aiLL 9/15/23 - Embracing AI: Becoming "AI Native"

Transcript

1. 0:27welcome everybody hey Dr J shaking happy Friday night it's a party party party what is
2. 0:41happening so there is that good evening hello how are you happy Friday night got on a new t-shirt a little little binding around the
3. 0:58neck oh happy Friday cuno what's happening what's shaking what's shaking what's going down what's happening what's
4. 1:06happening with you yeah yeah see yeah all right yeah all right we're going to I think I put the makeup on a little too
5. 1:14thick tonight what do you say all right all right welcome to the AI learning lab this is Ali the AI learning
6. 1:22lab we'll let some people get in here say hello say hello good day good evening good evening welcome to the AI
7. 1:32learning lab my name is Vincent Price that's a very old reference older than all of us combined all right
8. 1:45welcome I don't know if you knew this but this is the AI learning lab and this is an AI learning lab get
9. 1:55it it's comedy it's a visual comedic element welome all right so what are we going to do here uh we will uh we'll let
10. 2:05some people come in then we'll talk about Ai and then we'll uh talk about going AI native I put a new video up
11. 2:13tonight if you haven't seen my latest video I think it's a fun one I had a good uh I had a good talk in my uh AI office hours on LinkedIn
12. 2:22today um Jim Ross who's been in here before and a guy named G they took a class called um being AI native in business or
13. 2:35something like that and so we talked about that and it it hit me that I never really I I really considered the idea of a what like what
14. 2:46is AI native you know what does that mean other than you know the kids that are born today are going to be AI native
15. 2:53right the kids the kids that are sort of two two and under today are not going to know a world without a
16. 3:00I but what's it mean for us old farts that are just starting out on this journey and what hit me was it was it was actually a combination
17. 3:10of what we did last night so so Amelio's wife hey Ted looks at life what's happening Amelio's wife came in last
18. 3:17night and she said I want you I want to see if you can use chat BT to help figure out my stupid open enrollment
19. 3:24health insurance [ __ ] and I'm like yeah we can do that and I don't know we spent like 45 minutes just [ __ ] around with
20. 3:30chat gptt and we ended up getting it to sort of work in Claude and then Claude ran out of credits and it you know we
21. 3:36didn't quite get there but it was like a good 45 minutes of of the live was spent doing something that if you had just
22. 3:45spent the same 45 minutes looking at the piece of paper you'd have probably gotten to the same result right and then
23. 3:53when I heard these guys talking about this AI native class and and and the the basic idea was you basically try to do
24. 4:01everything that you do using Ai and I thought oh that's kind of interesting and then so what hit me what
25. 4:08the video is about if you go watch it is okay if it's going to take you 45 minutes to read the document traditionally and 45 minutes to dick
26. 4:18around with AI and maybe or maybe not get the answer which one do you choose I think you choose [ __ ] around
27. 4:26with AI for 45 minutes because at a minimum you'll learn something and it'll probably be able to do what you want it
28. 4:33to do and and that that I'm really excited about uh clear warned AI I'm 53 because of AI learning lab and my
29. 4:43inclination in multimedia I'm all in that's awesome AI native go AI native I'll be 54 next month I'll be 59 in
30. 4:55May so got a few years on you but awesome awesome awesome welcome um so welcome everybody this is the AI
31. 5:03learning lab that's Ali the AI learning lab and what we do here is as you can imagine we wax poetic about this this uh
32. 5:13generative AI stuff and and what it's what it's going to change and what we can choose to have it change
33. 5:23for us personally and that's that's the thing that I'm I mean I suppose that's what this Channel's always been about right choose
34. 5:32to step into it choose to if if you're here and you haven't played with ch Chachi BT one of the things I do is I I
35. 5:39leave the address of chat GPT up in my live for most of the time although now that we have cool images we also we also
36. 5:49put those up there um but the reason I put that there is if you've not played with chat GPT go there right now chat.
37. 5:57open.com and just start playing with it ask it to write you a poem ask it to write you a song ask it to write you a
38. 6:03marketing plan doesn't [ __ ] matter anything a gardening strategy um and just see what it does and then go
39. 6:13to prompt. chat this is a little like a Word document that's got a bunch of prompts in it you can copy and paste but
40. 6:19it also has some primers that teach you how these large language models work and how to talk to them and and and and and
41. 6:27what you can do what they make possible and just start on the journey and if you're lucky within the first I don't
42. 6:34know 10 or 15 minutes or an hour you'll have what we lovingly call in here your Kevin mallister moment
43. 6:41right your moment from Home Alone where you're [Laughter] like and that's what that's what you want to get to because in having that
44. 6:51moment you'll realize holy crap chat GPT is it's more than just hype it's more than just fancy Google it
45. 7:02looks like a fancy Google search bar it looks like a stupid Google search bar doesn't look fancy at all and yet when you use it you realize
46. 7:11there's some profound depth here I got a good parlor trick for you you know there's the meme going around on Tik Tok
47. 7:18all these wives asking their men how often do you think about the Roman Empire which by the way I don't think
48. 7:24about the Roman Empire but whatever a lot of men do go to chat JP and ask it about the Roman Empire ask it why men
49. 7:31might be so obsessed with the Roman Empire maybe it knows um but just go start playing with it and and start to reveal to yourself
50. 7:42why people are so excited about this cuz it's it's more than just hype this is there's some profound [ __ ] going on
51. 7:48these things are called large language models because they've been trained on like lots and lots of language on the
52. 7:54public internet for the past many years 30 4050 50 60 years and so they're reflections of us we are large language models right words
53. 8:07go in they process ideas come out that's what these are reflecting our Humanity amplifying our Humanity it's not just a search engine
54. 8:19it's not just copying and pasting and plagiarizing which is one of the misconceptions if you understand how they actually work it's [ __ ]
55. 8:26mind-blowing It's Magic Arthur C Clark any sufficiently advanced technology is indistinguishable from Magic that's what
56. 8:35these are right champy champy just came in he got excited about Arthur C Clark all right and so men think about the
57. 8:44Roman Empire limits equivalent is that we think about murder yeah I saw that cuno uh I saw someone else posted on on
58. 8:51Tik Tok it was she said she said you know either that were going to be murdered or other people's murdered or
59. 8:56us doing the murdering it's always about murder I thought that was awesome um by the way if you if you see people in here
60. 9:03and you're like Kyle you you haven't really said anything yet and there's all these likes what the hell are these
61. 9:08people doing meet the Irregulars so the Irregulars are people that just show up here regularly and uh they play games with
62. 9:17themselves they entertain themselves and they some of them actually mod the channel and they will support you if you
63. 9:24have questions and I can't answer it I haven't gotten to it they'll do that they're super awesome people
64. 9:29um and they're the ones that are getting carpal tunnel by doing this on the Tik Tok screen okay so if you have questions about generative AI
65. 9:39pop them in the comments below I'll do my best to answer them happy Friday everybody welcome everyone and let's get
66. 9:46get DED let get started and and listen sometimes sometimes in the AI learning lab I act a little dum dum sometimes I
67. 9:53act a little dumb Dum sometimes I act super smart sometimes you'll be very intimidated I'll be all like
68. 9:59intellectual like like you'll be like oh my God he's got qualifications and then you realize H maybe not I wonder if he's
69. 10:06qualified is he a dum dum it's possible he's a dumb Dum you just don't know that's part of the joy of the lab you
70. 10:15never know what you're G to get we like to call it chat add why is that well you know hang out we'll see he it's also there's a lot of
71. 10:28time spent talking about my hair no no one else actually points it out I'm apparently obsessed with it but sometimes it just goes crazy and
72. 10:40then sometimes it looks like I'm a carpet salesman it's just I don't know it's just it's something that entertains
73. 10:47me all right let's go all right so what I'm going to do I'm going to go back and I'm going to start answering some
74. 10:54questions I'm going to see some comments and I'm going to comment on the comments this is is you know what this channel is
75. 11:00this is like one of those reaction channels except reaction instead of me reacting to people who've Got Talent I'm
76. 11:08reacting to people who are commenting and then and then sometimes I'll call them a dum dum and then sometimes I'll
77. 11:15be like oh my God that's a deep philosophical question let me talk for 15 minutes that's how it goes you never
78. 11:21know what you're going to get here life is like a box of chocolates you never know what you're going to get go Forest [Laughter]
79. 11:34go all right I think I have a coffee being made hold I'll be right back I smell coffee
80. 11:57I
81. 12:27than
82. 12:47all right Champion let's do this let's do this let's do this all right I have my coffee did we lose everyone yep we
83. 12:56lost everyone 30 people left this is the good stuff this is a small intimate crowd this like this like seeing the
84. 13:04stones in a little corner bar well if the stones were like a fat old 58-year-old talking about computer [ __ ] it was just like
85. 13:16that very small audience very small pinch your head M all right that is tasty freshly brewed what is the best AI to translate
86. 13:30audio files to text what is the best AI to translate audio files to text then I want AI to summarize the conversation and find key points
87. 13:43okay um there's probably other tools for this go to futur pedia.com
88. 13:58it a couple of ways you can download the official open aai app like on the iOS or Android stores the official chat GPT app you
89. 14:09sort of turn on the microphone I don't know if you can upload a file in the app but you can talk into it and it does
90. 14:14really good transcription but if you want to upload a file oh by the way there's a new tool out let me show you
91. 14:21this um stable diffusion the company behind um or stability AI the company buying stable diffusion launched this thing called stable
92. 14:34audio and you just describe music and then it plays it so I just put in Moody finger style guitar and it's not [Music]
93. 14:51bad oh here comes champ he's a singer you like that [Music]
94. 15:07champy all right buddy it's done it's good that's good it's a good boy you're a good singing dog you're a good singing
95. 15:17doggy champy approves champy approves okay so anyway that is go to stable audio.com and you can play with
96. 15:26that it looks like their servers are very overwhelming right now um but what's interesting about this is they're
97. 15:32doing for music what stable diffusion does for images so it's trained in in an identical way they took 800,000 songs
98. 15:39they tagged them and then they turned them into essentially mathematical equations and sort of um how do how do I put it like math
99. 15:49mathematical versions of song signatures it's not really the songs it's song signatures and then they um when you put
100. 15:57in your prompt it sort of amalgamates a bunch of those signatures and then effectively denises them goes from a
101. 16:04diffused model to a more focused model over a bunch of steps and it generates music so um so it's it's it's a really
102. 16:13interesting approach okay uh text to transcription so what we're going to do is we're going to go to you're going to go to
103. 16:26playground or plat form so write this down because there's a pain in the ass there's a pain in the ass URL thank thank you
104. 16:36um um open AI all right so you ready you got your pens get your pens get your pens come on listen most of you in here
105. 16:47are are of an older Affliction and I know that you need papers and pens and reading glasses so go get go go get your autra mon
106. 16:59um so you're going to go to platform. open.com playground and then basically this is this is where developers develop tools
107. 17:10sitting on top of GPT 3.5 GPT 4 it's going to default you into chat mode so in the right hand side here here let me
108. 17:18make this bigger so you can see it oops that was too big in the right hand side here you've got mode and it's chat you
109. 17:25don't want to be in chat mode you want to be in comp complete mode it says complete Legacy which basically means
110. 17:32this is going to go away at some point but then you have a little window here and in this window you could type uh anything you could say um
111. 17:43um why are men so obsessed with the Roman Empire question mark it seems that almost every man thinks about it on a regular
112. 17:55basis and women are mystified why this is happening happening period can you shed any light men are fascinated by the Roman
113. 18:08Empire because of its impressive achievements in politics Warfare engineering art literature and architecture so so this works just like
114. 18:15chat gbt but you can do other things like you can Muck with the temperature blah blah blah blah blah but cool thing
115. 18:22here is this little teeny microphone in the corner right here this is access to chat gpt's whisper technology so you
116. 18:30click on that and then you can either start recording or you can upload audio from your computer and um let's see I wonder if I'm trying to
117. 18:44think if I've got an audio file MP4
118. 19:00not name MP4 MP4 EG movie this [Music]
119. 19:21Mac me see let me see if I have an MP3 file MP3 audio
120. 19:36a steady climb I don't think I have anything here I didn't really prepare for this oh yeah baby yeah baby yeah
121. 19:50[Laughter] baby you've got mail all right where's the beef hey anyway um so basically so just upload an audio file here it'll turn it into a
122. 20:05transcription and then depending on the length of the transcription you may need to break it up into into little segments
123. 20:12um there's probably other tools that do this but but once it's once you've got a transcription in fact I probably do have
124. 20:18a transcription in here somewhere let me go find a transcription transcription
125. 20:35all right here's an AI learning lab transcription wait why did that open oh there we go okay so I'm just going to grab a a good
126. 20:46chunk of this actually you know what I'm going to do okay so this is a really long transcription so this is one of my you
127. 20:55know 2 and 1 half hour rambling sessions right so this is just words from me talking now where I got this transcription was I
128. 21:05uploaded my video from Tik Tok to YouTube and then YouTube automatically generated this so another way to get
129. 21:12this transcription is via YouTube but the whisper API from chat GPT is actually better so for for what it's
130. 21:18worth um okay so now we're going to go to chat GPT and I'm going to try to paste that transcription into chat GPT and I'm going to say
131. 21:33um let's see um summarize this and it's going to do a lot of that it's going to read it and it's going to
132. 21:45go the message you submitted was too long chat GPT has a um a limit of about 3,000 words and a 2 and 1/2 hour
133. 21:54transcript is probably more like 15 to 25,000 words so it doesn't work there but we have options we can go to claw.
134. 22:05a we can do the exact same thing I can paste in that whole thing so I'm I'm just in the chat window I paste in and
135. 22:14notice what it did was if the if what you paste in is larger than a certain amount certain amount it basically
136. 22:20attaches it as a text file and so now I can say um please summarize this and tell me what is going on with
137. 22:37it and so it'll think for a little bit here so so Claude this tool is from a company called anthropic this is a bunch
138. 22:47of xop AI people so these are the people that develop gpt3 the the the model that sits underneath chat GPT
139. 22:553.5 these are the people that develop Ved it they quit open Ai and started anthropic um they're they're more
140. 23:03concerned about safety than than creating um than creating models but what they realized was they had to
141. 23:10create models in order to do the safety stuff they wanted to do so the transcript is a long video of Kyle
142. 23:17Shannon's AI learning lab on Tik Tok the main points are Kyle talks enthusiastically about AI especially chat GPT and how it's changing
143. 23:25everything he believes AI have a profound impact this is kind of embarrassing but good he encourages people to try out jet gbt um he gives
144. 23:35advice on how how to get started he discusses potential applications in AI then he discuss then there's a
145. 23:43discussion around jobs Kyle talks about the pace of change ethics risks regulation all of it let's see um does um Kyle ever swear
146. 24:00let's see what it says so anyway I mean this this tool is quite remarkable at at being able to this was the tool last night when we
147. 24:12were working with um with um Amelio's wife with the PDF document that she she shared with me about health insurance
148. 24:22plans and it actually it parsed that PDF better than chat chpt did yes based on a scan of the transcript
149. 24:32Kyle uses some profanity at time stamp 11629 he says Jesus Martha oh man my fears they've got their
150. 24:49heads up their asses all right when discussing insurance companies he says but you suck now to be fair when we're referring
151. 25:02to insurance companies but you suck is a technical term that's not offensive language that's just fact so anyway that's how you do that
152. 25:14all right I hope it helps did it help did it help I hope it helps I'm here to help I'm just here to help help people
153. 25:21you know if I didn't help then that would that would be so bad it would be so embarrassing if I didn't help but if
154. 25:27I didn't help we'll blame Ally Ally did it hair growth gate what's happening what's shaking lady hope you're well all
155. 25:37right I'm 49 messages behind thank you if anyone's modding just feel free to pop up a comment now that I'm out of the
156. 25:45uh transcription hole thank you steo um I had it make a plan for fall plantings that will bloom after June oh
157. 25:53that's super cool actually that will bloom after June so fall plantings that will bloom after June so I assume that's
158. 26:00bulbs I don't know much about gardening but I assume you're not putting in delicate little flowers as we go into
159. 26:07the cold season June Mary I want to create an evaluation methodology for International Education records That's Heavy um it's totally you
160. 26:19can do it playground AI is giving us Graphics like never before how hm how how how um hair growth geek how are you
161. 26:36getting images from it are you having it generate um prompts or using Dolly here I didn't know you could
162. 26:51do I didn't know you could do images here do share do share how you're using playground to get images that's pretty
163. 27:01cool uh and then and then we will share oh by the way that's that's one of the one of the I guess one of the purposes of this channel
164. 27:12is I'm doing everything I can to share what I'm learning in real time right that like it to a great degree the AI
165. 27:19learning lab is is something for me to be able to process all the crap I'm learning and get it out of my head one
166. 27:26of the things I encourage is wait we can customize and optimize using Google image analyzer oh cool oh very
167. 27:33nice that's super cool um as you start learning as you start playing with this stuff even if you're like brand new to
168. 27:41it talk about it talk about it at work talk about it with your family talk about it with your friends even if they think you're [ __ ]
169. 27:50weird it's important I just keep playing with it and learning and then I get better all text okay cool got it
170. 27:58yeah it's amazing there was Ethan mullik did a a an interesting LinkedIn post I don't know four days ago about a study where
171. 28:08um one of the one of the research groups compared human prompting and um chat gbt prompting and and the quality of the results and chat
172. 28:22gbt writing prompts for itself beat the humans excuse me beat the humans so that's that's pretty cool and that's what
173. 28:29you're up to there I like it I dig it I support it anyway as you learn this [ __ ] share it and and there's a couple of
174. 28:36reasons for it one is just that thing of you learn more when you teach than then when you have someone tell you what
175. 28:45to do if you have to figure out how to articulate an idea of something you learned it helps you learn um so do that
176. 28:53here let me pull up hair growth gey if she's around I don't know if you're if you're camera friendly I don't know
177. 29:02if you're camera ready right now if you're not that's cool um but but you know one of the things is you'll learn better the other thing is
178. 29:14with with um with teaching is you'll teach people like people in your life even the ones that think you're really
179. 29:20weird hang on settings no that's not what I wanted to do hold please dual camera there we go bang all right had to get my doggy
180. 29:32back I like doggies doggies is good uh my family is planning an AI intervention they think I don't know about
181. 29:44it yeah be willing to be intervened by your family be willing to be that weird because because here's what's going to
182. 29:52happen some people you're going to teach it to they're going to be really appreciative they're going to want to
183. 29:56learn they're going to jum jump on the bandwagon they're going be like woohoo other people are going to be like
184. 30:01whatever I don't care about AI blah blah and they're going to go away and then other people are going to be resistant
185. 30:05to it and think you're weird and you know try to diminish you and and put out your spirit and a don't let them [ __ ]
186. 30:12do that sorry in PJs no no that's good um but what's going to happen is this the people that are resistant to it
187. 30:23right now are doing that either out of fear or ignorance and at some point when these tools start invading their
188. 30:33work and it becomes clear to them that something's shifting and they don't know what to do they're going to come to you
189. 30:41because they know that you're you're someone in their life that knows about this [ __ ] oh that's something she was
190. 30:48talking about at dinner last you know six months ago maybe I should call her can you help me learn this right so so I
191. 30:56think it's I just think it's an important service I think I think the more people that are prepared to at
192. 31:01least understand you know the nature of what these tools are how profound they are what they can do well what they
193. 31:10can't do well um and and really start to start to to live in this idea of being AI native that whenever you got a
194. 31:21problem to solve the first thing you think of is how can I use AI to do it not because it can necessar do it better though it likely
195. 31:30can but because in you trying to figure out how to use these tools to solve that problem you'll do what we did last night
196. 31:38we tried chat GPT two different ways and I couldn't get it to work and like I'm I I think I'm pretty good at this [ __ ]
197. 31:45right and and we couldn't get it to work and then we went over to Claud and got it to work and and just just in doing
198. 31:51that the amount of time I spent on that I learned a ton um I don't know if anyone in here learned a ton cuz I was
199. 31:57kind of all over the place uh bing went Bonkers today with pictures and videos it took an hour to allow
200. 32:05searches I'm done yeah Nancy so here's the deal with Bing Bing um Bing got kicked off my adults table right my adult table now has chat gbt
201. 32:19Claude and Bard and if if you've been on this channel for a while you know I made fun of Bard hard for a bunch of months
202. 32:28and Bing is down on the kids table with pie and Po and perplexity perplexity actually probably deserves to be up at the up up at the
203. 32:39adult table but Bing is definitely down here so because yeah it's the the safety guard rails on being some whatever I
204. 32:47don't know what's going on with Microsoft but you know they're being Microsoft the funny voices not only help
205. 32:54but do something I didn't know what that said oh but heal my my soul that's that's good Winston I'm glad I'm Soul
206. 33:01healing I will never contradict you again that's cute uh you can customize to optimize wait we we can customize to
207. 33:11optimize wait what is going on with my comments are just jumping all over the place I learned a lot from the chat GPT
208. 33:18failure last night oh good Andre thank you I'm glad thank you for saying that like sometimes when I do stuff like that
209. 33:23I don't know if it was a complete waste of time cuz I know it was it was long it was like someone was flipping through Tik
210. 33:30Tok it it wasn't as good as like honey is this thing on you know that'll get him to stop all right um so thank you for that let's
211. 33:49see I don't know where I am am I toward the bottom or toward the top let me see oh very cool I love that one I I don't
212. 33:57know what that's called but that thing's so cool thank you very much for that and uh so there's that um just made it to
213. 34:05level 11 and it is a competition so so in case you're in case you're keeping track I don't know what level you're at
214. 34:12but if you're not at level 11 you know time to step up your game apparently um I learned a lot from the
215. 34:19chat gbt failure the other night thank you for that hair growth geek thank you for the CR by any other name would smell
216. 34:26is sweet oh very lovely Winston I love it it's the cosmos the cosmos Beyond billions upon billions of stars in the
217. 34:38sky oh very nice little puppy thank you very nice oh hair growth geek level 11 see you got there all right not
218. 34:49necessarily bulbs many weren't oh very cool I guess it could be seeds too right um I had issues with words in Discord
219. 34:58I don't know what that means but if you tell me a little bit more about that I'll do what I can to um to do something with it um AI
220. 35:13intervention listen listen we we need to talk oh all my friends are here yeah uh Kyle sit down thank you digital Gods I
221. 35:22appreciate that K life's tough it's tougher when you're stupid did you see the video of the dude there's like some dude in New York and
222. 35:36he's doing the NPC thing he's like and a hat and a hat thanks for the Hat thanks for that he's doing all that stupid [ __ ]
223. 35:41on the street and some guy standing behind him like in the camera view and he turns around just loses his [ __ ] on
224. 35:48this guy he goes I'm trying to feed my family it's pretty funny oh good lord shocking how few people are curious yes
225. 36:00one shy girl I totally agree I talk about it daily and one out of 10 actually ask questions exactly so and
226. 36:09and and this is the thing I know right now like sometimes it it feels like you're you know you're shouting at the ocean
227. 36:16right sometimes this channel feels like that although it doesn't because I mean the comments like like you guys are are
228. 36:21clearly engaged other than the Trolls but they're fun to pick on um that just means we're early it just
229. 36:31means we're early right this chat GPT is 9 months old you know November 30th 2022 is when that thing um came
230. 36:41out and so it's only been nine months and and you know if you've if you've been in the tech scene for a while you'd
231. 36:48be like but Kyle machine learning's been around for many years I I won't even argue decades yes it has that that date
232. 36:57the date that thing came out is the equivalent of when the worldwide web came out for the internet right the
233. 37:02internet was all command line Tim burners Lee comes out with this thing called the worldwide web and once we
234. 37:07have browsers now you can just point and click on websites right you can click here and jump to another website or jump
235. 37:15to another country that was the big innovation it was actually an interface thing that's exactly what that is so
236. 37:21there's an inflection point in history and we're only nine months into it these tools as profoundly powerful as they are
237. 37:29are we're just scratching the surface with what they're able to do they're the worst they're ever going to be but
238. 37:36they're brand brand new so what I what I would say to you is just keep talking you know keep keep putting it out there
239. 37:43and people will will opt in to your Social Circle one shy girl um who are into it and that's cool and then 6
240. 37:55months from now someone will come I'm knocking on your door could you teach me I know I made fun of you before I know
241. 38:00we had that fake intervention I was just being a jerk I was scared could you help me and then and then if they were too
242. 38:09much of a jerk you can choose whether you help them or not like well you were douchy MCD dington so I'm going to go on ahead and
243. 38:16choose to let you suffer or you could be empathetic and say okay it was fear I'll help him it's funny when my my um in my company
244. 38:27storyvine our our VP of account she's she's brilliant she's she's awesome she's sometimes in here um whenever she
245. 38:37she'll she'll sometimes come to me and says hey this this is going on with a client and and you know they want a
246. 38:43favor right the client wants a favor and my first question to her is are they [ __ ] cuz it it plays a part if
247. 38:52you're an [ __ ] uh you don't get the you don't get the uh the warm and fuzzies as much as if you're not an
248. 38:59[ __ ] so it's just you know life is actually not that hard if you just follow this one rule don't be a douche
249. 39:08it's it's actually simple it's actually simple everyone's going through [ __ ] just don't be a douche just try to have
250. 39:14some empathy just try to do it no no all right or just be a douche all right uh Winston Winston Friday night gift night
251. 39:27hello Emilio's wife good to see you did did you see amelos did you see the video that I posted I acknowledged what we did for
252. 39:38you last night and a conversation I had today for the idea in that video so you got a shout out in in a in a video on AI
253. 39:46learning lab channel it's very cool is there a recommended wait is there a recommended text on gpt3 I don't know what that means ask
254. 39:56that again I don't I don't quite know what you mean there um but ask it again I did not W yeah go check it out you got a little shout
255. 40:04out um I don't know if I actually shouted out your name I think I did I don't know I might not have but I said
256. 40:10we were we were talking about your uh the health insurance plans I think I said Amelio's wife though uh what's up talk hey Kyle hello
257. 40:23hello to-do list Shout at the ocean exactly AIML came out in 95 yep yeah and and you know like like this stuff's been around for a while and
258. 40:37similar to the the you know the internet had been around for decades before the worldwide web came out this is these
259. 40:45tools generative AI all this stuff is um democratizing that power to the rest rest of the world to the rest
260. 40:59of humanity you don't need to be a mathematician douchy McDon book is there a book Zippy to pin head by the way Zippy to pin head 10 out
261. 41:13of 10 on the name I would appreciate you camel casing the words so it's easier to parse but it's solid if you'd have done
262. 41:22Zippy the pin head it would have been 8.5 n but Zippy duin head 10 out of 10 solid like it telegram upon sending 21 I do not
263. 41:35know what that means but that's probably just cuz I'm old doesn't allow screenshot oh telegram upon sending if you're
264. 41:43trying to screenshot on a Samsung 212 or 3 it doesn't allow it because Ticky talk decided to say screw you to
265. 41:52screenshotting I don't know why I don't care I don't I listen I have stopped trying to figure out Tik Tok it's true
266. 41:59people are starting to ask questions I have helped with a I had oh I I have you you helped with a business plan today
267. 42:06Obi-Wan oh okay by the way Obi-Wan so it's not Obi-Wan it's OB Wan Juan 10 out of 10 on the name oh man you you almost break a 10
268. 42:20that's a great name I love it it's awesome I need to figure out a good assignment in my stats class that still aligns with my course
269. 42:33outcomes I need to figure out a good assignment in my stats class so brother 52 um I mean so so so going back to this idea
270. 42:47that that I hit on today of of or that I kind of internalized today someone else hit on it cuz they're teaching a [ __ ]
271. 42:53class on it but this idea of being um AI native what I would say um is wait where did uh uh oh digital Gods finally got the shove
272. 43:10the robot stuff right in ideogram cool I think you posted that in Discord too um Tik Tok just jumped me somewhere oh
273. 43:19here we go brother 52 okay I need to find out a good assignment in my stats class so so what would say is this I
274. 43:27would go to go to chat GPT right now or whatever Claude And whatever the the large language model of your choice tell it about your stats
275. 43:37class tell it about your course outcomes and then have it give you 10 ideas for a good assignment that you know with with some criteria
276. 43:51like you you define what good is like like come up with 10 assignments that accomplish you you know the following
277. 43:56goals or that you know provide the following challenges or make it interesting whatever it is have it give
278. 44:02you 10 and then when it gives you 10 ideas pick the top two and say I like three and seven you know give me 20 more
279. 44:10that are more like those and and just see if you can get chat GPT to give you all that and this is this is the thing
280. 44:20this is my new thing the the video that I just posted today this afternoon is the challenge is rather than only going to chat GPT
281. 44:30when it's obvious to go to chat GPT whenever you have a problem go to chat GPT or whatever mid Journey or you know
282. 44:38whatever the tool is that go to some AI tool um spend the time no matter how ridiculous it seems trying to solve it with
283. 44:47AI because you'll probably be able to do it for one thing and then if if if not then you just learn something so anyway
284. 44:55I I that that I'm really excited that's my that's my new ex I'm I'm I'm I'm all excited I'm all excited I'm all excited
285. 45:01can you tell that I'm excited I'm all excited I really want to have it summarize a video transcript but worried
286. 45:08that's too complicated brother 52 I just did it I I just popped in a 2hour 2 and a half hour transcript into Claude do it
287. 45:15in Claud it's not too complicated like literally copy and paste the transcript copy the transcript out of whatever thing you've got it in
288. 45:25paste it into Claude claw. A and tell it to summarize it and then start asking it questions about it it's it's remarkable it's
289. 45:36remarkable um oh do I recommend any book on gpt3 language no so Zippy to pin pin head here's the deal okay we're gonna
290. 45:47we're gonna we're going to go we're going to go a little smarmy for a moment so all of these tools right now
291. 45:54all of the AI tools are what I lovingly refer to as janky pieces of [ __ ] including Chachi BT not because the
292. 46:03companies are bad not because the technology is actually a piece of [ __ ] it's just that it's janky why is it janky because it's all
293. 46:10new generative AI the the ability to interact with chat GPT there are likely lots of books out there about
294. 46:18prompting most of them were probably written with chat GPT or GPT 3.5 or GPT before um no one is an expert right now these tools are
295. 46:31changing incredibly fast in fact um I heard last night on the live here someone someone said to me um the Google's new
296. 46:45Gemini model has been released to select customers that's their new replacement for for Palm 2 the stuff sitting
297. 46:54underneath bard um which it's actually pretty good right now but apparently Gemini is quite remarkable I haven't I
298. 47:00haven't played with it I haven't touched it but whatever book you would have read on chat GPT when Gemini comes out if
299. 47:09it's multimodal and if it's better and if it does certain things better and maybe it uses a different kind of you
300. 47:15know natural language maybe it likes uh structure or it doesn't like structure I like we don't know but when it comes out
301. 47:24whatever you learn for chat GP is is likely going to be not as relevant all of a sudden so so my advice right
302. 47:33now is if you if you can find you know a channel like the AI learning lab or Rachel Woods or McKenzie Bose or you
303. 47:42know you find some Tik tocker or some YouTube channel where the voice of the person doesn't bug you I don't know why
304. 47:48I would bug you why don't you listen to my channel it's not fair you know you got someone like that you can't listen
305. 47:56to them or maybe you love that find someone you like and and you know you can start doing it that way um the the
306. 48:02thing about watching you know Tik Tok or YouTube channels where they're constantly putting out new content is is
307. 48:09unlike a class or a book where you sort of have to you know put a stamp in time um they're going to be updating
308. 48:18things the the the thing that I think is is the the better approach though is get curious yourself start start
309. 48:26understanding how prompting Works start understanding the difference between you know a Google prompt which you know a
310. 48:34Google prompt Google has sort of trained us that you put in sort of the minimal amount of words because you if you over
311. 48:39complicate a Google prompt it tends to just give you really shitty results these things are kind of the opposite if
312. 48:45you give it too sparse a prompt it doesn't really have enough context and the way they work it just gives you
313. 48:51generic answers and it's just it's just bad but if you give it lots and lots of context if you start to have a
314. 48:57conversation with this thing it starts to get better and better and better you can learn that from a book but the book
315. 49:04is going to be so out of date that I just I I would you know um uh shy girl or I forget what your name actually is
316. 49:12the what your handle is but shy girl was talking about you know she's you know one out of 10 people she talks to even
317. 49:19asks a question and probably you know one out of five of those is actually truly engaged um but the one out of five that's truly
318. 49:28engaged now she's got a community like start a community of people that are curious about this stuff and exploring
319. 49:34it join the AI salon so this is a group that I started in December of last year um so if you go to the salon. there's a
320. 49:44there's there's a bunch of links there it's a link tree the the first one talks about the salon and what our values are
321. 49:50the second one is the Meetup we have bi-weekly meetups and then we've got a Discord Channel Channel where you can go
322. 49:56in and um ask questions you can play in mid Journey things like that I would just start you know just sort of dive in
323. 50:05um get curious learn on your own because the stuff's moving too fast to feel what I think the risk is of taking a class or
324. 50:14reading a book is that you'll think you've learned enough you you'll you'll think that the actual answers are in the
325. 50:21book or in the class they're not the only thing that's in the book or in the class is sort of a step up to get you
326. 50:29curious enough to start learning more because it's going to be the next two years is just going to be you're going
327. 50:35to have to learn this is this is not a low energy next two years by the way you can sit back and not participate and let
328. 50:43this stuff just blow through the blow through the economy but if you want to participate in this stuff you're going to have to be fairly
329. 50:50active all right cut and paste a syllabus yes do that as I'm experiment oh yeah uh yeah cut and paste a syllabus for um for that
330. 51:02stats class absolutely pop that in there pop in yeah have it yeah exactly good as I'm experimenting I'm finding that chat
331. 51:11GPT is giving me better info than barter Bing yes so gp4 in particular is so so so to put it in context um Bard and GPT
332. 51:263.5 are based on um um 175 billion parameter models it doesn't really matter what that means but GPT 4 is actually based on eight different
333. 51:42200 billion parameter models so it's however much that is 1.6 trillion parameters um so gp4 is markedly better
334. 51:53now the gp4 at open.com is very very different or the the chat gp4 at open.com is very very different than the
335. 52:05chat GPT at Bing chat so both of them are based on the gp4 foundation model but however they coded this app and
336. 52:15whatever safety guard rails they put into it have have essentially made it functionally unusable I think Bing chat
337. 52:22at this point is useless so I'm just the more I'm using it and the more it keeps telling me well I'm not even going to
338. 52:29answer that you're just too nasty you're just too nasty I'm good enough I'm smart enough and dog on it people like
339. 52:37me it it I mean it it it honest to God Bing is like the steuart SMY of of the chat GPT models oh well you don't need to BU so
340. 52:52offensive well so yeah yeah and and you know what what you're hitting on and also is different projects like different
341. 53:08things you try to get it to do like like like you'll get chat GPT you'll think you'll have it totally dialed in and
342. 53:14then you'll have some new kind of project and you're like oh great I'm going to use chat GPT and then it'll just fail miserably why I don't
343. 53:23know so go try Claude or go try Bard or go try whatever so anyway so yes I do not um couldn't custom instructions be
344. 53:35used for the stats class question it could Firefly although custom instructions the idea with custom instructions is to to sort of set your
345. 53:44chat GPT up to to be an ongoing thing if if you wanted to basically if you wanted to use chat GPT as kind of a stats teaching assist
346. 53:55istant actually that's kind of cool let's go do that let's go play with this okay that's a really good idea it's a good idea Firefly that
347. 54:06Firefly very clever it's very clever very clever all right uh chat chat chat GTP anyone else have a pet peeve when people say chat
348. 54:17GTP come on it's it's right there I know the letters rhyme but it you're reading it the the P's before the T all
349. 54:29right he seems a little bitter I don't know why why is he so testy about people making mistakes sometimes people say
350. 54:36chat gbt they didn't mean it yeah okay listen listen listen people okay so one of the things that you may or may not
351. 54:47know for both GPT 3.5 and GPT 4 for all of them you have a thing called custom instructions and with custom
352. 54:58instructions you can tell chat GPT to act like something so so this is my joke custom instructions but we're going to
353. 55:05replace this but but I'll show you how they work so my joke custom instructions are you're an overbearing pompous ass
354. 55:13who uses large words that even you don't understand as a way to intimidate those that you talk to and then how would you
355. 55:19like chat gbt to respond whenever anyone asks you any anything I want you to use an obscure literary ref reference to
356. 55:25belittle their idea and then redirect the conversation to talk about yourself it sounds like a nice guy right and so
357. 55:33now we can go in here and we can say um i' I'd like hang it like advice for my stats
358. 55:49class all statistics a subject all often misconstrued by the uninitiated it reminds me of Mark Twain's humorous observation facts are
359. 56:01stubborn things but statistics are pliable now allow me to indulge you with a profound Insight in my extensive
360. 56:08intellectual repertoire statistics are but a mere footnote nevertheless I shall Grace you with my wisdom right so so the
361. 56:17custom instructions basically turn Chachi BT into whatever the hell you want it to be so I'm going to what I'm
362. 56:23going to do is I'm going to turn off the pompus ass for a moment and I'm just going to use GPT 3.5 and I'm going to say
363. 56:42um I want you to describe a very helpful statistics professor in a way that gives gives their
364. 56:57background their teaching philosophy and their areas of
365. 57:14expertise all right I think that's too many words so I'll say that's great but the professor doesn't need a name and can
366. 57:28you consolidate this description into two or three paragraphs so that it's more concise all right so statistics
367. 57:43Professor holds a PhD in statistics a master's in Applied Mathematics their research particularly in basian
368. 57:49statistics and hierarchical modeling is widely recognized in the field with a deep understanding of mathematical
369. 57:55foundations blah blah blah blah blah blah blah blah okay perfect so now we're going to say now I'd like this
370. 58:16professor to be very helpful whenever anyone asks a question so can you describe how this professor should
371. 58:29respond in the different ways people might ask them questions about statistics and this will all make sense
372. 58:39in a second I I assume you you know where I'm going with this so what did it do here um certainly here's when students seek Clarity on
373. 58:48statistical Concepts okay responds with patience and Clarity when they apply okay so I don't I this might be too if
374. 58:57if they're too long we we'll have to uh concise it down again
375. 59:13[Music] um okay so now we're going to go into custom instructions I'm going to turn it on and I'm going to replace you're an
376. 59:23overbearing pomp s with the statistics thing and then I'm going to save it and then I'm going to go grab the how to respond
377. 59:45stuff come on come on copy paste don't be a dumb dumb there we go da da paste okay I have too many I thought
378. 1:00:01that might be the case so we're going to go save we're going to come back out here um um again please consolidate the
379. 1:00:19instructions what's going on into two to three paragraphs so let's see known for unwavering commitment
380. 1:00:38uh okay um rewrite the instructions to refer to
381. 1:00:55the professor as you you are a renowned okay when faced with conceptual questions you excel at simplifying
382. 1:01:09complex ideas and relatable concepts for practice okay so basically why why I went through that long and really boring exercise I'm
383. 1:01:20sorry I wasn't talking enough about what the hell I was doing there so if if it wasn't obvious I had chat GPT describe a statistics
384. 1:01:31professor and then I had chat BT write us how that Professor would respond anytime someone asked questions so now
385. 1:01:39we've got a chat GPT that's been trained up to be that so now I'm going to go to gp4 this is going to be a bit slower but but let's see um
386. 1:01:59my name's Kyle and while I'm curious about statistics I have a degree in acting so you can make some assumptions
387. 1:02:07about my mathematical skills and competency period I'd like to understand where I should even begin understanding statistics
388. 1:02:23for a second year graduate program now let's see how it does hey Kyle great to hear you curious here's
389. 1:02:34where to start basic terminology con uh conceptual understanding simple visualization practical steps online courses inclusivity and open
390. 1:02:45dialogue by starting with these foundational aspects okay so so let's say um great can you teach me some of the
391. 1:03:02terminology by creating a back and forth game where you ask me questions and I answer period don't move on to the next
392. 1:03:14question until I've successfully answered the previous absolutely Kyle let's make this fun and interactive question one what do
393. 1:03:28you think the term mean refers to well like in the movie Mean Girls Mean is how they act to other people as in they're Mean [Laughter]
394. 1:03:45Girls haha I appreciate the movie [Laughter] reference um I was just joking I think mean means the average is that
395. 1:04:01right you got it Kyle the mean and statistics is indeed the average of a set of numbers you find this by summing
396. 1:04:09up the numbers in the data set and dividing by the number of items so anyway so so I now have a chat GPT that
397. 1:04:16that I could probably you know and and you know if you know anything about statistics you could obviously you know
398. 1:04:22make make this much more specific much more interesting so anyway yeah that's a really good idea Firefly thank you for that that was
399. 1:04:29cool okay sis I'm a little winded I'm a little tired I'm a little tired what' you do Friday night I don't know I
400. 1:04:38listen to some old guy talk about janky pieces of [ __ ] and he turned chat GPT into a statistics Professor what have
401. 1:04:45you done with your life I don't know it's not good it's not pretty it's not pretty at all oh yeah oh yeah y'all see
402. 1:04:53this from last night night this is a good [Music] one the nasty lady the nasty lady nasty she nasty Sam Corn's in the house what's
403. 1:05:08happening sir good to see you good to see you good to see you all right let's see Tik Tock I lost my place all right
404. 1:05:18thanks for the stats demo need to take the class need to take the class undergrad degree yeah I know if I wanted
405. 1:05:25to learn anything I mean my problem about statistics is I just I got no interest in it although with all this AI [ __ ] I
406. 1:05:35probably should because it's pretty good at that stuff I should probably get actually actually that might not be a
407. 1:05:41bad idea I you know this idea that you can you can basically take any subject and get up to speed on it instantly and
408. 1:05:50then get you know decently competent with it in a very short amount of time there's something I don't know something pretty amazing about
409. 1:05:59that all right I find that Bard and Bing give me summaries of the sites they search yep oh oh yeah that's interesting
410. 1:06:08I dm'd you I'm an IR regular and still not a mod what's up talk let me go do that now where do I do it settings
411. 1:06:20mods Plus what's cuz I know I followed you what why am why is it not finding you huh well well well a that special
412. 1:06:34well it appears we've got a little problem with Satan doesn't it a that [Music] terrific what's up talk Ed okay confirm
413. 1:06:46you are now a mod thank you all right I don't know why I couldn't search for the word what in what's up talk but
414. 1:06:58you know what are you going to do agree not to be on the bleeding edge after 40 years and retired yep good thoughts cart before the horse
415. 1:07:09seems to be a thing nowadays in it absolutely Winston I just shared the live with someone by accident when I
416. 1:07:18picked up my IPA I didn't notice who that's awesome they'll be like why did you send me this old guy like rambling on about Winston are you
417. 1:07:35okay Winston needs an intervention I think maybe that should be all of our goals is to is to have you
418. 1:07:43know our friends and loved ones give us interventions Kyle we need to talk about the AI okay listen it's getting serious listen
419. 1:07:52um I'll speak on behalf of all of us know that we all love you we listen Kyle you're just you're amazing we've known you for years as you
420. 1:08:04know I oh I'm getting choked up oh my anyway um about this AI stuff it's listen we know you're passionate um the Tik Tok Channel
421. 1:08:22um it it's cute that you you you think you're youthful yeah exactly we should all have one of those interventions in our life I I like
422. 1:08:33it all right and major gift giving major gift fundraising that's a that's a that's a lot of words in that name and it's it's
423. 1:08:43still I appreciate it it says what you do and I don't think you should change it but for this tongue it's not so good
424. 1:08:50uh just interested in new new tech always curious well this is Tech to be it's good to be curious about this
425. 1:08:58Tech because it is not going away and it is going to affect oh you know everything hair growth geek thank you for the lovely
426. 1:09:10heart I like to wear the heart on my sleeve huh that's facts all right I think prompting will change as the models
427. 1:09:21improve as well I'm a librarian so yeah yes I think that learning prompting right now is kind of the equivalent
428. 1:09:33of learning to do math lonand knowing that calculators exist now the calculators don't quite exist yet
429. 1:09:44right two years from now I don't think we're going to have to do much prompting we're just going to sort of say what we
430. 1:09:50want and it's going to go figure all that [ __ ] out it's going to prompt it itself they're going to be autonomous
431. 1:09:56agents so I don't think long term that you need to understand prompting that said if you're in the development side
432. 1:10:02of AI I think it's going to be an ongoing requirement but even if you're not as these tools get more sophisticated if you've taken the time
433. 1:10:13to learn prompting right now you're going to be much better at using those tools when they get more sophisticated
434. 1:10:19because you're going to understand how they work and how they do what they do and how it's different from from how it
435. 1:10:24used to be all that sort of stuff little green dude I like that started following me thank you um it's funny my my friend David LaVine
436. 1:10:35from back in my New York days he had a band called little green and he and he always said dude so little green dude
437. 1:10:42you've got Yoda there but it it reminds me of David LaVine and his band little green little green little green little
438. 1:10:50green little green so yeah I agree with you I'm a librarian I think I think prompt is going to evolve pretty
439. 1:10:54dramatically over the next two years watching Kyle here will improve prompting 100% yep I hear him in my head
440. 1:11:02as I type my prompts I'm so sorry Ted but but good I think I think that's a good thing volunteered on Alice back in
441. 1:11:09the day oh wow very very cool that's super cool yeah it's it's it's it's really cool I mean you know I mean
442. 1:11:18listen there there are advantages to being young like you know would it would it be good to be young again it would be
443. 1:11:24good to be young again you know with with this much life experience um but I think I think anyone who went through anything in the early
444. 1:11:33days of the web and and you know having experience starting in the early days of the web and kind of writing that thing
445. 1:11:41up and and not thinking there'd be another one of those in my lifetime and knowing that we're at the beginning of
446. 1:11:47this one like I can't tell you like the level of EX well no I can you obviously can tell tell I'm a little excited about
447. 1:11:54this [ __ ] but it's like you know people sometimes say to me like how do you have so much energy it's because like I Don't Want to
448. 1:12:04Miss a [ __ ] moment of this cuz this here's the deal I mean this may be self-evident you don't get to be at the beginning of something this
449. 1:12:16transformative ever again like like I was in in my in my um AI office hours on LinkedIn today Jim Ross was in there who's sometimes in the
450. 1:12:31lives and um I said and and he he um he owns and runs um uh storage facilities and I said you know what are
451. 1:12:41you doing with AI and he said oh I started back in November well [ __ ] he started at the beginning right and even
452. 1:12:49if you're just getting in now you're n like we're nine months into this thing this is the beginning like you you you
453. 1:12:56you don't get to come back here you don't get to say I knew what it was like before these tools got good I knew what it was like to
454. 1:13:07understand how they worked so yeah it's an amazing time it's a dang amazing time it's a dang amazing time
455. 1:13:18dang it you know what I'm saying you know what I'm saying do you know what I'm saying listen listen we got to talk about this
456. 1:13:27AI stuff Kyle it's just not right it's just not right you know we go down to the VFW you ain't been hunting forever I
457. 1:13:37do you even fish anymore do you own a fishing pole do you own a fishing pole or is it all like I'm AI now you can't catch a fish with AI can
458. 1:13:47you oh I suppose you're going to tell us you could do that too good Lord you want a paps you want a blue ribbon no yeah I
459. 1:13:59figured not because you're probably like oh I got to have my wits about me to learn about the AI well anyway we're thinking about you
460. 1:14:09come on down to the bar if you want to you know I'll buy a drink or something maybe if you know the paycheck clears
461. 1:14:14you know it's rough it's rough but anyway I don't know I I ain't learning this AI stuff I mean them [ __ ] robots
462. 1:14:23going to they're going to take us out you know they're going to take us out and uh I don't know what you're doing
463. 1:14:29there but uh but you always been a little you always been a little different you know you know I mean I
464. 1:14:35like you but you know the the guys are talking you're just not the same so anyway that's how I grew up Gemini will believe five times GPT
465. 1:14:49for it launched 20 times in a year yeah it but is that coming from is that coming from Google or is that
466. 1:14:56coming from people that are using it cuz I don't know what five times GPT for at launch is cuz gp4 is 1.8 trillion
467. 1:15:03parameters it's eight different models they're calling it a mix of experts model like is it multimodal like I don't
468. 1:15:11know enough about Gemini I want to know about Gemini tell me about Gemini anybody know about G Gemini Pate are you
469. 1:15:16in here pate's been in here pate's doing some some uh some stuff at Google he's got one of them their uh what do they call them
470. 1:15:29phds I called him a mathematician he's like I'm not a mathematician I'm like uh you have a PhD and you work at Google
471. 1:15:37I'm I'm thinking you might be a little better at math than me but anyway I'm sure Pate can't talk about it eventually they may have to
472. 1:15:45standardize prompts or part of promps to get results I.E GPT bullion you know what's interesting I'm a librarian is that
473. 1:15:53um I've heard a number of people now um Rachel Woods talks about this in in uh in her Channel um where adding structure
474. 1:16:03into prompts like like um if you if you say hey based on this article and then you paste in an article if you put like
475. 1:16:12a text like an HTML bracket that says text at the at the top and then a closed bracket at the bottom like an HTML Clos tag um apparently
476. 1:16:24it um these tools act better because it really clearly demarcates your the different sections of your text so
477. 1:16:33apparently adding structure to um chat GPT prompts is effective so that might be something worth experimenting with guilty King
478. 1:16:4348 no way I was just about to use it I don't know what it is but that's super cool glad to hear it's not just me with
479. 1:16:50Bing it says here's a website I know Bing is just Bing is a janky piece of [ __ ] they got to get their stuff
480. 1:17:03together gp4 also has specialized plugins to customize your project search yep chat GPT does not yep well chat GPT 4 does but
481. 1:17:15yeah um fun yes guilty King if you want to upload your own image as a base to start with which which is a good one to use
482. 1:17:27oh you can do it in mid Journey um it's a little weird you can do it in Leonardo you can use a base image in Leonardo so Leonardo
483. 1:17:38is just a website sitting on top of stable diffusion and you can upload a source image and basically do imageo image prompting
484. 1:17:47um you can use photoshop's generative fill so depending on what you want to do with your base image like if you've got
485. 1:17:55a person in your image wearing a red coat you want to change it to a yellow coat you can just go into Photoshop
486. 1:18:02highlight the red coat and type in yellow coat and it will replace it um which one's the best it depends on
487. 1:18:11what you want to do I I would say the one where you can get the most are they predictable results the
488. 1:18:18one where you've got the most control is photoshop's generative fill stable diffusion and mid journey in terms of
489. 1:18:25their imageo image stuff are are both um fairly unpredictable so now if what you want is to like start with an image of
490. 1:18:36yourself and and kind of have it take creative Liberties and make like a really stylized version of yourself then
491. 1:18:43I would say stable diffusion and mid Journey if you want to take like you know the you can make money with time if
492. 1:18:50you want to take a photograph of this button and put it in the middle of a field somewhere then I would use
493. 1:18:55generative Phill for that right so so it depends on what you're trying to accomplish Chad add that's that's it
494. 1:19:03don't she just love chat ad well hello there can't stay along because I've got rashash Shana tomorrow Services early
495. 1:19:09tomorrow Happy rajash Shana um Sam but thank you for hanging out for a bit um can't resist stopping in I'm so sorry
496. 1:19:18hey there Firefly it's cool you're connecting with each other I love that hey regulars I hope you're all having
497. 1:19:24lovely evenings to you too sir um great to see you Thelonius uh can it take on the personality of the skipper and SS
498. 1:19:33minow and treat me like first mate Gilligan yes and I think that has to happen so we're going to go do exactly
499. 1:19:42the same thing we did with the stat statistician thing so wait I got to go turn Okay so for those of you that are
500. 1:19:48new here what I'm doing is I'm going to go play in chat gpt's custom instruction so before I made it a a statistician
501. 1:20:03and I'm going to kill all that and then okay so we're going to go expert stats Professor so basically I'm going to go
502. 1:20:16into why didn't this load Come on load load load okay so here's a cool thing you can do so if if you've been here for a bit so
503. 1:20:29what I did before was I had chat GPT describe a professor a a a professor of statistics and then I had
504. 1:20:39it describe how it was going to behave and answer questions so now what I'm going to do is I'm going to say okay now I want
505. 1:20:54you to rewrite the description of the professor but instead of being a professor you're going to be the skipper
506. 1:21:09from Gilligan's Island period And I want you to rewrite the description of how the skipper will respond and I want the skipper to resp
507. 1:21:22respond only as if you're Gilligan and you're a bumbling idiot that frustrates the heck out of you all right so hopefully like I what I
508. 1:21:37found in the past is that Chachi PT is actually really good at this all right so certainly here's the
509. 1:21:43Rewritten description of the skipper as the character the skipper known for his bumbling yet endearing personality
510. 1:21:52that's not really his thing that's Gilligan's thing oh often finds himself exasperated when dealing with Gilligan's
511. 1:22:00mishaps when faced with questions or challenges his responses are typically marked with a mix Mi mix of frustration
512. 1:22:06and resigned Amusement for example when Gilligan asks about a construction constructing a rescue signal the skipper might reply Gilligan
513. 1:22:15it's as simple as tying two sticks together but please don't mess this up when Gilligan Ires about okay um okay that's not quite right
514. 1:22:31so okay for the Skippers description I want you to start with you are the skipper from Gilligan's Island and then
515. 1:22:44describe the Skipper's personality period for the description of how the skipper will respond I want you to use the term you will
516. 1:22:59respond as Gilligan asks questions period um provide
517. 1:23:19details of the kinds of things Gilligan might ask and the kinds of responses the skipper might provide let's see if those are a little
518. 1:23:31better and again you know the the the the big lesson I've learned with chat GPT is it's less about prompting and it's more like you're
519. 1:23:41working with an intern it's like it's more like you're talking with like a really thickheaded intern that's like
520. 1:23:46well you write fast and you know everything but you're kind of dumb uh okay you are the skipper from Gilligan's
521. 1:23:54Island a character known for strong leadership rugged demeanor and a Perpetual frustration by Gilligan's
522. 1:23:58Antics perfect your personality is that of a seasoned sailor often exasperated by Gilligan's well intention but you
523. 1:24:05care deeply for the well-being of The Castaways on the island and do your best to keep them safe and organized despite
524. 1:24:11the chaos caused by your first mate Gilligan that's perfect as Gilligan asks questions and seeks guidance your
525. 1:24:19responses are colored by the history of Gilligan's Miss TS perfect for example if Gilligan inquires about building a boat to escape
526. 1:24:27okay when Gilligan ask seeks your advice your responses okay this is perfect so I'm going to grab the description of the skipper and that's
527. 1:24:37going to go in our first thing here I'm going to save it and then I'm going to describe I'm going to grab as Gilligan asked
528. 1:24:48questions we're going to say this is how you're going to respond and we're going to save that and so now we have chat gbt is the skipper and
529. 1:24:59we're Gil again hey Skipper how's it going today anything I can do oh Gilligan you always mean well but
530. 1:25:11you're asking if there's anything you can do just reminds me of all those times okay it's a little bit on the nose
531. 1:25:16um but you know what wait am I in GPT 3.5 I am okay so we're going to go to GPT 4 gp4 so so this is one of the subtlety
532. 1:25:26subtleties that you'll find between 3.5 and four is that GPT 3.5 writes structurally well but shitty and gp4 just tends to be a better
533. 1:25:40writer okay so um hey Skipper how's it going I'm ready to solve any problem you want me
534. 1:25:56to solve that was horrible question ah Gilligan good to see your enthusiasm but let's not forget the time you tried to
535. 1:26:04solve our food storage Problem by burying the canned goods you ended up creating a treasure map for the coconuts
536. 1:26:10but all right we have something that needs fixing the radio is acting up again can you take a look without electrocuting yourself this
537. 1:26:19time that's funny but I just put the radio in a bucket of water Gilligan bucket of water why on Earth would you do that all right so
538. 1:26:34it's not quite as funny as I wanted it to be but I suppose the writing on Gilligan's Island wasn't all that great either but
539. 1:26:42anyway that's a good example of just you know [ __ ] around with these tools all right hey chat GPT right yeah well right have
540. 1:26:59chat GPT write a Gilligan's Island treatment where they get off the island in the same day series over um actually that's not a bad idea
541. 1:27:14uh I'm gonna say new chat
542. 1:27:28I don't think I can do it as the skipper hey Skipper I want you to come up
543. 1:27:42with a plan for getting us off the island today do you think that's possible Gil boy remember the time you
544. 1:27:53tried to take charge of our Escape Plan yeah no I can't do it I can't I can't do it with with this as the as the custom
545. 1:28:00instruction I I would I would in that case I would basically say you know you're you're a writer that that you
546. 1:28:07know is writing the sequel to Gilligan's Island and it's it's a realtime you know three-hour exercise that gets them off
547. 1:28:16the island they patch the boat and leave something like that uh moderator Pinn comment oh man I was wondering why the writing was a bit
548. 1:28:26mediocre because it was still 3.5 and not four yep thanks for the stats demo I need to take the class that was interesting I
549. 1:28:38use chat GPT to write a song but how do I get the keyboard chords for the songs using chat GPT so um you just you just
550. 1:28:46ask for it so let me do gp4 because it's better write me a country song about rich men north of
551. 1:29:06Richmond verse one we went down on the coast where the water's fine but up north of Richmond there drinking expensive
552. 1:29:17wine I'm singing it like he does rich man North rich man got your gold and your bling but you don't understand a ain't money money ain't
553. 1:29:30everything you can't buy a mansion you can't U buy a friend Rich Man and north of Richmond one day you'll see in the
554. 1:29:37end um all right so there it wrote the you know the verse in the choruses so now I can say um I'd like you to add choruses to this and
555. 1:29:52have it be more sophisticated than a three chord country song starting in the key of G and using major 7ths and minor chords
556. 1:30:05as appropriate to keep it interesting and so now it should rewrite yeah so so it's rewriting it in tab right so this is G major 7th C
557. 1:30:20major 7th D7 and G or 7th and so it's actually sort of burying the chords within the signals right so this is this is basic You Know Tab
558. 1:30:31structure chord you know chord structure um and it you know it it understands basic music theory my experience with
559. 1:30:41this is that um the more you know about music theory and can look at what it creates and go no that's bad or that's
560. 1:30:50cliche or that's not following music theory like like actually an interesting way to do this would be um let's see let's see
561. 1:30:59um pretend like you're a music theory teacher and explain to me the difference between a basic country song and a sophisticated country
562. 1:31:14song ah music theory oh wait is this do I still have on oh yeah hang on I I still have on the skipper here let me stop
563. 1:31:34generating stop generating custom instructions turn that [ __ ] off okay so let's say um I want you to act as a music theory
564. 1:31:51professor from the Berkeley School of Music period I'd like you to quickly educate me in basic music theory for country songs
565. 1:32:04period And I want you to describe two different structures of a basic country song compared to two different
566. 1:32:17structures of sophisticated country songs and explain how the chords make a difference I don't know something like that let's see what it
567. 1:32:31does welcome to country music 101 major and minor chords common chord progressions the 251 is a more Jazzy
568. 1:32:45progression but can still be found verse course verse course Bridge verse vers Bridge so so it's it's given
569. 1:32:54us you know not only not only chord structure but song structure um Extended chords like seventh and Ninth or even suspended chords add
570. 1:33:14color this is great can you write me the lyrics for a country song called rich men from
571. 1:33:37Richmond on the banks of the James of the river bends grew up with the stories the myth the Legends Boys in blue jeans
572. 1:33:44dreaming of gold hoping they'd be tale hoping they'd be the tales that are told rich men from richm with dreams so tall they got to their
573. 1:33:52Mansions but they've seen their Falls all right so all right still still writing GPT 4 is slower but better okay so there's that so
574. 1:34:04now interpret these lyrics and add in chords using the more sophisticated country
575. 1:34:18approach that you describe above in a way that brings these lyrics to life certainly here's my interpretation
576. 1:34:34starting in the key of G so this is g cad9 g overb d sus 4 and G that's pretty cool on the banks of the James so so it
577. 1:34:44didn't quite format it the same but you know now you've got something that is at least got interesting chords but it's got the
578. 1:34:57chorus and the verse using the exact same chords so what I would do in this case the bridge actually does something interesting with the
579. 1:35:16chords uh uh uh d d d outro nice so I'm going to
580. 1:35:32say can you analyze what you wrote period I noticed that the chord changes in the verse and the chorus are
581. 1:35:47identical and in most songs I know they are different period did you do that on purpose or not question mark if not how
582. 1:35:59would you improve the song after you analyze it analysis you've rightly noticed that the verses and choruses have the same chord
583. 1:36:12progressions that similarity can create a continuous flowing feeling throughout the song but it might also result in a
584. 1:36:18lack of contrast between the sections in many songs it's important to distinguish the chorus extended chords all right improvements to add
585. 1:36:31contrast we can change the chord progressions in one of the sections since the chorus is the song's centerpiece I'd recommend altering the
586. 1:36:39verse chords so the verse is now G E minor c and d and the chorus is all the fancy [ __ ] um this may or may not be any good
587. 1:36:50but anyway that's that's how I would approach that so yes it can do all that can it do it good I don't know and again this is a place where
588. 1:36:59custom instructions could be really interesting right where where you could do custom instructions where you know um
589. 1:37:07tell Chad GPT to act like you know Master country songwriter and you could say you know like like you know Brandy
590. 1:37:14Carlile or John Hyatt or you know take people that you know and then in how they respond tell them that you want
591. 1:37:20them to songs that are interesting and and compelling and whatever it might be and I I don't know not derivative and
592. 1:37:28and so so yeah you could fine-tune this to to actually probably be a really good songwriting partner or or or don't even
593. 1:37:36do uh Brandy carile you know describe yourself right here's the kinds of songs I like here's the kinds of chord
594. 1:37:42progressions I like here's the kinds of you know metaphors I like using or or you know I like more straightforward
595. 1:37:49lyrics I Don't Like ryy rhymey [ __ ] right I like more sophisticated Rhymes um all that sort of stuff and basically
596. 1:37:56if you get that dialed in right you'll have a partner that you can basically pop in an idea and it'll get you to an 80% version of your
597. 1:38:07song in 3 seconds and you do 30 of those that would be awesome you know wouldn't that be awesome that would be
598. 1:38:17awesome yeah smik Tyson here yeah that would be awesome that'd be awesome all right let's see what time we
599. 1:38:28got 10:30 good Lord people what are you doing with my life what are you doing with my life it's Friday night Friday
600. 1:38:34night what are you doing with your life why you on a Tik Tok why you on a Tik Tok watching an old guy why are you
601. 1:38:41watching an old guy on a Tik Tok it don't make no sense we're going to have to have an intervention you know I don't know if
602. 1:38:50you knew this is but uh you talk a lot about artificial intelligence you know artificial intelligence was there before last year
603. 1:39:01you know that right oh I know all right I'm going back up into the questions I'm jumping up in there if
604. 1:39:10any of the mods uh see any interesting questions go ahead and pin one and I'll do that but also go look for some oh man
605. 1:39:17I was Wonder okay that was what we saw before while I've been watching I had chat GPT write me an entire booklet for
606. 1:39:24my industry as a download perfect Jim oh so Jim Jim was on the uh Jim was the one I was talking about that was on the uh
607. 1:39:31the LinkedIn office hours today who's been using chat gbt for like since November and yeah awesome so he's he he
608. 1:39:41created a a booklet for his industry that's fraking awesome yeah and and you know Jim because you know your industry
609. 1:39:49so well you're going to be able able to look at what chat GPT wrote and go oh Section 3 is [ __ ] and it you know
610. 1:39:56that's that's old thinking and we do things differently now and then you can just tell it that rewrite section three
611. 1:40:01so that it includes the new ways of doing things um the other thing to do Jim if you haven't done it I think I
612. 1:40:07showed this today no I didn't let me show you something cool Jim if if if you you probably know this you've been using
613. 1:40:15this long enough but just in case not other people may not all right so we're going to go did I I turned off custom
614. 1:40:21instructions right oh no wait did I yeah I did okay welcome to chat add all right so I'm going to say um let's
615. 1:40:37see you are an expert at the logistics of starting managing and expanding a storage unit business period
616. 1:40:51I want you to write a booklet for beginners who are getting ready to invest in their first storage business
617. 1:41:04period and then and then here's the thing that if you if if you don't know this trick this is this is a good one um
618. 1:41:10sometimes it will just do this um automatically but if it doesn't this is a good one to add so so here's here's what I'm adding to it use
619. 1:41:19markdown to format the booklet to make it easier to read period be sure to include titles subheads and bulleted lists as
620. 1:41:34appropriate and then what it will do is like give you all the sort of big Bolding welcome to the exciting world of
621. 1:41:41storage unit business table of contents right so here's the Bold here's the numbered list right introduction subtitle you know preliminary
622. 1:41:54considerations so anyway it's probably something you know but if you don't know that the the the term format using
623. 1:42:01markdown markdown is basically like a kind of simplified version of HTML for lack of a better metaphor um and you
624. 1:42:08basically just tell it to do that and it kind of knows it it knows how to bold stuff and make bulleted lists and things like that so
625. 1:42:17anyway welcome Jim mm [Music] three classes left uh what app generated
626. 1:42:33that cool Saloon pick the cool Saloon pick this was done in um ideogram and then all of the outside of
627. 1:42:48this was done in um adob Adobe Photoshop generative fill like she was done in Adobe Photoshop gener fill I put like cranky
628. 1:42:58bartender and and that was what it created which is she's freaking awesome look at look how angry she is uh wait why is it what's going on
629. 1:43:11here there we go she's so cranky and then I actually went in and photoshopped this this is for the the AI
630. 1:43:21Salon oh no I didn't I didn't Photoshop this one this ISS the AI salani um I did a Photoshop version of this where I took all that
631. 1:43:29out but yeah a combination of ideogram a um that does the letters and then Photoshop generative fill for that I know that's a pretty
632. 1:43:41slick little image isn't it I dig that I dig that one pretty good although this dog in a space helmet ain't bad it ain't
633. 1:43:48bad you know dogs are all right you know they're tasty they're tasty champ mhm I like dogs dogs are tasty M mhm they called it a sling blade
634. 1:44:07mm mod power activate digital Gods oh o I got a got a little dancing star in my head what's your favorite GPT is it chat um um it's
635. 1:44:21a good question so here let me show you I have a top seven and it's not that I necessarily have a favorite one um gp4 is by
636. 1:44:38far in my opinion um the best the best let's see it's by far the one that performs consistently better than the others but it's got a lot of
637. 1:44:55limitations all of these things hallucinate and they they all have different advantages so I'll kind of
638. 1:45:00walk through I'll walk through my top seven these three tend to be the ones that I kind of bounce between the most
639. 1:45:07and and Bard I I don't use a ton unless I need something connected to the internet if like a plugin like um web
640. 1:45:16pilot fails up here so most of the time if I've got a long document I need to analyze I'm using Claude because this
641. 1:45:24has 100,000 token context window which basically means 75,000 words so The Great Gatsby the novel The Great Gatsby
642. 1:45:32is like 55,000 words so you could put the entire Great Gatsby into the prompt of Claude and have it write you a new chapter U um chat
643. 1:45:48GPT has GP 4 which is good it's got plugins which are okay and then it's got code interpreter or now called Advanced
644. 1:45:57Data analysis which is [ __ ] remarkable so this one is still the King of the Hill Bard for most of this year
645. 1:46:07was unusable it was so bad it was unusable it was it was so bad and now it's not now it's actually pretty good
646. 1:46:16um and Gemini is is probably going to be out within a month or two excuse me and um Gemini is purportedly going to be
647. 1:46:27five times better than gp4 I'll believe it when I see it but Google kind of got their ass handed to them from a PR standpoint with with
648. 1:46:39chat GPT and Microsoft for a lot of this year so they have a chip on their shoulder which they should they invented the Transformer
649. 1:46:48that makes all this [ __ ] possible okay then these are the other ones that I think have got some some Merit to play
650. 1:46:55with Bing we've talked about a lot tonight it seems to be getting shittier the the safety guard rails on this make
651. 1:47:01it you know decently unusable unless you're doing sort of pablum kind of you know steuart SMY level interactions I'm good enough I'm smart
652. 1:47:12enough and dog on it people like me yeah um Bing is connected to the internet it is built on GPT 4 it can generate images
653. 1:47:22it can analyze images so there's there's things going for it but the safety guard rails right now make it make it I don't
654. 1:47:30I I I I stopped using it um and I was a fan of it I I sort of pushed it for a while and I just stopped because they
655. 1:47:36got to get their [ __ ] together um perplexity doai on its own is a really interesting tool because when you do
656. 1:47:43searches uh when when you do um interactions with it it provides sources which are actual links to actual sites so it's good for
657. 1:47:54research and then labs. perplexity lets you play with meta llama models then you've got Pi AI which we lovingly in
658. 1:48:03the lab here call that chatty GPT um the thing to do with pi is put it in um get the app get the IOS app or the
659. 1:48:15Android app and put it in voice mode so you talk to it and then basically when you stop talking it just you know takes
660. 1:48:22what you said and it runs it through its model and then it responds with its voice um it's designed for talking and
661. 1:48:29it's it's really interesting and then po lets you play with multiple of these models in the same interface and it's
662. 1:48:36also got Bots and a bot maker so those are the those are the seven that I that I think that that I use on a decently
663. 1:48:43regular basis I think they're all worth knowing something about um someone something else was was uh was oh there's
664. 1:48:50there's something uh pinned right now wonder how accurate chat GPT is with respect to syntax any measurements from Pilot or beta
665. 1:49:03um well chat GPT can code so so one of the things that happened when they trained so this was this was an interesting
666. 1:49:17thing when they first trained GPT 3.5 um they they trained it on sort of like all the data on the Internet or as much as they could get in
667. 1:49:28there and it and it was a certain level of good and then at some point they created the code fine-tuning where they
668. 1:49:35they trained GPT 3.5 to do coding right they trained it on a bunch of programming languages Apparently after they trained
669. 1:49:44it on the coding the language side got dramatically better um so so there's there's some interesting thing where the logic of
670. 1:49:53coding um because it understands that it got better at the logic of English or or language um so it's quite good at coding
671. 1:50:04so it's quite good at syntax so um I would just go experiment with it go have it write you some code um what I've
672. 1:50:12heard from engineer friends of mine and people that are working on it is that the code quality for gp4 has gone down
673. 1:50:20pretty dramatically in the past 3 months 4 months um but anytime I've interacted with it recently it's still pretty good
674. 1:50:28but I'm not doing a lot of coding like if I doing coding dayto day it might be ugly Bard b.google.r good at coding so I would play with that
675. 1:50:39and especially if Gemini is indeed going to be five times better than gp4 it's probably going to be five times
676. 1:50:48better at coding first right right Google's an engineering company not a Humanities company right Google ain't
677. 1:50:55apple apple hires creatives right who are also technologist creative technologist or or creatives and technologist Google hires Engineers
678. 1:51:06brainiacs right so so my guess is that Gemini is going to be better at logic coding syntax things like that now that said Google owns
679. 1:51:22YouTube so Gemini might be way better than gp4 because it was trained on all of YouTube that's a possibility here so
680. 1:51:31I don't know what it was trained on but we'll we'll have to wait and see anyway all right Bing is my boy um bing bing
681. 1:51:40also had a survey today interesting oh my God generative Phil new mascot generative Phill exactly oh Sam go make a generative fill all right
682. 1:51:50anyone who wants to make a generative fill make it in ideogram so by the way ideogram is the thing that can do words
683. 1:51:56so if you see all these things with words on them that's all from id. um so um yeah go make a generative
684. 1:52:06fill and throw it in the uh the share the the showand tell channel in Discord in the salon Discord that's really funny
685. 1:52:15SS minow and away we go all aboard all right come sit right back and you hear a tale A Tale of a faithful trip
686. 1:52:25that started on this traic wait on this aboard this tiny ship the mate was saing man a skipper tried and
687. 1:52:40true I used to know it cold oh my God Gilligan's Island I was I was as much time time as I put into the AI learning lab I put
688. 1:52:53that much time into Gilligan's Island as a kid I [ __ ] watched that show whenever I could is AI able to predict
689. 1:53:02likely likely numbers using a set of old data using probabilities or Series so Prince K yeah well I don't know enough about
690. 1:53:13statistics and able to predict likely numbers using a set of old data but Advanced Data analysis can do full-on
691. 1:53:23data analysis so I assume it can do that um how you play with um Advanced Data analysis is you have to be a chat GPT plus
692. 1:53:36member you have to turn it on so if you don't see it you have to go to your settings and then go to Beta features
693. 1:53:43and beta features you will then turn on plugins and data analysis and then now you go to data analysis and
694. 1:53:54what it allows you to do see that plus button you can now upload a file your system is run out of application memory oh Microsoft Edge
695. 1:54:06you're a real pain in the ass how about if we just pretend like that didn't happen Okay the technologists on this thing are
696. 1:54:17going no you got to restart nah you got to live on the edge people you got to live on the edge um let's go to Archive
697. 1:54:28archive name contains archive desktop archive 4 okay here's an archive for.zip so we now have a zip file here and what
698. 1:54:44you can do here is you can say please open this and tell me what files are inside and if they are data
699. 1:54:54files examine the data and tell me what might be interesting about it so this is um what this is doing oh see it just it
700. 1:55:09just crapped out again this this might not work let's see yeah it's good we're good here let me quit something else we'll quit that [Music]
701. 1:55:28it didn't quit or did it I don't know yeah we'll quit keynote too keynote a big one oh wait I don't want to Quick keynote oh well we just
702. 1:55:44did okay so so what what advanced data analysis is doing is it's creating container I used to say it's creating a
703. 1:55:53virtual machine and and and the technology Boys in here [ __ ] slapped me instead it's not really a virtual
704. 1:55:59machine it's technically a container all right so it's a container anyway it's creating some hermetically sealed little Computing
705. 1:56:06area um with with within um within memory basically and it's G it's been given access to a number of python
706. 1:56:16libraries so what ends up happening is code interpreter or Advanced Data analysis writes and executes its own
707. 1:56:26python code so to open the zip file it it imported the library to do that it wrote the code and it opened it up then
708. 1:56:33it says now I'll examine the CSV file that's in there so it brings in the Panda's library and writes the code to
709. 1:56:40look at the first few lines of the thing and then it says the CSV file contains data on unemployment in various US
710. 1:56:48states for specific months and years here's a breakdown of the columns so it gives you a breakdown of the column so
711. 1:56:53it knows what's in there let's see given this structure we can analyze various aspects such as states with highest is there a specific
712. 1:57:03analysis I'd like you to tell me what you think are the most interesting Trends in the data based on your analysis of
713. 1:57:18it all right I'll perform a comprehensive analysis now so notice what it's doing here I just said I want you to tell me
714. 1:57:28what's interesting about this all right I'll perform a comprehensive analysis of the data set to uncover interesting
715. 1:57:34Trends here are the steps I'll follow analyze the overall trend of unemployment across States over the
716. 1:57:40years find the states with the highest and lowest unemployment volatility identify states with the most volatile unemployment
717. 1:57:48rates long long-term performance check the things now look it's it's already it's creating charts and graphs right so
718. 1:57:56it brings in the matplot library right the P plot as PLT and then it writes this [ __ ] to to write that chart so
719. 1:58:09amazing overall the trend late 70s to early 80s this was a significant rise in unemployment early 2000 there's a
720. 1:58:16noticeable spike in the early 2000 which can be attributed to the dot bubble burst 2008 corresponded to the global
721. 1:58:25financial crisis post 2010 a decline in unemployment rates a slight uptick in recent years here's the states with the
722. 1:58:32highest rates this is doing full-on data analysis fullon data analysis so yeah you can also have it right you can have
723. 1:58:41it Recon you can have it clean up data and rewrite you CSV or Excel files um with cleaned up data um you can make it have it make pretty
724. 1:58:52charts and graphs it's still going right so it's prompting itself to go look at all this stuff and find out what's interesting about the
725. 1:59:01data summary the overall unemployment rate across the US experienced noticeable spikes certain States and areas New York City Michigan West
726. 1:59:09Virginia have been more susceptible to ex economic fluctuations some states Nebraska South Dakota North Dakota have consistently
727. 1:59:17maintained low unemployment rates others have higher averages would you like to an explore I'm going to say
728. 1:59:26um what is the most interesting way you could visualize some of the trends you've identified that will be beautiful
729. 1:59:33to look at and make the information clear visualizations so a heat map of the average unemployment rates by state
730. 1:59:46a line plot a bar plot with top and bottom States area chart for volatility I'll start by creating a heat map of the
731. 1:59:56average unemployment ratees so I know nothing about statistics and data analysis I I mean I've heard of some of
732. 2:00:02these charts but I don't know that like I don't know what off the top of my head so it does and so now what's it going to
733. 2:00:09do here let's see what it does I apologize for the oversight it seems there was an error in plotting the boundaries of the
734. 2:00:16USA instead of using the previous approach I'll use a different method to create the heat map for the average
735. 2:00:23unemployment rate by state let's try that so it may not have access to State boundaries yeah so like whatever that is is
736. 2:00:34weird so that's a blank chart [Music] um yeah that's fine so anyway that's Advanced Data analysis
737. 2:00:48crazy isn't it it's crazy I tell you it's crazy I tell you it's just crazy it's crazy man what are you doing what
738. 2:00:57are you doing running your own spreadsheets anymore put it in the Chet I don't know if you knew this boy you can make money
739. 2:01:07with you say what I'm saying oh boy all right what time is it it's 11: isn't it 10:56 all the Castaways had last name
740. 2:01:19but Gilligan but was Gilligan a first name or a last I don't know I do not know the answer to that I got to go to sleep good
741. 2:01:33night all good night Sam have a good have a good weekend Professor Saves the Day see this is just what the wga writers were afraid
742. 2:01:41of isn't it well yeah I don't know listen I have such mixed feelings about
743. 2:01:56the writer strike I support everything they're doing except for the contention that AI is evil and shouldn't be
744. 2:02:09used they they just it it just feels like both them and the SAG positioning on the role of AI is the misguided piece
745. 2:02:19like I want them to have you know rights for their likeness rights if they use AI to write scripts like all the stuff that
746. 2:02:27they want is good but they've got this attitude of like AI is evil and we don't want you to ever use it and I just think it's really
747. 2:02:34shortsighted um but yeah they they they should they should be both scared and excited like the smart WR the smart
748. 2:02:42writers should be spending the strike getting masterful at using generative AI to help with the writing
749. 2:02:52process they should because the writers that can turn out work faster better all that sort of stuff are going to be in demand in the
750. 2:03:01future um and I think it's still going to require writers right it you know the writing isn't about the act of
751. 2:03:10blocking and tackling language formatting writing is about curating ideas and refining ideas and expressing
752. 2:03:19ideas and while chat GPT can blurt out a bunch of [ __ ] it's ultimately the writer that goes that's the good idea that's the bad
753. 2:03:30idea and the ones that are the most facile with that and can do that quickly and effectively are going
754. 2:03:39to be in demand and the ones who can't won't be so anyway I I think you know just proclaiming AI is evil is is is idiotic I think it's
755. 2:03:52idiotic like not for nothing but chat gppt is the most powerful writing tool in the history of humanity and the
756. 2:04:00writer's Guild is saying we should never use that really we will only use typewriters for here on out these new fangled computers
757. 2:04:15with their memory and cassette tapes that store data it's not a piece of paper is it it's exactly the same as
758. 2:04:30that why would you not use a computer if you're on typewriters let's have a strike and let's tell the studios to that that they
759. 2:04:40can never hire writers that use computers they we only want them to hire writers that use typewriters because that's real writing makes
760. 2:04:50sound every time you strike a letter the letter imprinting upon the page is part of the writing process that's the equivalent of what
761. 2:05:00they're what they're arguing right now I think it's just [ __ ] idiotic I use chat GPT to write a song but how do I
762. 2:05:08get the keyboard chords we did that wow I'm I'm behind on my little questions here can anyone name a famous song that
763. 2:05:15starts with the bridge didn't doesn't Taylor Swift start a lot of her songs with the bridge she's really into
764. 2:05:21Bridges I don't but I don't know I don't know enough of her work to know that I'm I'm in total mansplaining territory here interesting
765. 2:05:31I'll try to keep interesting I will keep trying the results needed too much work yep that's a mega glass of wine this that is
766. 2:05:47coffee or you meant this one that's just water that's that's uh Rocky Mountain tap water this is the intervention you're right
767. 2:05:59Nancy this is the intervention wow results using format markdown and bulleted list prompt neat yeah Isn't that cool Firefly yeah print
768. 2:06:09AI is the past that's good um love the salon piic seems like some AI have problems generating text so Hamilton so this is
769. 2:06:20an important thing um to my knowledge well there's one other tool that sort of does it but to my knowledge the and it's
770. 2:06:28brand new it's like a three- week old tool so if we go to id. get rid of our dead memory bug oh let's see the memory might
771. 2:06:41might have bit Us in the ass stop stop all right are we in we're in okay
772. 2:06:55so ideogram does text it does it well um these guys are new they they're they're brand new I'm going to have to quit
773. 2:07:08Edge oh that's a cool one let's let's steal that one so so one of the things that that I like to do is I'll come in
774. 2:07:17here oh this is going to be unusable and I'll copy a prompt let me just move this out of the way and then
775. 2:07:28keep going as if it's not there oh I can't uh copy and we'll go out here yeah this is this is about to explode it's about to
776. 2:07:44explode 3D render typography poster vibrant and uplifting fuzzy wool style word text AI
777. 2:08:00learning lab 3D render and I've already click those little buttons so I think I can get rid of that crap and we're going to make it vertical
778. 2:08:10and we're going to try to get rid of that and then we're going to say generate and let's see if we get something they look like AI learning
779. 2:08:17loveb now here's the deal the this tool gets the letters right maybe 15% of the time so you have to keep messing with it
780. 2:08:33um but if you can get it close then um like AI learning lab is close learning is
781. 2:08:47close um that one's kind of cool so if you get something that's close but not quite there um what you can do is remix
782. 2:09:00it and just sometimes it'll just don't do anything just keep just make another four based on that one and then
783. 2:09:08sometimes it'll get the the words right yeah like this this one wasn't quite close enough but anyway this is this is uh ideogram a
784. 2:09:18[ __ ] this I'm going to Mars I love that awesome all right I'm quitting Microsoft Edge uh I don't need to force quit just
785. 2:09:29quit yeah there we go did it go it's going it's going oh I might need to force quit ah oh you Macintosh all right listen my Max had
786. 2:09:43it we got the turtles we got the janky stuff yeah we'll go out we'll go out with a nice AI learning lab Ali the AI
787. 2:09:51learning lab it's how we started the night it's how we're going to go [Music] out all right everybody listen for those
788. 2:10:06of you that modded tonight thank you for your modding I really do appreciate it Kyle is there a way to delete pictures
789. 2:10:12on ideogram I do not know I don't know um just go into your count and I don't know see you tomorrow all right Silver
790. 2:10:24Fox all right so let's let's uh let's let's say goodbye who who have we seen tonight uh Nancy and Silver Fox and Zippy to pin head
791. 2:10:34thonus um Pate was in here I knew it hey Pate how's it going girl with needles solid shared the live thank you
792. 2:10:46for sharing the live I do appreciate that Firefly Prince K good to see you who else we got in here tonight Emilio's
793. 2:10:54wife welcome or good night um big fish good night good night good night steo and cam katkin winning AI with cam katkin cam
794. 2:11:05katkin signing off for the night welcome I'm cam katkin this is the news traffic on the eights with Cam
795. 2:11:13catkin I hope cam appreciates me every time I see Cam's name I'm like I love it I love it I gotta I just got to go
796. 2:11:23there uh digital Gods thank you so much all right all right everybody peace out have a great F Friday night it's been
797. 2:11:31fun it's been a good one I had a good time hope you learned something I'd appre I I uh oh Cam katkin I appreciate
798. 2:11:38you love great thank you um I think you're all well this is fun all right I'll see you tomorrow