2/3/2025 - Musical Update & OpenAI Deep Research: A Quick Look

Transcript

1. 0:00w [Music]
2. 0:28[Music]
3. 0:42yo yo yo everybody what's happening what's shaking what's going down good people on the internet I am in New York
4. 0:56City in uh Time Square area I got I got busyness I got business to do I'm doing business did that work yes no
5. 1:14why I'm dealing with technical situations here people calm down all right um I can't talk super loud because
6. 1:22I'm in a hotel so I don't want to get told to shut up so give me a little dude uh we're not doing a deep dive
7. 1:32tonight um just because I'm too busy and traveling and stuff like that so these will be the travel ones will be like a
8. 1:41little little they'll be a little chill they'll be chill they'll be chill so got a handful of folks it looks like we've
9. 1:48got it's just a pile of Irregulars here so um if you have anything you want me to talk about pop them in there
10. 1:56um we will we'll make a jelly pod tonight because why wouldn't we oh we've got some exciting news jelly pod is loving
11. 2:06the AI learning lab making making podcasts for them um so we've got some good news there um
12. 2:21the new deep research from open AI is out I've used it for two pieces of research um it was quite good one of the one of the pieces of
13. 2:34research I used it for was nothing wrong with chill exactly James um one of the things I use deep research for
14. 2:46was what was the first thing I did oh a sales thing this was actually really cool and you could if I the problem the problem with both
15. 2:57operator and deep research which are these new kind of of powerful things is they're only available in Pro um so
16. 3:04talking about them is kind of shitty because I have access to it but but most people don't it's it's like well it's
17. 3:11nice nice for you Kyle isn't isn't that nice you get to play new exciting things um and you're correct you should talk to
18. 3:20me like that um let's see I did um so the way deep research works I think I've got one left I think you're
19. 3:33allowed three a day like even for the 200 bucks a month or it's a hundred a month or something like that I don't [ __ ] know
20. 3:42um I'll find out um but it goes off and it researches right it finds websites it reads websites wait you're still in New
21. 3:53York yeah so I was in Upstate New York for the reading tonight I'm in New York City we've got city meetings tomorrow then we go to Jersey
22. 4:06for half of Tuesday that's tomorrow then Wednesday we're supposed to go to South Jersey but that meeting
23. 4:14got canceled so I think we're coming back to the city and then Thursday we're somewhere in Jersey again and then late Thursday
24. 4:22we come home and then we've got virtual meetings when it doesn't [ __ ] matter but yeah I'm in New York uh um so
25. 4:30excited about the music how did the music Go the music reading go tell us okay we are fans and the team team
26. 4:37Irregulars um so the the um the reading was Saturday um it went very very well like better than I expected so when you
27. 4:49do a reading like that there's kind of three three possibilities um one of the possibilities is the thing you thought was great
28. 5:03isn't right you you have a piece of crap the other possibility is the thing you have has possibility but it's
29. 5:19structurally it it needs rearching and then the third possibility is you have a show and I was pretty confident we
30. 5:29weren't going to be in the first category but you really don't know um I actually was leaning toward I thought we were gonna have to
31. 5:38restructure the show um because in reading the show it reads like act one doesn't end strong enough strong enough
32. 5:47um and when we heard it we've got a [ __ ] show um the woman that played Sydney was awesome it was a husband and wife team played Kellen and
33. 5:56Sydney um she was great he was really good good um the woman that played terara was also really good they brought all sorts of cool choices
34. 6:08um there's one there's one scene that we wrote so so there's there's a scene where um where Sydney says I have a I
35. 6:18have a confession to make thank you Silver Fox Sydney says I have a confession to make you know I have a I
36. 6:25have a secret I want to tell you and the secret is she loves him right and so he's like like freaked out and he he
37. 6:31says just I'm going to change the subject like just talk about gutters and so and so there's this funny
38. 6:42scene of like I don't know three or four you know little monologue little speechy things from Sydney where she's talking
39. 6:49about gutters and they're funny like they're really funny um you know you know gutters gutters are an often
40. 6:57overlooked but important part of your house you know you should clean them regularly things like that um and so it's this
41. 7:06funny little scene and the way the actress played it was um she played it as if she wanted to change the subject back to
42. 7:19love but couldn't because she's programmed to give him what he wants and it was [ __ ] tragic it was really like oh so
43. 7:30sad um so that was it was a really good choice and then there's there's another scene where Kellen and his wife are
44. 7:36fighting they have a fight in the First Act and they make up in the second act and so we wrote it like he brings
45. 7:43her a sandwich and she instantly forgives him and you know and they have a nice scene together and the actress
46. 7:50played it like she didn't let him off the hook like all the lines are like she let him off the hook but she played it
47. 7:56like no I'm still [ __ ] with you I'm still pissing off that was really good look like you're in a holding cell
48. 8:05pretty much that's welcome welcome to the the world of corporate hotels it's pretty much a holding cell I have little little
49. 8:21coffee hermetically sealed paper cups because you know you wouldn't want an unsanitary raw paper cup all
50. 8:35right um okay couple of things so on YouTube now we've made it to the point that that we can do business stuff on
51. 8:47there so there's things like super thanks and stickers and things like that there's also um two different subscription levels if you want to
52. 8:55support the channel if you want to subscribe on YouTube you can do that now um we're not we we haven't put enough
53. 9:03time into like I just haven't had time to design what it means to subscribe but if you want to support the channel you
54. 9:09can just know that that's there now um yeah Brandon do you want to pop up on stage and just talk about it a
55. 9:17bit yeah hey Kyle uh and hey everyone uh live from Cleveland uh not in a holding cell uh at the switchboard here the cup
56. 9:28cont I just read that and I was trying trying to pretend I didn't um so first of all thank you to everybody
57. 9:39who uh got viewing hours up that liked that followed the channel um it was a Herculean effort over the past couple of
58. 9:47months to get to the point that we could get to the level where we could support it financially and um Now We're Here
59. 9:55YouTube um said that we've done enough and they're going to let us uh open so what we've got we've got two different
60. 10:03ways um that you four different ways really that you can engage with us and this is going to be an evolution of
61. 10:09process uh as well just like everything else and subject to change without notice but the um as Kyle said there's
62. 10:16two different levels of subscription uh and memberships um which you can join at The Irregular level of $3.99 a month and
63. 10:25these are all monthly memberships um and then and you could be a learning lab loyalty member at $9.99 a month and um
64. 10:34you get to even uh have a star next to your name you get things that some of the things that we've talked about um
65. 10:40member only live chats if we do any uh demos or previews we're going to offer those to Members First and then the fun
66. 10:49one for me is you get if you click if you're on the YouTube live stream and next to the chat and next to the hearts
67. 10:57and other um motive icons there's a new dollar sign and if you click on that you've got four options you can give
68. 11:04super stickers which I gave one earlier in the chat just to see what it did uh you can also do super chats which you
69. 11:13have my word as your producer if you do a super chat I will break into whatever Kyle is doing and make him read and pay
70. 11:21attention to what you're doing at that moment um because you're paying to play effectively uh and then you can also uh
71. 11:28join memberships in you can also gift memberships so if you've got somebody in your life that you want to or somebody
72. 11:33you don't like and you want to subscribe them to AI learning lab out a spite um that's also acceptable it's a good strategy I like
73. 11:42it so yeah uh it's going to be I would love your feedback um in the AI Salon as far as anything else that you'd like to
74. 11:49see from that in their regulars channel uh and yeah we'll really more to come after um uh after Kyle gets back from New York and
75. 12:02uh Vicky just had a comment about YouTube lives automatically posting to my channel that it's weird they're not
76. 12:09saved into the videos page they're saved into the lives page so they're there we might want to look at Brandon is is
77. 12:16there a way to move them around I don't know if there is or not okay sounds good I'm gonna back into hiding for sure okay cool um and uh
78. 12:27let's see Taki Taki said I joined both YouTube and Tik Tok just in case I need to switch sides awesome thank you very
79. 12:33much um I think Tik tok's going to stay but we're still in this weird thing where if someone doesn't buy it I think
80. 12:40that the current Arrangement is I think it was 70 days so we're probably 10 days into that so there's probably two months
81. 12:48left of Tik Tok so we'll see whatever I you know the there was someone wrote legislation um that if you download deep seek R1
82. 13:00it's punishable by 20 years in prison so it's like you know I it's just going to be an exhausting couple of years with legislation because
83. 13:13the legislators this thing is moving so fast right like we're on this bullet train and the legislators are sort of looking
84. 13:22at us as it passes Cleveland and they're like let us write the law about Cleveland and it's like already in Pittsburgh you
85. 13:30know so it's gonna be weird was that Elon I don't think so it's going for cult status a tax deduction was what Elon I don't
86. 13:45know I don't I didn't I didn't know what that me law lawmakers have no way of keeping up they don't I mean they're going to
87. 13:56make it illegal to download deep seek and then what three days after deep seek was quen which is apparently better than
88. 14:05deep seek and then three days after that is the new 03 mini High chat GPT still still has remarkably bad yeah oh oh Elon did not do that
89. 14:19legislation elon's not writing legislation um he's just got control of the purse strings of the government it's
90. 14:26nothing to worry about there I'm sure I'm sure it's going to be fine I'm sure there's it's going to be [Laughter]
91. 14:45just um Bam um the yeah the naming conventions that open AI with their with their stupid [ __ ] names is they they
92. 15:02like honest to God like I think they think it's cute at this point they know it's a [ __ ] nightmare gp40 great for most questions GPT 40
93. 15:15with scheduled tasks thanks um Brandon let me I'm gonna go to this view because it's because because I'm
94. 15:25sharing a window instead of a a tab it's too skinny so that puts me a little closer to the window anyway um let's go look at these again GPT
95. 15:3840 right that's your main model and then there's GPT 40 with tasks not to be confused with with reasoning or research or
96. 15:51operator then there's 01 for advanced reasoning they got rid of 01 Mini but they replaced it with 03 mini and then
97. 16:01because I'm on the Pro Plan I've got 03 mini high and then I've got 01 Pro mode and then I can still go to GPT 40 mini and gp4 the
98. 16:17original mixture of experts model um at this point what what started as a a
99. 16:34good um simple interface for everyone you now essentially have to be a developer right like you have to have
100. 16:44a developer mind you have to give a [ __ ] about learning what these different models do you have to figure out what's
101. 16:50possible um they need to create a chat GPT model that understands the GPT models okay they just name them after Disney
102. 17:03characters I mean Disney characters would be better well frankly Source Camp what they need to do what they need to do is we shouldn't
103. 17:13have to [ __ ] choose right we should be able to just type [ __ ] in and it'll go oh would you like me to go do
104. 17:22research on that for you or do you just want an answer oh go do research great right Disney would suit well but
105. 17:29just the the the basic idea of just doing a name um the other thing you can use now is co-pilot So within co-pilot you now have
106. 17:39reasoning um you don't have deep research um but they do have reasoning Source Camp I think that's the only you
107. 17:45can't do Disney doesn't pay yes or name them something we can keep straight yeah like like reasoning or or research
108. 17:59like descriptive names but again like I I just I go back to the place of and I think I think Google's going to need to
109. 18:05deal with this Google's got this really bad right now because their products is so fragmented at least chat gpts is sort
110. 18:11of in the same place you've got operators a separate application you've got Sora as a separate application and
111. 18:18then within chat GPT you've got 12 applications Google's got applications you know scattered all over the place I
112. 18:26think all of these companies need to create [Music] um AI simplification front ends that just figure this [ __ ] out for us but
113. 18:37we'll you know whatever whatever we're not there yet you know that's what what this tells me is we're still
114. 18:44early right this is not going to be a mass adopted thing while it's this janky and and fragmented so but that's all
115. 18:53right um we have PhD level problem solvers in our back pocket now all right um so let me show you something really
116. 19:03cool I'm going to go over to jelly pod and if you go to Jelly pod you go to the like the main page and then there's this discover
117. 19:14podcast little section wait hang on make this bigger oh no I can't and you go to discover podcast and look what the fact is is a featured
118. 19:31podcast right in the top right in the top our cute little thing my tabs are all over the place my tabs are a mess um
119. 19:40but but yeah what the uh what the fact is is we made the big time on the jelly pood homepage which you know I mean they're
120. 19:48not gonna they're not you know I asked the uh the founder the other day are they going to become a media company and
121. 19:55he said no but if you look at this this kind of looks like you know a podcast site doesn't it and then they've got
122. 20:03categories a bunch of different podcasts but we're featured Numero Uno so let's go make ourselves a new podcast so if you haven't seen this
123. 20:14before this is jelly pod they're breadcrumbing it out to us yeah exactly uh FY FYI notebook LM the speakers are somehow even better can't
124. 20:25tell their AI wow okay interesting all right so we're going to go into what the fact and we're going to go new
125. 20:34episode and then we're going to go to chat je we're going to go here we're going to go podcast so if you don't know how we did
126. 20:44this um the way jelly pod works is you design a co-host or two you design what you want them to sound like so we had chat GPT do that for us
127. 21:00and then you create a show concept which we had chat GPT do we had it come up with I don't know 20 something like that
128. 21:09and then what the fact was the one that we we liked and then you basically just copy your little outline from from the chat
129. 21:24je boom Oh wait what is this one haunted dolls and creepy toys where playtime becomes nightmare cool all right so we're going to pop all
130. 21:37that in there we're going to do episode 7 we going to do add Source then we're going to go back to chat Jeep we're g to grab what's that
131. 21:54oh an alternate Wikipedia Source that's kind of cool cool grab that did you bring any construction paper with you I didn't but we've got
132. 22:06we've got a low angle on the keyboard so we're not getting bad uh bad reflection so it's not [Music] horrible Okay so we've got three
133. 22:22sources so we got two two Wikipedia pages and our little outline from chat GPT and then we're going to say create
134. 22:33so the first thing it's going to do is create an outline chat GPT just puts a reasoning button or deep seek just puts a
135. 22:45reasoning button which is that's what open AI had that for a day and then they took it away the more crap they add
136. 22:53without a good UI the more they create the illusion of barrier to entry that's an interesting thought I don't know if I agree with
137. 23:12that I just think it's I just think it's they're a research agency I mean they're they're a research company and they you know they're not a
138. 23:23they're not a product company okay um welcome people to to the to episode
139. 23:377 welcome all right so we're going to move this up here we're going to say episode
140. 23:527 okay welcome people to episode 7 remind them that that this podcast is sponsored by the AIS salon atth
141. 24:15salon. and then we've got to go work on our our little images what the fact image is we got to get number seven haunted dolls Oh this
142. 24:29is cool oh wait except hang on uh I got to change my sharing for the YouTube people we're g to go we're going to go
143. 24:42full screen people we'll do it live look how cool that is the uh the lovely graphics for the what the fact um
144. 24:56podcasts were made by Corey Sandler she made them all in one night she just like blasted out 20 of them I sent her I sent
145. 25:04her all the outlines and she just blasted them all out but what we have to do here is we have to resize this
146. 25:10because whatever you can't upload an image this big so I go 768 this is the kind of stupid [ __ ] you
147. 25:18discover when you're trying to you know you're trying to move quickly and then you realize oh that's stupid
148. 25:28uh uh this is one of my um desktop jpeg one of my skills as a product guys I can usually find a bug within within five minutes of the
149. 25:42engineers telling me that they've got the perfect build and it's always stupid [ __ ] like uh could we upload an image bigger than you
150. 25:54know the thumbnail file too large still too large oh did I do the what what did I [Music]
151. 26:10do is that not the right um I saved the wrong one hang on duplicate as Tiff there we go export that
152. 26:26as JPEG with that with that go back over here go back in here get there all right is that
153. 26:41working is that working it does not seem to be working
154. 26:59M thanks Hotel internet oh my God I just checked out Jelly pod and I'm floored floor River Willow neon Beats I know right I hate
155. 27:11not using a mouse I know it sucks doesn't it why is this not uploading hang on let me let me reload the page maybe I'm having some you have
156. 27:20too many tabs open sir sir sir we're going to need you to close some tabs yeah listen that's the file's too large I got the wrong
157. 27:35file okay hang on hang on what what is going on here I'm going to do export I'm going to do higher compression I'm going to grab that I'm
158. 27:50going to do um we'll call this SM for small all and we're going to save
159. 28:06it on the desktop save come over here we're going to grab this where's the SM one I don't get it
160. 28:21man where' it go man um SM contains SM the
161. 28:35[ __ ] this is cool you taught me about GPT last year I use it all day every day awesome River Willow welcome back
162. 28:43welcome back and I'm I'm so glad that that the channel helps generate script audio ad music yeah yeah okay I I
163. 28:54gotta what what's happening this is making me very sad um I'm going to export
164. 29:10to why did it not save it I do not understand we'll do seven SM dash we'll put the SM at the beginning this time we'll make it smaller
165. 29:30350k okay do we have it it's there says it's there we're over here sm all right it's going good
166. 29:48Lord okay okay so this is one of those this so reminds me of the early days of the worldwide web where you were on dialup modems and and
167. 30:01you you went to show someone a website and you'd be like I oh I got this cool website I want to show you I got this
168. 30:07cool website I want to show you and then um you'd be like okay it's it's gonna load in a sec it's gonna be so good okay
169. 30:14just hang on just it's gonna be so good and then like 15 minutes later everyone's left you know you're sitting
170. 30:19there with your stupid [ __ ] computer all right we're going to generate the script now bang all right let me go change my sharing again
171. 30:29can delete that copy let's go back to this [Music] um present I think I should do I'm going to do this as a tab so we can share the
172. 30:49audio remember Joe Pizza's open till about 3:00 a.m. I may just head out I'm I'm in Time Square so I may just head
173. 30:57out after this and go walk the streets because why [ __ ] not um transcript editor what
174. 31:12why oh it's still generating okay I could do a live I could do a live uh Q&A from Time Square but I don't want
175. 31:23to I don't want I don't want to get mugged with my little with my little Gimbal and my little stupid [ __ ] iPhone in my face and someone
176. 31:36comes up and just punches me in the head and steals my phone that would make for a bad [Laughter] trip 93 94 something around there yeah
177. 31:4894 was when I started in all this web stuff I was just talking about the worldwide web early days this I mean
178. 31:55listen we are very much it yeah you said 93 994 I think we're closer to maybe 95 96 like it's decently it's decently
179. 32:04evolved at this point but we are nowhere near the hype cycle not anywhere near it this [ __ ] is still so janky and slow and
180. 32:15damn weird all right so Lily Vogue welcome to episode seven of what the fact so let's make generate the audio
181. 32:28I can save you about three hours and Chase down the mugger New York walk and talk we could I could do a New York
182. 32:36walking talk I like I should do it at night because that's when that's when the [ __ ] crazies are out I'm just
183. 32:44gonna be H I'm gonna have to just be hyper aware because you know it's it's [ __ ] tourist Ville right like if we
184. 32:52were in the village or we were in some cool neighborhood I wouldn't be as worried but you know it's time Square so
185. 32:58they're out to they're out to get all the midwesterners and I look pretty Midwestern you know I have New York attitude but but I've got
186. 33:06Midwestern chops I've got Midwestern DNA we should put an Apple Air tag on [Laughter] Kyle okay are we generating audio is
187. 33:23that what we're doing yes that's what we're doing Lily V oh are we there I think we're there where's the play
188. 33:33button oh here all right ready welcome to episode seven of what the fact where we dive into the Eerie world of haunted
189. 33:41dolls including Robert and Annabel dolls possessed ones never thought I'd be talking about that on a Tuesday but here
190. 33:49we are this episode and the chills it may bring is brought to you by the AIS salonthe salon. a place where technology and
191. 33:58creativity meet to spark inspiration that's fancy speak for they're pretty smart folks over there right exactly
192. 34:05Buster now back to the scary stuff let's set the stage for Robert the Doll one of the most famous haunted dolls of
193. 34:13all oh that's pretty good all right so we need to what the fact MP3 so we'll go mp3 mp3 audio
194. 34:28too good to be true opening that's from our musical where's the what the
195. 34:43fact I I have a feeling it got eaten by my iCloud storage hang on what search contains what what the fact
196. 34:59what's the fact intro push play there we go so we do that we'll do that we'll do
197. 35:14that 30 years being on the forgot the lead in music not yet I haven't morning what tool are we exploring so what this tool is is called
198. 35:25jelly pod and this is now our seventh episode of what the [Laughter] fact into facts that'll blow your
199. 35:41[Music] minding welcome to episode seven of what
200. 35:56the fact where we dive into the Eerie world of haunted dolls including Robert and Annabel dolls possessed ones never
201. 36:04thought I'd be talking about that on a Tuesday oh Bucky mclin talk oh it's Bucky McGill Bucky McGill Cy um Kyle did
202. 36:16you check out open AI deep research yet yeah I have we'll play with that tonight I think I have one one research left
203. 36:23today I think you you get a 100 a month or you get three a day and I don't know if it's three a day or 100 a month like
204. 36:30could I do 99 in a day and then just one more for the rest of the month I don't know um okay publish episode ready to go I'm good right I'm
205. 36:42pretty sure I'm good boom and then what we're going to do man we really need to update Tik Tock I don't know what that
206. 36:54means oh my God the voices aren't they hilarious well so and River Willow we designed those like we had chat
207. 37:02GPT I I told chat GPT that I wanted it to design voices that were comically different from one another and that's what it came up
208. 37:12with they're pretty good he's like buer buer okay episode seven so we're going to click on this we're going to say c
209. 37:29copy and then we're going to go to the x x is what they call Twitter oh there's a priest piing a baby how nice um let's
210. 37:46see oh you know what I need to do wait can you all see this oh no tick to or YouTube can't hang on hang on people you think this is easy
211. 37:58I'm on a shitty little desk in a shitty little hotel room um with a single monitor trying to do all this window streamyard Studio no that
212. 38:12window yes okay there we go um don't we recompile late on Thursdays okay so let me go over here go to my little profile find one of
213. 38:27these other things so we can copy all the text from it what the fact boom copy go here we'll do a new one we'll get rid of this
214. 38:43one we'll add in our little image with the fact we'll put all these on the same line
215. 38:58all right what the fact episode seven haunted
216. 39:16dolls and toys I think all right what the fact episode seven okay so here's what I need you to do people um 100 queries per month okay great
217. 39:30here's what I need you to do go to the Tik Tock no go to the Twitter go to the X did you link the episode no I didn't
218. 39:41did you want me to do that I could do that shut up oh I got to go get it again hang on hang on people just calm down
219. 39:54cool your flipping Jets everybody's like I he did it wrong he's doing it wrong this is really painful watching an old man try to use the
220. 40:03computer this is very painful all right what the fact episode seven looks great so you're going to want to go to the X the Twitter the X
221. 40:17the Twitter the Twitter that where you tweet where you post and you Z you Z and you post the Z the tweets the twitters
222. 40:30and I want you to tag this and I want you to like like it thumbs up and I want you to share it all right it's live we're live we're
223. 40:42live we did it all right all right I feel like K Kyle reads Tik Tock check more than us not true maybe I don't know
224. 40:57not saying a lot but still James low shut up shut up shut up I do what I can um you might need to unretire Frank
225. 41:10while you're in New York that's not a bad idea actually let me tell you something let me tell you something
226. 41:16about this whole naming convention open AI got going with their [ __ ] 01 03 we got the 40 the 01 the 03 the 01
227. 41:25mini the 01 mini High what are they [ __ ] high yeah we could do that I can't be that loud in my little hotel room they're going to yell at
228. 41:34me um so we're live so that's good there we go what the fact is out there did any of you like it let's see Kelly Camp she
229. 41:46gets the gold start tonight she was first she was first she was first people [Music] G okay so here's what we're going to do
230. 41:59we're going to make that go away we have a deep research button that we're going to turn on and I think we're going to do a little deep
231. 42:10researching I miss Frank laughing my ass off hey hey hey let me tell you something this [ __ ] this [ __ ] Sam Alman with his little
232. 42:21[ __ ] cute little names the cute little letters he's got like some sort of [ __ ] code it's like if we stick
233. 42:27all the letters of the names together it's going to it's going to make some [ __ ] password that's going to unlock
234. 42:32the [ __ ] Universe [ __ ] yeah exactly all right anyway I don't know if deep research Auto turns it to 03 mini I don't I think
235. 42:45you have to turn it I saw some people bitching about that today so so yeah so I'll go if if you just leave it on 01 or o0 3 mini the
236. 42:58Deep research doesn't work yeah I was using it today it was it was pretty bad it was pretty janky it's so confusing it
237. 43:04is Dr Bud frost bitten but you can use 03 mini high for it that's what I'm going to do some AI guy put Sam a in a
238. 43:13deep seek t-shirt that's pretty good that's solid Kyle can you explain the difference between the models in open AI okay I'll try
239. 43:29let me start at the bottom let me start at the bottom so gp4 Legacy model the one that they put all the way at the bottom that's
240. 43:40actually a really good model that is the that was sort of the Pinnacle of of chat gpt's um you know pure large language
241. 43:56model it was it was actually 10 different 170 billion parameter specialized models in what's
242. 44:11called an a mixture of expert models so it was 1.7 trillion parameters why they've got it at the bottom there is I think it's really
243. 44:20expensive to run but it's really good so if you're not getting good results with the other things you can try it I mean
244. 44:26the thing about all of these models at this point is they're all so good like I I think it's nuanced differences between
245. 44:34them the I mean the reasoning engines are very very different but like gp4 gp4 mini and GPT 40 like I don't know anyway um GPT 40
246. 44:50mini so 40 the O standed for Omni so it's it's a multimodal model meaning it it can see images hear audio
247. 45:06speak um make images but they still haven't released it fully we're still making images not with 40 not with the image generation part of
248. 45:18it so it's weird anyway um so 40 is just your basic run-of-the-mill chat GPT 40 was scheduled TKS is um if you want to say every
249. 45:32morning you want an inspiring little they have more models than you have tabs they do they really do it's exhausting it's
250. 45:44exhausting um tasks is really cool like if you wanna if you want to do an every morning inspirational quote and have
251. 45:52chat GPT write it for you you can do that it's pretty slick it's a expensive to run compared to 40 yep exactly sunno
252. 45:59AI cussed in my song a few times today yeah I got um refusion I want to go play with refusion as well tonight um we
253. 46:07we'll go play with that we'll make a song um I'm gonna let you listen to a new Sydney song we we um there were two songs in the script
254. 46:17that didn't exist um all the other songs we had um and I did a really cool quote a really cool prompt that I didn't think
255. 46:26would work and it worked beautifully and we actually included the song in the in the table read um did you get deep research yeah I
256. 46:35have deep research we're going to play with that in a second someone's just asked me to explain all these models okay
257. 46:4101 um is the advanced reasoning it's really really good 0 one's really really good um it's available for free users
258. 46:53and then down here at the bottom is 01 Pro so because I've got a pro account that's the 200 Buck a month account and
259. 47:00you're like Kyle is it worth 200 bucks a month not really but particularly for for open AI I want to I want to have
260. 47:09access to the new thing because they're kind of leading the charge with this stuff so I want to have access to it so
261. 47:15thank you AI salon for sponsoring that um and then um 03 mini and 03 mini High these are the new super fast super good
262. 47:29better than deep seek models apparently I don't [ __ ] know like everything is they're all just sort of slightly better
263. 47:37in most of the benchmarks than each other so at this point it's just for all of us it's it's way more than we deserve is is how I would put
264. 47:49it okay so that's all that stuff um Okay so with deep research you you flip it into gpt3 gp03 mini high and then one of the
265. 48:07things that I read on X today was that when you use deep research it's actually using the full 03 not not 03 mini I
266. 48:18don't know if that's true or not I assume it's not or they would say that Kyle have you played with or seen Quinn
267. 48:26to .5 I have I have seen it if you go to chat I think it's chat. quen lm.com chat. quen quen lm.com
268. 48:46no chat do quen ll.com no is it chat quinn.com can't be couldn't it couldn't be actually
269. 49:04good I bet it's quen ai.com NOP let me try quen ai.com I bet that's it nope qu. a I don't [ __ ] know
270. 49:28anybody does anybody know what the Quin thing is I I was playing with it last week chat dot quen quen
271. 49:43lm. move your move your mouse Brandon I can't see the quen LM so I had it I just didn't have I just didn't have it I didn't
272. 49:54have God damn I didn't have the the 17th iteration of the stupid [ __ ] URL um there it is if you go to it's a really
273. 50:04easy URL it's super easy to remember chat. quen lm. axyz doart I am so sick of n.com things isn't it on hugging face yeah but
274. 50:20so hugging next week I'm going to do a deep dive on hugging face hugging face for for normies and this channel is
275. 50:29really for normies for the most part hugging face is confusing as [ __ ] I had a doctor with easy Early Access to heds
276. 50:40on X and its results are amazing okay great I I saw your I saw your post thank you for tagging me in that
277. 50:52Danielle and and what is heds that's the Deep Oh that's oh Early Access to chat GPT deep research yeah it's I did um I did a
278. 51:05sales thing today and then I did oh you know what I did which is really cool I had it for Sydney for the for the
279. 51:14musical I had to go find any Play movie or book that had a relationship between humans and Ai and do research on that and then come
280. 51:28back and and give us a whole report on it it it did it really really well oh HDs is that that connective tissue disorder yeah yeah
281. 51:37yeah which yeah that's probably something that I need to need to look into okay um let's go back to chat JP so
282. 51:48we got to come up with can you please explain the Deep research absolutely I wish could explain I I wish this
283. 52:01[ __ ] weren't so nested but it's unfortunately I need pizza boxes I do
284. 52:18okay let me let me build up the let me build up the Russian dolls here for you so at the core at the core of what these things
285. 52:31do you have a you have a large language model like GPT 40 and that's that's the thing that we've been using where you know you've
286. 52:43you've got your large language model which is the cup and a condom and they've put all of human knowledge that
287. 52:49they could get their hands on they've packed it all in there so this is chalk full of human knowledge and when you use
288. 52:55chat GPT you say hey do you know anything and it does its probability engine thing and it spits out an answer and you're like
289. 53:05great and it hallucinates and it doesn't know what it said and it doesn't know how many RS are in Strawberry right all that
290. 53:13[ __ ] trying to think how to do this well H metaphors are hard when you don't have props when you don't have garbage all
291. 53:29around you that you can you can pull um so that's the base level the reasoning engines use what's called Chain of Thought
292. 53:43reasoning so it's still making a call to an llm but it's doing what they call thinking before it gives you the answer
293. 53:53and it do it it it writes its own prompts so so you give it a prompt and then it takes your prompt and it sort of
294. 54:01thinks about it and it says well Step One is I should it basically does planning step one I should do this and
295. 54:07then step two I should do that and then I should think about this and then I should think about that and so as it's
296. 54:13writing prompts it's getting back answers and then it looks at that answer and it creates the next prompt and so it
297. 54:19basically goes through this reasoning series of prompts that you don't have to write you basic basically give it a prompt and then it thinks
298. 54:32and it comes back with better answers because it has looked at did I hallucinate whatever deep research takes it one step
299. 54:46further so you've got your basic prompt you've got your reasoning right Your Chain of Thought reasoning and then what
300. 54:55deep re search is doing is it's going out to websites reading a single we so it goes and finds a website it reads the
301. 55:03entire website comes back with a summary and then based on what it found it adds more steps into the um the Chain of Thought
302. 55:14reasoning and then it goes out to another website and it comes back and it adds more and it's and so it's it's
303. 55:20literally doing what a researcher would do right if you worked at Morgan Stanley or you worked at some consulting company
304. 55:28and you would say hey super Highly Educated researcher go find all of the mentions go find all of the books that
305. 55:38mention humans and AI working together and go find all of the movies and go find and then come back and analyze them
306. 55:44and tell me which are the most important ones and tell me what are the themes in them and you know look at go read the
307. 55:50books and watch the movies and do all that [ __ ] or read all the websites or all that stuff's done and then so it
308. 55:56goes off and does that and it can it can think for as much as 30 minutes where it's out looking at websites analyzing it and and then what
309. 56:06it does is it writes up a research report so we'll we'll do it right now so what I'd like to do is give me some
310. 56:13ideas on what I could research and then we'll go we'll go do that we'll just go do one um I'm I'm feeling guilty about um
311. 56:23the YouTubers because they're all they're all Tik Tock jealous there's there's you know they're like
312. 56:28our screen's bigger but Tik tok's cooler so there's some insecurities going on in YouTube so let me just go make sure they're
313. 56:36okay um Rick olssen greetings Kyle hugging face is confusing thank you thank you for validating my like I want
314. 56:45to love hugging face but it is confusing as [ __ ] you think you think chaty PT's confusing go look at the go look at the
315. 56:52model names on hugging face try to figure out what a [ __ ] model is if you're not a programmer um IKR always been where C in a
316. 57:07cond how to make money with chat gbt I didn't bring my button you can make money with um all right let's see research Twi
317. 57:19training within industry I don't know what Twi training is I'm down with that research AI agents good point Frost research AI agents oh
318. 57:30that's kind of interesting kifs on both YouTube and Tik Tock yeah the really cool people are on both if if you're not dual screening Tik
319. 57:40Tok and YouTube you're not hip because what you do you put the YouTube channel up on your 80 inch plasma qled screen that you got it
320. 57:52curved qled screen so I'm up there because you want to see me as big as possible because I mean look at
321. 58:00this and then you you come to Tik Tok for your chat that's that's what the cool kids are doing Danielle's on both
322. 58:08she's you know she's a showoff Vicky's on both K was uncool long before YouTube Vicky Vicky concurs
323. 58:27triple screen valy Valerie Cox say triple screening she doesn't want to miss any comments on X it's a good thing I turned off the uh
324. 58:36the LinkedIn stream um all right idea from last week find a product on ClickBank tell it to create a sales
325. 58:47funnel well I just want to do it I just want to do the research part of it um generates a response in a particular format with without
326. 58:57hallucinating I'm on both hang on I have to do a physical dexterity challenge hey I did it my record so far on deep seek is 17
327. 59:10and a half minutes that's pretty cool AI agent um research the condition that your boys have yeah maybe I should do
328. 59:19that let's let's do that it's gonna be depressing but um okay so let's see um bear with me while I while I type I guess I could use
329. 59:30the talk the talkie talkie typy thing so one thing that I noticed is when you do when you do a deep research
330. 59:40query the next time you query the system it has turned it off so if you want it to keep researching you have to keep hitting the
331. 59:50button the the UI for this is a mess research the logistics of import importing a 100 Irregulars from around the world to an exclusive
332. 1:00:02event in Colorado that would be cool but it when we're going to do an inperson event it's going to be in Vegas by the way because I'm a
333. 1:00:13degenerate oh is Pluto a planet that's a really interesting one let's do okay let's do
334. 1:00:27look for potential cures that AI can create this reminds me the old old days when Kyle took 45 minutes to get back to us sh up sh
335. 1:00:43up Vegas for the win yeah baby I'm going to Vegas [ __ ] [ __ ] Colorado come to Denver Watch the watch the Cowboys um
336. 1:01:02okay let's see here's my my reticence to doing research on stuff that I not so I'm trying to find something that I
337. 1:01:17could get my head around hang on we could do a meeting at Casa Bonita in Denver that's pretty funny doing something at the sphere would be
338. 1:01:30cool all right my record so far 17 minutes we saw that just finished an online AI certification course use a AI
339. 1:01:41to answer answer most of the assignments that of course yeah did did you actually learn it Rob or did you just want the sticker to
340. 1:01:54put on your LinkedIn profile because it's all fun in games until someone actually hires you to do the thing that you used AI to learn I mean
341. 1:02:05for students that's one thing I don't know how old you are Rob I assume if you're on this channel you're probably a
342. 1:02:10little older than a student but I don't give a [ __ ] if you cheat I first of all AI detectors don't work and second of
343. 1:02:18all I don't think it's anyone's [ __ ] business how you answered the questions that's on you you're willing to put your
344. 1:02:23name and say you have that certification and you didn't learn the crap cool go for it empowered fundraiser an Murphy was shaking was shaking
345. 1:02:33lady um welcome back I'm I'm now on my trip you were back from your trip trip now I'm on my trip an and I are behind
346. 1:02:41on our homework assignment for the um are you ready for AI training session the only thing we have to do the only
347. 1:02:49flipping thing we have to do are we talking AI Readiness yeah what I'm talking about I'm shaming us I'm
348. 1:02:55publicly shaming us all we have to do is make videos where we're like hi I'm Kyle I'm an and you know we do the thing but
349. 1:03:03you know an was like I want to take a vacation and I'm like I want to go get business for my company and so we haven't done
350. 1:03:13it no we're playing in with um deep research from uh open AI I have access to it then we're going to research
351. 1:03:22something I got to figure out what the hell we want to research um [Music]
352. 1:04:04H I'm a doc research a medical condition all right Dr bod I'll do that I'll do the I'll do the um I'll do the Lyme disease thing so
353. 1:04:18I've got if you don't know I've got a family full of people with with long haul lime and now all sorts of of other
354. 1:04:26special diseases like neuropathy inspired by the lime um okay I want you to write me a
355. 1:04:40report on as much research as you can find on L haul Lyme disease comma the related co-infections like babesia bonella
356. 1:05:09Etc I want you to find research on related conditions like inflammation neuropathy joint pain autoimmune
357. 1:05:24disease cused by these little [ __ ]
358. 1:05:43period once we learn about the research into the organisms I want you to split your research into two categories one would
359. 1:05:58be treatments for the related conditions and one would be research on treatments that look
360. 1:06:14promising especially focusing on recent advancements accelerated by AI period
361. 1:06:32finally I want you to identify clinical
362. 1:06:51trials whether existing or upcoming that we can look into okay I think that's good so far so
363. 1:07:05good Dr Kyle in the house yeah I feel like I feel like house let's do a differential diagnosis I love that show that guy was
364. 1:07:17so [ __ ] bitter he made me happy all right here we go so thanks for your detailed request oh and then it does this which I think
365. 1:07:27is kind of interesting thanks for your detailed requests to gather the most relevant and comprehensive research on
366. 1:07:33Long hul Lyme disease and its co- infections I'd like to clarify do you want damn there's a lot of questions do
367. 1:07:41you want both mainstream and alternative so let's see one yes both main stream and alternative that's actually a really good question
368. 1:07:55because if you don't know anything about Lyme disease our lovely uh insurance industry um basically wrote the legislation
369. 1:08:06on the treatment for Lyme disease and any doctors that go outside of the six weeks of doxy cycling for Lyme disease
370. 1:08:15uh get sued to have their medical licenses removed so no doctor will touch it so there's this whole underground
371. 1:08:24medical community nothing is covered by insurance but if you want your kids to say you know not [ __ ] suffer um you
372. 1:08:32have to go to this underground so the fact that it asked for that good sign are you interested in specific types of
373. 1:08:38studies uh let's see uh yes both mainstream and alternative all [Music] studies um studies are fine I want a comprehensive
374. 1:08:56um report we'll go two time frame should I focus on recent advancements last 5 10 years uh last five to 10 years is fine
375. 1:09:15since they will likely site other research okay three do you want to focus on
376. 1:09:30conventional medicine um three both and then four clinical trials should I prioritize us um let's
377. 1:09:46see um prioritize us and um no
378. 1:10:02inclusion criteria but I want you to clearly summarize the inclusion
379. 1:10:17criteria C wait cry tier e and even categorize the trials into logical
380. 1:10:32categories so like this is this is pretty remarkable right that that you can like this is what you would ask if you hired someone to go do
381. 1:10:45research for you on stuff let's see wait does YouTube not have hearty Hearts I'm still a Tik Tok refugee it's got thumbs up an you can do
382. 1:10:57thumbs up on tick on YouTube and we now have you can do super stickers or I don't Brandon knows I don't know what they're
383. 1:11:08called this is what the doctor did with with my report oh that's awesome okay cool okay so here we go boom
384. 1:11:27great I'll conduct a deep dive research in the past 5 10 years the report will be divided into these sections research
385. 1:11:34on the organisms covering their biology mechanisms and impact on human health treatments for related conditions
386. 1:11:41mainstream and alternative approaches for inflammation neuropathy joint pain and autoimmune diseases promising AI treatments and
387. 1:11:49advancements clinical trials as I gather information okay so starting research and then let me do so once it starts
388. 1:12:04here it's it's going to pop up a little research thing on the right that's where we'll see its Chain of Thought reasoning
389. 1:12:29oh Danielle suggested a nod to effective pain management in the most recent two years of research okay so there's the thing why am I not
390. 1:12:40seeing the little thing over on the left or on the right maybe I have to do this huh understanding babiosis
391. 1:12:56got it can I click on this yeah there we go okay so here's the activity so this is pretty slick let me let me how how do I want to do this
392. 1:13:13let me just move the window okay so for you ticktockers so here's the activity so I'm piecing together Insights search for bellia Borg Deery
393. 1:13:27immune evasion review yeah these little [ __ ] they they go into these things called bofilm they migrate out of your
394. 1:13:35bloodstream into your muscles and they hide in this [ __ ] called bofilm that you can't get to them so you test negative
395. 1:13:42for them but they're sitting in there like little [ __ ] and then um when your immune system drops they [ __ ] go hey party [ __ ]
396. 1:13:57um so it's doing all that stuff look at all that stuff look at all the work it's doing it's found 15 sources so far so here's the
397. 1:14:06sources the NIH the mdpi frontier science hooks Hopkins medicine neurology Hopkins lime Hopkins is one of the big research centers
398. 1:14:23online we'll just let the activity keep keep rolling so anyway so this is what it does pretty pretty [ __ ] slick man it really
399. 1:14:44is $2 super sticker cool I guess I'm imagining that commercial with the Derma fights green slimy boogers and
400. 1:14:57party hats it's it's very much like that and they and they have oh thank you very much digital Gods I didn't realize
401. 1:15:05you sent me a super sticker I thought you were just telling people how much they were thank you sir very much appreciate the
402. 1:15:11support that's very cool you just shared the article in a regulars okay great yeah so for for Danielle's condition that that her
403. 1:15:20doctor researched she said he basically did what I did the one thing they added was pain management which is that's
404. 1:15:27probably a smart thing to do oh that's pretty cool look at look at this one comment I I love being able to see the thinking of this Chain of
405. 1:15:39Thought reasoning is pretty cool chat gbt look it says it read withp power.com and then the The Prompt that
406. 1:15:48it gave itself I'm curious about the with power s's mention of a trial in Maryland testing an antibiotic for
407. 1:15:56inflammation reduction for long-term Lyme patients I'm looking into specific details for that trial name then it went back to the
408. 1:16:13site so it found one of the lime trials the nct4 38 346 Lim trial at trial x.com like this is stuff I mean you know this
409. 1:16:25is stuff I've been dealing with for a while but like if you want to research this like these are the rabbit holes you run
410. 1:16:30down and it's it takes weeks months and this is just doing it we're at 28 sources oh it wasn't your doctor I was
411. 1:16:43with a doctor with over 300,000 followers on X who has open AI Early Access that's super cool Danielle that's very very cool
412. 1:16:58that's incredible it's incredible I mean and again like I you know I've done this is now my third search using deep research I did
413. 1:17:12um deep research with was it R1 had it I don't know something else had it and that was kind of cool like what it did was cool um
414. 1:17:26I still haven't really played with operator to the point that I understand what it's really good at that's why I was so excited and
415. 1:17:34tagged you that's awesome yeah thanks sorry I've been a little distracted good Tik Tock question can deep research
416. 1:17:42access all scientific papers because not all are available for free right I honestly don't know I am um I
417. 1:17:51assume that most of the research papers that are behind pay walls stay behind pay walls um but I don't know I don't know how it
418. 1:18:01works I don't I I mean I assume it's honoring those pay walls I'm 97% confident that the original llms
419. 1:18:18from all of the companies were trained on data that was scraped from behind pay walls there's two ways to do pay
420. 1:18:27walls the one way is you actually encrypt the [ __ ] and you can't you literally can't access the stuff unless
421. 1:18:35you have the key triggered by your pay wall the other way to do a pay wall is you basically just the shit's all sitting there in
422. 1:18:43folders and you have a pay wall in front of it that just doesn't give you the hyperlinks for all of those pay walls
423. 1:18:51that weren't doing the encrypted the fancy version pay walls uh I'm pretty sure open AI scraped all that [ __ ] and
424. 1:18:59it probably scraped the other [ __ ] too um I I why would you not if if you're if you're out trying to get as much
425. 1:19:07knowledge as you can I'm sure they did all sorts of interarc things as did everyone else I don't think they're just
426. 1:19:13particularly evil Kyle can you give me the tldr on on deep research I used it today by accident yeah so so what it is an is
427. 1:19:24it it's um it's going to be really good for your for your fundraising stuff it's going to be really good for it so I was
428. 1:19:32describing it earlier it's a combination of well that's doing that let me just I'll just pop my my big fat face up
429. 1:19:38there again let me flip this over here um you've got your traditional like chat GPT query you ask chat GPT it gives you an answer then
430. 1:19:53you've got reasoning model models where you give it a prompt and it prompts itself it does Chain of Thought
431. 1:19:58reasoning where it prompts the large language model and it responds to it and it it basically makes a plan right before it gives you an
432. 1:20:08answer deep research takes it one step further and goes out and searches the web summarizes the websites it reads and
433. 1:20:18and is dynamically writing based on what it learns in article one or website one it changes its um Chain of Thought
434. 1:20:29reasoning so it's basically as it's learning it's repr prompting itself and and sort of refiguring out what to do
435. 1:20:36it's out doing what researchers do right let me go look at this website oh like the thing it said where it was
436. 1:20:43like oh it mentioned something about some clinical trial but I didn't get the name of that so let me go back and look
437. 1:20:49at that like it's comprehending what it's reading oh we've got a report let's go back
438. 1:21:04[Music] oh good lord look at our little report damn dag boy
439. 1:21:24[Laughter] Long Haul lime disease a comprehensive literature review research on the organisms bellia Borg
440. 1:21:38Deery lime disease Spyro you ever seen little Lyme disease Spyro they're like little cork screws little [ __ ] swim through your
441. 1:21:48bloodstream and then they can cork screw their way out of your bloodstream a spiral-shaped bacterium transmitted by ticks
442. 1:22:03um tick born coinfections babesia Barella Etc same tick that carries lime can also transmit other pathogens called which often
443. 1:22:13co-infect patients and complicate the illness notable infections babesia M Micra malaria like protozoan bonella Hensley
444. 1:22:29babesia it causes biosis so I know all this stuff this stuff I all know related conditions persistent
445. 1:22:43inflammation neurologic issues and neuropathy that's what my guys are dealing with joint pain and Lyme arthritis autoimmune
446. 1:22:58sequi beyond the joints there's a concern that lme and Co infections May precipitate other autoimmune immune
447. 1:23:16phenomenon the main issue is once you believe they've gotten rid of all these pathogens then your body still interacts
448. 1:23:22with the remnant causing autoimmune diseases yeah I know it's it's a [ __ ] nightmare that needs to go in Notebook
449. 1:23:30LM oh that's a great idea let's let's definitely do that oh this is so cool dudes okay copy go copy this big
450. 1:23:45[ __ ] boom go to notebook LM oh no wait I got to go to docs first let me turn turn this into a Google
451. 1:24:01[Music] doc have you noticed by interacting with
452. 1:24:16AI that you're processing information similar I've noticed that working with AI I am it looks like I crashed my Google doc here did
453. 1:24:30I why can I not paste I've noticed that working with AI I've got way more I'm more creative I have more ideas
454. 1:24:48I have a more active intellect it's it's it's the opposite of what you think it will do it's bizarre because you think I think I I
455. 1:25:03thought um here we're not going to do this here um I thought if if that stuff's doing
456. 1:25:20everything for you that you you'll be less creative you'll be less intellectual you'll be all that stuff
457. 1:25:34wait why it did not copy everything what the
458. 1:25:49[ __ ] oh I see what it it okay wait can I is they've got some weird ass formatting
459. 1:26:03in their um in their documents is there a way that I can hey Kyle yeah there's a copy button at the bottom that you might want
460. 1:26:15to tap at the end of the at the end of the research paper is there oh yeah look at that that's producer Brandon coming in for
461. 1:26:26the win it it doesn't mean it's going to work it's it's still going to probably copy the weird ass uh formatting but let's go
462. 1:26:36see I've also got a Tik Tock pin for you um okay did it work oh yeah it worked all right good
463. 1:26:51job thank you so what Brandon said said each of these trials assistance alternative therapy immunotherapy okay all right so there's
464. 1:27:03a lot of [ __ ] in here good Lord with all these all of these look at all the citations all that yellow [ __ ]
465. 1:27:11are citations [ __ ] amazing all right so now I'm gonna export this as a PDF so this is going to be um oh wait hang on let me get rid
466. 1:27:27of all that [ __ ] okay export as PDF L haul Lyme disease we're going to drop that on the desktop
467. 1:27:43boom did the AI just find all the work that someone else did that answers your prompt well it I asked it to go out and
468. 1:27:55find all of the research like all of all of the research papers on no is the AI cheating off a human no no
469. 1:28:07no no empty bubbles no this this report that it wrote it wrote it went out and it read all of the papers like whatever
470. 1:28:18it went and found it went and read all those websites it read the papers it read the research papers and then it wrote this
471. 1:28:25report this is what you would have a researcher do if you wanted a researcher to give you an update on what's going on
472. 1:28:33with Lyme disease you would pay a researcher to probably spend I don't know two three months six months going
473. 1:28:40and reading all this stuff categorizing stuff putting it together writing it into a thing it wrote this from all the
474. 1:28:49research it did so so just think of this as this is consolidating you know two months three months six months oh I'm being a little bit
475. 1:28:58humorous okay good yeah yeah yeah I figured you I figured you were yeah it's yeah I listen I we're in a weird we're in a weird ass
476. 1:29:11place where people are going to start getting freaked out like if you're say a researcher that spends six months making
477. 1:29:19these things and you watch what I just did you be like um okay we're going to go to notebook LM do uh. goole.com
478. 1:29:31if you haven't seen this by the way oh I didn't say notebook I said note good Lord book LM all right we're going to create a new
479. 1:29:47one we're going to upload our source document which we can't find of course um
480. 1:30:03lime Long Haul lime disease let's make sure it's our thing yep there it is how many pages is it I don't know lot
481. 1:30:17okay so there's that boom Untitled notebook Long Haul lime disease the
482. 1:30:33literature examines Long Haul line focusing on persistent mechanisms of lime disease all right so we're going to go to the
483. 1:30:47studio and we're just going to say generate so what I'm going to do is it generating yeah so I'm going to generate an audio
484. 1:30:59overview it's basically doing a literature review those are very time consum very time intensive reports yeah as someone with
485. 1:31:09ADD I couldn't read the first four paragraphs of one study much less go read all of them it's like it's just unbelievable
486. 1:31:21the world we live in right now it really is is it's pretty pretty remarkable stuff okay where um oh there it is
487. 1:31:38okay Becky Ru my youngest has lyme disease he's 21 well that [ __ ] sucks I'm sorry Becky yeah my guys are
488. 1:31:4725 yeah my guys have had it for I don't know 12 years now sucks it sucks sucks bad what do you think about attacks on
489. 1:32:01llms creating nukes destroying Humanity what about llm security that's a lot of attacks on llms do you mean jailbreaking them I mean you've
490. 1:32:14got I don't you don't really need to jailbreak them you've got open-source models like llama and deep seek where
491. 1:32:22you can now go do whatever you want to do with them you don't you can you can train around their safety guard rails so
492. 1:32:35um what I think about it is you can do the same [ __ ] with existing technology like any technology can use be used for good or for
493. 1:32:48[ __ ] um and you know this channel is about why don't we try using it for for good you know you know like you
494. 1:33:00do we're not we're not making dirty bombs here on this channel um let's see um what are
495. 1:33:14the most exciting new studies that point to an actual
496. 1:33:30[Music] cure we have issues and problems that we keep pointing the finger at each other in any type of technology that we don't understand I
497. 1:33:47agree okay what's it say several new studies persist cell treatments a three dug three drug
498. 1:34:02cocktail of dapto myosin doxy cycline and whatever I think that's from um horor wits I think that's horowitz's
499. 1:34:16thing D suaram traditionally used for alcoholism huh AI accelerated drug Discovery ah here's
500. 1:34:32our thing hang on let me change the thing let's listen to a little of [Music] this is this it yeah that's it
501. 1:34:47yeah lme disease and co- infection oh yeah topic that's uh more incate than a lot of people realize it is you sent
502. 1:34:56over some really fascinating research for this deep dive a comprehensive literature review on Lime the co co-
503. 1:35:03infections and their long-term effects right so let's dive in yeah let's break it all down okay so the research really
504. 1:35:10digs into the bacteria behind Lyme disease yeah brilia bergd door fery what makes this bacteria and modulates the nervous
505. 1:35:19system which can have a positive exactly if hyy a proves effective in human trials it could be a real breakthrough
506. 1:35:27in Lime treatment that's great to hear and there are also advancements being made in treating lime Co infections
507. 1:35:33that's right what are some of the promising developments for babesia and bonella for babia scientists are
508. 1:35:39exploring new antiparasitic drugs okay and even using computational methods to identify potential compounds from
509. 1:35:47natural sources computational methods that sounds Hightech it is artificial intelligence is playing a growing role
510. 1:35:53in drug Discovery wow AI can analyze massive amounts of data to identify potential drug candidates much faster
511. 1:36:01than humans can so AI is speeding up the process of finding new treatments exactly and this is crucial for Lyme and
512. 1:36:07other complex diseases where we desperately need new and effective therapies right and for Bartonella researchers are finding that some
513. 1:36:14existing drugs not what if you have certain research papers as PDFs in one folder how do you ask chat GPT to review
514. 1:36:21them well you can if you've got a bunch I am if you've got a bunch of PDFs the the tool that I'm in right now is
515. 1:36:30notebook LM you can upload I think it's 50 documents and they can be up to 500,000 words each so it's like a
516. 1:36:40ridiculous amount of data you can put in here it's like 25 million words and then you put the the sources
517. 1:36:49in over here you can chat with them in the middle here and then you can turn them into documents over on the right so
518. 1:36:56you you can create study guides a briefing document a timeline um and this audio overview and then check this
519. 1:37:06out Welcome to our Deep dive today we're going to be tackling Lyme disease and co-infection I'm going into into interactive
520. 1:37:20mode Welcome to our Deep dive today we're going to be tackling l mind disease and co-infections oh yeah a
521. 1:37:26topic that's uh more intricate than a lot of people realize it is you sent over some really fascinating research
522. 1:37:33for this deep dive a comprehensive literature review oh hey there how can we help yeah hey I um I'm the one that initiated this
523. 1:37:44research study and I'm doing it because I've got twin 25y old boys who've been dealing with Lyme disease for about 12 years now
524. 1:37:56wow that's um that's really something yeah thank you for sharing that with us he gives us a lot of perspective knowing
525. 1:38:01the personal Drive behind this research absolutely and it's important to keep in mind the real life impacts of this
526. 1:38:07disease we really appreciate your insights and well we'll definitely keep your sons in mind as we go through this
527. 1:38:13yes oh back to our listener what's up yeah so they've got one of the one of the primary things they're struggling
528. 1:38:19with right now is myin sheath neuropathy so is there anything in the study that points to uh ways to deal with
529. 1:38:32that okay so milin sheath neuropathy that's definitely a tough one and it's something that the research touches on
530. 1:38:37though not always by that specific name right and what's fascinating here is how lime can affect the nervous system in
531. 1:38:43various ways yeah it seems that the research talks about how lime can cause damage to the nerves specifically small
532. 1:38:50fiber neuropathy yeah small fiber neuropathy or sfn is characterized by damage to the thin nerve fibers that
533. 1:38:56transmit pain and autonomic signals and from what I was reading this can cause a range of symptoms exactly it can lead to
534. 1:39:01things like widespread pain Burning Sensations and disautonomia like lightheadedness or heart rate issues so
535. 1:39:09while the research doesn't use the term milin sheath neuropathy all right so anyway so that's that if you haven't seen that
536. 1:39:15like as cute as the podcast is it actually does really good analysis of the do doents you put in there and the
537. 1:39:24fact that you can jump into it and ask it questions and have them explain it to you like this is you know we're in a
538. 1:39:32whole other thing all right listen I gotta get going it's it's late here it's late where you are um I'm Crispy and
539. 1:39:39dehydrated um so um yeah so Becky Becky didn't know my kid said Lyme disease yeah I know it sucks and I'm sorry that
540. 1:39:48yours does too like this is this is a club none of us want to be in it's like you get in this club you're like we I've got community um
541. 1:40:02okay couple of things um You probably have all heard of the Webby Awards the Webby awards are um you
542. 1:40:14know awards for websites and [ __ ] like that right they they came out during the hype cycle of the dot bubble
543. 1:40:24everyone is like you know the we the web's exciting Let's do an award show um they just added an AI category so I
544. 1:40:33think they extended the deadline of the Webb's to Friday that's what Brandon told me if it's wrong blame him
545. 1:40:43um but like Joy pie Corey I mean you could maybe I'll maybe I'll submit Sin City sipper no one will care but it's
546. 1:40:56funny um I'm going to do Quantum Cipher I'm GNA do the one I think I'll do the one with uh with uh Kelly Bosch unit maybe I'll do the
547. 1:41:11Civil War project I did I don't know but but anyway go do the go do the wees um okay if you don't know go to
548. 1:41:27[Music] wait go to oh please not banners media assets go to the salon.
549. 1:41:43a see that ticktockers um if you haven't joined the AI Salon go to the salon. click on the join our community button uh um and and
550. 1:41:55and go join it's a community one of the things that's going to be important for everyone in the next two years three
551. 1:42:02years as this stuff gets more and more and more intense 2025 mark my words 2025 is going to be one of those years where
552. 1:42:14um A lot's going to change and this technolog is going to get really good really fast like it it already is like
553. 1:42:23what we did tonight is kind of [ __ ] Bonkers what it just did and that was like that was without even really trying right that was just
554. 1:42:34like me sort of pulling a prompt out of my butt and just kind of asking it for what I thought would be good imagine if
555. 1:42:40you actually put you know like a week into using that tool um a lot of shit's going to change um there's going to be a lot of disruption
556. 1:42:52so get your ass in community the AI Salon is a community of AI optimists and creators who are generous and um helpful and smart and
557. 1:43:06accomplished and humble something you don't get [Laughter] here in spite of me there're a really humble bunch there's not a bunch there's
558. 1:43:22not a lot of ass holes in there they're like sweet empathetic supportive human beings that are trying to figure this AI
559. 1:43:30[ __ ] out too and they happen to be some of the people doing the most interesting work not building the AI we got a couple
560. 1:43:38of those we got like Pate who's building this [ __ ] but but most of the com the the community of the AI Salon it's like
561. 1:43:463,000 people are people that are there to just try to figure out how do we use these tools how do we use them in our
562. 1:43:52Liv how do we use them to help do the [ __ ] we do so go do that tomorrow night is the first Tuesday of every
563. 1:44:01month we have a meeting called AI Salon presents and tomorrow night we're going to have a return guest we had an artist named black
564. 1:44:12Blac um who's an nft artist he was a he was a consultant for many years business consultant he was really good at Art and
565. 1:44:21he got way into digital art got way into nfts all that sort of stuff and now he's coming back around he's he's done all
566. 1:44:28that and he's going to talk about coming back around and now he's starting an AI consultancy which I think is absolutely
567. 1:44:34fascinating so um so that's tomorrow night from 5:00 pm to 7 PM mountain time um Brandon and Eric Sanner uh are going
568. 1:44:45to be co-hosting the event Leah can't make it and I might be able to make it but I don't think I can because I'm
569. 1:44:53doing business [ __ ] um so so we're going to have guest co-hosts cool and we're going to have black speaking and there's
570. 1:45:00just so much news has happened you should go to that thing and you should meet the people there and join the salon
571. 1:45:07and be cool all right that's it I'm gonna get out of here um I've got a cool I think I'm gonna take that um that
572. 1:45:17podcast and publish it um I'll I'll put that up on LinkedIn and I'll put it up on uh Twitter X you know the thing all right everybody
573. 1:45:32Coolio peace out I will be back here tomorrow night um yeah it's I'll be back here I'll probably be a guest on the
574. 1:45:44salon presents and and then I'll do this probably normal time 8:00 p.m. mountain but I'll be it'll be 10 p.m. East Coast
575. 1:45:54why do I do this to people why do I start these things at 10 o'clock at night for East Coasters I'm an [ __ ] that's why okay
576. 1:46:04everybody has deep seek put every other competitor out of business no Snowman no no no the Deep seek thing was completely
577. 1:46:13blown out of proportion is it a good model yes did they train it for $6 million no it looks like it was closer to 1.5 billion it looks like
578. 1:46:23um deep seek has about 50,000 h100s which they're not supposed to have so that was that was some uh how
579. 1:46:34you call it in America a little bit of misinformation to drive the stock market crazy no um quen came out with a better
580. 1:46:42model three days later um open AI just came out with this deep research thing and and 03 mini and 03 mini high and no
581. 1:46:50it's that that is a small blip in a much larger War um it's like that I don't I wouldn't even consider that a battle did
582. 1:46:58it did it drain whatever it was 1.2 trillion out of the markets in a day yeah it did do that is that going to be
583. 1:47:07a long-term impact I don't think so but we'll see who knows I don't [ __ ] know anyway all right grock 3 will be very late
584. 1:47:15yep um all right peace out everyone have a good one I'm going to say goodbye to the YouTubers and xers peace out
585. 1:47:23everyone hope you had fun tonight