1
00:00:00,000 --> 00:00:01,920
Like you look behind me, you see an outlet.

2
00:00:02,780 --> 00:00:04,780
You see a little cat hair on my hat.

3
00:00:05,740 --> 00:00:06,700
This is me.

4
00:00:06,840 --> 00:00:09,120
This is not some, I'm not freaking AI.

5
00:00:09,640 --> 00:00:13,200
And people know that I'm not freaking AI because of those things.

6
00:00:15,520 --> 00:00:16,200
All right.

7
00:00:16,900 --> 00:00:21,360
That's one of the key things because people, they find out it's AI, they just turn it off.

8
00:00:30,000 --> 00:00:53,680
all right everybody welcome back to another episode of the space tomatoes podcast i'm one

9
00:00:53,680 --> 00:00:59,240
of your hosts havoc i'm the other guy lx hello welcome back all those things

10
00:00:59,240 --> 00:01:02,640
And today we have a repeat guest back with us.

11
00:01:02,800 --> 00:01:08,720
It's Russell Van Brocklin, a dyslexia professor out of New York.

12
00:01:09,440 --> 00:01:10,540
Albany, New York.

13
00:01:10,640 --> 00:01:12,900
I'm nowhere near that crazy place downstate.

14
00:01:14,240 --> 00:01:15,440
It's not all the same.

15
00:01:16,500 --> 00:01:18,540
I thought the whole state was just that.

16
00:01:20,480 --> 00:01:22,240
I live in a small town.

17
00:01:22,380 --> 00:01:24,420
I'm looking out on a lake right now.

18
00:01:24,620 --> 00:01:27,140
I'm not down in the crazy place in New York City.

19
00:01:27,740 --> 00:01:28,580
Oh, good.

20
00:01:28,580 --> 00:01:31,300
I don't know how people do that, honestly.

21
00:01:31,460 --> 00:01:33,260
There's so much stimulus there.

22
00:01:33,740 --> 00:01:34,500
I can't.

23
00:01:34,820 --> 00:01:38,580
It is amazingly uncomfortable for me.

24
00:01:39,500 --> 00:01:43,460
I go down there to present it to everyone reading conference every year.

25
00:01:44,320 --> 00:01:47,320
And let me just give you a quick story about New York City.

26
00:01:47,880 --> 00:01:51,500
A couple of years ago, I really tried to get good hotel deals.

27
00:01:52,260 --> 00:01:59,600
There was a Hyatt place, like a block away from the City University of New York Graduate Center where the conference was held.

28
00:01:59,900 --> 00:02:02,380
I paid $113 all in.

29
00:02:03,520 --> 00:02:04,060
All right.

30
00:02:04,240 --> 00:02:05,320
Really comfortable.

31
00:02:05,500 --> 00:02:09,040
People were shocked I could afford the place because they didn't know how I paid.

32
00:02:09,100 --> 00:02:13,120
When I told them how much I paid, they went crazy because they were paying a lot more for a lot worse place.

33
00:02:13,620 --> 00:02:14,880
I got up the next day.

34
00:02:14,980 --> 00:02:16,300
I had a really good breakfast.

35
00:02:16,620 --> 00:02:18,080
It's an extraordinary hotel.

36
00:02:18,680 --> 00:02:19,880
It's two and a half stars.

37
00:02:19,880 --> 00:02:22,480
But for that, it's the best third in the two and a half star range.

38
00:02:22,960 --> 00:02:24,400
Then I tried to book it the next year.

39
00:02:24,820 --> 00:02:27,460
Before taxes, it was $429.

40
00:02:28,720 --> 00:02:29,360
Oh, yeah.

41
00:02:29,520 --> 00:02:31,720
Which is the normal price for the place.

42
00:02:32,700 --> 00:02:33,380
So, yeah.

43
00:02:34,520 --> 00:02:36,300
Not to mention, I didn't do that.

44
00:02:36,660 --> 00:02:38,040
I went down in the morning.

45
00:02:38,360 --> 00:02:40,400
I did what I had to do, and I came back.

46
00:02:40,460 --> 00:02:42,160
I just wasn't going to spend that type of money.

47
00:02:42,700 --> 00:02:44,840
Just grabbed the Accela Express, right?

48
00:02:45,560 --> 00:02:46,780
Oh, we don't get that.

49
00:02:46,820 --> 00:02:47,880
I've come down from Albany.

50
00:02:48,000 --> 00:02:48,640
So here's the thing.

51
00:02:48,640 --> 00:02:57,560
The state capital of New York, its unofficial nickname is Smallbany because there's one street that's like a half a street that looks like a city.

52
00:02:58,820 --> 00:03:02,960
But the Albany Rensselaer train station is one of the 10 busiest in the country.

53
00:03:03,780 --> 00:03:08,080
So we had the head of the Senate for about 15 years ago, and he rebuilt it.

54
00:03:08,420 --> 00:03:14,320
It's magnificent, but it takes two hours and 40 minutes to go down, two hours and 40 minutes to come back.

55
00:03:14,580 --> 00:03:17,360
And it's a long day to do that.

56
00:03:17,360 --> 00:03:25,300
oh yeah that is yeah for sure absolutely yeah i've got my own hotel stories i i i remember my

57
00:03:25,300 --> 00:03:30,040
first time in new york on business travel so it wasn't my money but the hotel was like

58
00:03:30,040 --> 00:03:38,500
600 plus dollars this is well over a decade ago and i come into this courtyard by marriott

59
00:03:38,500 --> 00:03:49,080
it's it's in time 600 bucks it's it's in times square and i go and check into my room i open the

60
00:03:49,080 --> 00:03:56,380
door and i see a bed and i walk in and there's literally a bed and about a foot of clearance on

61
00:03:56,380 --> 00:04:02,260
each side of the bed okay there's there's a little stand-up wardrobe that you can't open

62
00:04:02,260 --> 00:04:09,260
the doors to all the way because they contact the bed. But anyway, so that was my first experience.

63
00:04:09,260 --> 00:04:15,940
And I remember very clearly I'd slept maybe 20 minutes that night because of the noise and the

64
00:04:15,940 --> 00:04:21,260
light. And even with the blackout curtains closed, it was just, I was blown away.

65
00:04:21,540 --> 00:04:26,480
Anybody going there now, there's something called a pod hotel. It's much cheaper. It's around 100,

66
00:04:26,580 --> 00:04:31,740
150 a night. All right. Get a room away from the street, pay retail. Don't go through a discount

67
00:04:31,740 --> 00:04:37,080
sites so you can get a room away from the street. And it's a shoe box, but it's much cheaper.

68
00:04:38,420 --> 00:04:44,260
All right. What you want to do is put a noisemaker on your fan and put that up and put in those

69
00:04:44,260 --> 00:04:50,640
earplugs that do the 33 decibel things. That works. But today you wanted to talk about

70
00:04:50,640 --> 00:04:57,140
artificial intelligence. Absolutely. We'd love to hear how you're making use of that in your work,

71
00:04:57,140 --> 00:05:03,820
how you're using it to improve your life and, you know, how that is used to help people with,

72
00:05:03,820 --> 00:05:07,260
you know, learning disabilities, dyslexia in particular.

73
00:05:07,920 --> 00:05:12,020
Well, just so everybody knows what it is that I do with this.

74
00:05:13,740 --> 00:05:17,080
I'm giving you an example of a student who I'm working with. His name is Grayson.

75
00:05:18,900 --> 00:05:22,420
Like you tell me how old you think he is. Grayson's not dyslexic in the least.

76
00:05:23,380 --> 00:05:25,880
He's in the 99th percentile in math and science.

77
00:05:26,000 --> 00:05:30,740
He wants to get his PhD and work at NASA to terraform Mars.

78
00:05:31,820 --> 00:05:37,000
Grayson, we found an academic journal where he needs to publish in to do that.

79
00:05:37,000 --> 00:05:46,880
And they do accept submissions from junior researchers, which they pretty much peg as PhD candidates.

80
00:05:47,380 --> 00:05:49,080
How old do you think Grayson is?

81
00:05:52,420 --> 00:05:54,400
I'm going to go with 16.

82
00:05:54,960 --> 00:05:55,580
He's 10.

83
00:05:56,540 --> 00:05:57,280
Oh, wow.

84
00:05:58,800 --> 00:06:00,260
That's quite amazing.

85
00:06:00,720 --> 00:06:04,720
Yeah, we're hoping to have him submit an article by the time he's 14.

86
00:06:05,780 --> 00:06:06,000
Okay?

87
00:06:06,420 --> 00:06:09,680
So AI is going to be critical in that aspect.

88
00:06:10,340 --> 00:06:13,700
But I'm not going to tell you how I deal with students because they're very restrictive.

89
00:06:13,700 --> 00:06:23,780
But for adults, I just had a transformation that I'm seeing all these people saying that they're having AI do things for them overnight.

90
00:06:23,960 --> 00:06:25,820
I'm like, what the hell?

91
00:06:26,120 --> 00:06:26,900
What is this?

92
00:06:27,320 --> 00:06:31,040
So the first thing that I need to tell you is just my background with AI.

93
00:06:31,980 --> 00:06:38,940
When on December of 2024, ChatGPT had its 12 days of Christmas, whatever stuff.

94
00:06:38,940 --> 00:06:43,260
and they came out with ChatGPT Pro, which at that point was $200.

95
00:06:43,780 --> 00:06:46,640
And people were shocked because they're used to $20 a month.

96
00:06:46,840 --> 00:06:48,040
I bought it that day.

97
00:06:48,180 --> 00:06:51,520
I used it through December, and then I canceled it,

98
00:06:51,540 --> 00:06:56,540
and I moved over to Anthropic because I found the $20 thing on ChatGPT

99
00:06:56,540 --> 00:07:01,760
Thinking was I got it up to 5, 10 minutes, and that's all I really needed.

100
00:07:01,760 --> 00:07:06,160
I was like, I'm not going to spend another 10 times that amount

101
00:07:06,160 --> 00:07:07,540
when I got good enough for me.

102
00:07:07,540 --> 00:07:16,080
so and then I finally about a week ago I said I want to do what these top one percent people are

103
00:07:16,080 --> 00:07:21,500
doing and it's not exactly what top one percent but to me it's close enough I'm not one of those

104
00:07:21,500 --> 00:07:29,160
guys that is killing myself to be on AI eight ten hours a day I'm doing it to do twice three times

105
00:07:29,160 --> 00:07:33,560
as much work as what a normal person would do and I'm using the rest of the time to you know

106
00:07:33,560 --> 00:07:40,500
have a life and to do podcasts and that sort of thing. So here's what I basically did. First thing

107
00:07:40,500 --> 00:07:46,540
that you're going to need to do is I'm going to tell you, these are the 10 things that you're

108
00:07:46,540 --> 00:07:52,900
generally going to need to do. And you're going to want to replay this to write all this stuff down,

109
00:07:52,960 --> 00:07:58,180
but here they are. These are the generally the top 10 things that the top 1% generally do.

110
00:07:58,180 --> 00:08:24,400
They treat AI like a team member. It's not a freaking search engine. All right. So you're using this to do real work, to have agents do real work. I do that through coworker. That is Claude coworker. But my instructions and the thinking, I use chat GPT 5.4 thinking extended. Okay.

111
00:08:24,400 --> 00:08:30,660
Okay, number two, you build memory and continuity.

112
00:08:31,160 --> 00:08:38,080
So what I mean by that is you need to get on Claude Opus 4.6.

113
00:08:38,200 --> 00:08:39,080
That's their best model.

114
00:08:39,640 --> 00:08:43,040
Don't, and yes, you're not going to get away with 20 bucks a month.

115
00:08:43,040 --> 00:08:46,520
After you start getting used to this, you're probably going to spend at least 100.

116
00:08:47,180 --> 00:08:52,360
But this is why you need to get on Opus 4.6 for at least a month.

117
00:08:52,860 --> 00:09:08,200
What I did is I've been on it for several months, and this has allowed me, when I had to start asking questions, unlike any other AI platform, it can go and search and see what I'm doing right, what I'm doing wrong, and it can actually answer the questions.

118
00:09:08,620 --> 00:09:12,200
So, like, for example, I had a client that I have to do customized work for.

119
00:09:13,020 --> 00:09:17,520
He wants to be – it's a college kid who wants to go into sports recruiting.

120
00:09:18,100 --> 00:09:22,780
So I said, okay, I did a fourth Harvard Business School case study like him.

121
00:09:23,060 --> 00:09:24,020
Pull all those up.

122
00:09:24,280 --> 00:09:25,960
Now, what do you think the fifth one should be?

123
00:09:26,140 --> 00:09:28,020
And we go back and forth with that.

124
00:09:28,940 --> 00:09:31,180
Okay, you can just remember things nobody else can.

125
00:09:31,840 --> 00:09:36,540
All right, number three, you put tasks together.

126
00:09:37,400 --> 00:09:40,360
Okay, don't do this little thing, this little thing, and this little thing.

127
00:09:40,360 --> 00:09:47,280
I'm trying to combine tasks so I can get the agent to run about 30 to 45 minutes.

128
00:09:47,280 --> 00:09:58,520
I find that's about the optimum time before I recheck in and run it through ChadGBT 5.4 thinking extended.

129
00:09:59,380 --> 00:10:02,240
And then what I do there is I'll put it back there and say, how do we do?

130
00:10:02,720 --> 00:10:04,260
What changes do you want to see?

131
00:10:04,380 --> 00:10:10,660
Okay, now tell me in an exhausted narrative every change you want to see to the smallest detail.

132
00:10:10,660 --> 00:10:19,600
then I'll say for the prompt, give me a prompt to have Opus 4.6 co-work actually do this and write

133
00:10:19,600 --> 00:10:25,520
it out so it covers everything, but look at what the best engineering prompt, before I have it run

134
00:10:25,520 --> 00:10:31,300
any prompt, I don't expect it to remember things. I say, look it up again. What are the best steps

135
00:10:31,300 --> 00:10:35,960
to write the most effective prompts? And then it does that and I have them combine it. And then I

136
00:10:35,960 --> 00:10:40,180
just send it back to co-work and it runs it again. And then I just go back and forth until

137
00:10:40,180 --> 00:10:47,480
chat GPT 5.4 thinking extended thinks that it's done. All right. Other thing, number four,

138
00:10:48,200 --> 00:10:55,040
people use AI to prepare, not just to produce. So before I get on any podcast,

139
00:10:55,840 --> 00:11:02,080
I will actually ask like 12 questions that as 12 standard questions that I came up and I'll say,

140
00:11:02,080 --> 00:11:06,980
run this through. I'll take literally our pod match discussion and I'll dump it in. I'll say,

141
00:11:07,080 --> 00:11:13,080
run through this so I actually know what the podcaster is generally looking for,

142
00:11:13,160 --> 00:11:18,740
where they're coming from, so I can take what I'm looking to do and customize it to their needs.

143
00:11:19,520 --> 00:11:25,720
Okay. The fifth thing is people do is they create reusable prompts and systems.

144
00:11:26,560 --> 00:11:36,580
So if you're doing the same thing over and over again, you want to go and you want to put this into a project folder and keep all of this stuff here.

145
00:11:37,040 --> 00:11:42,380
So you can just say, this is what we're going to do this again, run it through this.

146
00:11:42,380 --> 00:11:44,700
And it already has all that memory.

147
00:11:44,820 --> 00:11:49,360
So you don't have to keep retyping the same freaking thing in again and again.

148
00:11:49,360 --> 00:12:00,220
And number six, people correctly use AI and they make it to help them learn.

149
00:12:00,780 --> 00:12:03,360
And they have AI learn from their own mistakes.

150
00:12:04,020 --> 00:12:13,300
So, for example, after you get done with these 10 things, what I would do is I took this over to ChatGPT 5.4 Extended Thinking.

151
00:12:13,300 --> 00:12:21,220
and I would say, I want you to come up with a list of questions that can help me figure out what I'm

152
00:12:21,220 --> 00:12:27,440
doing right and what I'm doing wrong. And it came up with 196 questions over 16 areas. Then this is

153
00:12:27,440 --> 00:12:34,480
why you have to spend at least a month in Opus 4.6. I went back to Opus 4.6 and I said, I want you to

154
00:12:34,480 --> 00:12:38,740
answer these questions, look through all our conversational history. Again, the other AI,

155
00:12:38,740 --> 00:12:43,440
the other ones can't do this. And it came back and answered them. Then I would take those answers

156
00:12:43,440 --> 00:12:51,840
and I went back to ChatGBT 5.4 Thinking Extended and I said, here's the answers you wanted. Does

157
00:12:51,840 --> 00:12:55,060
that change the questions you want to ask for the next group? And they would say yes.

158
00:12:55,440 --> 00:12:59,160
And then I'd put those next ones in, but I'd put them in 10 questions at a time.

159
00:12:59,520 --> 00:13:03,360
And then I would bring it back. And I kept doing that until it created,

160
00:13:04,520 --> 00:13:08,180
until we went through about half the questions. And then I opened up a new conversation.

161
00:13:08,180 --> 00:13:19,360
but in the same project. And then I would ask Opus 4.6 to take everything and to put it into

162
00:13:19,360 --> 00:13:26,920
a Word document. So I have two sets of questions and each one's about 60,000 words. All right.

163
00:13:26,960 --> 00:13:33,220
Once that was done, I would then take, I would then put it all that into a new

164
00:13:33,220 --> 00:13:41,360
chat GPT 5.4 extended thinking, and I would say a new window, and I'd say, go ahead and create this.

165
00:13:41,900 --> 00:13:46,680
Then I would create my process, and then I would ask, then I would test it out, like I'm testing

166
00:13:46,680 --> 00:13:51,220
it out on my book that's coming out, and I'd test it out a few times, make the recommended changes,

167
00:13:51,220 --> 00:14:00,720
then everything that I do runs through that. All right. Number seventh, people use AI for the

168
00:14:00,720 --> 00:14:06,040
boring stuff so that they don't focus on the human stuff, okay?

169
00:14:06,500 --> 00:14:08,760
Data entry, schedule, reframing.

170
00:14:09,120 --> 00:14:11,740
Spend all the time doing this boring stuff.

171
00:14:12,140 --> 00:14:16,900
Have ChatGPT, sorry, have Opus 4.6 co-work do this for you.

172
00:14:17,460 --> 00:14:25,700
All right, so for example, I have to go and set up a time to set up a podcast.

173
00:14:25,700 --> 00:14:27,080
I pick the time.

174
00:14:27,320 --> 00:14:29,860
So what I do is I basically use this workflow.

175
00:14:29,860 --> 00:14:41,660
I will tell Chachi, I'll tell a co-work on Opus to find days that I have nothing going on or one thing going on.

176
00:14:42,140 --> 00:14:46,140
And then I'll say match it up with what the podcaster has available.

177
00:14:46,580 --> 00:14:49,520
And then I would pick a day and I spread things apart.

178
00:14:49,620 --> 00:14:55,460
I want to have between noon and four o'clock free so I can go to the YMCA and swim for an hour, hour and a half.

179
00:14:55,920 --> 00:14:56,560
All right.

180
00:14:56,560 --> 00:15:00,420
So it'll find those dates and then I'll find a time that fits for me.

181
00:15:00,480 --> 00:15:02,440
Then I pick that, then I tell it to book it.

182
00:15:03,160 --> 00:15:08,580
And honestly, it takes a while to do this, but I can have four or five things going at

183
00:15:08,580 --> 00:15:09,120
the same time.

184
00:15:09,160 --> 00:15:11,040
So if it takes longer, who cares?

185
00:15:11,540 --> 00:15:12,660
It gets it done.

186
00:15:12,820 --> 00:15:17,720
That allows me to spend more time actually being on podcasts and speaking to people.

187
00:15:18,520 --> 00:15:18,860
All right.

188
00:15:19,340 --> 00:15:22,720
People also combine AI with their own expertise.

189
00:15:22,720 --> 00:15:29,560
piece. So what you want to do is find out what is your speciality? What is your area of extreme

190
00:15:29,560 --> 00:15:36,320
interest and ability? What are you really, really good at? And you want to feed that into specific

191
00:15:36,320 --> 00:15:42,980
projects. This is how, this is your secret sauce. This is what you're better at. And that's why

192
00:15:42,980 --> 00:15:49,000
you want to use this for Opus 4.6 for at least a month, because as you're going through these,

193
00:15:49,000 --> 00:15:54,800
quite frankly, when I was done, it was over 400 questions. It can take your unique skill set and

194
00:15:54,800 --> 00:16:00,000
everything you have and put it into this one general service document. For me, that runs about

195
00:16:00,000 --> 00:16:07,940
5,000 pages. Number nine, it allows you to move faster than everybody around you. So a lot of

196
00:16:07,940 --> 00:16:13,600
people are literally working at this more than less. They used to put it eight hours days. Now

197
00:16:13,600 --> 00:16:18,820
they're putting in 10, 12 hour days. They're being five, 10 X more effective, more productive than

198
00:16:18,820 --> 00:16:25,820
everybody else. Honestly, I don't do that. I use this to do two, three times as much work as most

199
00:16:25,820 --> 00:16:31,300
people do in eight hours, five years ago. And I take the rest of the time to have, you know, a life.

200
00:16:32,140 --> 00:16:38,920
All right. And finally, people are not afraid to look weird using it. Okay. What do I mean by that?

201
00:16:39,740 --> 00:16:44,920
Opus just came out with this thing and I'm just starting to get to get to use it where you can

202
00:16:44,920 --> 00:16:52,160
actually tell it, or I can actually tell it on the phone to keep doing things. So I'll be out at a

203
00:16:52,160 --> 00:16:56,900
social event. I'll go in the corner, get on my phone, and I'll say, Opus, okay, go do this, and

204
00:16:56,900 --> 00:17:01,680
I'll run for 45 minutes, and I'll check on it for five minutes and tell us the next step. People ask

205
00:17:01,680 --> 00:17:06,960
me what I'm doing. I said, well, I'm getting a lot of work done, and yes, I'm taking five minutes out

206
00:17:06,960 --> 00:17:16,060
of every 45 to do that, but it just allows so much more stuff to get done. So that's the general

207
00:17:16,060 --> 00:17:28,900
overview. All right. So you're using it really to tailor documentation and I mean, essentially

208
00:17:28,900 --> 00:17:32,880
you're using it as a personal assistant, but also as something that will help you

209
00:17:32,880 --> 00:17:42,960
gather research, tailor a document and process possibly, and put that out. Is that a correct

210
00:17:42,960 --> 00:17:49,320
assumption? Yeah, let me try to, I was kind of, being the dyslexic I am, I was trying to go through

211
00:17:49,320 --> 00:17:54,300
the 10 rules and the 10 major things and discuss it along the way so it gets kind of scattershot.

212
00:17:54,820 --> 00:18:01,180
What I basically did is you take those 10 rules that I just gave you and you type them out,

213
00:18:01,180 --> 00:18:12,990
And then you go to ChatGPT 5 thinking extended Yes you need to spend a month on this And then leave it not on thinking but extended thinking

214
00:18:13,110 --> 00:18:16,970
And people are going to shout out, but it took seven minutes to get an answer.

215
00:18:17,270 --> 00:18:18,770
Yes, but it's a good answer.

216
00:18:19,930 --> 00:18:22,910
I'm doing this for work, not a hobby, okay?

217
00:18:23,450 --> 00:18:28,090
So it goes ahead, and I have multiple ones of these things going.

218
00:18:28,090 --> 00:18:31,230
I usually do three things at once, typically, at least two.

219
00:18:31,230 --> 00:18:50,090
All right. So while it's doing its thing, I'm doing other things. I'm just not sitting there waiting for it. So as it goes through, then I say, how many questions do you have for me to go through this? And I use, I use Anthropic Claude Coworker to do the work.

220
00:18:50,090 --> 00:18:57,790
but I use chat gpt 5.4 thinking extended to do the thinking and to write the prompt

221
00:18:57,790 --> 00:19:05,890
okay so for me it came up to a list of 196 questions over 16 areas I don't know how much

222
00:19:05,890 --> 00:19:11,050
it'll come up for you but you want to be on opus 4.6 for at least a month then you take those

223
00:19:11,050 --> 00:19:16,150
questions you put them in 10 at a time and you have them answer it and then once you're done

224
00:19:16,150 --> 00:19:21,330
answering all the questions in that section, bring it back to ChatGPT for 5.4 extended thinking

225
00:19:21,330 --> 00:19:25,630
and say, here's the answers you wanted. Do you want to change the next set of questions,

226
00:19:25,630 --> 00:19:30,970
which it will? And you do that until the whole thing is done. I had to split it up into multiple

227
00:19:30,970 --> 00:19:36,090
conversations because it was just getting too long. Once you have all the answers, then you say,

228
00:19:36,190 --> 00:19:42,210
okay, go ahead and create my customized process. And it'll do that. Then you need to test it out

229
00:19:42,210 --> 00:19:48,270
on some of your documents. I did a three test before I started using this. Now, every time I do

230
00:19:48,270 --> 00:19:57,670
anything, I put in this 5,000 word document. It knows the mistakes that I normally make. It knows

231
00:19:57,670 --> 00:20:06,070
what my strong points are. And it just keeps me from making the same stupid crap over and over

232
00:20:06,070 --> 00:20:13,150
again. So when I have a customized document that I need to create, this goes through and it gets it

233
00:20:13,150 --> 00:20:19,830
much closer. Now, once the document is done, then I bring it back to 5.4 extended thinking on

234
00:20:19,830 --> 00:20:25,450
ChatGPT and I'll say, how did we do? And then it'll tell me and I'll say, okay, what changes do you

235
00:20:25,450 --> 00:20:31,150
want to make? Tell me in an exhaustive narrative to the final detail to get to 100%. And it will do

236
00:20:31,150 --> 00:20:38,070
that, that every time I'll tell it, redo internet research to figure out how to write the best

237
00:20:38,070 --> 00:20:46,270
prompt instructions for, in this case, chat GP, Claude Cowork using Opus 4.6. All right, then it'll

238
00:20:46,270 --> 00:20:51,650
go ahead, figure that out, apply what it wrote out exactly and write out the prompt. I copy it,

239
00:20:51,750 --> 00:20:56,290
I dump it into Cowork, and I have to check on it every so often because sometimes, you know,

240
00:20:56,390 --> 00:21:01,110
Cowork wants to ask permission every two seconds, which I tell it not to do. I just say, go do it

241
00:21:01,110 --> 00:21:03,630
It still does it so often until it's done.

242
00:21:03,690 --> 00:21:05,990
And then I run it back through chat GPT until it's done.

243
00:21:06,070 --> 00:21:06,790
Then I ship it out.

244
00:21:08,330 --> 00:21:08,810
All right.

245
00:21:08,810 --> 00:21:17,370
So how do you, how do you counteract what, so, you know, as part of work, I use AI quite

246
00:21:17,370 --> 00:21:17,890
frequently.

247
00:21:20,330 --> 00:21:24,790
It's kind of ingrained in what we do and yada yada.

248
00:21:24,790 --> 00:21:34,790
But how do you counteract, especially with this much information being, you know, gathered together and put into a format?

249
00:21:34,790 --> 00:21:46,750
How do you counteract the AI hallucinations where it will, hey, I can't find an answer, but this is similar, so I'll go ahead and just create an answer out of this.

250
00:21:46,910 --> 00:21:54,570
Or maybe I got a hold of some reference that wasn't related to the topic at hand, but it seems similar enough.

251
00:21:54,610 --> 00:22:02,910
And I'm going to give you a very confident answer about something that is absolutely incorrect.

252
00:22:03,830 --> 00:22:08,890
I can't even get it to tell me what battery my key fob on my car remote uses sometimes.

253
00:22:08,890 --> 00:22:10,310
It tells me the wrong one.

254
00:22:10,350 --> 00:22:12,610
And then I open it up and like, that ain't what it said.

255
00:22:13,390 --> 00:22:19,210
Well, to be honest with you, number one, I'm an expert in what I do.

256
00:22:19,710 --> 00:22:21,590
So I just run it through.

257
00:22:21,810 --> 00:22:27,270
I don't even look at it until ChatGPT 5.4 thinking extended.

258
00:22:27,770 --> 00:22:32,130
And honestly, when I do the extended thinking and then I have the work done on a different

259
00:22:32,130 --> 00:22:38,750
AI, between the two, I'm finding the hallucinations are down to virtually nothing. Okay. But as I'm,

260
00:22:38,890 --> 00:22:44,250
but after it says it's done or close enough, I get it. When it gets to about 95%, I just stopped

261
00:22:44,250 --> 00:22:51,330
working on it. Then I read it over. And if I spot something, I'm like, yeah, good one. You missed

262
00:22:51,330 --> 00:22:57,130
this. Or yeah, where the heck did this happen? And then I'll let it know. And it's like, yeah,

263
00:22:57,130 --> 00:23:02,110
my bad. And it goes and tells it, we go through the process until it's fixed. But again, I don't

264
00:23:02,110 --> 00:23:09,210
do the final review until it says it's ready to go. But most of the time, honestly, it's maybe one

265
00:23:09,210 --> 00:23:16,790
out of one in 10, two in 10, that I have to call it out on any of that stuff. So you're the human,

266
00:23:16,890 --> 00:23:22,590
you just can't ship what the AI told you. But I get it to what it calls essentially a final draft,

267
00:23:22,650 --> 00:23:27,630
and then I send it out. The other thing is, people working with me know that I'm dyslectic.

268
00:23:27,630 --> 00:23:29,870
I'm just selective as hell.

269
00:23:30,350 --> 00:23:30,550
Okay?

270
00:23:31,070 --> 00:23:33,290
So I ship, the people are happy

271
00:23:33,290 --> 00:23:34,130
what I'm sending them,

272
00:23:34,270 --> 00:23:35,890
but a lot of times,

273
00:23:35,970 --> 00:23:37,210
like when I'm doing anything

274
00:23:37,210 --> 00:23:40,610
on the internet or stuff,

275
00:23:40,650 --> 00:23:43,550
I will just go and have it change.

276
00:23:43,670 --> 00:23:45,070
And I have a web expert

277
00:23:45,070 --> 00:23:47,930
and he checks it twice a week.

278
00:23:48,410 --> 00:23:50,150
And, you know, any little minor changes,

279
00:23:50,390 --> 00:23:52,010
he'll fix.

280
00:23:52,350 --> 00:23:54,290
I kind of got that idea from Cursor.

281
00:23:54,290 --> 00:23:56,790
They just ship things like every five hours.

282
00:23:56,790 --> 00:24:02,110
Like you start working with them at breakfast, the same model is different when you're doing things at dinnertime.

283
00:24:02,390 --> 00:24:04,410
It literally changes every five hours.

284
00:24:05,070 --> 00:24:11,450
So the little stuff that they fix when I'm doing online courses, I ship it to the instructors who are teaching it.

285
00:24:11,870 --> 00:24:16,470
And they tell me that the stuff I'm shipping them now is like 85%, 90% there.

286
00:24:16,670 --> 00:24:21,270
And they make whatever changes that they want to finally fix things.

287
00:24:21,850 --> 00:24:24,310
Humans have to be the final people to look around.

288
00:24:24,310 --> 00:24:33,070
But for me, I'm under the idea of just make sure it's run 85%, 90%.

289
00:24:33,070 --> 00:24:35,230
Ship it out there.

290
00:24:35,450 --> 00:24:42,210
And if one of my customers is going to complain because I didn't put a comma in the right place, well, fine.

291
00:24:42,570 --> 00:24:43,530
Go to my competitors.

292
00:24:43,750 --> 00:24:44,510
Have a nice day.

293
00:24:45,290 --> 00:24:46,210
I'm not that guy.

294
00:24:46,290 --> 00:24:48,210
If you're that crazy, then I don't want to work with you.

295
00:24:48,250 --> 00:24:48,510
Next.

296
00:24:51,370 --> 00:24:51,810
Nice.

297
00:24:52,310 --> 00:24:52,610
Okay.

298
00:24:52,610 --> 00:25:00,130
So you're primarily using it for your own internal processes, projects, documentation, etc.

299
00:25:00,410 --> 00:25:04,250
Are you using it with your students at all?

300
00:25:04,670 --> 00:25:05,610
Are you having them?

301
00:25:05,990 --> 00:25:08,490
Yeah, here's the thing when it comes to AI.

302
00:25:09,110 --> 00:25:18,790
I don't want the kids using AI until they can write probably around the 9th or 10th grade level for real.

303
00:25:19,350 --> 00:25:20,850
I don't want them touching it.

304
00:25:20,850 --> 00:25:25,730
I want this done all by hand because you need to learn how,

305
00:25:26,050 --> 00:25:28,630
it's kind of like an example I like to give people.

306
00:25:29,570 --> 00:25:33,490
I was a really good math student in high school.

307
00:25:34,130 --> 00:25:37,610
I wasn't the best of the best, but I was a very good student.

308
00:25:37,610 --> 00:25:42,170
You know, through geometry, trig, algebra,

309
00:25:42,750 --> 00:25:46,570
and then I went into calculus-based physics, okay?

310
00:25:46,570 --> 00:25:52,170
And from there, this was early 90s.

311
00:25:52,230 --> 00:25:53,910
We had those graphical calculators.

312
00:25:54,050 --> 00:25:55,470
Yeah, my father was fun with that one.

313
00:25:55,790 --> 00:26:01,790
He's like, 300 bucks for a damn calculator with eight megabytes of RAM, eight megabytes of memory.

314
00:26:01,950 --> 00:26:02,530
I said, yeah.

315
00:26:02,990 --> 00:26:05,970
So then we started using the graphical calculators.

316
00:26:05,970 --> 00:26:12,210
But if we didn't understand the principles of how this worked, I just remember what my professor said.

317
00:26:12,270 --> 00:26:15,570
Somebody complained, well, the rich kids can have these eight megabytes.

318
00:26:15,570 --> 00:26:18,670
He's like, can we bring in some worksheets with some formulas on it?

319
00:26:18,990 --> 00:26:20,030
He said, you know what?

320
00:26:20,070 --> 00:26:24,250
For those who want it, I will give you the exam in my office.

321
00:26:24,450 --> 00:26:25,510
It's not for three hours.

322
00:26:25,550 --> 00:26:26,350
I'll give you eight.

323
00:26:26,810 --> 00:26:28,030
Bring in your textbook.

324
00:26:28,350 --> 00:26:29,850
Bring in anything you want.

325
00:26:30,350 --> 00:26:31,910
Bring in anything but Uncle Bob.

326
00:26:33,070 --> 00:26:35,870
He said, it's not going to make a damn bit of difference.

327
00:26:35,970 --> 00:26:37,730
You know how to apply it or you don't.

328
00:26:39,190 --> 00:26:45,230
So at that point, it made sense for us to have the machine start doing the work.

329
00:26:45,230 --> 00:26:48,370
But if we didn't understand the principles, we couldn't apply it, we would flunk.

330
00:26:49,030 --> 00:26:59,310
So with AI, until you understand how to write, you know, 9th, 10th grade level and read at the 9th, 10th grade level, I don't think the kids should go near it.

331
00:27:00,010 --> 00:27:05,210
And then I love when teachers tell me this, but then the kids are going to be far behind.

332
00:27:06,110 --> 00:27:11,310
I'm like, no, because you don't have a freaking idea how to use AI anyway.

333
00:27:12,230 --> 00:27:14,550
And that, I don't tell that to their faces,

334
00:27:14,550 --> 00:27:17,090
but they have no freaking idea how to use AI for education.

335
00:27:17,710 --> 00:27:20,390
So again, what I said I'm using is the craft of research.

336
00:27:20,470 --> 00:27:23,010
So let me again explain what that is.

337
00:27:24,250 --> 00:27:26,350
Context is to get everybody on the same page.

338
00:27:26,490 --> 00:27:29,010
That's what I show for ninth, tenth grade level writing.

339
00:27:29,010 --> 00:27:31,970
Then you have to come up with a unique problem statement,

340
00:27:32,630 --> 00:27:34,750
a problem that actually needs to get solved.

341
00:27:35,250 --> 00:27:38,030
What I added to it is something you can also do

342
00:27:38,030 --> 00:27:39,370
and then coming up with a unique solution.

343
00:27:40,350 --> 00:27:42,770
This is what you have to do to like publish an article.

344
00:27:43,190 --> 00:27:44,490
That's the exact process.

345
00:27:45,550 --> 00:27:50,250
Now, the book on that is The Craft of Research has sold over a million copies since 1995.

346
00:27:50,610 --> 00:27:56,650
Now, the kids that I know that know that, some of them freaking hate AI, but their bosses love it.

347
00:27:57,010 --> 00:27:58,010
So they go into it.

348
00:27:58,090 --> 00:27:59,130
They go into a job.

349
00:27:59,290 --> 00:28:00,930
They don't know a damn thing out of it.

350
00:28:00,930 --> 00:28:03,490
They don't even know what a chat GPT is.

351
00:28:03,970 --> 00:28:07,870
Within two weeks, they're training their peers in AI.

352
00:28:08,410 --> 00:28:09,910
I've had that happen multiple times.

353
00:28:10,750 --> 00:28:12,610
And people are like, how did that happen?

354
00:28:12,750 --> 00:28:15,470
I show them the craft of research through high school.

355
00:28:16,130 --> 00:28:19,410
Just so you know, no private school in the United States touches it because it's too advanced.

356
00:28:19,870 --> 00:28:23,430
And then my kids go off to college and they practice it for four to five years.

357
00:28:23,950 --> 00:28:25,410
Then they go out in the workforce.

358
00:28:25,810 --> 00:28:28,950
And one example I like to give, the kid calls me up.

359
00:28:28,990 --> 00:28:29,990
He says, I don't know what to do.

360
00:28:29,990 --> 00:28:31,890
I said, well, pick an AI model.

361
00:28:31,990 --> 00:28:32,690
I don't care which one.

362
00:28:33,210 --> 00:28:34,150
Figure out context.

363
00:28:34,430 --> 00:28:35,310
And he's going back.

364
00:28:35,390 --> 00:28:36,210
He's talking to it.

365
00:28:36,250 --> 00:28:36,690
He's typing.

366
00:28:37,210 --> 00:28:38,430
He's frustrated.

367
00:28:38,570 --> 00:28:41,250
He goes back and forth for about a day.

368
00:28:41,330 --> 00:28:42,370
Then he figures out context.

369
00:28:42,870 --> 00:28:45,230
Then I say, come up, go and figure out a problem statement.

370
00:28:46,150 --> 00:28:48,950
That's unique, something you can do that needs to get done.

371
00:28:49,270 --> 00:28:50,510
So he does that for about a day.

372
00:28:50,950 --> 00:28:52,630
Then he comes, then I tell him to do the solution.

373
00:28:52,730 --> 00:28:53,570
He figured that out.

374
00:28:53,990 --> 00:28:58,070
Then he explains it to his boss, and the boss thinks that this is freaking genius.

375
00:28:58,770 --> 00:29:01,430
And then he's showing his peers how to go through this,

376
00:29:01,550 --> 00:29:04,070
but they don't know what the heck the craft of research is.

377
00:29:04,150 --> 00:29:05,690
He's like, how did you get through high school?

378
00:29:05,730 --> 00:29:06,510
I get through college.

379
00:29:06,690 --> 00:29:09,050
some with grad degrees

380
00:29:09,050 --> 00:29:09,970
don't know what is

381
00:29:09,970 --> 00:29:12,150
so he has to teach them what that is

382
00:29:12,150 --> 00:29:14,450
explain how to do the process

383
00:29:14,450 --> 00:29:15,410
that he came up with

384
00:29:15,410 --> 00:29:16,910
and then the boss is thrilled

385
00:29:16,910 --> 00:29:19,390
because everybody's drastically more productive

386
00:29:19,390 --> 00:29:24,170
pretty cool

387
00:29:24,170 --> 00:29:25,230
pretty cool

388
00:29:25,230 --> 00:29:28,130
yeah and that's from a book that came out in 95

389
00:29:28,130 --> 00:29:32,790
oh yeah I mean

390
00:29:32,790 --> 00:29:35,050
there's a common misconception

391
00:29:35,050 --> 00:29:36,470
that AI is new

392
00:29:36,690 --> 00:29:48,290
it was, I want to say 1950s, late 1950s, uh, first introduced, uh, I believe, uh, might've

393
00:29:48,290 --> 00:29:54,210
been early sixties, but it's been around for a very long time. I actually, uh, worked for,

394
00:29:54,590 --> 00:30:02,350
well, when I was at one of the facilities I worked for, uh, we, a big part of what they

395
00:30:02,350 --> 00:30:09,470
were doing was, was, uh, was voice patterning and recognition. And that was, that was all,

396
00:30:09,490 --> 00:30:15,470
you know, machine learning and AI as well. And, and what, again, people don't realize that this is,

397
00:30:15,470 --> 00:30:22,410
this is not new. Uh, just some breakthroughs have occurred that were bigger than anyone had

398
00:30:22,410 --> 00:30:30,890
predicted or hoped for, uh, in the last few years. So. Yeah, we're on a trajectory just,

399
00:30:30,890 --> 00:30:40,630
I mean, I'm very skeptical when people report about leaked things, but when it's like a major, I think it was like Fortune or one of those major publications.

400
00:30:42,910 --> 00:30:52,370
Anthropic had a little boo-boo where they left like thousands of documents that people could download for somebody at the company who made a mistake.

401
00:30:52,370 --> 00:30:59,250
they're coming out with a new model called something m something or other that is so powerful

402
00:30:59,250 --> 00:31:06,810
that it's finding cracks in on in online security that have been around for decades nobody could

403
00:31:06,810 --> 00:31:12,670
find so they're sending it out to all of these security people before they release it to the

404
00:31:12,670 --> 00:31:18,510
general public so they can harden everything all right and it's just going to come more and more

405
00:31:18,510 --> 00:31:25,010
powerful. Now, what I'm trying to tell the college kids, here's what's going, there's an economic term,

406
00:31:25,170 --> 00:31:32,590
40% of them are in sub-employment. My first degree was in macroeconomics. Essentially,

407
00:31:32,730 --> 00:31:36,790
40% of recent college grads are getting jobs that don't require anything close to a college

408
00:31:36,790 --> 00:31:45,090
education. All right, and why? Because, oh, I use ChatGPT to help me with my homework or this or

409
00:31:45,090 --> 00:31:52,490
that or the next thing. I said, yeah, who cares? I'm using it to the point now where the stuff that

410
00:31:52,490 --> 00:31:57,910
I would normally have a new college grad do, I'm just having it do it itself. And then I'm having

411
00:31:57,910 --> 00:32:03,170
senior people that I partnered with do the reviewing who just know what the problem is.

412
00:32:03,170 --> 00:32:08,410
And they say, I make these changes. And then I come back and I tell AI to, I worked through

413
00:32:08,410 --> 00:32:17,010
ChatGPT extended thinking 5.4. Why? Well, a couple of weeks ago, I was using Google 3.1 Pro.

414
00:32:18,270 --> 00:32:23,270
And then ChatGPT with 5.4 extended thinking came out. It was just a much better model.

415
00:32:24,330 --> 00:32:30,710
And then every week I keep changing my process. Literally, I spend a half an hour to an hour a day

416
00:32:30,710 --> 00:32:36,170
looking at new AI stuff and playing with it to make it a little better. Why? Because I got to

417
00:32:36,170 --> 00:32:42,750
tell my kids so they can get a job. So that process that I just told you, I will tell the

418
00:32:42,750 --> 00:32:48,430
kids when they're a senior in college, I said, okay, you're a senior. You're essentially a college

419
00:32:48,430 --> 00:32:56,810
graduate. I want you to start using Opus 4.6 on everything. Okay. You got to build this up for at

420
00:32:56,810 --> 00:33:03,250
least a month so we can go through this process. And then what I'm finding is I just showed some of

421
00:33:03,250 --> 00:33:09,470
my former kids who are out of college do that. And they're in positions where they have a lot

422
00:33:09,470 --> 00:33:14,530
of autonomy in their jobs. They have permission. I said, you have to get everything approved by

423
00:33:14,530 --> 00:33:20,350
your company. And they did surprisingly very rapidly. And they said, do you have any idea

424
00:33:20,350 --> 00:33:27,790
how much more productive I am? I wake up, I ran this stuff all night. They're paying 200 bucks a

425
00:33:27,790 --> 00:33:34,390
month out of their own money for this. But they said it is so worth it. And they work from home,

426
00:33:34,470 --> 00:33:37,470
they're finishing stuff up and their boss is contacting them and they're saying,

427
00:33:37,650 --> 00:33:41,450
well, where's all this? Where's this? Where's this? They said, well, it's in your inbox.

428
00:33:42,970 --> 00:33:48,270
But they don't send it when it's done because they're so productive. They're just kind of

429
00:33:48,270 --> 00:33:53,550
holding on to it. And the boss still thinks they're productive and they're out for a walk.

430
00:33:53,550 --> 00:33:57,630
or one of them said he went to the movies,

431
00:33:58,010 --> 00:33:59,550
told his boss he was in a meeting

432
00:33:59,550 --> 00:34:01,710
and everything he wanted was in his inbox.

433
00:34:03,370 --> 00:34:07,130
That's how productive this process is making folks.

434
00:34:07,390 --> 00:34:08,830
And isn't it the best one?

435
00:34:08,890 --> 00:34:10,010
Not by a long shot.

436
00:34:10,770 --> 00:34:12,390
I'm still learning that.

437
00:34:12,650 --> 00:34:14,390
But one thing I've learned is

438
00:34:14,390 --> 00:34:16,570
you can forget about keeping up

439
00:34:16,570 --> 00:34:17,850
because there are a few podcasts

440
00:34:17,850 --> 00:34:19,450
that I listen to every day on this.

441
00:34:20,230 --> 00:34:23,170
And these guys are the top of the top

442
00:34:23,170 --> 00:34:25,190
and they can barely keep up.

443
00:34:25,990 --> 00:34:26,710
And I'm just like.

444
00:34:29,410 --> 00:34:31,190
Do you think there's ever going to be a point

445
00:34:31,190 --> 00:34:34,190
where certain fields of study or employment

446
00:34:34,190 --> 00:34:36,830
are just not even needed anymore

447
00:34:36,830 --> 00:34:39,070
because AI goes from an assistant

448
00:34:39,070 --> 00:34:41,430
to just replacing that person altogether?

449
00:34:42,270 --> 00:34:45,130
Because I'm just going to put on it.

450
00:34:45,270 --> 00:34:46,510
Let's take what you do.

451
00:34:46,950 --> 00:34:49,330
Tell me a speciality that you do for work

452
00:34:49,330 --> 00:34:51,770
that you're really freaking good at,

453
00:34:52,210 --> 00:34:53,010
that you love doing,

454
00:34:53,010 --> 00:34:53,830
that you're really good at.

455
00:34:53,890 --> 00:34:54,350
What is it?

456
00:34:55,650 --> 00:34:56,310
Auto mechanic.

457
00:34:56,510 --> 00:34:58,410
I don't think there's an AI that could do that.

458
00:34:58,730 --> 00:34:59,030
Okay.

459
00:34:59,030 --> 00:35:02,230
Maybe the diagnostic part, but not the actual wrenching part.

460
00:35:03,330 --> 00:35:10,930
Now, I may diagnose certain things, but how, if you're putting in some gizmo into another

461
00:35:10,930 --> 00:35:14,910
gizmo, I'm just saying that because with auto mechanics, I am grade A stupid.

462
00:35:15,250 --> 00:35:15,370
Okay.

463
00:35:15,370 --> 00:35:16,550
I'm just horrible at it.

464
00:35:16,690 --> 00:35:16,810
Okay.

465
00:35:18,170 --> 00:35:21,810
Your judgment on does this work?

466
00:35:21,810 --> 00:35:23,030
Does this make sense?

467
00:35:24,330 --> 00:35:32,630
The AI can get close to what you do, but it can't understand your own style of how to make things run better, more efficiently.

468
00:35:33,250 --> 00:35:33,910
All right.

469
00:35:33,950 --> 00:35:36,470
Or what to do when the damn thing just looks crazy.

470
00:35:36,890 --> 00:35:39,570
Like, for example, there are people whose job it is.

471
00:35:40,070 --> 00:35:48,710
They have these autonomous cars and then the damn thing just gets stuck and they go and they say, oh, they clip on and then they get it out.

472
00:35:48,770 --> 00:35:50,690
They get it on stuck and then they go to the next one.

473
00:35:50,690 --> 00:35:51,830
That's literally their job.

474
00:35:52,610 --> 00:35:55,910
So just, you know, everybody thinks where AI is going right now,

475
00:35:55,990 --> 00:35:57,790
we're pretty much doing nothing.

476
00:35:58,390 --> 00:36:01,010
We're not even at 1% of where we're going.

477
00:36:01,130 --> 00:36:02,790
We're at like 0.001.

478
00:36:03,630 --> 00:36:04,070
We're nothing.

479
00:36:05,170 --> 00:36:08,930
Because right now, the next major thing is with the chips

480
00:36:08,930 --> 00:36:21,240
is we going to cool them with hot water That just coming out But these aren the real chips It not hot water And it actually been around for quite some time

481
00:36:21,680 --> 00:36:25,760
Fully submerged blade systems.

482
00:36:29,060 --> 00:36:33,560
This is one of the things that the company I work for does, right?

483
00:36:33,700 --> 00:36:36,160
And so they've been around for a while.

484
00:36:36,160 --> 00:36:43,340
they're in data centers and that is a huge thing it is cool cool water that gets turned into hot

485
00:36:43,340 --> 00:36:48,940
water uh but yeah they're they're fully submerged it's quite an interesting way to work on a machine

486
00:36:48,940 --> 00:36:56,840
because liquid and machines typically are are a big no-no right yeah well that was just one thing

487
00:36:56,840 --> 00:37:01,040
that jensen was saying how they're gonna the next set's gonna be hot well whatever but the main thing

488
00:37:01,040 --> 00:37:05,240
is these are still chips designed for video games.

489
00:37:05,340 --> 00:37:06,880
It's just a souped-up version.

490
00:37:07,000 --> 00:37:11,060
The real AI chips were hoping to get into production by 2029,

491
00:37:11,380 --> 00:37:13,880
and instead of being the size of our fingernail,

492
00:37:14,120 --> 00:37:15,760
they're going to be the size of a dinner plate.

493
00:37:17,640 --> 00:37:18,160
Okay?

494
00:37:18,440 --> 00:37:20,820
So when that happens,

495
00:37:21,440 --> 00:37:26,460
the cost of cooling these things will drop drastically.

496
00:37:27,600 --> 00:37:30,680
The amount that they can do will go up tremendously.

497
00:37:31,040 --> 00:37:36,040
And people, I just want you to think of an analogy in history.

498
00:37:36,980 --> 00:37:41,220
The steam engine, it made things drastically cheaper.

499
00:37:41,600 --> 00:37:44,280
So we used much more of it.

500
00:37:44,540 --> 00:37:46,700
We had much more of abundance.

501
00:37:48,060 --> 00:37:54,020
So if you can, if you know the craft of research and you can go into apply this, this was just

502
00:37:54,020 --> 00:37:57,660
some, this was just from one podcast I recently listened to.

503
00:37:57,660 --> 00:38:00,780
and they did some analysis of the job market.

504
00:38:00,920 --> 00:38:03,820
What they were finding is people that were senior,

505
00:38:04,260 --> 00:38:05,840
that really knew what they were doing,

506
00:38:05,960 --> 00:38:07,180
really knew how to use AI,

507
00:38:07,600 --> 00:38:10,060
literally to 5, 10x their productivity, literally,

508
00:38:11,060 --> 00:38:14,480
is that for every 3.1 jobs that were out there,

509
00:38:14,560 --> 00:38:15,800
there was one person to fill.

510
00:38:15,880 --> 00:38:18,140
It took about 146 days to fill a position

511
00:38:18,140 --> 00:38:21,420
and people were asking for a quarter million to 400,000

512
00:38:21,420 --> 00:38:23,620
and they were getting it.

513
00:38:24,880 --> 00:38:26,960
But this was not a nine to five job.

514
00:38:26,960 --> 00:38:31,560
They wanted them there a lot longer, but they knew how to come in.

515
00:38:31,760 --> 00:38:33,960
This is just not coding, engineering.

516
00:38:34,320 --> 00:38:35,360
This is marketing.

517
00:38:35,660 --> 00:38:37,360
This is a lot of like, let me give you an example.

518
00:38:38,160 --> 00:38:44,760
There was a guy who just was good at spreadsheets, but not like the expert of experts.

519
00:38:46,220 --> 00:38:53,880
Anthropic injected things into, Anthropic is now in Microsoft Excel.

520
00:38:53,880 --> 00:38:55,080
All right.

521
00:38:55,080 --> 00:39:04,500
And he did literally a financial sheet that would take a Goldman Sachs analyst like eight hours to do.

522
00:39:05,000 --> 00:39:09,380
He did it in 40 minutes and then had it checked by a Goldman Sachs analyst.

523
00:39:10,420 --> 00:39:14,200
He said, yeah, this would have taken me about a day to do about eight hours.

524
00:39:15,380 --> 00:39:15,580
All right.

525
00:39:15,620 --> 00:39:17,640
That's how powerful these things are.

526
00:39:17,640 --> 00:39:22,340
And it just means we're going to be able to get more and more stuff done.

527
00:39:22,820 --> 00:39:27,620
But it doesn't mean that there's going to be less and less needs for humans.

528
00:39:28,100 --> 00:39:32,420
But there are going to be a huge need for well-trained humans.

529
00:39:33,140 --> 00:39:39,640
And if you don't want to make this shift to using artificial intelligence, I'll just let you know, my younger sister hates the damn thing.

530
00:39:40,620 --> 00:39:45,740
So she got a job that will not be replaced by AI for decades.

531
00:39:45,740 --> 00:39:51,360
people are afraid that AI robots are going to come in and become car mechanics.

532
00:39:52,220 --> 00:39:57,840
So Boston AI robot, whatever, I forget the name of it.

533
00:39:58,000 --> 00:39:58,600
Dynamics.

534
00:39:59,020 --> 00:39:59,380
What's that?

535
00:40:00,420 --> 00:40:01,020
Dynamics.

536
00:40:01,180 --> 00:40:01,620
Dynamics.

537
00:40:01,760 --> 00:40:02,300
Boston Dynamics.

538
00:40:02,380 --> 00:40:04,860
It's 88% owned by Hyundai.

539
00:40:05,780 --> 00:40:10,280
So they've got a man-sized robot in a factory right now.

540
00:40:10,520 --> 00:40:11,660
They're testing it out.

541
00:40:12,320 --> 00:40:13,500
Five foot nine.

542
00:40:13,500 --> 00:40:21,400
It can lift 70 pounds and it can pick up something, turn it around and put it on the next table.

543
00:40:22,140 --> 00:40:27,100
And then pick up another thing and put it on the table like six inches apart or whatever.

544
00:40:27,780 --> 00:40:28,260
That's it.

545
00:40:28,380 --> 00:40:31,320
Not the complicated stuff of being a mechanic.

546
00:40:32,060 --> 00:40:39,340
I know that Elon Musk who runs, for his robots, the biggest problem they're trying to figure out is the hand.

547
00:40:39,780 --> 00:40:40,540
How do you grab?

548
00:40:40,660 --> 00:40:42,140
How do you do this?

549
00:40:42,140 --> 00:40:44,180
they just figured that out.

550
00:40:44,260 --> 00:40:44,940
It wasn't them.

551
00:40:45,140 --> 00:40:48,680
I believe it was a company in China that figured it out.

552
00:40:48,820 --> 00:40:52,200
And they just introduced a robot that has the,

553
00:40:52,560 --> 00:40:54,820
it has amazing dexterity.

554
00:40:55,340 --> 00:40:58,240
And part of that was putting a nut,

555
00:40:58,820 --> 00:41:00,220
threading a nut onto a bolt,

556
00:41:00,360 --> 00:41:02,080
and then screwing it all the way down.

557
00:41:02,340 --> 00:41:06,340
And then doing that with another different size nut

558
00:41:06,340 --> 00:41:07,340
and different size bolt.

559
00:41:07,340 --> 00:41:09,820
And then with a wing nut on a bolt, right?

560
00:41:10,320 --> 00:41:11,140
And then,

561
00:41:11,140 --> 00:41:21,040
they went through the whole thing again and with one hand it did all three and i mean blazing facts

562
00:41:21,040 --> 00:41:28,940
so if if you think about that i i honestly i think the idea that there is a safe job is

563
00:41:28,940 --> 00:41:37,620
absolutely not not true because robotics are are now i mean now you've got you know people can buy

564
00:41:37,620 --> 00:41:43,260
a robot for their home that will do a lot of tasks in a home for about $10,000.

565
00:41:43,820 --> 00:41:45,300
Give it another 10 years.

566
00:41:45,380 --> 00:41:47,860
That'll be a, you know, $1,000 robot.

567
00:41:48,120 --> 00:41:51,040
I mean, you pay that for your phone, right?

568
00:41:51,560 --> 00:41:53,780
You think about industrial application.

569
00:41:54,080 --> 00:41:56,160
You can actually have these things.

570
00:41:56,160 --> 00:42:03,300
If you go far enough, you can actually have an AI retool a manufacturing plant, create

571
00:42:03,300 --> 00:42:09,840
the parts that the robots can assemble to make a better robot that can then go and retool the

572
00:42:09,840 --> 00:42:16,860
factory with new robots that do what is needed and create another set of robots that could go out and

573
00:42:16,860 --> 00:42:23,300
be plumbers that could be mechanics that could be painters that could be welders uh etc right so

574
00:42:23,300 --> 00:42:31,240
i i think uh you know hopefully you know there's a lot of thought that goes into this and and people

575
00:42:31,240 --> 00:42:41,460
understand that there's a need to, there's a, how do I want to say this? There's a need to protect

576
00:42:41,460 --> 00:42:50,620
the ability for the human to have something to do that is worthwhile for their being, right?

577
00:42:51,200 --> 00:42:55,480
And a job is a part of that, right? It doesn't necessarily have to be for a company. It could

578
00:42:55,480 --> 00:43:01,180
be entrepreneurial. It could be, you know, just, this is the stuff I do on a day-to-day basis.

579
00:43:01,180 --> 00:43:06,160
And like, you know, living on a farm, right? Those are tasks that you have to do every day,

580
00:43:06,160 --> 00:43:11,220
right? But we get great satisfaction as human beings out of accomplishing these things.

581
00:43:11,960 --> 00:43:17,400
And so hopefully people are actually really thinking about this because I don't really see

582
00:43:17,400 --> 00:43:23,660
any boundary that AI with machine learning and robotics cannot take on.

583
00:43:24,420 --> 00:43:26,900
Well, let me give you an example with advanced manufacturing.

584
00:43:27,060 --> 00:43:31,440
I live right outside of Albany, New York, and there's a guy called Jeff Lawrence who

585
00:43:31,440 --> 00:43:35,220
quarterbacked what we started about 25 years ago in advanced manufacturing.

586
00:43:36,100 --> 00:43:41,620
So they actually started at our local community college, Huston Valley Community College,

587
00:43:41,800 --> 00:43:43,320
with a lot of state funding.

588
00:43:43,820 --> 00:43:49,200
And I can tell you people that go into that two-year degree will come out and there are

589
00:43:49,200 --> 00:43:51,640
four to five jobs waiting for them.

590
00:43:51,640 --> 00:43:58,040
right now that I think they're starting at around 50, 60 grand within whoever amount of time it

591
00:43:58,040 --> 00:44:02,360
takes become senior, they're making like 120 grand. And those figures are a couple of years old.

592
00:44:02,760 --> 00:44:09,940
All right. So you have to know how instead of 5,000 people from the fifties, it's now like 200

593
00:44:09,940 --> 00:44:15,840
people and you have to know how to work with the machines. All right. But everybody's having this,

594
00:44:15,840 --> 00:44:23,880
this view that, oh, AI is going to take over everything. I don't see that. I see that the

595
00:44:23,880 --> 00:44:29,680
biggest issue with humans is, do we know, like for, I'm just talking about intellectual work now.

596
00:44:29,960 --> 00:44:36,400
Is this correct? What changes need to be made so this is correct? I think AI can get better and

597
00:44:36,400 --> 00:44:42,940
better at helping us do drafts, but we have to make the final decisions. And yes, it's making

598
00:44:42,940 --> 00:44:51,740
it's making me vastly more productive. Okay. Now my team, I have two assistants. One has a master's

599
00:44:51,740 --> 00:44:56,780
in teaching English. One is a PhD candidate in education. I used to have a bunch of other people,

600
00:44:56,780 --> 00:45:02,160
which we no longer use. They were college kids, grad students, and they decided to self-select

601
00:45:02,160 --> 00:45:08,660
and move on. And have I gotten rid of my people? No, I'm actually going to have to bring some more

602
00:45:08,660 --> 00:45:14,320
people on, but I'll tell you exactly what I'm looking at. Next person I'm looking to bring on

603
00:45:14,320 --> 00:45:19,080
has a master's degree as an English teacher and seven years of teaching experience, teaching

604
00:45:19,080 --> 00:45:23,420
middle school English. Those are the type of people I'm going to bring on because I can show

605
00:45:23,420 --> 00:45:30,220
them how to do the AI stuff very quickly. But when it comes to customer service, I want to provide

606
00:45:30,220 --> 00:45:37,780
a very high level of customer service. So I can have the AI suggest things to make them a lot more

607
00:45:37,780 --> 00:45:43,200
productive, but I want an experienced human looking at everything that comes through to

608
00:45:43,200 --> 00:45:43,840
send out.

609
00:45:44,620 --> 00:45:46,500
And that's, I don't care.

610
00:45:46,880 --> 00:45:49,440
I'm not doing the, one person said, well, here's the idea.

611
00:45:50,100 --> 00:45:56,360
You can record an answer to a question, because the people ask almost the same questions over

612
00:45:56,360 --> 00:45:56,800
and over again.

613
00:45:56,960 --> 00:46:02,380
You can record that, make it really good, and then restrict the AI to answering just

614
00:46:02,380 --> 00:46:04,940
the exact questions you're asked and send the rest to a human.

615
00:46:05,300 --> 00:46:07,060
And I'm like, no.

616
00:46:07,780 --> 00:46:13,660
I'm having a human review everything because I want, and I can tell you that's important to people.

617
00:46:13,880 --> 00:46:15,740
Let me give you another example about.

618
00:46:16,520 --> 00:46:31,940
That is, that is one, one area that at least to date, and, and hopefully it will remain that AI, it's the area that centers around care, empathy, not sympathy, empathy.

619
00:46:32,220 --> 00:46:34,300
Cause AI can sympathize with you.

620
00:46:34,340 --> 00:46:37,420
It will, it'll do a great job of sympathizing with you.

621
00:46:37,420 --> 00:47:00,060
But it has no concept of what empathy is or really morals or ethics either other than what's been programmed into it. Like, ah, that's immoral. Don't do that. Right. And here's a list of the immoral. But it doesn't have a concept of, well, that's not in the list. Therefore, it's not immoral. Right. It's not unethical.

622
00:47:00,060 --> 00:47:22,900
And it doesn't have a concept of care. So when you look at things that are more, and by golly, it's really good at creating images and videos and things like that. But the, the, again, the human portion of that, right, the ability to actually create from imagination is, I think, at least now far superior.

623
00:47:22,900 --> 00:47:29,040
So those are the areas I see humans as being largely untouched in this, right?

624
00:47:29,140 --> 00:47:31,880
This will be an augmentation tool only.

625
00:47:32,140 --> 00:47:36,380
And I think in general, it'll be used as augmentation.

626
00:47:37,020 --> 00:47:39,680
But, you know, there is some concern, right?

627
00:47:39,740 --> 00:47:43,380
I mean, companies don't get into business to make their employees happy.

628
00:47:43,680 --> 00:47:45,440
They get into business to make dollars.

629
00:47:45,440 --> 00:47:59,860
And if I can spend a million dollars on an AI and a billion dollars on a factory retool that uses AI and I can replace 100,000 employees, I'm going to do it.

630
00:47:59,860 --> 00:48:08,740
Right. I wouldn't personally, but that's I guarantee you 99 by five nines of the companies out there will absolutely do that.

631
00:48:08,740 --> 00:48:10,760
Right. That being said.

632
00:48:10,760 --> 00:48:18,420
we're nowhere near that, right? We will get there at some point and maybe people will change minds

633
00:48:18,420 --> 00:48:25,820
and maybe thinking will change who knows. Right. But, um, yeah, I think, uh, you know, it's, it's,

634
00:48:26,180 --> 00:48:32,000
it's one of those things we, I, I caution people about being too short sighted about what it can

635
00:48:32,000 --> 00:48:37,640
do or what it will be capable of because of the fact that you have to consider the fact of

636
00:48:37,640 --> 00:48:41,560
robotics and machine learning as well as AI, right?

637
00:48:42,600 --> 00:48:48,800
Those three things together are exceedingly powerful and the capabilities far

638
00:48:48,800 --> 00:48:54,500
exceed our ability to do things like produce stuff, right?

639
00:48:54,980 --> 00:48:55,600
Or can.

640
00:48:56,920 --> 00:48:57,080
Yeah.

641
00:48:57,240 --> 00:49:04,980
Well, the, what I, what I try to do with my, let me get an example.

642
00:49:04,980 --> 00:49:06,760
there's one family I'm working with.

643
00:49:09,320 --> 00:49:14,120
The woman's daughter is a very gifted painter.

644
00:49:14,540 --> 00:49:20,400
She's already doing individual commission paintings in her early teens.

645
00:49:21,300 --> 00:49:22,840
And she doesn't want to turn into a business.

646
00:49:23,040 --> 00:49:23,760
But I said, okay, great.

647
00:49:23,880 --> 00:49:25,400
Keep her working on that.

648
00:49:25,840 --> 00:49:29,120
But when she's ready, you know, you can just go to,

649
00:49:29,120 --> 00:49:37,920
go to co-worker and have it create a custom website for you.

650
00:49:38,000 --> 00:49:40,340
People willing to pay for her work.

651
00:49:41,280 --> 00:49:43,580
Her next oldest son is, you know, the traditional,

652
00:49:44,860 --> 00:49:47,720
he's going to be an engineering science student at Big State U.

653
00:49:48,620 --> 00:49:48,920
Okay.

654
00:49:49,860 --> 00:49:53,000
And when I'm looking at that, I said, okay, you need to,

655
00:49:53,200 --> 00:49:54,860
the first two years are the same for everything.

656
00:49:54,860 --> 00:49:59,760
The engineering science hasn't changed really in the last 30, 40 years.

657
00:50:01,320 --> 00:50:08,380
Once you get him to be able to read and write at the 10th grade level, start having him go through the craft of research, get used to it.

658
00:50:08,760 --> 00:50:13,520
Because then he'll be fine because once you start publishing and you can add value, he'll be fine.

659
00:50:14,920 --> 00:50:18,760
Her next son is probably going to be a head and hands type guy.

660
00:50:18,760 --> 00:50:25,080
And we're, I mean, he's not going to enter the field for probably another 10 or 12 years.

661
00:50:26,060 --> 00:50:31,460
And one of the surprising things I found is the quickest way to a six-figure income is being an electrician.

662
00:50:32,340 --> 00:50:39,860
And I know AI is going to have things and there will be some robotics, but I don't think, we're not going to trust him.

663
00:50:39,900 --> 00:50:41,360
Here's some things that I found.

664
00:50:42,280 --> 00:50:47,660
I told, I think at his point, he'll probably be guiding robots and telling them what to do.

665
00:50:47,660 --> 00:50:51,960
But here's one interesting thing about families, about lives on the highway.

666
00:50:52,980 --> 00:50:57,620
Tragically, if you get in your car, and let's just wave the magic wand.

667
00:50:58,100 --> 00:51:02,960
You're a CEO of a Fortune 1000 company, and you got a lot of insurance.

668
00:51:03,680 --> 00:51:08,320
And you kill somebody by just literally a car accident.

669
00:51:09,340 --> 00:51:11,280
They might go after you for a couple of mil.

670
00:51:11,280 --> 00:51:17,880
when AI does it, when you have one of those driverless people do that,

671
00:51:18,920 --> 00:51:22,460
Tesla just got hit for, I think, 230 million bucks.

672
00:51:23,520 --> 00:51:24,700
Literally, it's like 100 times.

673
00:51:25,480 --> 00:51:30,180
We have, as a species, such zero,

674
00:51:31,360 --> 00:51:36,660
just zero, any acceptance of getting killed or hurt by one of these things.

675
00:51:37,140 --> 00:51:40,940
So you see all those little things running up and down the road delivering food?

676
00:51:41,280 --> 00:51:42,840
Imagine they hit somebody.

677
00:51:43,880 --> 00:51:44,020
Okay.

678
00:51:44,220 --> 00:51:47,600
I mean, it's just, it's crazy what people do going after that.

679
00:51:48,120 --> 00:51:49,160
So that's one aspect.

680
00:51:49,320 --> 00:51:54,800
The other one is one of the stupidest things I've seen out there with people with money

681
00:51:54,800 --> 00:51:58,560
is they will go and have an AI avatar themselves.

682
00:51:58,820 --> 00:52:02,740
They'll work with AI to come up with a script and they just dump it in

683
00:52:02,740 --> 00:52:05,220
and they don't have to spend the time getting on camera.

684
00:52:06,440 --> 00:52:06,880
Okay.

685
00:52:06,880 --> 00:52:12,100
And there was, and people hate it once they find out.

686
00:52:12,520 --> 00:52:15,640
So there was an economics professor said, I'm going to stay ahead of AI.

687
00:52:16,000 --> 00:52:18,600
His writing has dropped by 75 to 90%.

688
00:52:18,600 --> 00:52:21,100
And he does a lot of podcasts.

689
00:52:21,440 --> 00:52:22,540
He does a lot of in person.

690
00:52:23,060 --> 00:52:28,920
What I noticed when he's doing podcasts is there's always something about him that looks a little disheveled.

691
00:52:28,920 --> 00:52:29,660
So you know it's real.

692
00:52:29,980 --> 00:52:31,700
Like you look behind me, you see an outlet.

693
00:52:32,620 --> 00:52:34,560
You see a little cat hair on my hat.

694
00:52:35,280 --> 00:52:36,460
This is me.

695
00:52:36,460 --> 00:52:38,920
This is not some, I'm not freaking AI.

696
00:52:39,420 --> 00:52:42,980
And people know that I'm not freaking AI because of those things.

697
00:52:45,300 --> 00:52:45,980
All right?

698
00:52:46,680 --> 00:52:50,200
That's one of the key things because people, they find out it's AI,

699
00:52:50,340 --> 00:52:51,140
they just turn it off.

700
00:52:52,720 --> 00:52:52,900
Yeah.

701
00:52:54,200 --> 00:52:56,680
Yeah, well, and there is a lot of that, right?

702
00:52:56,680 --> 00:53:19,920
It's, uh, you know, I think it, it incenses us a bit, uh, in the fact that it's, it's, it's a fake creativity, right? There's something to the creativity of a human being, being on camera, speaking to another human being, we're speaking about something to, you know, an audience.

703
00:53:19,920 --> 00:53:22,120
whereas once it becomes AI

704
00:53:22,120 --> 00:53:25,100
then it takes that creativity out of it

705
00:53:25,100 --> 00:53:28,160
and it just becomes a machine, right?

706
00:53:29,160 --> 00:53:32,160
Yeah, and people ask me, do I study this?

707
00:53:32,900 --> 00:53:36,280
And what I tell them, is this the book that I

708
00:53:36,280 --> 00:53:39,020
this is the most popular book I use to teach dyslectics to read

709
00:53:39,020 --> 00:53:42,820
Walt Disney, The Triumph of the American Imagination

710
00:53:42,820 --> 00:53:47,480
Thousand pagers, okay

711
00:53:47,480 --> 00:53:50,760
It takes me one to three years to get a kid through this book.

712
00:53:51,380 --> 00:53:53,640
People are like, one to three years?

713
00:53:53,760 --> 00:53:57,060
Yes, but then that's written for 17-year-olds.

714
00:53:58,080 --> 00:54:01,960
So when they get done with this, it's like a 17 – they know how to read.

715
00:54:02,860 --> 00:54:06,280
And I literally show them how you go into Disney World.

716
00:54:07,840 --> 00:54:09,340
There's the Disney magic.

717
00:54:09,460 --> 00:54:12,280
It's two universal themes, and I show them how to discover both.

718
00:54:12,280 --> 00:54:27,790
And then I show them I had kids that have went to apply to work for Disney And they can go in and say well this is what Walt would have done to apply this traditional Disney magic and here how you do it

719
00:54:27,910 --> 00:54:29,390
And they would start spewing out ideas.

720
00:54:30,410 --> 00:54:31,990
Now, if one of those ideas hit,

721
00:54:32,830 --> 00:54:34,290
that might be worth a half a million bucks.

722
00:54:34,370 --> 00:54:35,190
It might be worth a million.

723
00:54:35,270 --> 00:54:36,210
It might be worth 10 million.

724
00:54:38,310 --> 00:54:40,650
That's something that I doubt

725
00:54:40,650 --> 00:54:42,350
we're going to see AI creating.

726
00:54:42,350 --> 00:54:44,510
Now, what this is going to do to us

727
00:54:44,510 --> 00:54:46,850
is this is going to push us

728
00:54:46,850 --> 00:54:50,910
to be more and more creative.

729
00:54:50,910 --> 00:54:53,990
I, my workflow compared,

730
00:54:54,230 --> 00:54:58,770
I was stuck in October of 2025 for months.

731
00:54:59,550 --> 00:55:01,230
Now, just in the last week,

732
00:55:01,270 --> 00:55:02,270
I've now moved to this.

733
00:55:02,330 --> 00:55:02,910
And now I have, you know,

734
00:55:02,930 --> 00:55:03,870
I wake up every morning

735
00:55:03,870 --> 00:55:06,210
and I have AI give me a lot of drafts to work on.

736
00:55:06,630 --> 00:55:09,090
And it's just, I get a lot more done

737
00:55:09,090 --> 00:55:10,610
and I'm a lot more efficient.

738
00:55:10,770 --> 00:55:12,250
And so I have a lot more free time.

739
00:55:12,870 --> 00:55:13,530
All right.

740
00:55:14,130 --> 00:55:16,590
And this is going to change in the next three months

741
00:55:16,590 --> 00:55:17,370
for the next six months.

742
00:55:17,450 --> 00:55:20,030
I mean, six months now feels like an eternity

743
00:55:20,030 --> 00:55:23,630
and most people have no idea what I'm talking about.

744
00:55:23,650 --> 00:55:24,690
What do you mean you wake up

745
00:55:24,690 --> 00:55:26,450
and the AI has your work done for you?

746
00:55:27,370 --> 00:55:29,850
Look, if you don't kind of keep up with this,

747
00:55:30,510 --> 00:55:32,950
then you're gonna be one of those people

748
00:55:32,950 --> 00:55:34,010
with these horrible jobs.

749
00:55:34,110 --> 00:55:35,510
Let me give you an example with China.

750
00:55:37,710 --> 00:55:41,010
Everybody's so afraid China's gonna take things over.

751
00:55:41,010 --> 00:55:51,010
China now, I think last year they had 12.5 or something million college graduates and 7.9 million births.

752
00:55:52,370 --> 00:55:52,690
Okay.

753
00:55:53,210 --> 00:55:58,970
And 40% of the jobs in major Chinese cities are delivering food.

754
00:55:59,170 --> 00:56:01,910
They're gig jobs, which aren't jobs.

755
00:56:02,570 --> 00:56:02,970
Okay.

756
00:56:03,410 --> 00:56:04,870
It's just really tough.

757
00:56:05,150 --> 00:56:09,030
And then they're creating all these robots to replace things in factories.

758
00:56:09,030 --> 00:56:14,670
so they need less. Yeah, it's all sorts of problems over there. So what I'm trying to say

759
00:56:14,670 --> 00:56:20,330
is in the United States and other Western countries, if you kind of keep up with AI,

760
00:56:20,610 --> 00:56:24,830
I'm just talking about for intellectual workers, and you can be very productive,

761
00:56:25,530 --> 00:56:30,410
is that the best data I could find is there's three jobs waiting for you for each person who's

762
00:56:30,410 --> 00:56:37,250
qualified to do it. If you're new coming out of college, this shocked the hell out of me.

763
00:56:37,250 --> 00:56:42,270
There was a major podcaster who a couple of months ago walked into a bunch of senior people.

764
00:56:42,950 --> 00:56:53,430
And they said, if you're doing something new, something you haven't really, you don't know everything about, have you run a deep research report before you start doing it?

765
00:56:54,170 --> 00:56:56,710
And out of hundreds of people, two people raised their hands.

766
00:56:56,830 --> 00:57:02,110
So I tell the kids, anytime you're doing anything, run a deep research report and read it.

767
00:57:02,470 --> 00:57:05,790
I limit mine to 3,500 to 4,000 words on Anthropic.

768
00:57:05,790 --> 00:57:08,170
and read it.

769
00:57:08,650 --> 00:57:10,770
But then if you're doing really important things,

770
00:57:10,950 --> 00:57:12,510
I will do a deep research report

771
00:57:12,510 --> 00:57:13,830
sometimes at the paragraph

772
00:57:13,830 --> 00:57:17,430
and sometimes even at the sentence level as a guide

773
00:57:17,430 --> 00:57:20,010
when I absolutely have to get it right.

774
00:57:20,430 --> 00:57:22,990
And yes, when you're new, read the things.

775
00:57:23,110 --> 00:57:24,450
For me, I just summarize it

776
00:57:24,450 --> 00:57:26,770
because I know what to do

777
00:57:26,770 --> 00:57:28,750
because I really know my field really well.

778
00:57:29,610 --> 00:57:31,290
And then one of my kids do that.

779
00:57:31,290 --> 00:57:33,730
And he said he became,

780
00:57:33,730 --> 00:57:37,690
his boss thinks he became relatively senior in about nine months.

781
00:57:38,870 --> 00:57:42,850
And he's ahead of me on doing this AI thing where it's doing things overnight.

782
00:57:43,310 --> 00:57:45,910
And guess who keeps getting promoted with a big salary?

783
00:57:46,690 --> 00:57:49,810
He actually left his company because he went in and said,

784
00:57:49,950 --> 00:57:51,650
I want you to double my salary.

785
00:57:51,730 --> 00:57:52,330
And they said, no.

786
00:57:53,230 --> 00:57:56,690
So then he went out in the marketplace and he found somebody that would,

787
00:57:56,930 --> 00:58:00,010
after testing him, who had said, we'll be happy to double your salary.

788
00:58:00,530 --> 00:58:02,530
They were thinking they're going to have to go higher.

789
00:58:02,530 --> 00:58:04,110
and he moved on.

790
00:58:04,190 --> 00:58:06,150
He said, well, maybe I'll be here for a couple of years

791
00:58:06,150 --> 00:58:07,810
and have somebody else double it again.

792
00:58:09,470 --> 00:58:12,030
But he's way beyond me in AI right now.

793
00:58:12,710 --> 00:58:14,550
And as I said, I just started doing this,

794
00:58:14,590 --> 00:58:16,590
but the people who aren't trying to keep up,

795
00:58:17,370 --> 00:58:19,170
I don't know what's going to happen to you.

796
00:58:22,330 --> 00:58:24,430
Yeah, absolutely.

797
00:58:24,430 --> 00:58:29,710
And especially if you're in any kind of a STEM field, right?

798
00:58:29,710 --> 00:58:56,990
That's, that's the, the key piece there. If you're in one of those fields and you're not, you know, gaining experience playing with models and, you know, different ones, uh, you know, by, I would strongly urge you to do so. Right. And there, there are plenty of them out there. You can, you know, if you, if you don't have the, the funds for, you know, say a, you know, big subscription or something, you can go and self host.

799
00:58:56,990 --> 00:59:06,210
You can go to a virtual, you know, a virtual private server is what they're called or a VPS.

800
00:59:06,670 --> 00:59:09,830
You can get one for five bucks a month, right?

801
00:59:09,930 --> 00:59:11,610
That's a quarter of the cost.

802
00:59:11,670 --> 00:59:16,610
And most of them have a Docker panel that you can just load Claude into.

803
00:59:16,890 --> 00:59:21,930
So you've got your own self-hosted version of Claude right there for five bucks a month.

804
00:59:21,930 --> 00:59:27,470
and then you can, you know, you can kind of avoid some of the costs, right?

805
00:59:27,470 --> 00:59:32,470
So obviously you're limited in what it has available for information.

806
00:59:33,250 --> 00:59:36,370
But again, you know, to start out, that is something.

807
00:59:38,810 --> 00:59:42,330
Yeah. And sorry, I was getting a little dark where I am.

808
00:59:42,810 --> 00:59:48,370
Yeah. So one thing that I tell some of my students who really wanted to learn,

809
00:59:48,370 --> 00:59:57,910
I would tell them to go over to Udemy, U-D-E-M-Y, pick something they're interested in or pick something they're good at.

810
00:59:57,950 --> 01:00:02,270
Like one of my kids says, I don't like what I do, but I'm damn good at it.

811
01:00:03,290 --> 01:00:03,610
All right.

812
01:00:03,710 --> 01:00:04,670
Take the courses.

813
01:00:05,710 --> 01:00:10,350
Go to one of those job boards, figure out, you know, put up some things, get some jobs.

814
01:00:11,010 --> 01:00:13,450
And then one of my kids did that.

815
01:00:13,530 --> 01:00:17,190
And before you knew it, he was actually making enough money.

816
01:00:17,190 --> 01:00:20,090
and right now he's on a $200 a month Claude subscription,

817
01:00:20,270 --> 01:00:24,870
I mean, a Claude subscription and $20 a month on ChatGPT.

818
01:00:25,370 --> 01:00:28,670
And he's making thousands of dollars a month.

819
01:00:29,510 --> 01:00:32,570
And he says, I'm not going to college

820
01:00:32,570 --> 01:00:34,010
because on my current pace,

821
01:00:34,150 --> 01:00:36,970
I should be making $20,000, $30,000 a month

822
01:00:36,970 --> 01:00:38,250
within the next year.

823
01:00:39,870 --> 01:00:41,490
You know, he's moving past.

824
01:00:41,650 --> 01:00:41,910
Nice.

825
01:00:42,610 --> 01:00:44,330
He's moving, but he's just getting so much,

826
01:00:44,390 --> 01:00:46,410
but you can start doing it that way.

827
01:00:46,410 --> 01:00:51,430
I knew so many people that trained themselves to do the basics really well with Udame.

828
01:00:51,610 --> 01:00:53,510
And you never pay full price on Udame.

829
01:00:53,770 --> 01:00:55,110
There's always a sale.

830
01:00:55,270 --> 01:00:57,890
I had one of the kids I recommended.

831
01:00:58,410 --> 01:01:01,670
Mom came back and said, they have an 80% off sale.

832
01:01:02,530 --> 01:01:05,750
And she said, give me a list of courses for them.

833
01:01:05,750 --> 01:01:06,750
I said, okay.

834
01:01:07,750 --> 01:01:09,090
It was 23 courses.

835
01:01:11,090 --> 01:01:11,790
All right.

836
01:01:11,790 --> 01:01:16,030
And each course she spent about 20, 30 bucks on, which was a huge rate.

837
01:01:16,030 --> 01:01:22,570
a huge write down. She said for less than a thousand bucks, she went through and he went

838
01:01:22,570 --> 01:01:27,830
through all the courses several times, took like six months and came out and he started

839
01:01:27,830 --> 01:01:37,010
sending stuff out. One of the companies says they like the work. Are you local? He said,

840
01:01:37,090 --> 01:01:41,090
well, yeah, I'm 10 miles from you. They said, come in. They gave him another test. They just

841
01:01:41,090 --> 01:01:48,890
offered him a job. They said, hey, we'll pay. He's dyslectic. And they said, yeah, we'll pay

842
01:01:48,890 --> 01:01:55,450
200 bucks a month for this and 20 bucks for that. You need anything else? Why? Because it used to

843
01:01:55,450 --> 01:02:03,430
be we need an administrative assistant. You know, 50, 60,000 total cost, probably 80 grand a year

844
01:02:03,430 --> 01:02:08,810
now with health insurance and all the other stuff. They're happy to pay a couple hundred bucks a

845
01:02:08,810 --> 01:02:14,810
month. They're happy to pay 500 bucks a month. They said, compared to the quality of work they're

846
01:02:14,810 --> 01:02:20,070
getting, because it's so much more productive being, when you're dyslectic, when you're in

847
01:02:20,070 --> 01:02:23,950
your speciality and you know what you're doing, we have two and a half times that part of the

848
01:02:23,950 --> 01:02:28,450
neural activity in the brain, it becomes really unfair. And a lot of my former students are

849
01:02:28,450 --> 01:02:34,470
telling me that their employers are so happy to pay the extra money. They said, this is nothing.

850
01:02:34,470 --> 01:02:35,750
What else can we do for you?

851
01:02:36,850 --> 01:02:39,690
Because they're so much more productive.

852
01:02:40,370 --> 01:02:46,390
One of them came back and said, they went to a new employer and they said, I'm sorry, I'm so bad with my writing.

853
01:02:46,870 --> 01:02:48,050
They said, is there anything we can do?

854
01:02:48,090 --> 01:02:49,570
I said, well, if I could use Anthropic.

855
01:02:49,770 --> 01:02:53,090
They said, they looked at his work, looked at the productivity.

856
01:02:53,690 --> 01:02:58,730
They came to him and said, yeah, after a couple of weeks, we'll, what do we say we do the $200 seat version?

857
01:02:58,730 --> 01:03:07,070
And he's just thrilled because his work is still reviewed by an administrative assistant.

858
01:03:07,870 --> 01:03:16,850
But it's so, she uses AI, but it's so quick now because the quality, so they thought it was a, you know, they said, listen, for what you're producing, this is nothing.

859
01:03:16,850 --> 01:03:25,110
yeah and on top of that i mean a lot of companies nowadays if they aren't already there they're on

860
01:03:25,110 --> 01:03:33,410
their way to having their own uh you know ai depending on what version they go with etc but

861
01:03:33,410 --> 01:03:39,350
you know there's there's certainly that at a an ever-increasing number of organizations

862
01:03:39,350 --> 01:03:44,470
that have their own, at least one model internally, right?

863
01:03:44,570 --> 01:03:49,670
So, yeah, I definitely think it's something people should focus some time on.

864
01:03:49,670 --> 01:03:54,170
Well, the thing is, most people, most companies,

865
01:03:54,310 --> 01:03:57,650
if you actually go in and see what they're actually doing,

866
01:03:58,050 --> 01:04:00,590
the vast majorities of companies are,

867
01:04:01,070 --> 01:04:05,810
it's like they're operating in 2024 if they're lucky.

868
01:04:05,970 --> 01:04:07,290
Most of them are 2023.

869
01:04:09,350 --> 01:04:12,470
I mean, it's just they know nothing about it.

870
01:04:12,630 --> 01:04:17,970
And when they start, it's not even, the spread out there is next to nothing.

871
01:04:18,910 --> 01:04:20,490
And that's what shocks me.

872
01:04:22,150 --> 01:04:25,570
Yeah, that's changing rapidly.

873
01:04:26,150 --> 01:04:30,110
It's rapidly changing, but it's still so far.

874
01:04:31,110 --> 01:04:35,350
I go and I'm trying to explain the workflow I had back in October,

875
01:04:35,670 --> 01:04:38,250
which to me, again, is ancient times now.

876
01:04:38,250 --> 01:04:43,990
and people are like what do you mean what is this what is all it's just shocking

877
01:04:43,990 --> 01:04:52,630
okay i'd honestly be quite amazed if by 2030 over 90 of the companies out there aren't hosting

878
01:04:52,630 --> 01:04:59,270
their own models i really would be very amazed i think you're going to be amazed because while

879
01:04:59,270 --> 01:05:07,470
it's spreading a lot of people just don't they know a fraction of what i know and my web guy

880
01:05:07,470 --> 01:05:12,490
who's just brilliant at all forms, technical and marketing.

881
01:05:12,930 --> 01:05:15,550
I know a fraction of what he knows.

882
01:05:17,070 --> 01:05:20,790
Okay, I'm just, this is the workflow that I have to build myself,

883
01:05:20,890 --> 01:05:24,230
design for myself as a dyslexic so I can show others how to do it,

884
01:05:24,510 --> 01:05:25,710
especially my clients.

885
01:05:26,570 --> 01:05:33,130
And, but what I'm finding shocking is the kids that go out that I've trained in AI,

886
01:05:33,130 --> 01:05:40,290
They will go in, and as they said, they know a decent amount of what I know, but they don't keep up at the level that I am.

887
01:05:40,650 --> 01:05:47,630
They go in, and now all of a sudden they're training their whole team how to do AI because they don't even know what they know.

888
01:05:48,050 --> 01:05:53,330
They start talking about context, problem, solution, original problem statement and original solution.

889
01:05:53,510 --> 01:05:54,990
They're just shocked.

890
01:05:56,030 --> 01:05:59,810
And they go in, and their bosses think that they just hired some genius.

891
01:05:59,810 --> 01:06:03,810
Like this book came out 31 years ago.

892
01:06:03,970 --> 01:06:04,090
What?

893
01:06:04,870 --> 01:06:07,490
So yeah, the adoption is still very far.

894
01:06:07,590 --> 01:06:10,570
You have some that are really at an elite level.

895
01:06:11,010 --> 01:06:12,730
I mean, heck, there's one company I heard about.

896
01:06:13,030 --> 01:06:13,870
There's three guys.

897
01:06:14,590 --> 01:06:16,950
They spend $1,000 a day per employee.

898
01:06:17,810 --> 01:06:23,490
And the AI not only creates everything, it checks everything.

899
01:06:23,670 --> 01:06:24,630
And they just ship it.

900
01:06:26,030 --> 01:06:28,990
I mean, a lot of these companies, what really got me in sales

901
01:06:28,990 --> 01:06:30,950
was there's one particular company,

902
01:06:31,030 --> 01:06:34,270
they used to do $300,000 to $500,000 per employee,

903
01:06:34,610 --> 01:06:36,650
and the sales guy got paid like $80,000 a year.

904
01:06:37,130 --> 01:06:39,650
Now the same sales guy is doing $3,000 to $5 million,

905
01:06:40,390 --> 01:06:42,610
and they're paying him $300,000 a year.

906
01:06:43,950 --> 01:06:47,790
Because AI is doing everything to just keep putting them

907
01:06:47,790 --> 01:06:48,970
in front of people to speak to.

908
01:06:50,490 --> 01:06:51,890
It's doing all the other...

909
01:06:51,890 --> 01:06:55,550
So yeah, that's where it's going.

910
01:06:55,770 --> 01:06:57,750
I just want people to go ahead and prepare

911
01:06:57,750 --> 01:07:01,250
So you can be one of those on the right side of this.

912
01:07:01,250 --> 01:07:03,390
So you can do much better economically.

913
01:07:04,110 --> 01:07:08,130
And for those who us who are on the right side, we'll be just fine.

914
01:07:08,430 --> 01:07:10,410
I'm just scared about everybody else.

915
01:07:13,050 --> 01:07:14,270
Yeah, that makes sense.

916
01:07:15,330 --> 01:07:19,950
Well, hey, one more wonderful conversation.

917
01:07:20,350 --> 01:07:23,450
I think this is probably a good place to start winding it down.

918
01:07:23,450 --> 01:07:30,270
And, you know, just like last time, one of the questions we always ask our guests and we're going to hit you with it again.

919
01:07:30,470 --> 01:07:35,850
So you can keep the same response or different throughout our lives.

920
01:07:35,850 --> 01:07:42,750
There's someone that's made an impact, historical figures, sixth grade science teacher, author, mom, doesn't matter.

921
01:07:43,270 --> 01:07:48,350
But we like to give our guests the ability to shout out a person that was influential to them.

922
01:07:49,770 --> 01:07:53,030
His name is Professor James Holmes.

923
01:07:53,450 --> 01:08:10,930
And he's just, everybody knows quickly about his story. He didn't learn to read until he was in fifth grade. No way was he going to go on to college until in 11th grade he entered science fair competition in the United States in the 1950s, won the state of Indiana, got third place in the country.

924
01:08:10,930 --> 01:08:20,910
Then all of a sudden, he was an A student, went on to college, went on to get his PhD at the top economics program in the World University of Chicago with their top scholarship.

925
01:08:21,490 --> 01:08:24,730
He said he stopped counting rewriting his dissertation at 100.

926
01:08:26,670 --> 01:08:28,250
After that, he always had a co-author.

927
01:08:29,010 --> 01:08:31,730
And he taught me how to think in causality.

928
01:08:31,970 --> 01:08:35,170
If it wasn't for him, I would be nothing, nowhere.

929
01:08:35,170 --> 01:08:39,310
I wanted to go to SUNY College at Plattsburgh.

930
01:08:39,650 --> 01:08:40,610
It fit me perfectly.

931
01:08:40,930 --> 01:08:43,590
I went to Buffalo instead to work with him.

932
01:08:44,090 --> 01:08:55,290
And it was his intermediate macroeconomics class where we had analyzed the Great Depression and explain it, which we all did very poorly, even the best among us.

933
01:08:56,330 --> 01:09:04,770
And if it wasn't for him, I would have ended up, no, I would have probably been sweeping floors at McDonald's if it wasn't for him.

934
01:09:06,530 --> 01:09:07,410
Well, awesome.

935
01:09:08,450 --> 01:09:10,530
That's a good shout out right there.

936
01:09:10,930 --> 01:09:15,570
All right. What do you have going on? What are you working on? Where can people find you?

937
01:09:16,470 --> 01:09:21,470
Well, my book's about ready to come out. Famous last word of any author. We're doing our last

938
01:09:21,470 --> 01:09:27,410
round of edits. It's called, if you can write, you can read. It's coming out on Amazon Kindle

939
01:09:27,410 --> 01:09:34,630
within a couple of weeks. And for those of us who want to contact me, the best thing to do is go to

940
01:09:34,630 --> 01:09:40,010
dyslexiclasses.com. It's with an S, dyslexiclasses.com. There's a button that says download free guide.

941
01:09:40,010 --> 01:09:42,650
just fill out three questions and set up a time to speak with me.

942
01:09:43,010 --> 01:09:44,550
It's a half hour. It doesn't cost you anything.

943
01:09:46,370 --> 01:09:48,870
That's awesome. Awesome.

944
01:09:49,490 --> 01:09:52,950
Well, thank you once again for giving us another chunk of your time.

945
01:09:53,050 --> 01:09:55,390
We really appreciate it. And I think, you know,

946
01:09:55,410 --> 01:09:57,630
we'll probably have you back on here again as well.

947
01:09:58,070 --> 01:09:58,970
I'd be happy to do that.

948
01:09:59,830 --> 01:10:01,730
Yeah. We'll wait for your book to come out.

949
01:10:01,810 --> 01:10:06,730
Maybe we can talk about your book and get into that and listen to what that

950
01:10:06,730 --> 01:10:08,630
process was like and everything else.

951
01:10:08,630 --> 01:10:13,470
I will tell you as a dyslectic it was very interesting

952
01:10:13,470 --> 01:10:16,330
oh I I can imagine

953
01:10:16,330 --> 01:10:25,470
this is very structured writing and and yeah okay you've got the creative form and yeah it's kind

954
01:10:25,470 --> 01:10:32,310
of a mix of the frontal and the the you know the rear look yeah you're you're probably oh no

955
01:10:32,310 --> 01:10:34,790
Really what I did is I wrote it with AI.

956
01:10:35,010 --> 01:10:37,790
It was ChatGPT 03 Pro at the time.

957
01:10:38,470 --> 01:10:42,010
Then I had two ghostwriters completely rewrite it.

958
01:10:42,470 --> 01:10:44,510
And then I went to one of the top co-authors.

959
01:10:44,750 --> 01:10:45,930
Her name was Evelyn White Bay.

960
01:10:45,990 --> 01:10:48,250
She was on the New York City Dyslexia Task Force.

961
01:10:48,750 --> 01:10:49,950
And then she rewrote it.

962
01:10:50,270 --> 01:10:51,810
And now we're going through a final edit.

963
01:10:51,910 --> 01:10:53,570
So that was generally my process.

964
01:10:55,030 --> 01:10:55,990
Well, that's awesome.

965
01:10:57,130 --> 01:11:00,070
And no, AI is nowhere close to writing a book on its own.

966
01:11:00,070 --> 01:11:04,430
it's good to get the first instead of going to a ghostwriter where they asked

967
01:11:04,430 --> 01:11:04,590
you,

968
01:11:04,670 --> 01:11:06,230
what are your visions for the book?

969
01:11:06,290 --> 01:11:06,990
What are you thinking?

970
01:11:07,170 --> 01:11:09,210
I just wrote it with AI and I said here.

971
01:11:09,210 --> 01:11:12,890
And then she came back with something that it didn't look anywhere close to

972
01:11:12,890 --> 01:11:13,130
it,

973
01:11:13,130 --> 01:11:14,930
but it gave her a lot.

974
01:11:15,050 --> 01:11:16,750
It was a lot easier for her to work with.

975
01:11:16,810 --> 01:11:17,310
She just said,

976
01:11:17,370 --> 01:11:20,250
there's a long list of stuff that needs to be corrected,

977
01:11:20,470 --> 01:11:22,150
but at least she had a draft to go from.

978
01:11:23,410 --> 01:11:23,850
Nice.

979
01:11:24,510 --> 01:11:24,950
Wonderful.

980
01:11:25,930 --> 01:11:26,590
All right,

981
01:11:26,610 --> 01:11:26,990
everybody.

982
01:11:26,990 --> 01:11:32,910
thank you for joining us for another episode of the space tomatoes podcast we're everywhere you

983
01:11:32,910 --> 01:11:39,550
know where to find us we are now also hosted off of fountain.fm go check us out we've got a chat

984
01:11:39,550 --> 01:11:48,370
group in there as well if you'd like to join just go log on join our chat that'll be fun if you like

985
01:11:48,370 --> 01:11:53,870
us give us thumbs up if you didn't give us thumbs down leave a comment tell us we suck i don't care

986
01:11:53,870 --> 01:12:03,870
right uh and remember wherever you are whenever you're watching this have an amazing day

987
01:12:03,870 --> 01:12:06,870
thanks for coming russell thanks for having me

988
01:12:23,870 --> 01:12:25,150
You