced Camera control so I'm gonna I'm gonna
344. 54:08open these Advanced Camera controls so there's horizontal vertical zoom and roll and then the speed so I'll leave the speed at 0.5
345. 54:20and I guess we'll go we'll pan it to the right so let me go horizontal to the right and so now I say generate so now it's
346. 54:28going to give me movement of 8 out of ten it's going to pan it to the right and it's going to use my cloud photo as the start image
347. 54:36one of the things I've noticed about the videos that are created from images is that the the sharpness of the image
348. 54:43degrades really quickly it makes it very blurry but it does create for the most part stuff that is interesting and
349. 54:53usable so so this is absolutely one of those tools you know I kind of jokingly talked earlier about oh all these tools
350. 55:00are janky pieces of [ __ ] um they are they're all janky pieces of [ __ ] right now because all this stuff is
351. 55:07so new so the art of playing with them right now is kind of learn where their boundaries are what are they good at
352. 55:15what are they not good at right now and then figure out ways to work around that and don't get too married like if you
353. 55:23find the perfect way to make Runway do in you know videos better um don't get too married to that because
354. 55:31within months they're going to come out with a new version that's probably going to solve all the problems that you've
355. 55:36created fancy workarounds for in fact one of the things that the the runway ml people were doing um that were making some of those longer
356. 55:45form movies is that they would they would they would render a video then they would take the last frame of
357. 55:52the video use that as the start frame of a new video so they they were artificially extending the length of the
358. 55:59videos by doing that and then Runway just came out with now you can extend these up to um up to 18 seconds so so all of those
359. 56:09workarounds that people created uh Runway just launched a new version and it did that so let's see what it looks like
360. 56:18oh that's really good that's really good that's way more coherent than I've seen before now what I would do is
361. 56:30I would actually start this at the end I would start it where it's bright and then end it where it's more Sunset right
362. 56:37so I would sort of I would flip this in post if I wanted to use it because this seems like it's getting earlier it's it's like the uh
363. 56:47I guess it could be sunrise couldn't it yeah this could be sunrise um but I took it at night but that's
364. 56:54there you go that's pretty good and then you can extend it for four seconds and extend it again and so from an image so that was from a
365. 57:02photograph I took but that could just as easily have been mid-journey so that's that wild huh isn't it crazy
366. 57:08isn't it crazy it's crazy it's crazy crazy crazy crazy crazy crazy crazy all right let's see
367. 57:23Share Bear hello Irregulars good day to you good day do they milk the finances wait this could be so brilliant for
368. 57:32b-roll yeah exactly I mean so so think about this okay okay yeah big future of work thing I don't know if and funding fundraising
369. 57:42is in here but she wanted me to talk about future work future of work stuff 18 months ago the image generation tools
370. 57:54were so bad that they were laughable stable diffusion mid-journey you'd say make me a horse and it would
371. 58:02make you some sort of four-legged thing that looked like it was a genetic mutation of something from another
372. 58:07planet that kind of resembled the horse um within 18 months mid-journey now is doing you know full photographic realism
373. 58:18for the most part they fixed the hand problem um they haven't quite dealt with the with with letters and words but that's
374. 58:26coming in Dolly three if I need an image for a presentation or for a concept I historically would have gone to stock
375. 58:38photo houses I'd have gone to Google I'd have gone to you know I'd go to whatever the resources are that I go to and I would have spent
376. 58:49hours trying to find an image that was acceptable for the thing I wanted and then if I actually wanted to get
377. 58:58something that was really close to what I wanted I would have to go do a photo shoot right I have to hire a freelancer now
378. 59:09if I'm facile with mid-journey or stable diffusion or adobe Firefly I go in and I type exactly what I want and it gives me
379. 59:17exactly what I want and if you see what's coming from Dali it's going to give you exactly what you want because
380. 59:24it can take all of the context of a big long conversation in chat GPT and incorporate that into an image with text
381. 59:32that's coming in like a week or two and then you look at Runway ML and you've got the same exact thing if you
382. 59:38wanted to comp up to mock up a video of hey here's here's the the ad concept that we've got and you want to put
383. 59:45together a bunch of pieces of b-roll and you know record a nice voice over and go pitch that to a client
384. 59:52you would have had to go pay for Adobe stock photo you know stock video it wouldn't have been close to what you
385. 59:59want you'd have to you'd have to spend hours looking for the right thing and then tweaking it and color grading it
386. 1:00:04and all that sort of stuff and now you can just ask for it you can take the images from a storyboard and use them as the start
387. 1:00:14frames of Runway ml videos I told a bunch of fellow students at fit 30 years ago that this day would come Winston Grant you are correct so
388. 1:00:27um um I was 30 years ago oh share bear thank you that's my favorite of the uh of those thingies the dangly lights I love that um
389. 1:00:37I was I I was a co-founder of agency.com back in the early days of of the World Wide Web so I was I I I loved that
390. 1:00:46neighborhood where fat was but um but yeah yeah this the day's here the day's here it's like like we knew in the
391. 1:00:54early days of the web that this stuff was that that that some version of this was coming right but the tools of of the
392. 1:01:02early web were just they were so primitive right it was literally like click on a hyperlink and it will take
393. 1:01:08you to a document and what you could kind of Imagine at that point was this interconnectedness right that that came
394. 1:01:15to be right instant Communications um you can you can be anywhere in the world you know instantly um and our ability to connect with one
395. 1:01:28another is is you know seamless now it's now just the fabric of society for the most part and we kind of knew something like this
396. 1:01:39was coming but I think all of us including a lot of people in machine learning an artificial intelligence
397. 1:01:45assumed that the stuff that we get to play with today that stuff as powerful as chat gpt4 would be 30 years from now 50 years from now
398. 1:01:59probably not in my lifetime and it just like kind of materialized it's amazing it's amazing right and it's you know
399. 1:02:11it's not that it's not that chat gbt is this new thing it is a new thing but it's just an interface on a foundation
400. 1:02:19that's being built for 30 or 40 years neural networks and machine learning and Transformers in 2017 and gpt1 and gpt2 and gpt3
401. 1:02:30and then GPT 3.5 they build this proof of concept kind of thing called chat GPT so they have something to show when
402. 1:02:38Microsoft makes their big announcement of their investment and the world lost their mind I'm spending less time in Adobe stock and
403. 1:02:48now creating magazine covers for my business this is cool Lauren right in mid-journey yeah it's amazing it's that yes
404. 1:03:04I'm curious what kind of magazine what's what's the business that's really cool um do you have parallels
405. 1:03:20I've installed parallels occasionally in my life it's just maybe it's easier than to use than it used to be um I have a
406. 1:03:31I don't want to maintain two operating systems it's rough enough maintaining one if I did install parallels then I could
407. 1:03:42run The Tick Tock livestream thing and maybe I could stream this to more places than just Tick Tock I don't know I'm
408. 1:03:51hives as they are right now these more intimate things you know just just just a 17 000 people hanging out if you
409. 1:04:00weren't here earlier last night we had an infiltration of of uh it was first five and then ten and then
410. 1:04:0915 and then 17 000 people and then it dropped down to 10 and then it went back up to 15 and it was crazy but it just ended up being some
411. 1:04:17engagement farming which was interesting this is cool but can it make you money oh I don't know print on demand can it [Laughter]
412. 1:04:29if you're wondering how to make money with chat EBT I have opinions on that it's one of my side businesses a monthly
413. 1:04:36publication called the dance calendar oh super cool that's that's cool um oh yeah Winston Make It Rain thank you appreciate that
414. 1:04:50will there ever be an AI that can turn your script idea into a TV episode or movie yes it exists it's right here
415. 1:05:00wanna see I've written seven screenplays I I've these things are [ __ ] amazing Claude Claude you can just oh so here let's just watch
416. 1:05:13these this is just stupid [Music] um we'll do gpt4 since it's the best um
417. 1:05:27[Music] I have an idea for a screenplay and I want to make sure that I write the best screenplay possible
418. 1:05:39can you recommend the five best story Frameworks used by successful screenplays written by Hollywood a-listers
419. 1:05:56oh sorry so um the 3x structure the hero's journey
420. 1:06:10little Joseph Campbell action mono myth I've never I've never heard that phrase the mono myth that's cool
421. 1:06:20and there's that save the cat that's another famous One the sequence structure divides the screenplay into eight sequences about
422. 1:06:30about 10 to 15 minutes each and then the 5x structure tips when structuring while structures are helpful as guides
423. 1:06:40it's important to let your unique voice go through okay that's great um so I'm going to ask um all right what's the advantage of sequence
424. 1:06:55structure and are there examples of films that were made where this is the primary framework manageability pacing each sequence has
425. 1:07:10its own tension and resolution which is cool cool I kind of like that so so notice What's Happening Here is there's
426. 1:07:18two things happening as I'm talking with this I'm narrowing the context of what what Chachi PT and I are talking about
427. 1:07:26so it's understanding more and more what the nature of the conversation is um I'm also getting an education I'm also getting ideas
428. 1:07:39um so I might not have thought about like the sequence structure wasn't really one that I dug into and heard about Star Wars is one North by
429. 1:07:47Northwest Raiders of the Lost Ark Silence of the Lambs Pulp Fiction okay so I'm gonna say this is great my screenplay idea
430. 1:08:00is in the style of Pulp Fiction so that's exciting period can you create a framework for me that combines the save the cat
431. 1:08:09framework with the sequence framework you just described [Music] if you're going for a style similar to
432. 1:08:22Pulp Fiction you might be interested in a non-linear narrative or intertwining storylines the sequence structure can
433. 1:08:28help by breaking down these individual story lines while save the cat Canadian ensuring each sequence for the overall
434. 1:08:34story is a compelling development of satisfying Resolution Act One sequence one sequence two right so now I've got a framework
435. 1:08:46and now I can just say all right [Applause] that's terrific can you now give me an outline based on that framework about
436. 1:09:02a movie about a hedgehog named Larry who gets killed early in the film and his brother Lenny seeks revenge
437. 1:09:18and it turns out that Lenny's girlfriend Laura was the real culprit all along period [Laughter]
438. 1:09:33I'll show I'll show you the inspiration for that the title of this movie revenge of the quills um so so sequence Larry the Hedgehog
439. 1:09:49enjoying his life in the Woodland oblivious to the lurking dangers introduction of Lenny Larry's brother and Laura Lenny's girlfriend
440. 1:09:57theme stated the deceptive nature of appearances and the pursuit of Truth inciting incidents Larry is mysteriously killed
441. 1:10:08Lenny discovers Larry's body and vows to find the killer Lenny faces an internal conflict the Woodland creatures debate
442. 1:10:15whether vengeance is the right path right so why am I why am I starting here there are probably if you want to go look for
443. 1:10:29things that are specifically tied to screenwriting with AI if you go to futurepedia.io there are probably tools
444. 1:10:37at futurepedia.io that are like we'll write your screenplay for you here's the problem all of those tools all of them
445. 1:10:46are basically what I just did here but they're pre-prompted so someone like me wrote a bunch of those prompts stuck
446. 1:10:53them in a framework and put a button in front of it so you press that button and it's actually making calls to open AI or
447. 1:10:59one of these other large language models and it's doing what we just did there it's probably broken it down into a
448. 1:11:05bunch of steps here's the problem with it you may have a very specific way that you want your thing to work the way they set up that program
449. 1:11:19probably isn't that so what ends up happening is you spend more time working around the limitations that their framework that their software
450. 1:11:30handcuffed you with then you would just exploring chatgpt so here's the other thing that I'll say about chat GPT and and it might not be
451. 1:11:40the best tool for long conversations Claude is better for that but Chachi BT is is by far the best at writing so what I would do is I would
452. 1:11:53chunk up if rather than imagining that Chachi GPT is just going to write you a whole screenplay it's not
453. 1:12:02for a bunch of reasons it can't it's only got a context window of 4 000 tokens which is about three thousand
454. 1:12:09words so that's kind of its memory so very very quickly the early stuff that you talked about with it it will
455. 1:12:17forget and it'll kind of lose its mind Claude is another tool like chat EBT that won't lose its mind because it's
456. 1:12:26got a hundred thousand token context window or about 75 000 words Claude AI um let me see did I the hell was the movie I got did I get
457. 1:12:41Godfather no did I get Scarface no
458. 1:12:56oh what was the movie that I got um well anyway it doesn't matter you can upload
459. 1:13:10you could upload an entire screenplay into Claude so one of the ways that you could and and if you don't think that screenwriters do
460. 1:13:21this you're wrong because even like Star Wars like everything's based on everything else right there's only nine stories or
461. 1:13:30whatever it is right there's only nine story types you could take us an existing screenplay that's that's in the in the hood of what
462. 1:13:40you like upload it here tell it to abstract out the the framework of that screenplay and then fill it in with your story
463. 1:13:51you could also have Claude do an interview a back and forth interview with you so let's let's try that um so let's see oops
464. 1:14:09I want you to act as a screenwriting coach like Sid field and interview me
465. 1:14:24over a series of questions to get enough information about my screenplay idea so that you can create an outline with me period
466. 1:14:36do you understand yes I understand let's let's begin the session by discussing your story could you please provide a brief premise of
467. 1:14:48your main storyline my story is about an animated Hedgehog named Larry who dies early in the movie and his brother Lenny vows to
468. 1:15:05avenge his death only to realize that his girlfriend Laura was the real culprit Hedgehog love triangles hit hard period
469. 1:15:27uh okay so so check what it's doing here it's doing what I hoped it would which this is not what I want it to do
470. 1:15:35here are some follow-up questions to help develop the outline what genre is it where does the story take place what
471. 1:15:41are the main characters the antonym as someone with add the thought of filling out that [ __ ] set of questions makes my head hurt right now
472. 1:15:52like it makes me want to just turn this thing off and go away but we don't have to live with this [ __ ] anymore because
473. 1:15:57this is this is what you do if you get Sid Field's book he asks all these [ __ ] questions and then you got to
474. 1:16:03sit there and go well I don't want to answer all those questions I just want my screenplay idea to be brilliant I
475. 1:16:09just want to start writing this seems like work okay so how are we gonna fix that what we're going to do is we're going to say
476. 1:16:18while those are great questions I want this to be an interactive experience period can you please ask me one
477. 1:16:25question at a time and don't move on to the next question until I've answered the previous and you're clear that
478. 1:16:33you've gotten all the information you need for that question period absolutely the interactive approach is
479. 1:16:43very helpful for developing a strong story outline let's go through this one at a time what genre is the film is it
480. 1:16:48an animated comedy drama or something else I'd like it to have the look and feel of an animated comedy but the hard-hitting
481. 1:16:58story of an animated drum like an animal documentary see you know what we're gonna do one night in here we're gonna inadvertently
482. 1:17:10create something [ __ ] brilliant and then we're gonna go make a movie and and everyone's gonna go see our movie
483. 1:17:17okay that gives me a good sense of tone and style we're envisioning an animated film that visually looks like a comedy
484. 1:17:23but has a weighty narrative of a drama like it reinterpreted and mirrored that back to me and it did it well it's good
485. 1:17:29with some documentary style narration as well that's an interesting genre blend right so it understands that what I just
486. 1:17:38asked for is not a standard thing moving on where does the story take place is there a particular setting or location that's important
487. 1:17:46the setting is the woods near Larry's house which is now abandoned because Larry's dead
488. 1:18:01got it the woods near Larry's house that helps set the scene next question what are the main characters Laura Lenny
489. 1:18:10and oh what are the main characters Larry Lenny and Laura like what other personalities backgrounds and motivations
490. 1:18:24Larry was a door-to-door grain salesman while Lenny fancied himself an entrepreneur although he never actually started a
491. 1:18:36company period he only talked about it Laura Laura was the true ambitious one in the bunch
492. 1:18:52and it seems like things just got done when she was around so I don't need to keep going on this it's just kind of a silly exercise but
493. 1:19:02with Claude because it's got such a long context memory you can now just have this thing interview you right you could also do
494. 1:19:12this with pi so if Pi Pi is another one Pi dot a I or the PI app pi on the App Store and this one down down there
495. 1:19:24if you go to the App Store search for pi personal assistant and the thing about Pi is it's connected to the internet so if you wanted to do
496. 1:19:33research on the internet you could do that but you can put it into conversation mode where you just talk to it
497. 1:19:40and it listens and then it'll talk back to you so you can actually have a conversation like I'm having there but
498. 1:19:45instead of having to type or or you know use the the microphone feature on a Mac you can just have a conversation with pi
499. 1:19:53and then it remembers all the [ __ ] you talked about so and then you know what I would do is you know out of this I would have it
500. 1:20:00generate you know a character sheet really explaining that I would have it create log lines I would start having it
501. 1:20:07create dialogue dialogue is going to be the toughest thing um because it's just so nuanced it it you know I would say gpt4 is going to be
502. 1:20:18way better at dialogue than Claude will like I would say Claude would be good to do all your all your framework and your
503. 1:20:24outline and then I would use gpt4 to do any any sort of um writing um and I would assume that for screenplay dialogue it will get the
504. 1:20:36structure right the scenes will tend to be too short and they'll tend to be too patent so I feel like with the outline part you
505. 1:20:47can probably get it to 90 pretty quickly with the dialogue part I wouldn't expect to get it better than sixty percent of
506. 1:20:55what you want so assume you're going to be rewriting a lot of the dialogue if not writing it you know wholesale
507. 1:21:02um and that's how I do it and and here's the deal like there may be a tool out there that does this really well but I doubt it what you
508. 1:21:10will learn in the process of figuring this [ __ ] out for yourself is invaluable it will serve you in all other parts of your life so anyway
509. 1:21:21go to the go go do that what's the app name uh I don't know what app I was showing um did the Mandarian come through yet
510. 1:21:34um I don't know what the mendarian is Emilio's wife anyone know the name what's the name of the site again can
511. 1:21:39someone name the app oh my God don't you know any other language well I do I I oh the Mandarin oh that's what
512. 1:21:47you're asking Emilio's wife no the Mandarin I'm only up to like 60 something thousandth in line hang on so
513. 1:21:55what these two are asking for is they're tired of they're tired of hearing me speaking Italian and I did I went to hey Jen and um
514. 1:22:08did a mandarin version of me I'm up to 63 000 in line so I'm gonna have to I'll I'll probably I'll probably buy a subscription
515. 1:22:19tomorrow I don't really want to buy a subscription but I think I will because that will accelerate me in line
516. 1:22:36[Music] came up with the name we're not a big Library I didn't want to take Kyle's name that's cool uh um I'm going to use this to reach out
517. 1:22:52wait I'm going to use this to reach out to VP of sales cool yeah I mean these tools to help you get your thoughts
518. 1:23:01together to write letters to um you know one of the things that you can do Franklin is Franklin Tech SDR one
519. 1:23:11of the things that you can do with these tools is go copy um copy like the work experience or what someone has written on their
520. 1:23:22LinkedIn profile of someone you want to reach out to pop it into Claude and then have it write you a personalized letter to that person
521. 1:23:31referencing stuff in their work history about the thing that you want to talk to them about and it will figure it out
522. 1:23:39and it'll write you something that'll take you know give it a little piece of writing about your [ __ ] give it some you
523. 1:23:46know background on that person like can literally be copy and pasted out of LinkedIn and then have it write a very
524. 1:23:53personalized letter and then if it's too long say I wanted in three short paragraphs but keep keep the most
525. 1:23:59important things that will build the relationship whatever you might want to do it's brilliant at that [ __ ] brilliant at it um
526. 1:24:08I tried to use basic logo today with ideogram and it was bad real bad EA Source Kemp I don't know if you've heard
527. 1:24:16this here at the AI learning lab but all these tools today they're janky pieces of [ __ ] anything anything that requires precision
528. 1:24:27these diffusion models are not great at right now it looks like Dolly three is gonna be good at Precision prompting
529. 1:24:37and it's it you're gonna it's going to allow you to use chat gpt4 to write the prompts for you and it's gonna it's like
530. 1:24:45I think I think it's gonna be good whether or not it can do logos well is another thing I learned of a thing last night called
531. 1:24:54um Trace Journey not Vector Journey like that I called it last night but there's there's a tool called Trace Journey that you
532. 1:25:05install into Discord in a channel that you've got mid-journey in and then when you make a mid-journey image
533. 1:25:15Trace Journey you can create vectors out of it so if you can get mid-journey to give you a decent logo and
534. 1:25:24vectorize it then take it into so mid Journey won't do text well but it will do images really well it'll do logos well and with this
535. 1:25:36vectorization tool you can vectorize the logo just pop it into illustrator and do your own text in illustrator or make
536. 1:25:49a janky shitty version of the logo in ideogram that's got the typography that you want use that as a source image in mid-journey
537. 1:25:59generate the image based on that Source image then vectorize it then take it into illustrator and redo all the font
538. 1:26:06but you'll at least be following a design then um that's how I would do that um we'll we'll see how good Dolly three is at um
539. 1:26:16at logos and stuff the the minute see without doing licensed deals with like the NFL like if you want NFL logos
540. 1:26:32without doing license deals with the NFL teams I don't think any of these um image generation tools are
541. 1:26:41I don't think they're gonna work that hard at doing logos of existing companies well because there's too much liability
542. 1:26:48there but we'll see I mean at some point someone's going to cut deals licensing deals with with big IP holders so that
543. 1:26:56you can do something that's like mid-journey but you want a Denver Broncos logo on it or you want the
544. 1:27:01Colorado Buffaloes on it you can do that so yep janky pieces of [ __ ] right now just do the best you can work around be patient understand
545. 1:27:14that these tools are going to change everything and it's all brand new it's all brand new stability uh AI unveiled a model called
546. 1:27:26stable audio for controlled audio generation yeah they did the the the the um all the audio things that I've seen right now are
547. 1:27:36they're kind of like do you remember when when digital audio first came out and and all the song like you you could hear songs in digital
548. 1:27:46and then they were also over compressed they all had these compression artifacts and they were just bad for a long time
549. 1:27:52that's kind of where all these uh all these audio things are including stable audio but we can go we can go look at it and play with it uh
550. 1:28:03last time I tried to play with this it was it was over busy it would not let me play all right let's see um
551. 1:28:24let's see melancholic Hedgehog
552. 1:28:39trance music with a thumping Jungle Beat so so this is text to audio and they're
553. 1:28:54doing this is this is diffusion so this is an almost identical technology to how stable the fusion generates images
554. 1:29:02through the diffusion model so it's using the same basic technology that's one of the remarkable things about
555. 1:29:10these tools is that they work on any kind of data they're kind of data agnostic in some sense that basically if
556. 1:29:18you train them on a certain data set then it'll it'll find patterns in that data set and if you put it in another
557. 1:29:24one it'll find patterns in that data set is this ready oh yeah no [Music]
558. 1:29:42[Music] I don't know that that's melancholic or Trancy I guess it's Trancy it's not melancholic that's pretty awful okay here's another
559. 1:29:56one I did called Moody finger style guitar [Music] okay the compression kind of artifacts
560. 1:30:12was like wow wow you get that swelling wow [Music] but again here's the thing even though
561. 1:30:27they're that this is as bad as it will ever be and if you go back 18 months to what stable diffusion was doing with images
562. 1:30:40this is the audio equivalent of that it's it's it's decent it's not bad but it's kind of like in the
563. 1:30:48neighborhood of Music 18 months from now this will be full on orchestrated layered no no compression artifact music probably with coherent
564. 1:31:01lyrics and singing and harmony seriously seriously people listen I'm very disappointed there's only 73 people
565. 1:31:15in here last night we had 17 000 people in here tonight it's like 70 what's going on people have you not told your friends last
566. 1:31:23night you told your friends and we had 17 000 people in here do you understand what I'm saying I don't think you do
567. 1:31:27I don't think you understand the gravity of the situation um I decided that I was a a celebrity last night and now all of a sudden it's
568. 1:31:37not looking so good okay it's a little awkward it's a little awkward okay so I don't know what you're doing
569. 1:31:44but I've got responsibilities you've got responsibilities and uh where are my 17 000 peeps they're not here are they oh we're not
570. 1:31:53special in that special I wonder who could have brought those 10 000 people who could that have been maybe Satan that was actually James
571. 1:32:03all right and my ten thousandth caller exactly we're gonna give away a Toyota in the ten thousandth caller this Saturday call in
572. 1:32:16wesn radio hard rock and hip hop we're giving away the Country Store you're going to want to come all right see you at the county fair
573. 1:32:26Saturday [Laughter] my ha moment I made a get well card for a friend that had a hip replacement GPT
574. 1:32:41nice of you to make a card GPT made it amazing oh that's awesome that that was your uh Kevin McAllister moment oh by the way so
575. 1:32:52look at the article that I wrote this is this is a nod to all of you uh I'm gonna go to LinkedIn LinkedIn bam
576. 1:33:11so this is the article that I wrote and this is a cool this was cool I did this in mid-journey and then it was
577. 1:33:20um like these these two dudes were there and then those people were there but this was just kind of this empty space
578. 1:33:26but the the desk was there but there was no person so I went into Photoshop generative Phil and I selected that area
579. 1:33:33and I put a woman sitting at a desk with a computer screen and so that was kind of cool so that was that but then I I talk about um
580. 1:33:42uh every Everything used to be about uh productivity and AI is gonna help with productivity but then I talk about my
581. 1:33:50Kevin McAllister moment and look laughs [Laughter] uh and then I the the caption is what it
582. 1:34:07feels like every day working with generative AI so so if you want to read this article I'm actually really happy with this
583. 1:34:16article the basic idea that that one of the things I talk about a lot in here is just go play with AI just go play with it
584. 1:34:25and I think people are a little weirded out by that like well oh thank you very much you know I think I I have to learn
585. 1:34:31it it's a technology it's a tool it's AI it's it's this important thing and I gotta no no just go play
586. 1:34:37just go play because everything that you learn will be some augmentation to who you are and some amplification of who
587. 1:34:46you are and it's and it will feed your work in non-intuitive weird ways this stuff is I it's we're just we're we're entering a weirdness
588. 1:34:59and so by by enter by by engaging with AI with a playful curiosity um you won't take it too seriously
589. 1:35:11because it's gonna get surreal very very fast and the more Nimble you can be with sort of bouncing between different tools and
590. 1:35:20learning how to use them and playing like joyfully between these different things and kind of forgiving the imperfections of it
591. 1:35:29um it it will serve you so that's what that article is about so um there's a nice little balloon that's a little balloon
592. 1:35:42oh you know what I'll show you while I have the uh well I have the screen up here we a couple of folks were asking about I
593. 1:35:50don't get the Larry the Hedgehog reference um open AI so if you go to openai.com dolly2 or doll d-a-l-l Dash E-3
594. 1:36:06Dolly three D A openai.com d-a-l-l-e-3 that'll take you to this page so this was announced three days ago
595. 1:36:20it's just going to show up if you're a chat GPT plus member if you're paying the 20 bucks a month you're gonna get access to this first
596. 1:36:28this will also show up in Bing's image generation so right now if you go to Bing and you say make me an image of a
597. 1:36:34hedgehog it'll do it it'll be fine but that's Dolly too and Dolly 2 sucks this is not good you know Bing continues to
598. 1:36:43disappoint and Dolly too is the is their image generation engine but it's gonna have this in it this looks remarkable so
599. 1:36:51a couple of things about this that I'll show you right now and then I'll show you Larry the Hedgehog at the bottom
600. 1:36:56um which is where that reference came from so so one of the things that that this thing does well is details so see all that stuff okay
601. 1:37:11and now I'm going to read you the prompt an illustration of an avocado sitting in a therapist's chair check
602. 1:37:21saying I just feel so empty inside check it did that text the text is perfect it doesn't look weird it's like well aligned
603. 1:37:31with a pit sized hole in its Center so if you were just drawing an avocado the it wouldn't be a hole it would be the
604. 1:37:40the pit would be there right no there's a pit sized hole done check the therapist is a spoon check scribbles notes check so
605. 1:37:53Dolly three is really good at taking instructions and detailed instructions and incorporating them all this one's
606. 1:38:01wild a 2d animation of folk music band composed of anthropomorphic Autumn Leaves each playing traditional
607. 1:38:09bluegrass instruments those are all traditional bluegrass instruments amid a rustic Forest setting dappled
608. 1:38:17with the soft light of a harvest moon so yep it did that so this is what's coming and this is not coming like a year from now this coming
609. 1:38:29like I don't know a week or two this is another one that's just gorgeous let me zoom in on her so she's a dancer
610. 1:38:39look at the dress like all this sort of porcelain looking stuff and then look at the the um The Prompt here is insane in front of a
611. 1:38:50deep black black backdrop a figure of middle years her Tongan skin rich and glowing is captured mid-twirl
612. 1:38:59her curly hair flowing like a storm behind her her attire resembles a whirlwind of marble and Porcelain
613. 1:39:06fragments illuminated by The Gleam and of scattered porcelain shards creating a dream-like atmosphere the dancer manages
614. 1:39:13to appear fragmented yet maintains a harmonious fluid form holy [ __ ] it's just like just amazing um so that's that stuff um
615. 1:39:27this is one that shows again the details of the image that there's all these details in the prompt that are all
616. 1:39:33captured and it kind of points those out um this was Dolly too and this is Dolly three and the prompt is an expressive oil painting of a
617. 1:39:44basketball player dunking depicted as an explosion of a nebula so that looks like an explosion of a nebula he's definitely dunking this
618. 1:39:55looks like a blob and he's he's about as far from the rim as you can be maybe he half dunked but you know he is not dunking he is
619. 1:40:08dunking right so the what they call the coherence here is really good um oh it's going to be native Within gpt4
620. 1:40:18and gpt4 will help you be able to do the prompting and then this was this is where Larry the Hedgehog came from and
621. 1:40:24this is why we were joking the other night and someone said something and then on the on the Discord in the irregular's
622. 1:40:33channel someone made an image of of a hedgehog by um a cemetery gravestone that said Larry and so I said it was Lenny at Larry's
623. 1:40:45grave and so that started the whole thing where Larry's dead and Lenny's his brother and Laura and then we came up
624. 1:40:52with a whole thing where we could do a whole story where all of the words begin with the letter L and it was just all
625. 1:40:59that but what check this out so one of the things if you've played with mid Journey or any of these things one of the things that's very
626. 1:41:06frustrating and it's very hard to do you can do it in some tools sometimes is if you create a character that you like
627. 1:41:16replicating that character across prompts is effectively impossible it sucks really bad so watch this video and
628. 1:41:24I'll pause it here in a second to show you something [Music]
629. 1:41:40my daughter says its name is Larry can I see more like this right so it understands what this is it understands
630. 1:41:49its name is Larry so it understands that the Hedgehog that I picked out of that lineup of four is the one I'm talking
631. 1:41:56about and its name is Larry and so watch the interaction from this point forward [Music]
632. 1:42:24thank you thank you [Music] right so that's where the Larry
633. 1:42:40reference came from and that's coming that's that's coming I don't know two weeks maybe maybe tonight who knows I don't know I
634. 1:42:49do not knew I do not knew I asked for the email addresses from a business marketing list I have on my drive it constantly Rudy
635. 1:43:00gave me the website Okay so Rudy you asked for the email addresses from a business marketing list I have on my hard drive
636. 1:43:11Chachi BT can't see in your hard drive bard as of three days ago can [Laughter] but it's got to be in Google Drive so
637. 1:43:28an important thing to understand what it's been trained on is they took as much of the public internet as they
638. 1:43:39could and they started training it on all this data as much as they could as much of the public you know public
639. 1:43:45databases and Wikipedia and just websites and blogs and just as much crap as they could put in there and they stopped training chat gbt in
640. 1:43:56particular September of 2021 because embedding all of this data into this magical mythical GPT data or vector space
641. 1:44:07is really time and GPU intensive so they it's not like this is an ongoing thing so when you're interacting with chat GPT
642. 1:44:18it's not looking at your heart it's not a it's not a normal database query it's not a Google query it is pulling from this lexicon of words
643. 1:44:27that are clustered by meaning in some sort of mathematical space and it and it's it's just answering what it
644. 1:44:36thinks from a probabilistic standpoint you want based on what it knows so it doesn't know your hard drive it doesn't
645. 1:44:44know anything after September 2021 things like Bing and Bard and Pi that are connected to the internet also have
646. 1:44:53a similar limitation but because they're connected to the internet they can search for more recent things
647. 1:45:00so Bard can connect to the internet as of three days ago there's a thing embarred so long as you're using an individual the
648. 1:45:10you will not get this anywhere else because I couldn't get the [ __ ] answer to this so this is a big one this
649. 1:45:15is this is the kind of value we deliver here at the AI learning lab it's a solid you can only use what I'm about to tell
650. 1:45:23you if you do not have a Google workspace account meaning if your work pays for Google you can't use this feature which is
651. 1:45:34really counterintuitive and it pisses me off if I'm paying for the thing I should get priority but that's not how it works at Google
652. 1:45:42so it has to be an individual Google account and then if you go to bard.google.com extensions it will take you to this thing called
653. 1:45:52extensions one of the extensions is Google Drive so you can click on Google Drive and then when you interact with it you
654. 1:46:00type at drive and then your prompt and it will actually go search your Google Drive that's a three day old feature for a
655. 1:46:09small subset of Google users it might be a big subset of Google users but I have to use an old account where I have I
656. 1:46:17have none of my current stuff in that version of drive so it's kind of useless to me so I'm a little cranky about that
657. 1:46:24particular business decision I understand it because I would assume that um dealing with the permissions across
658. 1:46:32an organization is probably a nightmare with this chat GPT [ __ ] so so there's that um it can also search Google travel or
659. 1:46:42Google flights Google hotels Gmail drive and Docs I think that's it there's five of them but that's where all this stuff's going
660. 1:46:54these tools are going to be able to um you're gonna be able to search your own data which is going to be amazing
661. 1:47:01Apple's going to do this really well by the way [Music] if it gave you the website that's probably because it's a company that it
662. 1:47:13understands how to website in its data set it does not it cannot see your hard drive I see we're back down to manageable
663. 1:47:22levels we're up to 71 people I think that's respectable it ain't seventeen thousand but you know but I mean here's the deal all those
664. 1:47:31[ __ ] 17 000 we we have more likes tonight than we had last night with 17 000 people in here the average view time of a
665. 1:47:40non-follower of the channel was four seconds the average view time of a follower of the channel was like 12 minutes or 14
666. 1:47:49minutes or it was like it was long um I missed tonight's live unless you're still there still there till there um jump in
667. 1:48:01uh last night viewer screenshot 17 000 12 21 a.m eight point seven thousand twelve twenty four a.m 14 000 12 27 a.m yeah I know it was crazy
668. 1:48:16well it was you know what was funny is like we were just going along like this and I I you know I normally look up and
669. 1:48:21I'm like you know is is is the thing tapering like is it is it time to wind it down is is essentially what I'm
670. 1:48:27thinking so we kind of started at I don't know 40 and then it went to 60 and then it went to 75 or 80 and then the
671. 1:48:35next time it looked up it was like it said four and I was like huh and it was 4.6 K and I was like wait is that thousand I was like right
672. 1:48:46and you saw if you were here last night it totally threw me so and then it was just like going it was crazy
673. 1:48:56uh Community Theater is about to get lit um I don't know what that's in reference to but I I agree I think these tools are
674. 1:49:05gonna help help all the creative arts hey Champion what's up you gonna go out [Applause]
675. 1:49:25sometimes a dog's gotta do what a dog's gotta do you know what I mean all right yes can I play DJ in the movie yes you
676. 1:49:36can yeah yeah we'll oh the the community theater thing was was uh was based on our little creative writing thing yeah
677. 1:49:45exactly uh everyone can play everyone you know we'll put on a show it'll just be like that uh that old and Tiny movie
678. 1:49:51I got a barn I got costumes let's put on a show so I just jumped on what happened last night with all the viewers so so it was
679. 1:50:05um there's this guy James um I can't remember his last name because I'm a loser and have ADD yeah James kamua
680. 1:50:20James kamua from um Kenya and he's been in he's been in before he's a fan of the channel and and James James last night like the numbers
681. 1:50:32went crazy I'm like what the hell's going on and James like I did that so I think James I think James dumped dumped
682. 1:50:40a little coin to do some engagement farming and I guess there are places where you can just have people log into
683. 1:50:46your Tick Tock lives at scale um they didn't stay very long and they didn't really engage but it was it was
684. 1:50:53exciting while it happened it was pretty cool how is that different from the current search in Google Drive
685. 1:51:00um that I don't know Karen I mean because it's Google here's here Karen you bring up a good
686. 1:51:16point here's my strong instinct my strong instinct hang on I'm thirsty my throat just closed up my strong Instinct and this could be
687. 1:51:32wrong I'm pulling this out of my ass if Pete's in here he might know Pate works for Google um my strong instinct
688. 1:51:40is that what they're probably doing with these extensions right now is um doing some form of a Google Search
689. 1:51:49and then taking that Google search and tossing it into the large language model to put it in plain English so rather
690. 1:51:57than just giving you a list of links it's probably incorporating some of what it found with a Google search
691. 1:52:04um and the only reason I say that is the the couple of times that I tried to play with it it was about as effective at finding
692. 1:52:12obscure content as as a Google search which sometimes Google's just not good at finding [ __ ] on especially in Drive
693. 1:52:21um so I kind of had I had a similar experience so my my guess is that they might be doing a semantic search as the
694. 1:52:27first pass and then incorporating those results into a large language model query and then giving you something that
695. 1:52:34looks like a you know a GPT like response I don't know what this Gemini thing is all about but I have a feeling Gemini
696. 1:52:45has been trained on a lot I have a feeling it's been trained on YouTube I have a feeling it's been trained on all
697. 1:52:50the Google searches I have a feeling in some way or other it's probably been trained on some amount of Drive data and docs data
698. 1:53:01and sheets data of course they would never steal our data definitely not they would never do that all of your data is safe with them
699. 1:53:09they're definitely not selling it um pay attention to Google because a couple of things about Google they invented the
700. 1:53:21Transformer they invented the core technology that makes all of these tools that we're talking about on this channel possible
701. 1:53:29they kind of backed off it because I think they knew it might threaten their business model and and so they kind of
702. 1:53:34sat back and then when chatgpt came out and was a big hit Google went into Red Alert mode they brought back Sergey and Larry and said help
703. 1:53:46uh we kind of [ __ ] up so for six months of this year Bard was laughably bad and then Bard got good it's not great but Bard right now is
704. 1:54:00in the neighborhood of GPT 3.5 it's connected to the internet and you can tell it's got some solid technology behind it Gemini they're
705. 1:54:10claiming is that Gemini is their new underlying large language model their claiming is five to I've heard two
706. 1:54:18different numbers five times better and 20 times better than gpt4 I doubt that very seriously but even if
707. 1:54:27it's as good as gpt4 that's a big deal so the the tools that are rolled into the office suite right now like Google Docs
708. 1:54:40and Google Sheets and things like that are very rudimentary they're very weak the extensions um for Bard right now
709. 1:54:50are kind of you know in this janky pieces of [ __ ] territory pay attention to Google I have a feeling it's going to
710. 1:54:58get very good very fast from them because they are throwing much money at it and many engineers at it and they
711. 1:55:05invented this [ __ ] so they they and and they've got plenty of gpus to be training this stuff as does meta
712. 1:55:15as does Microsoft and open AI as does anthropic as does inflection so there's some cool [ __ ] coming but I I
713. 1:55:23have a feeling we're gonna see we're gonna see dramatic improvements in in Google's technology by the end of the year
714. 1:55:32so pay close attention how is Lenny killed was his drink spiked oh Stevo with the dad humor oh that's solid
715. 1:55:41that's have you seen the Australian people that that tell each other dad jokes and then they have to take a shot
716. 1:55:48if they laugh at them that that one would get someone it's solid it's solid it's good somehow my Wi-Fi constantly disconnects
717. 1:55:56so that's why I have to reconnect oh Omar that sucks Chinese Mandarin that's uh oh I'm way behind on these I'm gonna
718. 1:56:03jump down to the bottom here I'm I'm very far behind on my questions so people in here are probably like why
719. 1:56:11isn't he answering my question you know I come in here um it seems like there's some sort of interactive component to this learning
720. 1:56:19lab first of all is he a scientist it's called a lab where's his white coat why is it like this he seems to be a bit
721. 1:56:27smart alecky I don't even know if I like that it's a little tacky and he he occasionally will even curse
722. 1:56:34this seems like a wholesome Endeavor and and then he says the naughty words and he's talking about killing hedgehogs
723. 1:56:44what have hedgehogs done to him I feel triggered what was the V word you were using earlier um
724. 1:56:58very bad place I don't think I was using that v-word I do not know I do not know bubbles what my V word was uh vernacular uh
725. 1:57:14anyone V is for L vector vector yes yes the V word vector why would you assume it would be something other than Vector
726. 1:57:23you naughty naughty people good Lord I'm not rad this is It's a wave of heat just flew through the just listen leave me alone people
727. 1:57:36stop picking on me oh man oh we're having a good time tonight the V word Vector yes the vector monologues exactly
728. 1:57:52um yeah so so what I was talking about was Trace Journey which is the tool that will vectorize mid-journey images now
729. 1:58:02here's here's another thing that pisses me off um Trace Journey which vectorized vectorizes mid Journey images sounds
730. 1:58:13super cool so after I left the live last night and sort of came down from my 17 000 person high and realized oh they really weren't people
731. 1:58:24um I went to trace journey I installed it on the the um what's it called the AI Salon Discord in our mid-journey channel so it's in
732. 1:58:35there right now if you go to if you go to the Discord Channel um this the salon Discord channel for mid-journey play
733. 1:58:44Trace Journey's in there but it's got no free tier for vectorizing images you have to pay for it it's like a 10 Buck
734. 1:58:50minimum but I'm like like give me two like give me two for free so I can at least see if it's any [ __ ] good I'm so sick of these software
735. 1:59:00companies that like have you go through a full [ __ ] installation and then and then in the case of Discord
736. 1:59:08you have to actually add their bot to your channel and give it permissions like it was it was a whole [ __ ] thing
737. 1:59:14and then I go to make my first vector image and it's like well you're not a subscriber I'm like well how about go [ __ ] yourself
738. 1:59:20give me two even hey Jen gave me two free translations the two Italian ones that I've got that's why I only speak Italian
739. 1:59:34oh man all right what'd Richard Punk have to say I've done a comparison oh yeah that's right okay cool awesome awesome awesome
740. 1:59:51I'm in me too yours oh wait oh that's cool dump the dolly three prompt into ideogram oh that's cute problem with the live with this live
741. 2:00:03He's a human exactly it'll be so much better when I'm when I'm the synthesized version of me on it hey Champion yeah Champion was here
742. 2:00:13Champion a good deal all right um I don't know what my name is oh wait I don't wait I don't know who what my
743. 2:00:23name is why do I need to vectorize okay so mandrid having having done desktop publishing as my day job when I was an actor I've
744. 2:00:34learned all about resolution and things like that so um Adobe Illustrator when it makes a circle it makes it out of math the the the
745. 2:00:45lines that that are made in Adobe Illustrator are called vectors they're math they're perfectly mathematical um
746. 2:00:54representations of whatever the shape is so you zoom in to an illustrator Circle it will always be perfectly smooth you'll never see pixels
747. 2:01:06a Photoshop image the more you zoom in on it you'll see all the pixels that make it up so the the thing that looks
748. 2:01:11like a perfect circle the more you zoom into it you'll see all these little stair step pixels mid-journey stable diffusion all of them
749. 2:01:21right now their their default image size used to be 512 by 512 pixels DPI or 512 pixels by 512 pixels
750. 2:01:33now they're doing 10 24 by 10 24 so so a thousand pixels but if you were to if you were to print that out you could
751. 2:01:42really only print that out to like I don't know a three or four inch square before you would start to see kind of softness and
752. 2:01:51pixelization and if you tried to print it really big it would just look like [ __ ] there are new upscaling AIS that there's one in
753. 2:02:02stable diffusion that you it's got all these settings and then there's one called topaz labs that are not just
754. 2:02:11um doing what Photoshop used to do which is called interpolation where they basically just try to fill in the stair
755. 2:02:17steps with extra pixels what's happening with these new upscalers is they're adding new detail but it's still a
756. 2:02:24pixelized image what vectorization does is it turns it into those hard lines so you could take a mid-journey image of a logo
757. 2:02:33use Trace journey to vectorize it then take it into illustrator and fine-tune all of the lines and it would be perfect
758. 2:02:41and you could print it on a billboard and it would look awesome right so that's that's the difference um don't ask me to explain the other V
759. 2:02:53word [Laughter] how is Bard's search of Google Drive different oh that's that's what we asked that's what I I speculated before that I
760. 2:03:05have a feeling they're doing a combination of traditional search and and uh GPT stuff there I I could be very wrong
761. 2:03:13on that though so do not do not hold me to that that's one of those answers I'm I gave you speculation right there and
762. 2:03:20that's one of those answers that if I were qualified I could actually answer that but I'm not so that's my my defense
763. 2:03:26here is that I'm completely unqualified [Laughter] um hey we're at 40K we're actually up to 42
764. 2:03:43now where's Amazon AWS with artificial intelligence AI so so goat Eric forner AWS announced three months ago four months ago
765. 2:03:58they've got a whole Suite of AI models available as part of AWS now including one of their own and including
766. 2:04:08essentially everything except open AI because they're because Microsoft's their big competitor on the cloud
767. 2:04:18business right Microsoft has Azure and they're they are deeply in bed with open AI so AWS will support everyone else go
768. 2:04:27[ __ ] yourself open AI um so they're there I heard I heard two things one I heard here in the live I heard that they're gonna come out with
769. 2:04:42a with a an Alexa that actually is like pie that you can talk that you can talk to it it's not hard-coded Alexa
770. 2:04:50like Siri and Alexa right now are these hard-coded piles of garbage um so I think Alexa is gonna get better
771. 2:04:58and then I I kind of recall hearing that AWS is is going to do a chat GPT like thing although you know what that might have been today
772. 2:05:10I was listening to the uh the marketing AI institute's podcast today and I think they were talking about meta
773. 2:05:18I think meta is coming out with a chat gbt like thing that they're claiming is going to be as good as chat CPT for
774. 2:05:24every everyone's claiming it's gonna be better than Chachi ebt4 they have to say that um but anyway AWS is in the game they're
775. 2:05:34just more on the so so remember well I don't know if how long you've been here tonight but but if you've you've you've
776. 2:05:41been in these things before when I talk about that inflection point of this channel is kind of from November 30th
777. 2:05:472022 forward and and you know the sort of whole history of machine learning and you know development and engineering and
778. 2:05:55research that that's really not this Channel That's the [ __ ] that came before a lot of AWS um what they're doing right now is
779. 2:06:03living more on the machine learning side of this thing so if you want to be building applications doing things
780. 2:06:09internally like I've got a proposal right now out to a big Pharma company um and we're going to be doing that on
781. 2:06:15AWS so so we'll be building on that infrastructure um but that's not what this channel is about so when that when they do a fancy Alexa
782. 2:06:25then they'll be entering this space they're just not there yet this is just the beginning and it's mind-blowing uh for their starting point
783. 2:06:34at that exactly that that's one of the things that that Paul writer says and that I that I say all the time as well is um
784. 2:06:44and Ethan Malik says this too these tools are the worst they will ever be they're the worst they will ever be I just feel empty
785. 2:06:57in the in the uh in the Irregulars channel in the salon Discord there's a good conversation going on with the
786. 2:07:04avocado so we've now moved from magic hedgehogs to to empty avocados oh my God this is fun [Laughter] um what town is it 10 13. I got okay I
787. 2:07:20gotta do a little bit it's Saturday night I gotta go soon let's see no you're not uh we've been pining like
788. 2:07:26crazy what was the V word you were using earlier thank you Sky Jimbo I appreciate that I didn't quite see what it was it was it a
789. 2:07:36pizza or I don't know what it was but it was cool thank you uh the big problem with this live He's a human
790. 2:07:42there's there's many problems of this life V is for Vector the V word you're very red the vector monologues vectorized that was it
791. 2:07:56oh man velor that was not it velor is a good one um we learned it from watching you Dad oh what's that I missed the damn day oh
792. 2:08:08I missed on on the damn day um velvet um we gonna go tonight last night oh we're gonna go to like last night 17K yeah ion
793. 2:08:22China that that was not the 17 000 people last night were were likely people um but they I think they're an
794. 2:08:29engagement Farm there was there was very little interaction with the channel like additional followers only went up like I
795. 2:08:37don't know 70 or 80 people I figured with 17 000 people you know I'd sort of flip in and we'd I'd be at 50 000
796. 2:08:44followers now there was that that was some sort of Engagement forming situation it was fun though it was
797. 2:08:52really fun it threw everyone for a loop like the mods were like what the [ __ ] is going on here like I don't know oh man
798. 2:09:15the odds speak for themselves over 83 is a given I got a good one for you eluna.ai let's go look at eluna.ai I
799. 2:09:26downloaded aura.ai based on Sam Corn's recommendation last night a o e e Luna wait we're gonna dot a i
800. 2:09:42create unlock the full potential of generative AI explore 42 000 images get started with 15 credits hum my stuff with GPT
801. 2:10:03[Music] [Music] yeah I'll receive product updates AI tools text to image reimagine motion
802. 2:10:18blend okay so this is this is yeah we're gonna start seeing a lot of sites like this there's a there's a company called
803. 2:10:28descript that kind of does this for video um what's the one from stability AI um um something.co
804. 2:10:42I don't have it in my list do I I did but I got rid of it um I can't remember it now um yeah anyway the this is oh fireflies
805. 2:10:59like this as well so if you go to Firefly so so what these what these sites are doing is they're taking a lot
806. 2:11:05of these different models that are you know there's a bunch of tools written that you can access um that are based on stability ai's
807. 2:11:17image Generation stuff for example and then and there's all sorts of models out there so if you go to hugging face
808. 2:11:23there's all these different models you can play with what these sites are is they're putting nice interface front
809. 2:11:29ends on a lot of those discrete tools so right now it right now the industry's in this weird place where all the tools are
810. 2:11:37discreet very few of them are actually integrated I think this is a very temporary phase because what's going to happen is the
811. 2:11:45the more sophisticated versions of these models like like gpt4 and Dolly three are going to be really tightly integrated
812. 2:11:54and then the trademark for gpt5 talked about text to voice and voice to text so so pretty soon GPT chat gbt is going to
813. 2:12:06be able to do you know images words um audio music right um animation all within the same model and then you'll be able to build
814. 2:12:17applications on top of that that are just pulling from a single model all these different things it can do so I
815. 2:12:23think we're I think these kind of sites are kind of a temporary stepping stone to a more integrated future but they're
816. 2:12:30they're interesting I mean my challenge with any of these kind of tools is all of them are 15 or 20 bucks a month they
817. 2:12:38all are like here's your 15 free credits and you can make like three images and then you have to pay for it and it's
818. 2:12:44just like you can't keep up with them all and things are changing way too fast Delhi
819. 2:13:00let me see retro do you have access to dial E3 question mark um
820. 2:13:17Janet Jackson is coming just announced I don't know what that's about I'm in San Mateo California where are you
821. 2:13:26oh you're talking to each other that's cool I basically had an anxiety attack today and used pi.a to calm me down who
822. 2:13:34needs real friends so so Jim that's so as as a as a intermittent I go through phases of having really nasty anxiety
823. 2:13:45and panic attacks and [ __ ] like that so so I I empathize with you and it sucks um at one of the salons two two or three salons ago
824. 2:13:55someone said you know they think that these chat Bots are going to be good at therapy and someone else in the meeting
825. 2:14:01really lost their [ __ ] and they're like they can't do therapy and while they are not human and they don't have real empathy
826. 2:14:14these tools are at Pi in particular is really good at having conversations and it's really good at providing context
827. 2:14:25and perspective and things like that so it's really good to hear you say that like thanks for sharing it first of all and
828. 2:14:32I really do think that this idea of virtual companionship is going to be huge and not like the creepy all [ __ ] up
829. 2:14:41you know you know I'm gonna be dating a robot but but you know if if you I've mentioned this in this live before if
830. 2:14:49you haven't seen the movie Larsen the real girl you should you should see it because it on its surface it's about a guy that's
831. 2:14:57dating a sex doll and when you hear that you're like oh I know exactly what that movie's gonna be but you don't know what
832. 2:15:05that movie's gonna be it's not that at all it is and it's not and it's an incredible story and it's an incredible
833. 2:15:15story of resilience and empathy and and community and I think there's a massive role for these conversational AIS to serve a role just
834. 2:15:30like you experienced Jim so thank you for sharing that um because because this is because as weird as it is it's really easy to
835. 2:15:37demonize it and dismiss it and go oh that's just freaky you're not really talking to a person it's a robot now
836. 2:15:45if you haven't played with seriously if you're on here and you haven't tried go to Pi dot AI or ideally get the PI app on the App
837. 2:15:57Store search for pi personal assistant the company is inflection and put Pi into conversation mode there's a little telephone icon and you
838. 2:16:07click the telephone icon and it just is this sort of animated vertical bar and you talk and then it goes and then it answers you
839. 2:16:18and it's really designed to have a conversation I had a long session with pi today not about mental health but talking about my
840. 2:16:27come on AI Biz ideas yeah exactly I had I've I've had conversations with pi now about my car's electrical situation what
841. 2:16:37were the what were the specials at Costco that was a good one while I was driving to Costco pie was telling me
842. 2:16:43what the specials were at Costco I'm on the AI therapist side the DSM the DSM-5 is the wrong way about mental health
843. 2:16:51issues anyway yeah and it's you know are these things going to be a replacement for human interaction
844. 2:16:58absolutely not but but let's take it out of mental health for for a moment um because I think it's I think you get
845. 2:17:07into weird slippery slopes when you start dealing with people harming themselves but let's take let's take
846. 2:17:14something in the medical space physical therapist so you know I blow my knee out and I go to the physical therapist and the physical
847. 2:17:23therapist does some stuff with me and then the physical therapist tells me 30 out thing 38 things yeah like the movie
848. 2:17:29her yeah exactly the physical therapist tells me 38 things that I need to remember between now and our next
849. 2:17:36meeting and and here's the here's the exercises and here's how you breathe and here's how you do this and here's how
850. 2:17:42you do that and then I leave and I wasn't listening to her anyway and now I can't remember the 38 things
851. 2:17:49she said well imagine now I have that same interaction with that physical therapist and at the end of the session she tells
852. 2:17:57me the 38 things and she goes if you don't remember them that's okay I've got a virtual version of me that's available 24 7.
853. 2:18:05you can talk to it anytime it it knows your history it knows my exercises it it knows what I've recommended
854. 2:18:16and you can ask it anything about your recovery and anything that you ask it it will summarize and tell me what you guys talked about and
855. 2:18:26if in talking to the virtual me it determines that I need to be involved it will let me know and I'll call you
856. 2:18:34within an hour but whatever it might be any time of day imagine that imagine a physical therapist being able to expand
857. 2:18:44their quality of service to 24 7 where maybe they get paid for those interactions and then just move that across you know
858. 2:18:53every modality of care including mental health I think that is something that's really interesting gpd4
859. 2:19:00is a great therapy for mathematicians I struggle oh how I struggle that's awesome no one understands tensors exactly laughs
860. 2:19:13her was a great movie I enjoyed Tom Hanks and Finch never dated one but had a crush on Rosie the robot made in The Jetsons that's
861. 2:19:21really funny um I said wait oh what was that one about the PT I just missed it that karuna had up there um Pie app not in play sky Timbo
862. 2:19:35um not p-i-e just p-i personal assistant see if see if let me let me see if it's available in Google Play is inflection PI app in Android
863. 2:19:52store Android version is not available but coming soon
864. 2:20:07inflection AI and Android version is coming soon when was this written
865. 2:20:22September 1st okay so yeah so so it's only for iOS right now thanks thanks for pointing that out I'm on the AI therapist side the yep the
866. 2:20:32DSM-5 is is pi on Android it is not but Pi is so you can just go to pi.ai on Android in a browser and it's a this is a bit of a pain in
867. 2:20:47the ass to get into voice mode but I think you can do it okay that's very helpful information based on what you've told me it sounds
868. 2:20:58like you would need to charge sales tax on the painting since you're selling the painting as a business and it's being
869. 2:21:04shipped by you that means that you're responsible for collecting and remitting the sales tax to the state of Missouri
870. 2:21:11there may be some additional rules or regulations that apply depending on the specific details of the sale but generally speaking you would
871. 2:21:21[Applause] can you help me think about some anxiety I'm having right now
872. 2:21:38absolutely I'm here to help with whatever you need let's talk about your anxiety and see if we can get to the root of the issue do you feel
873. 2:21:46comfortable sharing a little bit about what you're feeling anxious about and also has this been an ongoing issue for
874. 2:21:52you or is this more of a recent development so it's not quite it's not quite what the mobile experience is so I'm just
875. 2:22:00sort of putting it in voice mode manually but I I know at some point I got the website to behave like the the uh
876. 2:22:10the app where where you don't have to type into the box it's it's just listening but but anyway but but on
877. 2:22:18Android in a browser you you could at least you know have have some kind of an interaction like this um but but yeah but again Pi so here's
878. 2:22:28here's the trick with pi if you're trying to get [ __ ] done like like with chat jbt where you're like give me a list and it'll give you a
879. 2:22:36list I won't do that but Pi will just keep asking you questions and have conversations I really want to unders I I understand that you're
880. 2:22:46urgently trying to get a list of very specific things but let me just understand why you want to do that just
881. 2:22:52because I want them I understand you're getting frustrated so um but if you want to have a conversation with it it's really quite good
882. 2:23:02um I can't hear you I'm going through a tunnel um hey it's you again um I'm trying to get whisper to work
883. 2:23:10it's painful but love your knowledge oh that sucks sorry about that um I'm happy to show it again did you get
884. 2:23:18to the playground did you get to to open ai's playground and go into complete mode versus chat mode did you get to that little um microphone
885. 2:23:28and let me go see you know what now that I think about it is there a web interface for whisper yet and I'm going to say open AI
886. 2:23:46whisper is a free text
887. 2:24:11it looks like there's a Chrome extension called whisper to chat GPT but that's not that's I don't think that's the
888. 2:24:18direction you want to go or I don't think that's what you're looking for um web UI for whisper and audio
889. 2:24:34I built a web UI for open ai's whisper join now is there a link you can find the repo here screenshots
890. 2:24:50here issue tracker here now this you have to install I want one you can just play with people you know you you developers that are
891. 2:25:00like here I developed this web UI and then you got to go install the ship from GitHub it's like no that's not a web UI
892. 2:25:07that's something that someone who knows how to program can can easily get a web UI up and running ultimate open AI whisper tutorial open a
893. 2:25:19platform how to use whisper if anyone knows of an actual web interface for whisper
894. 2:25:33pop it in the comments below sorry about yeah I know that's frustrating it's frustrating that that's again right what what you're
895. 2:25:43experiencing is that these tools are janky pieces of [ __ ] right now what they do is remarkable but where
896. 2:25:52they are in their development is very very very early we're early we're early we're early we're early are you aware of the activists
897. 2:26:02behind the stochastic parrots paper I remember reading it or reading about it like six months ago but I don't remember um
898. 2:26:17so castic
899. 2:26:48I mean I I'm not a I'm not aware of the um of the activists behind it but but I understand the concern I mean the basic concern is that
900. 2:26:58is that these large language models while they seem human are not and yes I agree with that I don't think
901. 2:27:07they have sentience I don't think that they have empathy I don't think that they have feelings and yet they behave more empathetically than
902. 2:27:19most humans they're more capable than most humans they can respond with nuance and logic unlike almost all humans in
903. 2:27:28the United States right now uh Kyle as the creator of the janky image I should have it copyrighted can I get royalties yes as soon as we
904. 2:27:38well I'll tell you if we uh if we end up selling the janky pieces of [ __ ] t-shirt absolutely we'll get your royalties on
905. 2:27:46that as soon as I start making money on this thing we can we can figure out how to split it up um I think at this point with the amount of
906. 2:27:55hours that I'm putting in I'm probably I'm probably pulling in a you know solid Buck 50 an hour something like that
907. 2:28:04but but I'm not doing this for the money so um I do use the janky the janky piece is a [ __ ] image a lot it's a good one yeah
908. 2:28:13I mean the Emoji is the is really the reason why to keep using it this because it says everything it needs to
909. 2:28:22say without putting the word [ __ ] on the screen I paid five dollars to open AI oh that sucks because some other guy recommended
910. 2:28:33it oh wait oh here Emma I'm unsure at this point I'm on a Mac which is terrible but I'll keep trying I
911. 2:28:38managed to get whisper to run on Google collab okay because some other guy recommended it yeah yeah I paid five
912. 2:28:45bucks to open AI That's because you need access to the API to get to that to be able to use that interface um
913. 2:28:56it's all good the janky will get so much better it will um I look forward so Emma Emily is it Emily or Emma Lee um
914. 2:29:11follow along didn't have uh oh no I know we can do so let's go to um we're gonna go here so let's go to
915. 2:29:25there's champi platform.openai.com playground okay so Emily um if you're here if you paid them five bucks you should be able to get an API
916. 2:29:40key you've got you know an account here this button right here switch that from chat to complete have you done that and then
917. 2:29:54in this big window in the upper right hand corner is that little microphone so
918. 2:30:10all of the AI tools right now are janky pieces of [ __ ] but if we give it 12 to 18 months everything's going to be remarkable
919. 2:30:23so it says it's transcribing all of the AI tools right now are janky pieces of [ __ ] but we if we give it 12
920. 2:30:30to 18 months everything's going to be remarkable so that's working with recording um I don't know I don't see here any
921. 2:30:44um there's no controls here so if you got it running in Google collab there's probably all sorts of parameters basically go to
922. 2:30:57oh this is this is [ __ ] awful right now basically go to documentation and find the whisper documentation um
923. 2:31:08because if you have it running in collab you can probably do things like time code and things like that creating automated
924. 2:31:16creating an automated me meeting minutes generator with whisper and chat and gpt4 so here's all the here's all the documentation for whisper
925. 2:31:30but this is not what like I assume what you don't want to do is create a transcription application like
926. 2:31:38what you want is something that [ __ ] works this is the joke I have with my developers right now is I have very
927. 2:31:45simple acceptance criteria for their work it should [ __ ] work this doesn't [ __ ] work right now anyway um
928. 2:31:55yeah but but if you just want to transcribe something flip it into complete mode click on the microphone and then upload
929. 2:32:04something there so hopefully hopefully you haven't tried that and that actually helps but if you have tried that and it
930. 2:32:09doesn't work I don't know what to say I'm on a Mac so I got it to work here um all right let me do a couple more
931. 2:32:17questions and then I got a motor on out Robert Metzger wow did you know you could make money with chat gbt I did not know that Robert
932. 2:32:29oh yes I did it's comedy it's comedy we're having big time comedy all right Emily thank you very much
933. 2:32:43Winston oh yeah making it rain thank you very much sir I appreciate that thank you so much oh this will work good awesome
934. 2:32:59awesome good yay we actually did something I I now have a qualification Emily came in and wanted to get whisper working and
935. 2:33:09now it's working well no she said it was it will be working Wham Careless Whisper uh where is the microphone oh did you
936. 2:33:19get it I think you got it um this will work thank you so much you're welcome you're welcome you're welcome humans equal suing Americans who
937. 2:33:27listen to politicians who say inject bleach you did okay great awesome I'm a teacher and you did good I'm teach I'm teacher verified
938. 2:33:40all right listen I think let's end it on a high note again I'm disappointed that you didn't bring your 17 000 friends
939. 2:33:47tonight we had a solid 70 for most of the evening which actually I do prefer I like this more intimate thing um
940. 2:33:56really cool hanging out with you all tonight I know it's kind of low-key tonight hope you have a good Saturday
941. 2:34:00night go have nachos or tacos or whatever the hell you have now a whiskey um we'll bring them tomorrow yeah yeah
942. 2:34:08exactly forever hooked I need so listen our goal tomorrow let's try to go from 70 to 10 000. I that seems reasonable it
943. 2:34:16seems reasonable to me just invite your friends thank you Winston loving the dangly lights good night it's my anniversary what did
944. 2:34:24you do on your anniversary I'm so sorry I'm not on Tick Tock with some old dude that looks like Rosie O'Donnell what is my life coming to
945. 2:34:42goodbye everybody thanks for hanging out everyone Thanks for thanks for uh those of you who modded tonight and who've
946. 2:34:48been modding I really really appreciate it it makes a big difference peace out everybody I'll see you tomorrow