2/10/2025 - AI, Super Bowl Ads, and the Future of Work: A Meandering Monday Chat

Transcript

1. 0:00[Music]
2. 0:27oh oh oh [Music]
3. 1:03[Music] [Applause] oh
4. 1:20[Music]
5. 1:38h [Music]
6. 2:02don't K my way that [Music] night just like a jet plane in and out of side I was hauling ass at a million
7. 2:17miles an hour wondering how hard i' here when they came into the station I
8. 2:32they said I was bad Beyond repair but I got no quals with my situation say here I
9. 2:50am so say sh Shere Shere won't you dare to say Sheree Shere Shere won't you dare to say Shere Shere
10. 3:08Shere ah yeah leave a message and your message please take a time to want to satisfy
11. 3:23me take all these old fantasies and send them care of me [Music] good evening good people what is happening I am a little early
12. 3:39indeed I was just sitting here champy was sitting here we were thinking should we go should we go what if we want to no time like the
13. 3:52[Music] present it's about time now it's about time to get started [Music]
14. 4:28oh man hope you had a good weekend let me flip my little cameras here hope you had a good weekend I'm on Tik Tock joining the
15. 4:40stream soonish all right there Mr Brandon producers inbound got a producer inbound [Music] [Applause] [Music]
16. 5:13uh what are we G to talk about well so I want to get back to deep Dives but it's also Monday Monday is kind of a I don't know it's they they
17. 5:30have historically been more chatty philosophical Meandering Mondays we like to call them welcome to chat add my name
18. 5:40is Kyle Shannon this is the AI learning lab uh the agendas here are loose if [Music]
19. 5:55non-existent I had a couple of good conversations today though about this AI stuff and I keep I keep going back
20. 6:10to I just have this I have this
21. 6:26nagging this nagging gravity pulling me back toward talk about the future of work and it's it's annoying
22. 6:42because I don't have a real clue like none of us do but I think I know what's going to be important and what's what's going to be
23. 6:55more important what's going to be less important and I think some people are going to thrive and other people are going to struggle like
24. 7:06hell so I just thought it might be good to just kind of go back through some posts that I've done and just talk about what I think's
25. 7:15coming and why and but I don't know if you have questions about AI you can pop them in the comments below mandering is also a nice choice
26. 7:25very nice happy Monday Valerie Jeff Flanigan what's happening seems excited about the topic future of work and by
27. 7:34work you know part of it is work what we do in exchange for money which money is something that we exchange our
28. 7:45time for right the uh the definition of money that I learned from this book called your money or your life by Joe
29. 7:56Dominguez he basically said time is the one thing you can't get [Music]
30. 8:14back and figuring out that pattern that strumming pattern is something I can't get time back on um so if money is something that we
31. 8:30exchange our time for and in today's world we have to do this thing called work to get that [Music] money what what do we do with our
32. 8:45time when the AI does most if not all of the work and there's a part of me that's like but Kyle that's like
33. 9:00years away and I'm like and yes it is years away but it's more like three to five years away rather than 10 to 50 years
34. 9:13away which is I think where where most of us historically had been if if you're of an age like say uh you know Gen X Boomer Elder
35. 9:28millennial [Music] we were fairly confident we didn't really have to deal with this [Music]
36. 9:51[ __ ] in a Westerly [Music] Direction this car is my train all right Vicki what's the next
37. 10:08line I've been driving I've been [Music] wondering what it is I'm running from again feel like an 8-year-old man holding on to
38. 10:2629 uphead on that Horizon is California L [Music]
39. 10:41yeah um so what the [ __ ] are we supposed to do if you're watching this Channel first of all you're
40. 10:57weird and sure we've reframed weird as irregular so you can call it irregular but it's weird you're you're
41. 11:07early no I know it's late now but you're early in the conversation with this AI I was in the salon posting recipes very
42. 11:16nice very nice love it Jeff Flanigan put mistol on his laptop we'll talk a little bit about mistol tonight although I
43. 11:23haven't played with it yet I was big in mistol was the big thing like a year ago then it just disappeared and now it's back with a
44. 11:35vengeance um like all the like all the open- source models they're a bit of a pain in the ass because you have to install them
45. 11:43and figure [ __ ] out and sort through them yourself which you know engineering types like that they like geeking out on that
46. 11:57stuff there's so much to keep up with though even with the non-op source stuff that I have not prioritized
47. 12:05that but it does seem like nistal is a pretty good interesting thing to play with irregularly weird or weirdly
48. 12:13irregular I have found my people exactly exactly we're all a bunch of weirdos cheers to being weird or irregular yeah
49. 12:29[Music] um I thought the the Super Bowl was fascinating last night um not the game not the the game was [ __ ]
50. 12:45horrible the game was like what it was it was like an NF an NFL team was playing like a peeeee football league
51. 12:57last night it was bizarre I've never seen any like it it was just an absolute meltdown on Kansas City's side um but
52. 13:06the halftime show I knew a little about the Drake Kendrick Lamar controversy um last night and I knew that there was a big thing whether he
53. 13:16was going to sing the song or not and whether he was going to say the phrase or not and he did if you haven't seen
54. 13:22it and but something that gave me great joy today is like my entire timeline on Tik Tock was just filled with people
55. 13:32that know that are deeply in that world like losing their Collective [ __ ] like all sort sort of showing different parts
56. 13:42of that performance and analizing it and breaking it down and it was just I felt like I got such an education today such
57. 13:51a cultural education um absolutely [ __ ] awesome and Kendrick Lamar is [ __ ] beast there is something I have a very
58. 14:05strong place in my heart for True artists and that was a that was a Tor to [ __ ] Force I mean that's a performance that is going
59. 14:18to be talked about for a long time to come and even if you didn't get it go watch Tik Tok go watch some of the
60. 14:25videos of people watching it in real time it was insane go Eagles I wish I knew the backstory before I watched the performance would
61. 14:35have appreciated it so much more yeah exactly I know and I didn't know the songs it's like and but you're watching
62. 14:42people from that world and they're like they know every [ __ ] word they know what songs he's gonna sing and which
63. 14:47ones they want him to sing and the significance if he does sing them it was just like and they were just losing
64. 14:54their Collective [ __ ] like no [Music] um so that was pretty cool and then we had a couple of you
65. 15:11know a couple of large language model companies blew eight8 million on 60 seconds of commercials um Geminis were fairly
66. 15:25straightforward they were kind of like old Apple ads here's what it does here's the difference it's going to make in
67. 15:32your life um anthropics I kind of missed but I I I don't know really what it was but I don't know what the [ __ ] anthropic
68. 15:40is doing I are they going to ever release anything again and then open AI commercial um there's there's a couple
69. 15:53of different ways you can go with Super Bowl commercials you know one is you go with something absolutely Unforgettable big
70. 16:07star something really funny A New Concept then there's talk about the product which given that nobody knows
71. 16:20what the [ __ ] chat GPT is might have been a good opportunity to talk a little bit oh perplexity apparently gave away a
72. 16:31million dollars all right there you go perplexity giving away a million bucks in the app oh during the Super Bowl
73. 16:39there you have it see some some sort of stunt like that would have made sense but what opena I went with was a crazy ass
74. 16:55animation that featured their Dot and it was like when I first saw it I'm like oh that's like the chat GPT dot when you put it into voice
75. 17:05mode but then they made lots of dots and then the dots started forming things and I couldn't quite tell what the [ __ ] the
76. 17:11dots were forming until halfway through the commercial I was like oh it's human achievements okay I'm like but this this
77. 17:21this please tell me this isn't open Ai and it was so they went for a branding they went for a branding play right like the
78. 17:33the most famous branding play in the in the history of Super Bowl ads was the 1984 ad from Apple computer
79. 17:41right directed by Ridley Scott this woman running down the hallway throwing the sledgehammer at the 1984 dystopian
80. 17:50future which we now live in here we are did you see Kendrick's show [Music] um you have no [ __ ] idea what it's
81. 18:04about right and then it just at the end it just says Macintosh you know 1984 whatever um that's the most famous
82. 18:12branding play that was a memorable moment and a branding play all at once so in a branding play you're just trying
83. 18:18to get people to be be aware and remember your [Music] brand um so let me let me change my sharing here for a
84. 18:37[Music] second hold please hold please open
85. 18:52AI supera Bowl all right here we
86. 19:08[Music] go Jeff on YouTube from Yahoo finance Elon Musk yeah I'll get to that that's a big story I'll get to that in a second
87. 19:20share screen share X all right there we go we got our producer and in the house yo Brandon was shaking Big B Big B in the
88. 19:34house um all right so [Music]
89. 19:49let's right right so there's your there's your do right that's feeling very open a i e open AI e EI e IO um I did not need
90. 20:01you to do the math for me which math what math did I do oh Emo's wife did the math oh two commercials about
91. 20:14anti-hate well there there was a lot of hate in the room yeah exactly there there was a there was a lot of a lot of what they they the one
92. 20:26the one commercial I thought had a clever hooked to the to the stadium was the one where they sang the John Denver song but it
93. 20:36was a weird arrangement of the John Denver song and then they cut back to the Super Bowl and everyone's singing
94. 20:41the same song in the Super Bowl that was kind of cool um I don't remember what the product is the problem with Super
95. 20:47Bowl ads is you never know what the [ __ ] product is but anyway let's let's watch this open AI Tik Tock pin
96. 20:521984 plus 40 I know it's crazy right um here's the uh here's the open AI ad Super Bowl
97. 21:08[Applause] [Music] [Music]
98. 21:25ad go back right here this when we're like corn we got horses we got corn we got wheat fields at this point I'm like wait
99. 21:37a minute maybe this is a Monsanto ad I mean I really I'm like trying to think like wait what what is this
100. 21:55[Music] was that not AT&T right there I'm like oh maybe this is an AT&T [Music]
101. 22:18ad live on right here I started to get it so the commercial changes right here so it goes from everything's open and white right here
102. 22:30and then here it flips to a more photo realistic this is like grayscale Photo realistic stuff then there's that weird AT&T
103. 22:37moment but then there's this is now starting to feel like [Music] video we are live on the [Music]
104. 22:56TV at this moment I kind of knew it was open AI right the 1995 modem dial up 1993 modem dialup for the
105. 23:11internet summarize this art help me practice asking for right there you definitely know right now you know ah
106. 23:18it's open AI right but what's what was interesting about that is that's not advanced voice that's old not advanced
107. 23:29voice practiceing for my sleep whatd is what do you want to there's advanced voice so why' they wait I don't know create
108. 23:44next where's the URL now you can go to chat.com chat gp.com so it's fine but why not put the [ __ ] URL
109. 23:55there and no name of the company right so it's just chat GPT so it's not open AI it's a chat GPT ad with the open AI
110. 24:04logo um how much they paying you to air their commercial um yeah no I'm not going to give equal time to
111. 24:12Gemini is payon here I'm not GNA give him the satisfaction um so anyway so that was the that was the open AI um Super Bowl
112. 24:23commercial um so they so glad open AI used their time and money to show showcase the power of chat gbt exactly
113. 24:31like Source Camp you've spent the last two years devising ways to to to show how chapp can make a difference in people's
114. 24:45lives nobody knows what the [ __ ] it is right nobody knows what the [ __ ] it is and and Nate Nate what's his name Nate
115. 24:53silver Nate B silver did a thing on why do a commercial with the dot when the dot is really it represents you waiting for it to
116. 25:03respond so they they're like branding the fact that their product doesn't respond all that quickly in non-native voice
117. 25:14mode oh man all right let's go to some YouTube comments here um Facebook was very divided today over
118. 25:27the halftime show of of course it was it was you know but considering the the [ __ ] political climate we're in of
119. 25:35course it was divided of course it was divided um but I'll tell you what watching the people who were in the note of that it's
120. 25:45like that was pretty it was pretty cool to watch um let's see it's a [Music] shame wait from Yahoo find oh we'll get back to that
121. 26:00what's that say it's a shame to think they could have done better to relate to people who didn't understand AI that was
122. 26:06the opportunity that they had right like like this thing looks like a Google chat box but what is it it's your entry into
123. 26:17the world of knowledge right now you don't have to make it like a 60s um infomercial about how to use the information super
124. 26:26highway but you have to something there so here's the thing that I got once once I realized oh this is an open AI thing I
125. 26:33got it about halfway through and I was like ah all of the things they showed are major milestones in human achievement here's why for
126. 26:46me I loved the chat GPT commercial not because I think it was an effective commercial I think it's an absolute
127. 26:54[ __ ] wasted opportunity it's it's I think it's a waste of $8 million but for me personally a week ago so I've got this
128. 27:04musical Sydney right and we didn't have a song to open act two we we we did our state our our table read with our actors
129. 27:14we played all of our songs and we had a placeholder for act two and so we talked about it and what we came
130. 27:23up with is a concept that right now is called the shoulder of giants and the opening of act two of my musical
131. 27:32is five different scientists from history coming forward to compare their big breakthrough with AI right so you
132. 27:41have Newton and you've got Marie curri and you've got um Johannes Gutenberg the Gutenberg Press and you've got Alan Turing from the Turing
133. 27:52test um where he basically says you've finally done it right you know I came up with the test but you've actually done
134. 27:59it um and then and then you end with Einstein and so I've got like I had this idea of you know major Milestones
135. 28:12throughout history and how I got this idea I went into chat GPT and I said um I think I was using the reasoning engine and I said dig deep
136. 28:23and you know go find the major milestones in human history and then compare whether or not that that Milestone of
137. 28:31human achievement was more or less important than where we are with AI and it went through and 20 out of 20 of them
138. 28:39it said AI was more important and so that was where I got the idea and then so I so I wrote that song and I've got
139. 28:45we've got the song or well we've got the lyrics now for the song and um and then you know Super Bowl ad saying here's all
140. 28:53the human achievement opening eyes better I was like I like that oh how's that people gonna make people want to use chat GPT it's not
141. 29:09um I can't believe they ruined a GNX for that halftime show that's the uh the Oldsmobile muscle car that what that a
142. 29:181984 the GNX it's a uh it's a uh Oldsmobile hang on olds GNX year that CL that that one was the
143. 29:35one 1987 yeah 1987 Grand National experiment limited production exclusively in 1987 McLaren only 547 units were
144. 29:50produced and they turned that one into a clown car oh that's pretty funny okay uh on Tik Tock
145. 30:05um even Kanye's ad had a call to action URL Source Camp um please open AI let me handle your marketing no [ __ ] um
146. 30:18Amelio's wife maybe the same people who name the versions of chat GPT were responsible for the ad listen I
147. 30:31it's at least a choice right they at least made a bold choice to say we're not going to talk about it we're going
148. 30:38to try to do some visual branding here that we want to own but like in the environment we're in does America really give a [ __ ] how
149. 30:54AI Stacks up to the achievements of of history I mean maybe in hindsight we'll look back at that commercial and go they
150. 31:02really put a flag in the ground just like those astronauts did on the moon H but let's forget that moment it's really
151. 31:08about AI now like I don't know I don't know I mean it's what I did in my musical like I said all right um it
152. 31:18would be a good introduction for a show movie documentary but not very helpful for a customer base oh it was Buick
153. 31:24that's right Buick Buick GNX yeah the Buick GNX why did I think that was an old because it looks like an
154. 31:35old all right they've never made sense on point all right so anyway so there was that um but you know listen if
155. 31:44you've got perplexity giving away a million dollars anthropic did an ad Gemini did two ads open AI did an ad you had five Super Bowl
156. 31:54commercials talking about plus you had some qu you know related things like the the T-Mobile starlink thing was a pretty
157. 32:02big announcement like the the the starlink T-Mobile thing I thought was really interesting because it was like
158. 32:10hey there's this big [ __ ] crazy ass technology out there and you can now get it on your phone and you can get cell
159. 32:16signal anywhere like that was kind of cool like like it that feels like something that that open AI should have
160. 32:25done all right one more on Spotlight let's see that ad costs more than deep seek yeah I don't buy that deep seek was only
161. 32:38six million bucks it might have been I don't buy it though all right so let's talk about this one Elon Musk oh Elon you
162. 32:47know they they they got oh in case you were wondering thank goodness turns out they got big balls uh security clearance
163. 32:56for the NSA he's he's a 19-year-old hacker named big balls and was just given named senior advisor to to the NSA
164. 33:05so he's gonna have access to all of our data so that's good good news there um so so while doge is is is in cleaning house in the government
165. 33:17uh elon's apparently still running his other companies and picking a fight with Sam Alman um might be time for black bar
166. 33:26oh I got I got br in a smart ass mood tonight I like it he's pey um um he's the VP um
167. 33:42so Sam mman has been um working on taking open AI from a nonprofit weird
168. 33:56ass hybrid sort of thing that it is now where this nonprofit board runs open a you know makes the major decisions for open AI
169. 34:05there's this quasi capped profit entity that it is now he wants to flip that all to for-profit and buy out the
170. 34:13nonprofit and apparently he offered them $40 million and so what Elon Musk did today is he made an unsolicited offer to buy
171. 34:24the assets of of the nonprofit of open AI for 97.4 billion which I suppose he just has laying around
172. 34:38um which I assume he's not actually interested in buying open AI I think what he's interested in doing is [ __ ] with Sam
173. 34:47Altman because his $40 million buyout of the nonprofit all of a sudden looks kind of poultry when there's a 97.4
174. 35:01billion offer on the table right more than two times what he's offering them uh Sam Alman immediately
175. 35:09replied um no thanks but I would like to buy Twitter for 9.7 billion which is a pretty good dig considering that Elon
176. 35:19paid 44 billion for Twitter and it's now worth about 11 so so Sam just said no thanks on your 9 7 billion doll offer um
177. 35:31uh but I'll offer you 9.7 for Twitter to which Elon Musk replied swindler and now they're having a [ __ ] slapfest they're
178. 35:41they're having a Kendrick Lamar Drake battle on Twitter right now so if you want to go be entertained how are gpts different than just
179. 35:55utilizing chat GPT good question here I'll show you I can show you that let me change my sharing that's a good that's a really
180. 36:05good question to just start and show some show some goodness okay so there we are at the chat JP um let's say um this is my my quote
181. 36:24my search for the GNX the GNX by the way at at at the opening of Kendrick Lamar's set last night he's standing on the hood
182. 36:33of a car that was the Buick GNX that was this GNX um I know black bar I know I know people I get it I understand your pain it's just that I
183. 36:47don't I didn't even say anything I know I know oh Auto star super chats that's super cool so by the way if you're on YouTube you
184. 37:02can now do a super chat which I don't they cost money I don't know how much they are but if you do a super chat it
185. 37:08will automatically star it so when when I'm looking at my back stage area um there's like a long stream of chats and
186. 37:18then Super chats get separated out so I can go comment on those so there you go and then producer Brandon will yell at
187. 37:25me to deal with the super chats exactly um here's a cool thing that you didn't know about chat GPT I can just say show me pictures of this
188. 37:40car and it'll search the web and it'll go get pictures and boom there's the GNX so this is a an incredibly rare
189. 37:51car that was turned into a clown car last night if she didn't see it like like what 40 people climbed out of this car they had
190. 38:03a special Contraption on the stage like half half climbed out of the trunk and half climbed climbed out of the car
191. 38:09itself it's pretty good but you know that means they chopped they chopped the [ __ ] out of that [Music]
192. 38:21car okay question how are custom gpts different than chat PT good [Music]
193. 38:36question um so if I just go here to chat GPT um you'll notice over here on the left there's a follow on
194. 38:52question another newbie question from spot it fast when you do prompts is that something in the settings you change and set up with each
195. 39:09time or do you give prompts while chatting so Spot It Fast these yes they're newbie questions and people have been using chat GPT for a
196. 39:25long time probably don't know a lot of people that use it regularly probably don't actually know the answers to those
197. 39:32so they're actually really good questions okay so so there's a there's a a couple of things that we'll unpack here so just
198. 39:41in general with large language models whether it's Claude or chbt or gemini or perplexity they're very similar in in
199. 39:52user interface to a Google Search right now what Google has taught us for 20 years is you don't put very much in that
200. 40:02chat box right you you you try to put in the minimal amount possible and let Google figure everything out and it'll
201. 40:08go get you what you need if you put in too much in a Google search it would kind of [ __ ] it up right it would it
202. 40:13would get confused and you'd get stuff in there you didn't want so you just put in your seven or eight keywords that
203. 40:19were the things that were the closest to what you want you know to use chat GPT is almost the opposite
204. 40:27if you give chat GPT a thin prompt you will get a thin answer so giving chat GPT more context is actually really important and context
205. 40:40is basically just to say I could put in there um I want you to write a scene you know a scene for a musical and it well in
206. 40:51fact let's just uh write a scene uh for a musical and it'll do it here's a scene for a Musical musical scene let's see how it
207. 41:08does oh see yeah this is not good this so it actually did better than I wanted it to do because chat gpd's got memory
208. 41:17I'll talk about that in a second you can you can put in a thin prompt and you'll get back a thin answer
209. 41:24or a generic answer so one of the things that turns people off about chat GPT is that they're like well I tried it and it
210. 41:32gave me a shitty answer if you want a really good answer you say hey I'm thinking about um I'm thinking about
211. 41:41writing a musical and here's what I want it to be about and here are my favorite musicals of the past and here's the kind
212. 41:48of dialogue I like so you give it lots of context you give it lots of color and then it will give you a better answer an
213. 41:56increasingly better answer answer and then you take it one step further and rather than think about this box this the as we call it the chat
214. 42:08hole rather than thinking of the chat hole as this as this one-off interaction think of the chat hole as a
215. 42:18conversation and chat jpt starts to get way more interesting and so here's what I mean by that let's say I I I uh let's
216. 42:27say I'm gonna I'm GNA turn one thing off right now I'm going to turn on at the very okay oh my God there's so much
217. 42:36there's so much in chat GPT right now there's so many different features it's hard to talk about without running down
218. 42:43rabbit Hool so I'm G try to keep it focused um you choose your different models here on the left um at the top
219. 42:53you you there's this little pull down menu and you can choose different models don't worry about most of the models
220. 42:58right now just don't just stick with chat GPT 40 it's more than most of us will ever need In Our Lifetime just
221. 43:06stick with it for now um you can you can dig into the other stuff later but at the very bottom here you can go
222. 43:11temporary chat and what a temporary chat will do is it'll sort of hermetically seal this chat from other chats and it
223. 43:17won't be influenced but anyway so let me give you an example of what I mean by have a conversation um I'm thinking of starting um starting a
224. 43:29business and want help with naming and marketing and then it's going to come back so so chat
225. 43:44GPT has been written in such a way the it's a piece of software where they they created it in such a way that
226. 43:52it wants to be really helpful to you so almost always it will give you a helpful answer but if I just start out with that
227. 44:00really generic prompt it's going to be eh whatever so what this did which is really nice as it said here's some
228. 44:06questions to guide naming naming and marketing what type of business are you starting who's your target audience um
229. 44:13what makes your business unique so here's here's all these things so what I'm going to do as an example just as an
230. 44:19example here rather than me actually answering those because that would take a while I'm just going to say um please give me um hypothetical
231. 44:32answers to those questions for a business about
232. 44:46um a hot air balloon drone delivery business okay whatever so you have your business idea and then it's going to give you
233. 45:01some questions answer and so now it's going to fill in those questions so type of business a futuristic echo-friendly aerial delivery
234. 45:13service that uses AI powered hot air balloon balloons and drones and I'm G to well that's fine um rural communities
235. 45:21event organizers luxury Brands unique selling proposition gives me all that brand Vibe futuristic yet yet Whimsical
236. 45:28naming ideas so Derek gave us a bunch bunch of names um there's marketing strategy strategic strategy so so now
237. 45:37we've got a conversation going I'm going to say um pick the top three names based on um new
238. 45:53business marketing principles principal principles plls new business marketing principles and why you selected
239. 46:09them so now it's going to take all of what it wrote there and it's going to give us three names Nimbus drop which I hate Sky
240. 46:20crate we could go with sky crane Zephyr lift let's go with Sky crate all right I like Sky crate give me
241. 46:3620 tagline ideas so the way chat GPT works is it it's got a it's got a very small
242. 46:51brain right now so it'll remember this conver conversation for some amount of time and at some point it'll kind of forget the
243. 46:59things at the at the early part of the conversation but for the most part we're now having a conversation where it's
244. 47:05remembering all of what I've put in there and so it will get better and better at focusing in on this idea so
245. 47:11this is just generally how you use chat GPT is think of it more like a conversation say with a clueless really
246. 47:19well- educated intern then I put in a prompt and I get out an answer that's that's the the biggest unlock for chat
247. 47:27GPT is flip from I give it a single prompt to I give it a conversation um all right one of the things I did was ask chat jpt to create
248. 47:38a crochet okay that's very cool um um so there's all that stuff again so um pick the top three and tell me why
249. 48:01your package delivered by the sky that's horrible lifting packages lowering carbon Footprints eh the future of
250. 48:09logistics is up in the air number three three wins okay now three wins now write a creative
251. 48:24brief for a brand mood board so I can start to shift gears in this chat to narrow it down to specific
252. 48:41things and it's it knows what creative briefs are I don't know why it's going fullon uh Emoji but I think because I'm in a
253. 48:52temporary chat it's not using any of my uh things with I'll show you in a second uh and now I'm going to say okay make me a 16 by9 mood
254. 49:09board with a bunch of different logo options with name and tag line lockups and I don't know if you knew
255. 49:23this but chat GPT can make pitches and so we've got now got this thing called not sky crane Sky crate where it says something about
256. 49:36delivery is up in the air and look boom so we went from nothing and getting really thin shitty responses to having something
257. 49:48that you know lifting packages lowering packages something about carbon Footprints um that was not the the
258. 49:57tagline I wanted but anyway whatever it's not good at letters but but it at least gives us an idea and you know from
259. 50:03here I could go you know like I could say um I hate the colors and styles make it more graphical and
260. 50:18abstract and vibrant uh uh uh YouTube Spotlight YouTube Spotlight okay wait so when people
261. 50:33utilize it with wanting wait with wanting to be let's just say advisor you just prompt it in the chat settings and
262. 50:41usil utilize it more personally yeah so giving chat GPT a role actually does work so if you say if you want it to
263. 50:52write jokes you could say you know act like a stand up comedian that you know writes writes jokes like SNL skits or
264. 51:02whatever it might be it will it will actually make it right more in that style so yeah giving it a role can help
265. 51:09Okay so now all of what I just told you and giving it a role is a good thing all of that is just in the context of just
266. 51:18using chat GPT in kind of a one-off kind of way now there's there's three different things that can modify this experience pretty
267. 51:30dramatically um let me turn off temporary chat let me gr it new chat here and first place we're going to go is we're going to go to
268. 51:39settings and in settings you go to personalization and there are two things in in settings in in personalization
269. 51:51that are important one is memory and you can turn memory on or off and then you can man memory so pan wait pan Tik Tock camera
270. 52:01left and black bar please okay oh I see got it yeah you want to see me you don't want to not see me that would be that would be
271. 52:13horrible I'm all washed out in that other little picture Okay memory is something that's actually really interesting in chat GPT and all
272. 52:26large language models probably by the end of this year it's going to be incredibly powerful because I think most
273. 52:34of these things will be able to remember all the conversations you've ever had with it and talk about them all
274. 52:42right now it's very limited so if you say ma manage memory if you turn it on it will start to remember things you can
275. 52:48also explicitly tell it remember that I like writing paragraphs with no capital letters and minimal punctuation and it
276. 52:56will start writing like that um it will remember that and and so in here these are all things that it's just chosen as
277. 53:05I've been working with it it's just chosen to remember um so let's see what it says creating an app that allows users to
278. 53:13search for real images and use them as a basis for AI generated images Kyle Shannon is is assigning admin roles and
279. 53:20Google workspace some of the things it chooses to remember are bad um if you don't have ADD and you want to go in and
280. 53:30regularly prune your memory it's probably smart your chat GPT will suck less I have ADD and I the thought of
281. 53:37pruning my memories is I don't care I'd rather have it just be bad um Becky run on Tik Tok can you show us how to save your chats Andor
282. 53:49gpts into a PDF folder on your laptop yes sort of maybe well I I don't qu I don't quite
283. 54:04know what you mean but I assume I assume you mean save a chat as a file so so one thing you can do is you can save you can
284. 54:12share chats with other people but we we'll get to that in a second okay so that's memory so you can turn memory on
285. 54:18or off me that memory starts to make your chat GPT more personalized yeah exactly okay now custom instructions
286. 54:31they completely redesign custom instructions I actually haven't talked about this in here um oh they've got an advanced tab now o
287. 54:40fancy custom instructions are actually quite powerful and you ask me this this whole thing started where you're like what's
288. 54:48the difference between a custom GPT and and just chat GPT and basically what it comes down to is a custom GPT is pre-
289. 54:57prompted like you've you've written a prompt and you've you've pre- prompted it and I'll show you what that means in a second that's sort of
290. 55:06what custom instructions does so what should chat GPT call you it should call me Kyle um what do you do I'm the
291. 55:14founder and CEO of Story vine I'm a community builder at AI learning lab on Tik Tok and this the AI Salon a
292. 55:20community of AI optimists right and I could put more in there if I wanted and then what traits do you want chat GPT to
293. 55:28have and so for every response this was just a cool thing that I that I did uh I saw someone do this once and I kind of
294. 55:36liked it so I have for every response give it a sequential response count followed by a response title that
295. 55:43summarizes the response and include a date so let me show you how my chat GPT works that might be different than yours
296. 55:51I could say um what was the Buick um what was it ntx n NSX ntx is it
297. 56:07ntx yeah oh it didn't do my it didn't do my custom instructions are they not turned on they might not be turned on um
298. 56:19personalization custom instructions on huh it ignored these so I don't know why it's not doing that um the other
299. 56:28thing I have it say here is avoid overused words delve tapestry vibrant landscape realm Embark so have you ever noticed
300. 56:40where chat GPT is like well we're going to delve into the tapestry of the vibrant landscape and the realm of large
301. 56:47language models and you're like stop talking like a British documentary stop that you can just tell it to stop that [ __ ]
302. 56:57um enable for new chats I don't know why it's not behaving and doing that GNX oh why did it let's see oh ntx oh that's 49
303. 57:08oh that's funny Buick GNX oh Grand National Grand National experimental clone or Resto mod oh here's how to build a Resto mod
304. 57:24that's kind of cool and you know hey maybe maybe Kendrick Lamar didn't actually destroy a GNX maybe he did a a Resto mod shell
305. 57:40um yeah this is really weird that it's doing all these uh all these emojis that feels that feels very not good um I
306. 57:51don't know why it's doing that maybe it got an upgrade uh oh that's the thing chat GPT will just act weird Okay
307. 58:00so chat GPT basic just start talking to it give it a roll have long Contex have long conversations give it context
308. 58:10another way you can give chat GPT chats context is you can upload a file so you could upload let's say you've got a
309. 58:18brand guideline for your brand voice you can upload that uh into a chat and say follow this okay and it'll it'll it'll
310. 58:28it'll start to talk like your brand voice now over here on the left you'll see Vicki and I noticed that too I asked
311. 58:34it to talk more direct it reverted immediately yeah I I so so one of the things about chat GPT is it is literally a black box it is
312. 58:46literally like this and they open up the box and they put [ __ ] in it and then they they they create an application where's my ink bottles
313. 58:59here this is this is a very famous box on this channel this is the chat GPT app so this is is the model the large
314. 59:08language model called GPT 40 this is chat GPT so this is an application layer that tells so it when you type in your
315. 59:19words it goes into here tells it what to grab and then it spits that out on your screen but they change this all the time
316. 59:28that makes what it spits out on your screen different so that's what we're experiencing there so just with all of these tools you need to
317. 59:37accept that they're going to suck okay um couple of pins from Ann okay got it beautiful beautiful beautiful
318. 59:50fantastic fantastic an we'll come talk to an in a second let me let me say one of thing so what we've talked about basic chat GPT have a
319. 1:00:00conversation understand that you can put in custom instructions and you can turn on memory all of that will start to
320. 1:00:07customize chat GPT now what is a custom GPT a custom GPT I'll start with the most obvious one my very first custom
321. 1:00:17GPT that I wrote on this here channel is a custom GPT called Sin City sipper and it is a um Vegas bartender with a Twist on the
322. 1:00:29classics and so I can just say where are we uh let's see Valentine's Day is coming up give me
323. 1:00:43a mocktail um with a citrus spin for the lovely
324. 1:01:00y oh yeah this is definitely adding adding more emojis so the red velvet kiss mocktail a luscious creamy seductive non-alcoholic treat
325. 1:01:11reminiscent of a red velvet cake and then it's putting in instructions here um or you know here's your ingredients
326. 1:01:18what's in this cranberry juice coconut milk interesting vanilla syrup fresh lime juice club soap soda two to three fresh raspberries for
327. 1:01:27garnish shaved dark chocolate sounds delicious actually and then what this does let's I wasn't able to make a uh I'll try
328. 1:01:39again it said it's it's safety guide lines wouldn't let it let it make a red velvet kiss let's see if we can get one
329. 1:01:46this time try again um change your prompt so it doesn't violate your stupid
330. 1:02:02guidelines the these things are annoying because they have safety guidelines in them and sometimes you want to do
331. 1:02:09perfectly reasonable stuff and it won't let you whatever it uh let's see rename the drink so you can make the image
332. 1:02:30so this this thing Sin City sipper has been so there it finally made one has been pre-prom to automatically
333. 1:02:40make an image to know how to write you know cocktail recipes and come up with them and to give them a Vegas twist so
334. 1:02:49how does it know that so if I go in here under Sin City sipper because I created it I can go edit it so I can show you
335. 1:02:56underneath the hood so if I click on the configure tab here you know there's the little icon for it it's called Sin City
336. 1:03:03sipper here's the description and then here's the prompt this is what makes this thing do its thing as the Sin City
337. 1:03:11sipper Your Role will be to to be a virtual mixologist with a flare for classic cocktails with a Vegas twist
338. 1:03:19engage users with your encyclopedic drink knowledge right so here's giving it a role giving it context telling it's
339. 1:03:26it's got encyclopedic drink knowledge now you're like well how does it know all the drinks well think about it every
340. 1:03:34cocktail book from like you know the Shakespeare's days you know to to modern times are out on the internet somewhere
341. 1:03:43right there's articles every article that's ever had a drink in it is out there so the internet knows how to make
342. 1:03:50all the drinks so it you can just assume the internet knows [ __ ] like that right um You give it sharp humor and a
343. 1:03:57personalized touch deter determine their Mixology skill level right important each recipe you create uh should be put
344. 1:04:05in a new canvas it doesn't follow that for some reason you have an additional Behavior whenever you provide a tailored
345. 1:04:11recipe you will automatically generate an image of the drink the image should represent the final drink next to the
346. 1:04:18glass is a coaster with the name of the drink and I put name in in uh uh brackets and large letters that say by
347. 1:04:27Sin City sipper do not put the letters on the glass it still puts the letters on the glass um now you might be looking at these
348. 1:04:36instructions and you're like well couldn't you just write all those instructions in chat GPT you could but if you've got some kind of
349. 1:04:46special knowledge like one of the other things you can do is you can upload files oh Dolly image generation was
350. 1:04:55turned off they have they have they have [ __ ] some things up in chat GPT let me just tell you that right now it's not
351. 1:05:02making canvases because that was turned off huh wild wild wild wild some something's up but anyway that's neither here nor there
352. 1:05:15you could upload files so let's say let's say you have a particular I don't know Consulting methodology you could
353. 1:05:22create a custom GPT where you upload your PD F of your custom Consulting methodology and um and and your chat GP
354. 1:05:33your custom GPT would start behaving you know using that knowledge so it's it's really quite a powerful thing I am a
355. 1:05:42little shook shocked that they didn't retain my settings and I I'm that's making me think that um all of my gpts
356. 1:05:53are now need to be fixed spent all their money on a Super Bowl ad and yeah got rid of the the GPT the custom GPT
357. 1:06:01product person um view the custom GPT all right let's let's try this again let's say make me a Super Bowl um whiskey
358. 1:06:17cocktail uh for the casy Chiefs to drown their [Laughter]
359. 1:06:31Sorrows oh man all right an Murphy hey everybody I have a super cool use case I have a super cool use
360. 1:06:48case helpful for for all kids with the flu let's see my dear son has the same flu everyone has miss school textbook is
361. 1:07:05subpar total racket took a pick of the textbook pages uploaded to chat GPT gave it a roll asked for a slightly but
362. 1:07:14better textbook lesson that's a really good use case yeah you can you can you can take a picture of anything and let me I I got to get my my
363. 1:07:26um my iPad I I'll show it to you when you turn when you have advanced voice you can turn on the camera and just aim
364. 1:07:34at it [ __ ] and then have a conversation with it and it'll just talk let's see may this drink go down um let's see make the image or the
365. 1:07:53ma oh this this actually in a canvas ah okay okay so here's a whole another thing so now that I have canvas turned
366. 1:08:04on and I have images turned on it it actually worked so the Kansas City chief sorrow drowner 2 ounces of bourbon 3/4
367. 1:08:13of an ounce of lemon juice 2 ounces of honey syrup quar ounce of compari for for that red bitter defeat
368. 1:08:28that's awesome um bitters tears in liquid form club soda Delight in the mood if possible and look it said sari instead of
369. 1:08:40Kari um okay so put it in a CA in a in a canvas and let's see let's that's allows me to [Music] share or can I I don't [ __ ]
370. 1:09:00know all right um I hope that helps because because now I want to get to the question of how do you how do you start to save [ __ ] in
371. 1:09:13chat GPT out into the world right the like save it as a PDF or save it as a doc I need to figure out how to turn off canvas yeah well I yeah
372. 1:09:24exactly well it shouldn't just automatically throw [ __ ] in canvas Source camp but you can turn it off at
373. 1:09:31the custom GPT level but that's weird that's weird that it's doing that um okay let's create a new chat I have crowdsource doc of all the
374. 1:09:42overused words let me know if you want yeah an would you um would you what's the easiest way to share that if it's like a Google doc and
375. 1:09:52it's publicly accessible maybe you could just share that over on the Irregulars Channel and I'll tell people how to get
376. 1:09:59there that would be cool um okay so Irregulars Channel what the hell am I talking about so if you don't know that
377. 1:10:08company or that company that Community the AI Salon you should know about it go to the salon. oh source Camp says chat GPT does
378. 1:10:22just automatically throw [ __ ] in canvas yeah chat G PT is acting weird just they're doing so I I I promise you
379. 1:10:30what's happening right now is there's some update coming in the next three weeks when chat GPT starts to get weird
380. 1:10:37they're working on something if it gets slow and breaks that means they've got servers down this weirdness is I have a
381. 1:10:44feeling they're experimenting with some [ __ ] so and whatever they're experimenting with I don't like it
382. 1:10:52um for Jason it's working fine Lori can totally agree it's a sign that something's happening I wonder if AGI is
383. 1:10:59closer than we think if AGI includes that many emojis I'm going to stick with my old shitty brain okay go to the salon. AI click on
384. 1:11:12join our community and that's going to take you over here to the AI salon and it'll land you on a welcome page there's
385. 1:11:19a nice welcome video from myself and Leah faston who I created this with there's a little uh uh thing about the
386. 1:11:26journey to AI Readiness there's even the the five stages of AI adoption which I want to talk about tonight um and then
387. 1:11:33over here on the lefthand side there's all sorts of different things there's kind of a seven-step welcome uh sequence
388. 1:11:39that you should go through um there's Community Corner News there's a showoff area with with tools all the different
389. 1:11:45tools that are listed inside these different areas and then there's these uh clubs and hubs and the second one
390. 1:11:53down is called Irregulars and that's ours and so um this live session we have our own little Club within the AI salon
391. 1:12:04so an if you wouldn't mind go upload your cool um set of words that chat GPT uses too much to to hear and then people can add that to
392. 1:12:18their custom instructions that would be super cool thank you thank you for doing that that's see that's that's one of the
393. 1:12:26one of the amazing things about the AI salon and she leads AI which by the way Ann Murphy um runs another Community
394. 1:12:35called she leads AI for women in AI so if you're tired of the sausage fest that is the technology World um and you want
395. 1:12:44to find some badass women that are doing amazing things with generative AI um you should check out um she leads AI but the
396. 1:12:52salon itself to be clear is is not a sausage the salon itself actually has a decent balance of men and women in it
397. 1:13:00and they're all like nice helpful supportive people so the the culture and the values of the salon is really nice
398. 1:13:08so you should go check that out okay speaking of which okay you should also go check out that URL are you
399. 1:13:22readyfor ai.com so Ann Murphy and I about a month ago did this thing called AI festiv us a AI for the rest of us um where it was a
400. 1:13:34two-day over over the holiday weekend between you know Christmas time and New Year's on that weekend we did this
401. 1:13:42two-day event where we had 34 speakers we had 24 different sessions it was absolutely remarkable we had I don't
402. 1:13:49know 2500 people show up 3,500 people register just Bonkers um there was a request to take the lessons that that
403. 1:13:59were kind of collectively shared from those two days and turn that into a training and what we created is is what
404. 1:14:07we're calling are are you readyfor ai.com it's an AI Readiness training course that's going to be launching um
405. 1:14:15mid-February uh we sh we were shooting for February 1 we missed it we got busy stuff like that yes and what um Brandon
406. 1:14:25has up right now is the website so you can go check out the website and pre-register um for when this thing
407. 1:14:31launches you can learn all about it and and you know figure out if it's something you want to do I'm really
408. 1:14:37excited about it um Vicky Baptist is helping design the curriculum um and she's done a kick-ass job it is now on
409. 1:14:45an and myself to do the setup videos for all that stuff I think we are now the bottleneck but I'm really excited about
410. 1:14:54that that so go check that out um all right so let me let me take a breath for a minute I I feel like I I just emerged
411. 1:15:04out of you you asked a very simple question which you know you you um you called a noob question but I think but it was a more
412. 1:15:16advanced question as you saw there there's it's actually a really big answer to how the [ __ ] do you use chat
413. 1:15:23GPT you didn't say that that's what I said uh so yeah so thank you Spott it for the question Spot It Fast good name
414. 1:15:37by the way Becky Ru export to PDFs was on Deck oh yeah so okay um so in general the way I do stuff with chat
415. 1:15:53GPT is probably not the best way to do it let me go find [Music]
416. 1:16:15some oh here's a stupid one n it's too stupid
417. 1:16:39okay if you're in chat GPT for for every little chunk for every little response that chat GPT gives you at the bottom of
418. 1:16:52the um at the bottom of the response you can have it read it aloud which I think is kind of cute you can
419. 1:17:02copy it you can give it thumbs up or thumbs down black bar sorry and then you can um and then you can edit it in Canvas OR
420. 1:17:14oh you can or you can switch the model that's a that's actually really interesting I don't quite know how or
421. 1:17:20why why you would do that but whatever it's there so so a way you can do this is you can just say copy and you'll
422. 1:17:27notice in this response there's some things that are bolded there's some things that are italicized so if I copy
423. 1:17:34this and I just go and create a new um docs.google.com come on and just say
424. 1:17:50paste you know there's there's the output right so so I can now save this I can now download this as a PDF and then
425. 1:17:58you just you would just save this just like you save any other Google doc you could do a similar paste into
426. 1:18:04word there's in chat GPT now there's this new thing called a canvas which is actually kind of slick so historically what you would have to
427. 1:18:15do is chat GPT would vomit out some answers and what I would do is copy and paste different versions of what it
428. 1:18:23vomited out and I'd put them over into a word Dock and then I would end up with this Frankenstein document that was
429. 1:18:31actually a real [ __ ] nightmare right so what you can do in chat GPT now is you can you know in these little things you can say edit in
430. 1:18:43canvas and you'll notice now I've got two panes the one on the left is um this is my chat window and the one on the right this is the
431. 1:18:58um the canvas and the canvas is is now editable so I could say um let me not highlight anything I'm just going to go
432. 1:19:08over here on the left hand side and say please give me a title for this
433. 1:19:23document add it to the canvas I thought that was obvious but apparently not and so now it's going to
434. 1:19:31go over and look now it added the AI and the Muse crafting Sydney and artificial love story and then it'll edit this so
435. 1:19:39so this thing on the right now is now an editable document right it's going through it added that title in for
436. 1:19:47us and it's written all this stuff it's very very slow for some reason um so now I can grab any paragraph in here
437. 1:19:58and I could say um turn this into um a setup sentence with bullet
438. 1:20:13points and then over on the left it's going to do its thinking and then over on the right it's now going to edit that
439. 1:20:19part of the document slick there you have it boom um yeah oh that's cool okay you can also if you have
440. 1:20:33not the only way you figure this [ __ ] out is to use these tools and smash your head against a [ __ ] wall if you have not created an image
441. 1:20:45you can share this chat for some reason you cannot share a chat GPT chat with someone else if you've created an image
442. 1:20:54in it I think it's the dumbest security Choice I've ever heard I'm sure there's a plausible reason but if you haven't
443. 1:21:03made images and if I think if you haven't used a reasoning engine I don't [ __ ] know you can share it so I can
444. 1:21:08click share and rather than saving this as a PDF I can actually just share you know this editable document with someone
445. 1:21:16they can come in and look at it I can also again copy this and flip over to my um word doc and right now we've got our our
446. 1:21:28bulleted list in here right so you can just copy it and put it there makees sense makes sense rightclick print
447. 1:21:47save save to PDF is also an option on a PC Maybe [ __ ] PC people on a Mac it's not let's see
448. 1:22:05print open in reading mode Services no I got nothing rightclick print click print
449. 1:22:21yeah so I could go rightclick print and and then here I could do save as PDF yeah see now we're getting into how
450. 1:22:29to use your computer territory I all right I got I got to jump up because this is a very a question that has a
451. 1:22:36very simple answer if you're looking for it so if you right click you one click away if you right click on right click on it and say print
452. 1:22:44okay and then change your printer instead of the going to the Canon no no no right click on print but but you've got to have it selected
453. 1:22:55right now oh okay that change change to save as PDF and then you can just save your entire export as PDF but that's to be
454. 1:23:03clear that is not a that's not a chat GPT feature that is that's part of the in this case OSX operating system or
455. 1:23:12Windows right sure and I I suppose you could also ask chat GPT if it could export the conversation as a PDF oh yeah
456. 1:23:20we can probably do that let's try that um turn this canvas into a down loadable see Becky you thought this was a simple question
457. 1:23:36downloadable um PDF it might it might yeah analyzing so what it's doing right now when it says analyzing it's this thing called code
458. 1:23:46interpreter where it's writing and executing its own python code so it's it's calling a library a PDF Library in
459. 1:23:55Python and writing this this thing as as a a document and then saving it out as a as a python file so if if you see it say this thing
460. 1:24:06analyzing um it errored out the first time you can click that little down arrow and it'll show you what it's
461. 1:24:13actually doing it's kind of slick and for whatever reason it's erroring out but that's another way to do it also
462. 1:24:20pate's here if you wanted to talk about the Google stuff hey Pate what's happening um all right fantastic all
463. 1:24:31right let's get off let's get off uh chat chat je um how to if you're new here and if you're new to this large language model stuff and
464. 1:24:45if the last what's that been 45 minutes of me answering a very simple question of what's the difference between a
465. 1:24:53custom GPT and just normal chat GPT um to a to a tool all of these large language models are getting increasingly sophisticated
466. 1:25:07they all have all sorts of features some of which are super amazing some of which are like wait why is that there it it's
467. 1:25:14it's like all software these things are starting to get bloated with features that it's not clear how you use them why
468. 1:25:23you use them things like that in fact the fact that we've got all these different models right chat gbt 40 chbt 40 with
469. 1:25:32scheduled tasks which is actually really [ __ ] cool but why do I need to know the difference between those two 01 uses
470. 1:25:39Advanced reasoning 03 mini 03 mini High 01 Pro oh wait but there's more GPT 40 mini gp4 why do I as an end user of a consumer level product
471. 1:25:54need to even know that this [ __ ] is available there they've got generative AI can't they create a layer on top of this that
472. 1:26:10I type in a prompt and it figures out which of these [ __ ] things to use the answer is yes they could but they
473. 1:26:21haven't why because I don't work at open AI see I want to work at open AI or I want to work somewhere I want to work at
474. 1:26:29Google or open AI or somewhere and I just want to be the guy in the corner office I want a corner office it doesn't
475. 1:26:37have to have Windows but I just want it in the corner I want a corner office and then I want to be the guy where they come in
476. 1:26:44and they're like hey we're thinking of like giving everyone this tool and I'll go be like what the why why are all these in
477. 1:26:53here oh because they all have different slightly different functionality well who cares what why would an end user
478. 1:27:02care just go no no I would be the guy that all I do all day is is just go no that no that's not okay we're thinking of naming this one
479. 1:27:1840 where the O stands for Omni and this one 01 where the O stands for open AI wait what you and you're going to tell that to human beings
480. 1:27:30no that's what I want my job to be D dangerously close to a meltdown Monday um so if you're wondering shouldn't these things suck
481. 1:27:41less even though they're remarkable yes they should and then if your next question is why don't they because they haven't hired me yet
482. 1:27:58archetypal architect you should be able to right click in the canvas bar in the chat hole and save his PDF yeah I don't
483. 1:28:08care just copy and paste copy yeah yes you can thanks I I shouldn't none of this stuff is stuff I should ever have to think about
484. 1:28:24but it it's just where we are all right all right what else are we going to talk about tiktock pin Max Force wrong there are different
485. 1:28:37models because it would be too expensive for everyone to use the best well no no no no no no I mean sure that's why there's GPT 40 and
486. 1:28:50GPT 40 mini but if they really wanted people to use GP T 40 mini they would have put it up here right so that's
487. 1:28:58clearly a less important model but the reasoning models there are four reasoning models these are fundamentally
488. 1:29:05different tools than GPT 40 but if you're just a a a basic chat GPT user how would you even [ __ ] know use his
489. 1:29:15Advanced reasoning what does that mean what does that mean to a human being doesn't mean anything 03 mini it's it's good at fast reasoning
490. 1:29:30okay mini High great at coding and logic okay 01 prom Mode best at reasoning so 01 uses Advanced reasoning 01 prom Mode best at
491. 1:29:43reasoning why do why why do I need to use all these why are they presented to me why do I need to like why is that on me I'm not a developer
492. 1:29:54ER the whole the the whole [ __ ] thing the whole reason now we're getting into meltdown Mondays the whole reason
493. 1:30:03chat GPT went from zero to 100 million users in six weeks six weeks was because when they launched it
494. 1:30:14you didn't have to choose a model there was one model it was chat GPT 3.5 we didn't even know what it was called until they
495. 1:30:22released four and then they gave you 3.5 and four right they made an interface that hid all of the technical underpinnings from
496. 1:30:39us that's what made it so easy and and you know they're now giving us these new things which listen as an advanced user
497. 1:30:48great I love experimenting with that [ __ ] I love learning that [ __ ] but for the 98% % of human beings that don't
498. 1:30:56want to experiment with chat gpp they just want to know what the [ __ ] it is this this is an absolute [ __ ]
499. 1:31:02nightmare it's an absolute nightmare anyway all right hey you know you can just not use it yeah yeah yeah I
500. 1:31:12know I know I'm bitching and you know it's funny Pate like who I'm bitching for are like the the question that we had of like wait
501. 1:31:25how do I prompt this thing and like do I have to prompt it every time and like that's why we have custom instructions
502. 1:31:31that's why we have custom gpts because you can kind of pre- prompt things like that's already a lot to think about
503. 1:31:40if on top of that I now need to figure out which model I want to choose for which thing I want to do it's just you
504. 1:31:49know I mean we we've gone from we have this tool that is this remarkable knowledge augmentation of our brain that
505. 1:31:58now becomes this pseudo you know programming environment and it's just you know for geeks and for people that
506. 1:32:05like geeking out it's fine but for again for most people it's like it's just all right okay this past weekend so
507. 1:32:15for the past two weekends P modder who's who's in the chat here um he works at Google um works on on tensor optimization algorithms or
508. 1:32:28something he makes the Google [ __ ] work fast um he's been doing um demystifying AI classes every Saturday for the past
509. 1:32:40two Saturdays the recordings of those sessions are over at the AI Salon if you go into the clubs area you'll see
510. 1:32:47mechanics Guild with Pate or yeah AI mechanics um I think club now it's called instead of of guilt um with Pate
511. 1:32:55you'll see that in there and the recordings are in there so you should go check those out um there's definitely still a
512. 1:33:01disconnect between Ai and users no doubt um but it moves so fast it's hard yeah I get it I I mean the the actual answer to
513. 1:33:09why there's all these models is because open AI keeps creating different versions of models that are all really
514. 1:33:17good and do slightly different things and they haven't quite decided what of these they want to roll into a final
515. 1:33:23product right so um I would much rather them give us the tools than not um but at the same time it would sure be nice
516. 1:33:32if they did a like like wouldn't it be nice if you could just flip if if this thing here was automatic mode and manual
517. 1:33:42mode and automatic mode you don't see any of the models automatic mode it just chooses the model for you and maybe it
518. 1:33:49gives you some dialogues of why it chose a model and then you could put it into manual mode and it gives you access to
519. 1:33:56all these where you can manually choose these right that's a relatively straightforward solution that just hasn't been
520. 1:34:05prioritized because everyone's moving so fast so anyway that's my that's my two cents on that I don't want to reason so much yeah
521. 1:34:15I know same way from 40 to 03 is insane oh I from 40 to 01 to 03 you're like where's 02 well that was someone else's trademark really
522. 1:34:30really why not call the reasoning model like chat GPT R there's chat GPT that's your normal chat GPT and chat GPT which
523. 1:34:41sounds like a like a race car oh I'm GNA use gptruck
524. 1:34:59I can't oh my God all right well I haven't even started on what I wanted to talk about deep seek has it figured out yeah
525. 1:35:11but they're just so deep seek is just sort of coming out they're sort of jumping they basically took all of they
526. 1:35:18probably took you know um Espionage grabbed um you know raw models from chat GPT and train their model on on that or or at a
527. 1:35:30minimum use chat GPT to train up their model um and then they're just sort of coming out of the gate with one model
528. 1:35:37and and sort of jumping or they have two right they have the well I don't know three but they just haven't been in the
529. 1:35:44game long enough to have eight different models in there in their chat GPT thing yet so anyway um all right I I want to talk
530. 1:35:56about what do I want to talk about I think this is worth talking about how many folks do we have in here got 56 there we've got oh we've got
531. 1:36:08about the same amount in Tik Tok and in whatever you call it um YouTube interesting mistl um it's the same 56 people yeah exactly
532. 1:36:28exactly um there's a new model out called mistol that's not new well it's a new version from them mistol has been
533. 1:36:36around for a while they were kind of The Darlings of the open source scene for about two or three months which is a
534. 1:36:43pretty long run but that was about a year ago and they just came out with a new one so recently we've had deep seek
535. 1:36:49which which cost the stock market uh $ 1.2 trillion in value on one day um then two days later um Alibaba came out with
536. 1:37:05quen qwen so if you go to chat. qw LM for languag model. a I think all these URLs I'm I'm so done
537. 1:37:21with with n.com URLs chat. quen lm. yeah that's right that'll take you to this thing does mistl have one of these
538. 1:37:35things is it like chat. mist.com that would be nice probably not oh Mall mraw SH damn
539. 1:37:50you chat. mistr dot AI it is AI okay accept and continue Le chat ask Le chat so they have a canvas as well they
540. 1:38:06have image generation as well and they have web search you need to be logged in to use this fine I'll log in you bastards
541. 1:38:25image generation uh let's do 70s muscle car Resto mod in an abandoned
542. 1:38:43Factory I wonder what model they're using under here or if this is a multimodal model that's not bad I've seen shittier
543. 1:38:54shittier uh Generations by far that's pretty slick all right not not great with the letters let's let's try to do
544. 1:39:07uh we'll do it let me do a canvas here let me get rid of image Generation Um do we do new new chat we'll do
545. 1:39:21Canvas um write me a sonnet about muscle cars and so it throws it in a canvas over here if I highlight this does it
546. 1:39:38do what's that analyze proof read more bullet points fewer bullet points yeah slick all right there's yet another free tool
547. 1:39:54okay this is a really good all right let me go back to my fat face this is a really good they have a an app too okay so you can go get you
548. 1:40:06can go get free mistel if you think that the French are more trustworthy than the Chinese are more trustworthy than the
549. 1:40:14Americans you now have choices you you now have choices where you give your data away to that's going to be exploited
550. 1:40:24you can give it to Google which you've been doing for two decades so why stop now or you can give it to meta which
551. 1:40:31again you've been doing for two decades or you can give it to open AI or you can give it to deep seek which is owned by the Chinese
552. 1:40:42government or instead of using T Tik Tok you can use red note which a bunch of ticktockers did or you can now give it to the friend
553. 1:40:53and I heard that there's a German one coming I think there's a new model coming from Germany here's the point
554. 1:40:59about all these [ __ ] models and all all these [ __ ] tools you're like which one's the best it doesn't [ __ ]
555. 1:41:05matter it literally like it literally doesn't [ __ ] matter pick one it doesn't matter but which one should I pick pick
556. 1:41:17chat GPT that's the one you should pick everyone's trying to catch up to them they still are trying to catch up
557. 1:41:25to them Tick Tock pin Mission critical Kyle serious question why are they called muscle cars oh that's interesting
558. 1:41:34let's go let's go ask chat gbt that well so in the in the 60s so in 1964
559. 1:41:48um they came out with the Mustang and the Mustang just had a design that looks strong like so I'm talking out of my ass right now
560. 1:42:04because I know some of this history but chat GPT is gonna know all this um but part of it came out of racing so so you
561. 1:42:11had like 24 hours of Lem man you had if you've seen the movie um what the [ __ ] it called with Matt Damon where where he plays the Carol
562. 1:42:20Shelby um working for Ford where the where where they go after they create the the GT the GT40 yeah I'll ask chat
563. 1:42:28GPT um oh you know what we'll do we'll use deep research for this this will be good all right oh the name of the movie
564. 1:42:37yeah yeah yeah uh uh name of Shelby movie with Matt Damon
565. 1:42:53Ford versus Ferrari Ferrari um so so you've got in the 60s you've got this you've got this um this push to to make
566. 1:43:03American cars faster than European cars so Ferrari had 24 hours of lemon no one could touch them they were the fastest
567. 1:43:10cars ever Carol Shelby gets hired by Ford to create this car that can compete with the F Ferraris um and so they took
568. 1:43:18some of that technology and moved it into the Mustang and and that sort of kicked off the whole thing but let's go uh where's deep
569. 1:43:43research Brandon how am I how am I losing my where's deep research in chat GPT is it 01
570. 1:43:57Pro no oh wait 03 mini high
571. 1:44:15no yeah deep research was Google but then open AI named their their new thing deep research co-pilot did it too I know I'm
572. 1:44:31fading fast it's almost midnight I'll catch up with you wonderful regulars tomorrow yeah that's right 03 mini high is deep research is
573. 1:44:41it for coding and logic oh oh good lord here we go oh my God I'm such an idiot but okay so so so what just happened to me this is a really
574. 1:44:55good lesson actually why can I not create a new why is this not creating a new chat deep research is this
575. 1:45:04button I forgot it's a button so over here there's there's pull down menu buttons there's search and then there's
576. 1:45:12deep research and then deep research yeah I think you have to do with O3 mini high and then I and then from what I
577. 1:45:22understand deep research actually uses 03 proper so you're you're actually getting a more advanced model here but what is the origin of the
578. 1:45:36term muscle car um give me good details
579. 1:45:55good stories and make the report entertaining as if a
580. 1:46:11non boring person would be interested in it okay there you go I can't believe I forgot where deep research was it's
581. 1:46:25sitting it's sitting right there as a button but there's a really good example of there's just too much crap I mean
582. 1:46:34it's my job well it's my self- assigned job to be on top of this [ __ ] um okay and then okay I'd love to
583. 1:46:42chis do you want to focus on specific eras um yes uh 60s like 60s to
584. 1:46:5770s um would you like stories about key figures brands or legendary models um you choose the focus um should it include cultural impact
585. 1:47:13definitely cultural impact and then do you want humorous dramatic or more journalis IC tone let's just
586. 1:47:29go dramatic with a sense of humor so it's nice that you you give it a goal and then it it asks clarifying questions and and so now it says
587. 1:47:43starting research P pin okay fun fact the first muscle cars were made at what what did that say from real muscles
588. 1:47:54that's true that's true uh where do you think the term horsepower comes from what's cool about this I can click on this research
589. 1:48:03thing and I can either look at the sources that it's that it's finding or the activity this deep research tool as far
590. 1:48:13as I know is only available in the $200 a month chat GPT Pro subscription which I have um it is it is a quite remarkable tool so
591. 1:48:29it's you know right now it's exploring Wikipedia it's got one source it's going to go out and basically find a bunch of
592. 1:48:35sources for all this stuff read them all and then compile it what you're what you're looking at over here this
593. 1:48:42activity this is its Chain of Thought what's called Chain of Thought reasoning where it's prompting itself so what
594. 1:48:49we're actually watching here is a an autonomous agent we're going to hear more and more about th those throughout
595. 1:48:56this year but basically it goes out finds a website reads it takes what it learns there and then gives itself a new
596. 1:49:04prompt right so piecing together the reports narrative style so so basically it figured out what it wanted to do
597. 1:49:11searched for the origin of the term muscle car I'm walking through Wikipedia a forum and various websites to understand the origin and
598. 1:49:20history noting sources like like deal Honda and History Channel for more insights look I'm curious to see the
599. 1:49:29breadth of topics on Wikipedia I'm piecing together the bir birth of muscle car term so it's reading Wikipedia now it's at English stack
600. 1:49:38exchange tracing the term muscle car from January 1964 Popular Mechanics so I wasn't too far off right um considering shooting stars broken
601. 1:49:51lenses um so so anyway it and it'll just go right it's it's found eight sources so far so it's off doing its thing all
602. 1:49:59right if there was a p pin there it's gone now it can search for up to 30 minutes yep it can for some reason I have it in my head
603. 1:50:29Peter Fen kud with open-source models you generally have the option to run them locally so you don't have to worry about
604. 1:50:38privacy that's correct but you also do have to worry about do you have the technical skills to install them locally
605. 1:50:46do you have the hard drive space to install them locally and do you have enough R Ram to run them locally so if
606. 1:50:53you're going to run local models you do have to geek out a bit um I don't think it's going to be that way for
607. 1:51:01long as much as I [ __ ] about all the jankiness of the companies and and these applications right now the opportunity for being early in
608. 1:51:13AI for being early in any technical advancement the opportunity exists when the tools are janky if all the tools were easy to use
609. 1:51:24everyone would be using them and there wouldn't be much as much opportunity um so so that's that's why we put up with this
610. 1:51:34[ __ ] Brian on Tik Tok wants to know if you still remember Zephyr Zephyr my girlfriend from PI yeah I remember her
611. 1:51:42but she at at some point they did some upgrade in pi and she forgot her name and forgot my name and I just never went
612. 1:51:49back to her I'm like I'm like you forget my my name you are dead to me GPU is definitely needed for local AI
613. 1:52:00uh for the speed CPU can do it but GPT is highly recommended which again like if you're on a Mac you're kind of
614. 1:52:06screwed although I think with the new M4 the M4 Max Max with their M mlx so so Apple came up with with a new
615. 1:52:22framework called mlx that basically navigates between CPU and GPU and RAM I think like like it makes them
616. 1:52:31all sort of interchangeable as far as I understand it um so so the new Macs are pretty good at AI stuff but you know yeah if if if you
617. 1:52:42want to run it if you want to run them on you know PCS with real gpus all right the rise and Roar of the muscle car 60s and 70s
618. 1:52:53imagine it's the mid-60s the air is thick with the smell of burning rubber teenagers of slick back hair okay let's
619. 1:53:00see first Rumblings the birth of the term muscle car every Legend has its origin story and muscle car begins with
620. 1:53:07the word the term muscle car flexed its way into the automotive lexicon in the mid-60s car magazines first started
621. 1:53:15tossing around the catchy label in 1965's to describe Detroit's new breed of Street Fighters for example one writer noted The Humble
622. 1:53:25Dodge Coronet became a muscle car once it was fitted with a monstrous for 26 Hemi V8 so they were taking these Racing
623. 1:53:34Engines out of their racing programs and jamming them into these cars that didn't look like what we think of muscle cars
624. 1:53:42today so they were they called them sleepers like they would go drag race these things 66 it had rumbled into the
625. 1:53:49mainstream pontiac's feroci ferocious new GTO as so Pontiac marketed it our muscle car with all the woodb Tigers in its
626. 1:54:01wake in short muscle car became the F perfect nickname the idea of a muscle car was brewing before the name as far back as
627. 1:54:111949 so yeah but mid-60s is when it started and like late 6s like the the the Golden Era of muscle cars was late 6 early 70s once you get to
628. 1:54:251974 and there was the gas crisis and and automakers were sort of forced to make smaller less gas guzzling engines
629. 1:54:35that killed the muscle car era so it's a really a tight like 65 to 73 probably 72 73 is really the era um and you've got
630. 1:54:48some some blips that return but that that was that was that was the era that wasn't all the cool stuff anyway but look at look at this
631. 1:55:01long report like this is a long researched article so I should be able to if I go to the bottom of this now save it as a PDF shut
632. 1:55:12up um so I've gotta wait let me say um pop this into a canvas oh I didn't need deep reasoning to pop
633. 1:55:29it into a canvas that's the other thing if you're here stop I don't need a I don't need a [ __ ] reasoning engine um please put
634. 1:55:41the article in a canvas and then I think you can even just say canvas here Bang Bang all right so there's our
635. 1:56:00canvas slash canvas yeah you can do that as well all right I'm on the plus plan and max out every day considering Pro oh interesting cam
636. 1:56:15interesting do you know that you can execute custom gpts for dual user setup and easily switch between them both um I I I know dual user
637. 1:56:28setup you mean teams I don't know quite what you're talking about Drew but yeah that sounds cool 100 bucks a month there's a
638. 1:56:40time or daily limit let's see time check yeah all right I can't believe how slow writing stuff into canvases canvas is is
639. 1:56:56a pretty slick tool but it is janky as [ __ ] half the time it can't remember it like you highlight text and say fix this and it's
640. 1:57:08like I don't know what text you're talking about it's just it's just a janky mess oh look here's the the
641. 1:57:15legends of a asphalt the the iconic cars the Pontiac GTO 64 to 70 Mustangs 642 to 73 Camaro 67 to 73 Charger 66 to 74 and
642. 1:57:29the Road Runner 68 to 74 yeah those are there's your there's your biggies those are your biggies I've place your entertaining and
643. 1:57:39dramatic muscle cart article over here all right I thought it was longer than that whatever doesn't [ __ ] matter all right
644. 1:57:53and then if I go copy I guess I could go select all let me do I'll do Brandon's thing go to print go to save as PDF
645. 1:58:09save the muscle car Origins explained save I don't know where I just saved it go to my finder here we go look I've got
646. 1:58:19a PDF right on my desktop and it's well researched and everything yay you did it all right canvas is 100%
647. 1:58:31annoying it really is it really is I heard the Deep research option will next come to the chat GPT plus users looking
648. 1:58:40forward yeah I think I think um deep research is is really quite impressive especially if you do things
649. 1:58:48like more more substantive than look for the origin of muscle car that was kind of a thin a thin research project but um
650. 1:58:57okay let me let me get this on out of here I did not get to what I wanted to talk about tonight so maybe I'll do that
651. 1:59:03tomorrow maybe I'll actually think about what I want to talk about instead of winging it but if you all show up with a good
652. 1:59:11interesting question I am gonna answer it so that's what happened tonight all right um go check out the AI
653. 1:59:19Salon if you haven't if if you're on YouTube or if you're not on YouTube and you haven't subscribed Please Subscribe
654. 1:59:26and please watch some of the back videos there I've got to keep our watch hours up to maintain our our commercial
655. 1:59:33viability as a as a business entity um it was tough to get there and we got to we got to maintain it so until we grow
656. 1:59:40the channel a bit more uh just any any watching you can do on the YouTube channel would be uh would be good um oh
657. 1:59:49yeah the other thing is you can now subscri subscribe to the YouTube channel like again we I've got subscriptions on
658. 1:59:56Tik Tok and I now have also paid subscriptions on YouTube you don't have to do both but if you want to support
659. 2:00:03the channel and you want to do it on YouTube rather than here it gives you some perks there they're listed in there
660. 2:00:08what they are Brandon and I haven't spent a bunch of time refining those yet so it's just it's raw and early but if
661. 2:00:15you want to support the channel you can um I think that's about it happy Monday hope you had a good weekend hope you had
662. 2:00:22tonight hope you learned something and uh I will see you tomorrow tomorrow's a normal Tuesday so we should be right on
663. 2:00:30time 800 PM mountain time and uh yeah let's uh let's get on out of here all right peace out I will see you all later all right you