1
00:00:01,080 --> 00:00:04,839
Speaker 1: How'd you like to listen to dot NetRocks with no ads? Easy?

2
00:00:05,360 --> 00:00:08,560
Become a patron for just five dollars a month. You

3
00:00:08,599 --> 00:00:11,320
get access to a private RSS feed where all the

4
00:00:11,359 --> 00:00:14,599
shows have no ads. Twenty dollars a month will get

5
00:00:14,599 --> 00:00:18,440
you that and a special dot NetRocks patron mug. Sign

6
00:00:18,519 --> 00:00:34,719
up now at Patreon dot dot NetRocks dot com.

7
00:00:34,840 --> 00:00:37,159
Speaker 2: Hey Philly, it's dot net rock.

8
00:00:46,240 --> 00:00:46,479
Speaker 3: Wow.

9
00:00:47,079 --> 00:00:49,159
Speaker 1: Yeah, so long since we've been in front of a

10
00:00:49,200 --> 00:00:50,039
real crowd.

11
00:00:50,119 --> 00:00:51,200
Speaker 3: That a live show man.

12
00:00:51,880 --> 00:00:55,000
Speaker 1: While I drove down here from New London, it's about

13
00:00:55,039 --> 00:00:58,159
four and a half hours and I had road trip flashbacks.

14
00:00:58,320 --> 00:01:02,079
Speaker 3: Yeah, yeah, no kidding. It was one of the I

15
00:01:02,119 --> 00:01:06,120
think we stopped here on every tour, Yes, every single one,

16
00:01:06,120 --> 00:01:07,920
from the very first one two thousand and five, which

17
00:01:07,959 --> 00:01:11,840
is entirely your idea because I was brand new, twenty eight,

18
00:01:11,959 --> 00:01:15,439
twenty ten, twenty twelve, twenty thirteen. Yeah, yeah, everyone. We

19
00:01:15,519 --> 00:01:17,319
always came to Philly. Sometimes we got to do a

20
00:01:17,319 --> 00:01:20,760
whole day, like we do a day of dot net thing.

21
00:01:20,879 --> 00:01:22,799
Sometimes it would just be the evening show. But you know,

22
00:01:22,840 --> 00:01:24,719
you could always count on Philly folks to come out

23
00:01:24,840 --> 00:01:27,359
and have a great time. So this is a special one.

24
00:01:27,400 --> 00:01:29,400
It's a one off, it's one of a kind, and

25
00:01:29,959 --> 00:01:30,840
we're gonna have some fun.

26
00:01:30,920 --> 00:01:34,480
Speaker 1: We are going to have some fun. Before I introduce

27
00:01:34,599 --> 00:01:37,519
the other two people who are here with us, let's

28
00:01:37,599 --> 00:01:38,719
talk about.

29
00:01:39,439 --> 00:01:40,159
Speaker 3: Nineteen eighty one.

30
00:01:40,239 --> 00:01:42,799
Speaker 1: Nineteen eighty one. All right, so do you know what

31
00:01:42,840 --> 00:01:43,959
happened in nineteen eighty one?

32
00:01:44,040 --> 00:01:46,959
Speaker 3: A few things, tell us, but I focused purely on

33
00:01:47,040 --> 00:01:49,680
space and technology topics. Well, go ahead, Well it's nineteen

34
00:01:49,719 --> 00:01:51,400
eighty one, so it is the beginning of the Space

35
00:01:51,400 --> 00:01:55,959
Shuttle program. Well, the successful launch after years of delay. Remember,

36
00:01:56,000 --> 00:01:58,760
Shuttle was supposed to rescue sky Lab and they lost

37
00:01:58,799 --> 00:02:01,079
that in seventy nine. This is nineteen eighty one. They

38
00:02:01,079 --> 00:02:03,879
missed it by a few years. So STS one flies

39
00:02:03,879 --> 00:02:07,000
in April with Young Crippon on board. They for fifty

40
00:02:07,040 --> 00:02:13,240
four hours, thirty seven orbits and a totally successful flight, really,

41
00:02:13,400 --> 00:02:17,879
no no drama, great test flight. Famously, it was Crippen.

42
00:02:18,439 --> 00:02:20,680
Young was the guy who flew everything. He was in Gemini,

43
00:02:20,719 --> 00:02:23,759
he was in Apollo, but Crippen was the new guy.

44
00:02:24,080 --> 00:02:25,960
And as they were bringing it in for a landing,

45
00:02:26,039 --> 00:02:27,759
says how it FLI And he said flies like a

46
00:02:27,800 --> 00:02:31,599
brick shit house because it did right, This a one

47
00:02:31,680 --> 00:02:34,159
hundred ton vehicle with stubby little wingstune two hundred miles

48
00:02:34,280 --> 00:02:36,759
nuts at hour. You know, it's stuffed a landa thing. Anyway,

49
00:02:36,759 --> 00:02:38,800
they had to edit that so if you ever see

50
00:02:38,800 --> 00:02:40,919
the quote from Robert Crippen it's flies like a brick,

51
00:02:42,039 --> 00:02:45,879
that is not not what he said Later that year

52
00:02:46,039 --> 00:02:50,199
sts two. So they turned Columbia around in about six

53
00:02:50,240 --> 00:02:53,800
months and flew it again in November. This time it

54
00:02:53,840 --> 00:02:58,560
was angling truly. Engel was supposed to be on Apollo seventeen.

55
00:02:58,599 --> 00:03:01,000
He got bumped by Harrish and Schmidt, the only scientists

56
00:03:01,000 --> 00:03:04,240
to ever walk on the Moon. And so they flew

57
00:03:04,319 --> 00:03:08,400
Columbia again, and the main job of the second flight

58
00:03:08,560 --> 00:03:10,400
was to test out the Canada Arm because their first

59
00:03:10,560 --> 00:03:12,000
this was back when they thought they were going to

60
00:03:12,039 --> 00:03:13,719
fly these things every two weeks and they were going

61
00:03:13,759 --> 00:03:16,280
to deploy satellites from them, repair satellites from do all

62
00:03:16,319 --> 00:03:19,080
this cool stuff right. It was a space transport system

63
00:03:19,120 --> 00:03:21,159
except for all that part where did that didn't work?

64
00:03:22,199 --> 00:03:24,800
And so they this was the trials for the It

65
00:03:24,879 --> 00:03:26,599
was going to be five days up there working on

66
00:03:26,639 --> 00:03:30,520
the Canada ARM And on the first day they had

67
00:03:30,520 --> 00:03:33,120
a field cell failure and so they were low on power.

68
00:03:33,159 --> 00:03:34,919
They couldn't operate the arm, so they cut the mission

69
00:03:34,960 --> 00:03:36,800
down to two days and turn around came back.

70
00:03:37,319 --> 00:03:40,800
Speaker 4: Was that the flights with the white painted external tank.

71
00:03:40,840 --> 00:03:43,719
Speaker 1: This is when the external light was still Another voice,

72
00:03:43,840 --> 00:03:46,159
what's going on here? Oh, that's Jeff Fritz.

73
00:03:46,159 --> 00:03:48,719
Speaker 3: That's Jeff Frid's okay, Yeah, So that that's back when

74
00:03:48,719 --> 00:03:52,599
the tanks were painted white, and that paint weighed something

75
00:03:52,599 --> 00:03:54,800
like fifteen hundred pounds. So that's why they lost it eventually.

76
00:03:54,840 --> 00:03:56,879
Speaker 4: It was just weighs that much.

77
00:03:56,919 --> 00:03:58,240
Speaker 3: When you get that much of it on there.

78
00:03:58,120 --> 00:04:00,520
Speaker 1: Well, you can dump the cans overboard once you get space.

79
00:04:00,599 --> 00:04:04,240
So and Columbia was always overweight, right, she was. She

80
00:04:04,439 --> 00:04:06,479
was just the first flying vehicle. And that's why she

81
00:04:06,520 --> 00:04:08,080
never went to the space station. She couldn't get to it.

82
00:04:08,599 --> 00:04:10,919
That didn't have any that mass, didn't have enough energy

83
00:04:10,919 --> 00:04:11,280
get there.

84
00:04:11,479 --> 00:04:13,280
Speaker 5: That was the space shuttles that had.

85
00:04:13,159 --> 00:04:14,000
Speaker 3: Space shttle Columbia.

86
00:04:14,080 --> 00:04:16,480
Speaker 5: Yeah, and that's the one with the bricks underneath it.

87
00:04:16,800 --> 00:04:20,480
Speaker 3: With the well, and that would crumble in your hands

88
00:04:20,519 --> 00:04:23,000
like they would crush. Right, they're very delicate. It was

89
00:04:23,079 --> 00:04:23,839
ridiculous to delicate.

90
00:04:23,879 --> 00:04:26,000
Speaker 1: And another voice you heard was Bill Wolf.

91
00:04:29,560 --> 00:04:33,839
Speaker 3: And uh, the geek out start because of me ranting

92
00:04:33,879 --> 00:04:36,519
at Carl in twenty eleven when Atlantis landed for the

93
00:04:36,600 --> 00:04:38,040
last time. Well he was in a bar with some

94
00:04:38,079 --> 00:04:41,600
of this, yes, and uh and just how screwed up?

95
00:04:41,720 --> 00:04:43,639
You know, space Shuttle never really did what was supposed

96
00:04:43,680 --> 00:04:45,680
to do, all the missions that it could never actually do.

97
00:04:45,920 --> 00:04:48,319
Speaker 1: And and Richard will never tell you this, but he

98
00:04:48,439 --> 00:04:51,000
had the Space Shuttle manual and read it cover to

99
00:04:51,040 --> 00:04:55,480
cover so he could tell you everything about the Space Shuttle.

100
00:04:55,680 --> 00:04:57,519
Speaker 3: Uh, not entirely true, but okay.

101
00:04:58,000 --> 00:04:58,120
Speaker 1: Uh.

102
00:04:58,279 --> 00:05:01,399
Speaker 3: Anyway, he talked me into let's do that as a show.

103
00:05:01,639 --> 00:05:03,800
And I thought you guys would crucify us for that

104
00:05:03,839 --> 00:05:06,240
because it was not exactly software. And I was wrong.

105
00:05:07,399 --> 00:05:10,839
The geekout and I kept doing them. And this is

106
00:05:10,879 --> 00:05:14,360
eighty this is eighty one. This is the end of November,

107
00:05:14,639 --> 00:05:16,680
end of December show. The next two shows will be

108
00:05:16,759 --> 00:05:19,079
the annual Geekouts. So I've been working all my scripts

109
00:05:19,120 --> 00:05:21,680
for an update on space and an update on energy generation.

110
00:05:22,240 --> 00:05:25,079
So that's coming last space topic and then we'll talk computers.

111
00:05:25,839 --> 00:05:30,360
Voyage two does its flyby of Saturn in August, so

112
00:05:30,680 --> 00:05:33,560
one of the pioneers already coming gone. But this is

113
00:05:33,560 --> 00:05:35,160
the second time we got to look at Saturn Voyager

114
00:05:35,199 --> 00:05:37,839
when I gotten there the couple years before, and we

115
00:05:37,920 --> 00:05:40,079
got some better images, and that sets us up for

116
00:05:40,240 --> 00:05:44,800
eventually Cassini. And this is when they figure out that

117
00:05:44,800 --> 00:05:49,959
that titan very likely has a prototypical atmosphere and increases

118
00:05:49,959 --> 00:05:53,079
the excitement for building bigger missions around all right. On

119
00:05:53,120 --> 00:05:55,759
the computer side, nineteen eighty one is the year that

120
00:05:55,839 --> 00:05:59,000
Sony introduces the three and a half inch floppy disco.

121
00:06:00,079 --> 00:06:03,480
Beginning of all of that, there was competing sizes. There

122
00:06:03,480 --> 00:06:05,279
were three inch versions and three and a quarters as

123
00:06:05,279 --> 00:06:06,959
well as the three and a half and a three

124
00:06:07,000 --> 00:06:09,000
and a half ultimately be the one that was adopted everywhere.

125
00:06:09,040 --> 00:06:11,120
Not so floppy, though not a flop at all, quite

126
00:06:11,160 --> 00:06:12,839
rigid actually, which part of the reason it worked so well.

127
00:06:12,879 --> 00:06:15,240
They used the Bernoulli effect inside that rigid housing so

128
00:06:15,279 --> 00:06:18,000
they could increase I density. The original discs were only

129
00:06:18,000 --> 00:06:19,800
three hundred and sixty K, then they went to seven

130
00:06:19,839 --> 00:06:24,279
twenty and then one point four megabytes, so much storage,

131
00:06:24,439 --> 00:06:27,439
but by far easily most important story nineteen eighty one

132
00:06:27,680 --> 00:06:30,480
the release of the IBM fifty one fifty which you

133
00:06:30,600 --> 00:06:33,240
know is the IBM PC four point seven to seven

134
00:06:33,319 --> 00:06:38,839
MEGAHURTZ processor. The eighty eighty eight processor by default sixteen

135
00:06:38,920 --> 00:06:40,480
k ARAM, but you could expand it to two hundred

136
00:06:40,480 --> 00:06:42,560
fifty six. Later we would come up with an expansion

137
00:06:42,600 --> 00:06:43,480
burd to take to six.

138
00:06:43,439 --> 00:06:45,759
Speaker 4: Forty, but need more than two hundred and fifty six.

139
00:06:45,920 --> 00:06:49,079
Speaker 3: Well really sixteen k right, basic configuration, but you could

140
00:06:49,120 --> 00:06:53,360
run twos that's all there was, of course, shipped with

141
00:06:53,439 --> 00:06:57,600
CPM eighty six, or you could run this brand new

142
00:06:57,680 --> 00:07:01,000
shiny ms DOUS one point zero. Why did they do

143
00:07:01,040 --> 00:07:02,959
the eighty eighty eight instead of the eighty eighty six

144
00:07:03,240 --> 00:07:06,040
because it was cheaper and more readily available. The eighty

145
00:07:06,040 --> 00:07:07,680
six ers were in high demand and they wanted to

146
00:07:07,720 --> 00:07:09,240
build a lot of these machines. They were trying to

147
00:07:09,240 --> 00:07:11,600
get faster to the market, and so the eighty eighty

148
00:07:11,759 --> 00:07:14,000
eight was more readily available. It also made the machine

149
00:07:14,079 --> 00:07:17,000
cheaper because it only had an eight bit input bus,

150
00:07:17,000 --> 00:07:18,879
and so all of the bus structures were eight bit

151
00:07:18,879 --> 00:07:21,800
instead of sixteen bit. That will be changed with the

152
00:07:21,879 --> 00:07:23,800
later models, with the XT and so on.

153
00:07:24,000 --> 00:07:25,439
Speaker 1: The turbo, remember the turbo, but.

154
00:07:25,519 --> 00:07:31,120
Speaker 3: Turbo button go up to eight megahertz and dokey go like, yes,

155
00:07:31,199 --> 00:07:34,160
do not be processor cycle synchronized. That's not a right

156
00:07:34,199 --> 00:07:38,920
idea idea. It was always intended to be licensable. And

157
00:07:39,040 --> 00:07:42,600
so within a year Columbia Data Products makes the first

158
00:07:42,600 --> 00:07:47,360
of the PC compatibles and all of our careers. Yeah,

159
00:07:47,399 --> 00:07:50,079
also in the same year. And I never realized this

160
00:07:50,120 --> 00:07:52,399
happened in the same year. You always thought this was

161
00:07:52,439 --> 00:07:56,959
older the Osbourne one. The Osbourne one came out in

162
00:07:57,040 --> 00:07:59,399
nineteen eighty one. So this was what we call a luggable.

163
00:08:00,079 --> 00:08:03,720
It was literally like a suitcase keyboard attached onto the bottom,

164
00:08:03,759 --> 00:08:05,399
had a five inch displaying in the middle, a pair

165
00:08:05,439 --> 00:08:07,439
of five and a quarter floppy drives on each side,

166
00:08:07,439 --> 00:08:12,439
weigh about twenty five pounds speakers. Yeah, I remember having it.

167
00:08:12,480 --> 00:08:15,240
I had a compact, but not in Osborne in Osbourne.

168
00:08:15,319 --> 00:08:16,680
And there's a good reason for that, because Osborne were

169
00:08:16,759 --> 00:08:20,079
quickly out of business by Osborne in themselves. Yes, would

170
00:08:20,120 --> 00:08:22,120
just say as they started selling their first machine, they

171
00:08:22,160 --> 00:08:24,519
started talking about what was going to be a second machine. No,

172
00:08:24,560 --> 00:08:27,639
I was born too, and that stopped sales of Osborne

173
00:08:27,639 --> 00:08:30,240
one because everybody was waiting for that, and that bankrupted

174
00:08:30,279 --> 00:08:32,360
the company. So they never built the Osbourne two, and

175
00:08:32,399 --> 00:08:36,799
they became a verb. Do not Osborne yourself. Don't talk

176
00:08:36,799 --> 00:08:39,840
about your new product till you're ready to sell it.

177
00:08:39,840 --> 00:08:41,919
Speaker 1: It's okay that we talk about the geek outs because

178
00:08:42,120 --> 00:08:42,919
our show is free.

179
00:08:43,039 --> 00:08:45,679
Speaker 3: Yeah, there you go. One last thing on computing. We'll

180
00:08:45,679 --> 00:08:50,759
move on the IETF. That's the Internet Task Force publishes

181
00:08:50,840 --> 00:08:54,639
AREC Specification seven nine to one, which is a definition

182
00:08:54,960 --> 00:08:56,080
of IPv four.

183
00:08:56,480 --> 00:08:57,240
Speaker 1: Oh there you go.

184
00:08:57,519 --> 00:08:58,320
Speaker 3: Oh yeah.

185
00:08:58,399 --> 00:08:59,919
Speaker 1: A lot of things happened in nineteen eighty one.

186
00:09:00,120 --> 00:09:00,759
Speaker 3: That's the beginning.

187
00:09:01,000 --> 00:09:04,519
Speaker 1: And in the cultural side, I have three things. Ronald

188
00:09:04,519 --> 00:09:12,120
Reagan inaugurated in December, shot in March. Nice yeah, sorry sorry, no, no, no,

189
00:09:12,159 --> 00:09:18,279
it wasn't nice. I mean, yeah, it wasn't nice. MTV debuts.

190
00:09:19,320 --> 00:09:24,120
I want my MTV Gary Newman here in my cars? Right,

191
00:09:24,320 --> 00:09:28,600
was the first video Killed the Radio Star? And also

192
00:09:28,840 --> 00:09:32,159
speaking of Donkey Kong that came out in nineteen eighty one. Awesome,

193
00:09:32,720 --> 00:09:37,919
But if you played it in turbomodate one and that's it,

194
00:09:38,039 --> 00:09:40,919
So I guess we should go for better no framework, right,

195
00:09:41,480 --> 00:09:52,080
all right? Play the crazy music? All right, buddy, what

196
00:09:52,080 --> 00:09:57,519
do you got? So I've got an interesting GitHub repo

197
00:09:57,919 --> 00:10:00,240
that's trending, and I thought it was topical for what

198
00:10:00,279 --> 00:10:02,600
we're going to talk about today. This is from the

199
00:10:02,679 --> 00:10:07,080
dot Mac and it's called claud dash mem. So this

200
00:10:07,120 --> 00:10:12,639
is persistent memory compression system built for Claud code. So

201
00:10:12,799 --> 00:10:16,799
it's a plugin. It's a plugin for Claud code that

202
00:10:16,919 --> 00:10:22,600
preserves context across sessions by automatically capturing tool usage observations,

203
00:10:22,720 --> 00:10:26,600
generating semantic summaries, and making them available to future sessions.

204
00:10:27,480 --> 00:10:31,679
So it's like client side context, right. It squishes it

205
00:10:31,720 --> 00:10:35,000
all down and then on every other prompt it sends

206
00:10:35,000 --> 00:10:39,559
it back. So it allows claud to maintain continuity of

207
00:10:39,720 --> 00:10:44,480
knowledge about projects even after sessions end or reconnect. So

208
00:10:44,559 --> 00:10:46,519
I kind of like that because you know, keeping all

209
00:10:46,519 --> 00:10:49,399
that context is kind of expensive these days. But we're

210
00:10:49,399 --> 00:10:52,519
going to talk about all that stuff coming up here.

211
00:10:54,120 --> 00:10:56,320
But first, I guess the next thing we do is

212
00:10:56,559 --> 00:10:57,919
talk about comments.

213
00:10:57,919 --> 00:11:00,320
Speaker 3: Who's talking to us, Richard Hi grab a comment off

214
00:11:00,360 --> 00:11:04,399
show set nineteen seventy nine, So that's brand new, and

215
00:11:04,440 --> 00:11:06,759
that's the show we did with Callum Simpson talking a

216
00:11:06,799 --> 00:11:10,799
little bit about yak Shaver and our ongoing explorer defined

217
00:11:11,200 --> 00:11:14,080
apps that are actually out in the market using AI technologies.

218
00:11:14,519 --> 00:11:17,399
This comment comes from Blackweb, and I'm going to summarize

219
00:11:17,440 --> 00:11:19,440
them just a bit, but he says, on a recent project,

220
00:11:19,480 --> 00:11:21,679
I gave some AI tools that you discussed or try.

221
00:11:21,799 --> 00:11:24,960
We're trying to build at ten app using cloud and

222
00:11:25,039 --> 00:11:27,960
Aspire Native application in VS twenty twenty six, brand new

223
00:11:28,120 --> 00:11:31,919
VS code. The I tools included Copilot for Windows eleven,

224
00:11:32,000 --> 00:11:35,039
get hub Copilot with GPT five and Claude four point five.

225
00:11:35,120 --> 00:11:37,200
Copilot was helpful in all aspects of the project for

226
00:11:37,200 --> 00:11:39,919
coding and research, but it had trouble fixing errors and bugs.

227
00:11:40,320 --> 00:11:41,960
Often gets stuck in a loop and can't get out

228
00:11:41,960 --> 00:11:44,559
of it without some human help. When the Claude four

229
00:11:44,639 --> 00:11:47,159
point five plugin is at it, it gets even faster and

230
00:11:47,200 --> 00:11:49,960
better at all aspects of coding, including script generation, which

231
00:11:49,960 --> 00:11:52,399
it excels at. And now for the bad stuff. Get

232
00:11:52,480 --> 00:11:56,399
Hub's pricing model and billing is aggressive and predatory. It

233
00:11:56,440 --> 00:11:59,879
counts premium requests, which is basically anything you send in

234
00:12:00,039 --> 00:12:02,799
in VS and VS code. You starting free trial budget

235
00:12:02,879 --> 00:12:05,120
is twenty five requests, which most developers will use up

236
00:12:05,200 --> 00:12:08,159
in about an hour. You can set the dollar limit

237
00:12:08,159 --> 00:12:10,960
t additional requests for various purposes, including gethub actions and

238
00:12:11,000 --> 00:12:13,120
so on. For Copilot pro, you get three hundred quests

239
00:12:13,120 --> 00:12:15,759
for ten bucks, then of course they're all premium requests

240
00:12:16,120 --> 00:12:18,559
and you're showing a progress bar. I used up three

241
00:12:18,639 --> 00:12:21,879
point seven of my requests, so that would be like

242
00:12:22,039 --> 00:12:25,919
ten in a few hours. I quickly became paranoid about

243
00:12:25,960 --> 00:12:28,240
exceeding my request budget and running up a big bill.

244
00:12:29,360 --> 00:12:31,799
Claud's pricing model is even more aggressive than Githubs, starting

245
00:12:31,799 --> 00:12:33,799
it at seventeen dollars a month, it must be paid

246
00:12:33,799 --> 00:12:36,879
in advance for about two hundred bucks a year. I

247
00:12:36,919 --> 00:12:40,120
was impressed by Githubs, Copilot, and Claudes coding ability, but

248
00:12:40,159 --> 00:12:42,519
I immediately felt like it was aggressively trying to run

249
00:12:42,600 --> 00:12:45,840
up the request score as fast as possible. It is

250
00:12:45,879 --> 00:12:48,639
able to edit my code window, generate new files, enter

251
00:12:48,679 --> 00:12:52,759
in excute CLI commands, and rapidly generate tests, all of

252
00:12:52,799 --> 00:12:56,120
which eats into that budget. Get help Copilot did generate

253
00:12:56,159 --> 00:12:57,759
new code that worked, but it did it so fast

254
00:12:57,799 --> 00:12:59,240
that I had no idea what it had done, and

255
00:12:59,360 --> 00:13:02,080
had essentially no knowledge or understanding of the code degenerated,

256
00:13:02,279 --> 00:13:04,879
making me dependent on the AI for fixing any bugs

257
00:13:04,879 --> 00:13:09,799
it created, which of course consumes more credits. After a

258
00:13:09,799 --> 00:13:12,480
few hours with giphub, Copilot and Cloud, I felt like

259
00:13:12,519 --> 00:13:14,799
I'd been replaced by it and had become a little

260
00:13:14,799 --> 00:13:16,919
more than a human moderator or a monitor. Watched it

261
00:13:16,960 --> 00:13:19,399
work and gave minimal input in suggestions while running up

262
00:13:19,440 --> 00:13:22,679
AI tool generated bills of hundreds or maybe thousands of

263
00:13:22,679 --> 00:13:25,480
dollars per week. I also wondered if it was being

264
00:13:25,559 --> 00:13:28,279
used to train AI tools to replace me and other developers.

265
00:13:28,399 --> 00:13:31,759
AI coding tools are not creating bells developers. They're creating

266
00:13:31,840 --> 00:13:35,519
unemployed or underemployed AI coding tool addicts who are stuck

267
00:13:35,559 --> 00:13:39,360
in their exponentially increasing bill for those tools if they

268
00:13:39,360 --> 00:13:42,720
are independent coders. And I've builted the many podcasts on

269
00:13:42,720 --> 00:13:44,960
AI coding on dot in Rocks, I'm increasing concerned and

270
00:13:45,000 --> 00:13:48,679
baffled by the endless fanboyism for a by the two

271
00:13:48,720 --> 00:13:51,080
people I like and respect very much, Carl and Richard.

272
00:13:51,480 --> 00:13:53,559
You guys need to stop drinking AI kool aid and

273
00:13:53,559 --> 00:13:56,080
start talking to real developers who workflows and coding skills

274
00:13:56,080 --> 00:13:58,200
are being adversely affected by AA coding tools that you're

275
00:13:58,240 --> 00:14:00,919
providing with free blowing and biased reviews.

276
00:14:01,799 --> 00:14:04,000
Speaker 1: Yeah, you know, I've been waiting for a comment like

277
00:14:04,080 --> 00:14:06,720
this because there is that sentiment out there and it's

278
00:14:06,759 --> 00:14:11,440
probably underrepresented by our audience. So, first of all, a

279
00:14:11,559 --> 00:14:13,879
black Web, thank you very much for being brave enough

280
00:14:13,919 --> 00:14:17,080
to send that to us. And yeah, I mean, I

281
00:14:17,120 --> 00:14:21,039
feel your pain. However, I don't share your experience, and

282
00:14:21,080 --> 00:14:26,840
I shared my experience, which is, you know, mostly positive.

283
00:14:26,919 --> 00:14:29,320
There are times when I don't understand the code that

284
00:14:29,320 --> 00:14:31,759
the AIS written, but I usually ask it to comment

285
00:14:32,320 --> 00:14:34,759
any code that it writes so that I can understand it.

286
00:14:35,480 --> 00:14:40,279
I also develop in pieces. If I say need a

287
00:14:40,279 --> 00:14:43,399
full stack editor for some classes, I'll start with the models,

288
00:14:43,440 --> 00:14:46,200
than the data layer, than any kind of manager, then

289
00:14:46,240 --> 00:14:50,039
finally the UI. I might break all these down into

290
00:14:50,120 --> 00:14:53,240
different prompts if they're too complex. I didn't always think this,

291
00:14:53,440 --> 00:14:56,320
but now I believe in baby steps and it works

292
00:14:56,320 --> 00:14:58,559
for me and my agent of choice is the co

293
00:14:58,759 --> 00:15:02,320
pilot CLI and clauds on at four five. I never

294
00:15:02,399 --> 00:15:04,399
felt like it was taking over my job. It's only

295
00:15:04,399 --> 00:15:07,600
made me more productive, and he basically came back by saying,

296
00:15:08,120 --> 00:15:11,039
can we hear a balanced view of AI coding tools

297
00:15:11,080 --> 00:15:15,320
from actual developers, not AI apologists who are trying to

298
00:15:15,360 --> 00:15:17,559
sell a very expensive product. And I said, well, I

299
00:15:17,600 --> 00:15:21,559
am a real developer with real customers. The money they

300
00:15:21,559 --> 00:15:24,120
pay me and the productivity games far out weigh the

301
00:15:24,200 --> 00:15:27,759
costs of Copilot. I'm not an apologists for any technology

302
00:15:27,840 --> 00:15:30,840
or any company. I'm an independent developer and a podcaster.

303
00:15:30,960 --> 00:15:35,039
And he basically said, you know, you make a good point.

304
00:15:35,159 --> 00:15:38,360
Maybe I just need to figure out more efficient ways

305
00:15:38,360 --> 00:15:38,840
to use this.

306
00:15:39,080 --> 00:15:42,919
Speaker 5: Well, so time is money, yep. And if this stuff

307
00:15:42,960 --> 00:15:46,759
saves you time, that's worth a lot. And you could

308
00:15:46,759 --> 00:15:51,840
easily justify paying charges for things like Claude if it's

309
00:15:51,879 --> 00:15:53,679
saving you hours of time.

310
00:15:53,720 --> 00:15:55,240
Speaker 1: Yeah, and therefore making you money.

311
00:15:55,320 --> 00:15:58,240
Speaker 3: Yes, I'm working with teams that are spending thousand dollars

312
00:15:58,320 --> 00:16:01,840
a month. Yeah, but they're cranking through six weeks prints

313
00:16:02,240 --> 00:16:07,240
in days like they're moving code fast. It isn't. And

314
00:16:07,279 --> 00:16:09,720
then my problem is and there's two things that really

315
00:16:09,799 --> 00:16:12,600
hit me with black webs comment and all about billing,

316
00:16:12,759 --> 00:16:16,360
Which is first is this feels like telcos We're only

317
00:16:16,399 --> 00:16:18,559
going to find out after the fact how much it

318
00:16:18,600 --> 00:16:21,320
costs you, and the fact that you're learning on the

319
00:16:21,360 --> 00:16:24,320
back of it and it's fixing itself means it makes

320
00:16:24,360 --> 00:16:26,720
a mess then cleans it up for you and gets

321
00:16:26,759 --> 00:16:30,320
paid for it like that is pretty annoying. Yeah, right,

322
00:16:30,399 --> 00:16:32,679
But the bigger thing to me is we still don't

323
00:16:32,720 --> 00:16:36,720
have real pricing. Right. This is the blue ocean phase

324
00:16:36,759 --> 00:16:40,639
of this technology where they're low buying the figures just

325
00:16:40,679 --> 00:16:44,120
to get us on board. It will be very interesting

326
00:16:44,120 --> 00:16:46,679
to see what the real economics this looks like when

327
00:16:46,720 --> 00:16:50,320
the crazy ends and everybody actually has to pay the piper, Like,

328
00:16:50,559 --> 00:16:52,600
is this actually going to be viable? I don't know

329
00:16:52,639 --> 00:16:53,279
the answer to that.

330
00:16:53,639 --> 00:16:56,080
Speaker 1: It's so we have before we have a discussion that

331
00:16:56,200 --> 00:16:59,360
we need to wrap up the comment because that's basically

332
00:16:59,360 --> 00:17:01,159
what this whole show and.

333
00:17:01,200 --> 00:17:04,240
Speaker 3: Black Web you're timing is impeccable having provide us with

334
00:17:04,279 --> 00:17:06,880
a comment that you literally says a foundation for the

335
00:17:06,960 --> 00:17:09,480
show about what is AI going to look do to

336
00:17:09,599 --> 00:17:12,079
development in the next in these next few years. So

337
00:17:12,160 --> 00:17:13,640
thank you so much for your comment and a copy

338
00:17:13,680 --> 00:17:15,200
of music Code Buy. It's on its way to you.

339
00:17:15,359 --> 00:17:16,839
And if you'd like a copy of music Code Buy,

340
00:17:16,920 --> 00:17:19,119
write a comment on the website at dot netroocks dot com.

341
00:17:19,119 --> 00:17:20,799
When the facebooks you publish every show there, and if

342
00:17:20,839 --> 00:17:22,640
you comment there and ever reading the show, will send

343
00:17:22,680 --> 00:17:23,599
you a copy of music.

344
00:17:23,359 --> 00:17:24,599
Speaker 1: Go and if you don't want to do that, you

345
00:17:24,599 --> 00:17:26,480
can always just go buy it and music to code

346
00:17:26,519 --> 00:17:30,279
by dot net. It's twenty five minute tracks. There's twenty

347
00:17:30,319 --> 00:17:32,279
two of them. I'm working on twenty three now and

348
00:17:32,319 --> 00:17:36,160
you can get the collection in MP three, wave or flak.

349
00:17:36,240 --> 00:17:39,839
All right, now, let me formally introduce the other two

350
00:17:39,880 --> 00:17:42,480
guys that are here with us tonight, starting with Jeff Fritz.

351
00:17:42,960 --> 00:17:46,240
Jeff is a principal program manager in Microsoft's Developer division

352
00:17:46,240 --> 00:17:49,079
on the dot Net Community Team, where he leads development

353
00:17:49,119 --> 00:17:52,640
of live video and online content. Jeff is the executive

354
00:17:52,640 --> 00:17:55,839
producer of the dot Net coonf series of online events.

355
00:17:56,000 --> 00:18:02,319
Heard of that. He is also a Twitch and YouTube

356
00:18:02,359 --> 00:18:04,759
partner as well as the founder of the Live Coders

357
00:18:04,759 --> 00:18:07,559
stream team. You can catch Jeff writing dot net code

358
00:18:07,559 --> 00:18:10,160
with Gethub Visual Studio in Azure on his video stream

359
00:18:10,200 --> 00:18:13,799
called Fritz and Friends at Twitch, dot tv, slash c

360
00:18:14,000 --> 00:18:21,720
Sharp Fritz. Bill Wolf is here and this is an

361
00:18:21,759 --> 00:18:23,519
important date for Bill, isn't it?

362
00:18:23,559 --> 00:18:23,799
Speaker 3: Bill?

363
00:18:24,079 --> 00:18:28,440
Speaker 5: It is? Because so tonight I've been running Philly dot net,

364
00:18:28,480 --> 00:18:31,240
which is one of the largest and certainly one of

365
00:18:31,279 --> 00:18:35,519
the oldest dot net communities on the planet, possibly in

366
00:18:35,519 --> 00:18:39,079
the Galaxy and been running it for twenty four years.

367
00:18:39,079 --> 00:18:41,079
So I started it in two thousand and one.

368
00:18:41,759 --> 00:18:43,240
Speaker 1: And also you had something to do with this whole

369
00:18:43,240 --> 00:18:44,720
Inetta thing, do you remember I did.

370
00:18:44,799 --> 00:18:47,440
Speaker 5: I was one of the I was a VP at

371
00:18:47,440 --> 00:18:50,759
Anetta and I ran the speakers Bureau and my job

372
00:18:50,920 --> 00:18:54,279
was to send famous people all over the country to

373
00:18:54,400 --> 00:18:55,559
various user group meetings.

374
00:18:55,599 --> 00:18:56,279
Speaker 3: And that was all.

375
00:18:56,119 --> 00:18:59,519
Speaker 5: Funded through Microsoft at that time, and so that was

376
00:18:59,559 --> 00:19:03,880
an interest project. But I started doing user groups in

377
00:19:04,519 --> 00:19:09,920
nineteen seventy eight, so some of you may not have

378
00:19:09,960 --> 00:19:14,559
been around then, but it's been a long time. And

379
00:19:15,920 --> 00:19:19,519
Rob Kaiser, who's been by my side much of this time.

380
00:19:20,880 --> 00:19:25,359
We were very instrumental in something called PAX, which was

381
00:19:25,400 --> 00:19:28,359
one of the first user groups in the country, Philadelphia

382
00:19:28,559 --> 00:19:33,960
era Computer Society. And I'll just one quick story. One

383
00:19:33,960 --> 00:19:37,119
of my favorite meetings I ever ran at PAS. We

384
00:19:37,200 --> 00:19:41,079
were at LaSalle University. This was in the early eighties

385
00:19:42,119 --> 00:19:47,400
and I was because I'm very shy, I was on stage.

386
00:19:47,799 --> 00:19:51,559
I was on stage sort of moderating, and the panel

387
00:19:52,079 --> 00:19:57,160
there were the team from eniac OH and because some

388
00:19:57,200 --> 00:20:00,000
of them still lived in the Philadelphia area, we actually

389
00:20:00,200 --> 00:20:03,079
got like remember that they got like six people in

390
00:20:03,759 --> 00:20:06,119
and they explained to us how they built eniac and

391
00:20:06,160 --> 00:20:08,319
how they made it run and how they tested it

392
00:20:08,359 --> 00:20:09,599
and stuff. It was fascinating.

393
00:20:09,839 --> 00:20:13,839
Speaker 1: Was that related to the Jetson's uniblab because that's the

394
00:20:13,839 --> 00:20:18,359
one I remember when I was that old. Well, Bill,

395
00:20:18,440 --> 00:20:20,720
this is your last hurrah, isn't it. Yeah?

396
00:20:20,920 --> 00:20:24,200
Speaker 5: Yeah, I'm gonna step down as the Philly dot net leader.

397
00:20:24,240 --> 00:20:27,880
I'll still be involved in community. I still do speaking

398
00:20:28,039 --> 00:20:30,519
here and there, but certainly not what I used to.

399
00:20:31,279 --> 00:20:33,680
I can't keep up with these guys. They do a

400
00:20:33,720 --> 00:20:37,160
dozen conferences a year. I used to do that, you know,

401
00:20:37,160 --> 00:20:39,640
I used to do the v bits and you know,

402
00:20:39,799 --> 00:20:43,680
connect and all of those things. But I'm, you know,

403
00:20:44,440 --> 00:20:45,200
on my way out.

404
00:20:45,319 --> 00:20:46,880
Speaker 1: Well, we're going to give you a good send off.

405
00:20:47,000 --> 00:20:48,640
That's why we're here. So how about.

406
00:20:48,519 --> 00:20:51,319
Speaker 2: Hands straight down?

407
00:20:51,559 --> 00:20:54,599
Speaker 5: Yeah?

408
00:20:55,039 --> 00:20:58,039
Speaker 3: I want to go back to the to the comments

409
00:20:58,200 --> 00:21:00,319
from the black web because I think it's all this

410
00:21:00,400 --> 00:21:04,440
fear element that I think is pervasive. Right now, if

411
00:21:04,480 --> 00:21:07,359
you think about what he just said, and I paraphrase it,

412
00:21:07,480 --> 00:21:11,480
Middley's original email was longer. It's like the tool got

413
00:21:11,519 --> 00:21:15,160
away from them. Yeah. Claud's especially kind of notorious for that, right,

414
00:21:15,240 --> 00:21:17,839
you get this sort of agentic mode where it's starting

415
00:21:17,880 --> 00:21:21,039
to make changes in all kinds of places. And if

416
00:21:21,039 --> 00:21:23,079
you start really thinking about about tokens all the time,

417
00:21:23,119 --> 00:21:25,359
now you're asking you questions about what it's done, which

418
00:21:25,440 --> 00:21:27,359
is burning more tokens for you. Just get a picture

419
00:21:27,400 --> 00:21:30,279
what's going on. Like, I think you got to get

420
00:21:30,279 --> 00:21:31,799
away from the token trap part.

421
00:21:32,079 --> 00:21:36,039
Speaker 1: I would look into a good system prompt that puts,

422
00:21:36,359 --> 00:21:38,759
you know, cages, a cage around what you can and

423
00:21:38,839 --> 00:21:41,200
can't do and which you sure and shouldn't do. And

424
00:21:41,640 --> 00:21:44,880
I've learned from Jeff and other people that they're these

425
00:21:44,920 --> 00:21:50,599
great cope There's a Cope pilot, awesome repository that has

426
00:21:50,720 --> 00:21:53,559
all sorts of prompts for different things. There's a we

427
00:21:53,720 --> 00:21:56,960
just learned about this. There's a c sharp pro developer

428
00:21:57,039 --> 00:22:01,279
prompt that has all sorts of of you know, guidelines

429
00:22:01,319 --> 00:22:03,359
about how it should write code and all that stuff.

430
00:22:03,519 --> 00:22:07,119
If you're not using those things, you're missing out.

431
00:22:06,839 --> 00:22:09,160
Speaker 5: And I'm running up your charges. Yeah.

432
00:22:09,400 --> 00:22:09,880
Speaker 1: Yeah.

433
00:22:09,920 --> 00:22:13,000
Speaker 5: So I'm an enterprise architect. I actually have a real

434
00:22:13,079 --> 00:22:16,960
job these days. And one of the things fascinates me.

435
00:22:17,079 --> 00:22:19,319
I'm at a company that has six seven thousand people

436
00:22:20,720 --> 00:22:22,880
and part of my job is to try and keep

437
00:22:22,920 --> 00:22:25,160
track of Azure cost usage.

438
00:22:25,319 --> 00:22:26,599
Speaker 3: Oh wow, yeah, of course.

439
00:22:27,279 --> 00:22:29,559
Speaker 5: Very few people know how to do that, and you

440
00:22:29,640 --> 00:22:31,960
have to sort of have a fintech degree. Where is

441
00:22:32,000 --> 00:22:34,720
a Kouva over here. You have to have a fintech

442
00:22:34,759 --> 00:22:37,680
degree to actually figure out how much it costs to

443
00:22:37,759 --> 00:22:41,599
run a website connected to a database and all the

444
00:22:41,680 --> 00:22:44,440
security and all the monitoring and all that stuff. But

445
00:22:44,559 --> 00:22:47,039
I can see the same problems they're going to apply

446
00:22:47,119 --> 00:22:51,039
in the AI space and until they have some better

447
00:22:51,119 --> 00:22:53,200
tools that help you understand that.

448
00:22:53,240 --> 00:22:54,759
Speaker 3: Is part of the problem here is all this stuff

449
00:22:54,799 --> 00:22:58,000
is so masure like we don't have It's like Agile

450
00:22:58,039 --> 00:23:02,119
at the beginning, right where nobody really new well because

451
00:23:02,119 --> 00:23:03,519
you kind of got to do what you want because

452
00:23:03,519 --> 00:23:04,559
there was no plant.

453
00:23:04,839 --> 00:23:08,680
Speaker 4: So there are tools that are built into Visual Studio

454
00:23:08,759 --> 00:23:12,680
visual Studio code where there are gauges that you'll see

455
00:23:12,720 --> 00:23:15,559
down in the foot of the tool. And I'm sure

456
00:23:15,599 --> 00:23:18,599
everybody here is using Visual Studio twenty twenty six. And

457
00:23:18,680 --> 00:23:21,559
yes I'm even talking to you, dear listeners. You do

458
00:23:21,759 --> 00:23:25,039
on that you're using the new Visual Studio, But on

459
00:23:25,200 --> 00:23:28,920
in the footer there's a little getthub copilot icon, and

460
00:23:28,920 --> 00:23:31,079
when you mouse over that or you click on it,

461
00:23:31,079 --> 00:23:34,519
it'll bring up a little set of progress bars and

462
00:23:34,559 --> 00:23:39,000
show you just how much you're using your Copilot resources.

463
00:23:39,319 --> 00:23:43,160
Now to a little bit of our our commenter to

464
00:23:43,680 --> 00:23:47,839
their request requirements. There are different models that you can

465
00:23:47,920 --> 00:23:51,240
use with Copilot that have different we call them multipliers,

466
00:23:51,640 --> 00:23:55,480
right right, Okay, I see the agile one has has

467
00:23:55,559 --> 00:24:00,839
jimmed in here, yes, rate, so right, there are some

468
00:24:01,200 --> 00:24:04,599
more energy hungry models that you'll see a higher multiplier

469
00:24:04,920 --> 00:24:07,680
run with. And right now here at the time of

470
00:24:07,720 --> 00:24:11,599
this recording, right Claude Opus is up there as a

471
00:24:11,839 --> 00:24:14,400
I think it's a three time multiplier right now. But

472
00:24:14,640 --> 00:24:16,480
there are other ones that you can use that have

473
00:24:16,559 --> 00:24:19,880
a zero multiplier, like GPT Mini.

474
00:24:19,920 --> 00:24:22,279
Speaker 1: It's kind of like playing slots exactly.

475
00:24:22,799 --> 00:24:24,839
Speaker 4: So what you want to do is you want to

476
00:24:24,960 --> 00:24:28,640
choose a model that's going to help you appropriately for

477
00:24:29,000 --> 00:24:31,519
the types of tasks you're going after. If you're doing

478
00:24:31,559 --> 00:24:34,559
some text summarization, you're doing a little bit of small

479
00:24:34,599 --> 00:24:38,400
code generation, using that mini model is going to really

480
00:24:38,400 --> 00:24:40,519
help you and not cost you a thing. But if

481
00:24:40,559 --> 00:24:44,039
you're trying to one shot an asp net core controller

482
00:24:44,119 --> 00:24:46,160
with some great views to go along with it that

483
00:24:46,279 --> 00:24:49,599
use Bootstrap, break out that Claude Sonnet four to five

484
00:24:49,680 --> 00:24:50,400
and go to town.

485
00:24:50,720 --> 00:24:50,839
Speaker 1: Right.

486
00:24:51,160 --> 00:24:55,440
Speaker 5: So if I ask Copilot in visual Studio, is it

487
00:24:55,480 --> 00:24:58,119
going to tell me what's a way to save money

488
00:24:58,440 --> 00:24:59,599
by changing my model?

489
00:25:00,799 --> 00:25:02,400
Speaker 1: I would.

490
00:25:03,160 --> 00:25:07,440
Speaker 4: So the models sometimes they know about each other, sometimes

491
00:25:07,440 --> 00:25:11,480
they don't, and you can give them. There's a fetch

492
00:25:11,519 --> 00:25:15,559
command that you can give copilot GitHub copilot right hashtag

493
00:25:15,680 --> 00:25:18,480
fetch space and then you can give it a URL

494
00:25:18,799 --> 00:25:21,359
and it will go out and analyze that RL. So

495
00:25:21,400 --> 00:25:23,880
you can point it to to your point. Bill, you

496
00:25:23,880 --> 00:25:26,240
can point it to that pricing page on the GitHub

497
00:25:26,279 --> 00:25:29,640
website and have it report back here's the relative pricing

498
00:25:29,640 --> 00:25:32,839
of the models, and it might be able to give

499
00:25:32,839 --> 00:25:35,359
you some advice as to hey, let's use this model

500
00:25:35,359 --> 00:25:37,200
instead of this for certain.

501
00:25:36,920 --> 00:25:37,720
Speaker 3: Types of tasks.

502
00:25:38,200 --> 00:25:43,200
Speaker 4: So there is also auto mode, which is there's the

503
00:25:43,240 --> 00:25:44,000
real slot machine.

504
00:25:44,039 --> 00:25:46,119
Speaker 1: Right, Yeah, that's a weird one, that's right.

505
00:25:46,200 --> 00:25:50,079
Speaker 4: Hey, do choose the model appropriate for this task that

506
00:25:50,119 --> 00:25:52,720
I'm asking you about, and it will go and figure

507
00:25:52,759 --> 00:25:54,480
it out. And the best parties, it's not going to

508
00:25:54,519 --> 00:25:55,319
tell you which one that.

509
00:25:55,279 --> 00:25:57,400
Speaker 3: You that you haven't even turned it all the way up.

510
00:25:57,400 --> 00:26:00,240
Automode is one thing. Automode with unsafe.

511
00:26:01,839 --> 00:26:06,039
Speaker 4: So curl and curl, and I might have some experience.

512
00:26:05,640 --> 00:26:06,599
Speaker 3: Doing a little bit of that.

513
00:26:07,400 --> 00:26:08,720
Speaker 5: Please don't do this at home.

514
00:26:08,880 --> 00:26:11,759
Speaker 3: No, no, no, but again I have seen folks dialed

515
00:26:11,799 --> 00:26:15,079
in like that. Yes, and the burn rate is spectacular.

516
00:26:15,279 --> 00:26:18,559
But you've written your prompts well, and these are not

517
00:26:18,880 --> 00:26:22,839
These are multi page prompts. These are not small things. No. Yeah, yeah,

518
00:26:22,880 --> 00:26:25,920
the burn significant that the co generation is astonishing, and.

519
00:26:26,000 --> 00:26:30,039
Speaker 5: So having a very explicit instead of instructions, the prompt

520
00:26:30,319 --> 00:26:33,039
yeah can reduce the token.

521
00:26:32,880 --> 00:26:36,880
Speaker 1: Yeah, especially with Claude because, like Richard said clause Claude

522
00:26:36,960 --> 00:26:39,799
is kind of like Scooter the intern that's over eager

523
00:26:39,920 --> 00:26:43,160
to please the boss. Miss Frankly, shopping your Patsy is

524
00:26:43,200 --> 00:26:47,359
twice I already shopping them again, So it goes out,

525
00:26:47,400 --> 00:26:50,319
and you know, I thought you might need some more paper,

526
00:26:50,440 --> 00:26:52,880
So I went to Staples and I bought fifty reams

527
00:26:52,880 --> 00:26:55,240
of paper that kind of stuff. And like, I didn't

528
00:26:55,400 --> 00:26:58,039
ask you to do that. So in the sipstem prompt

529
00:26:58,039 --> 00:27:02,119
you have to say, don't do anything except exactly what

530
00:27:02,160 --> 00:27:03,519
I tell you to do, right.

531
00:27:03,720 --> 00:27:05,880
Speaker 4: And this is this is one of the reasons why

532
00:27:05,880 --> 00:27:07,960
when I when I give talks that that get into

533
00:27:08,039 --> 00:27:11,039
AI topics like this, I like to come back to

534
00:27:11,279 --> 00:27:14,839
I believe that the best folks at writing those prompts,

535
00:27:15,240 --> 00:27:17,720
it's going to be the elementary school teachers.

536
00:27:17,720 --> 00:27:20,440
Speaker 1: Right, because are used to talking to kids. You have

537
00:27:20,559 --> 00:27:21,759
to hear it's splicit.

538
00:27:22,799 --> 00:27:25,039
Speaker 4: They've got a room full of kids that have ADHD,

539
00:27:25,279 --> 00:27:28,319
they've got opposition defiance disorder, and you don't know what

540
00:27:28,359 --> 00:27:28,960
they're going to do.

541
00:27:29,079 --> 00:27:32,519
Speaker 1: Get in the tub, the water use.

542
00:27:37,359 --> 00:27:38,839
Speaker 5: We forgot that step.

543
00:27:39,880 --> 00:27:44,240
Speaker 4: Right, So it's it is one thing to be very explicit.

544
00:27:44,640 --> 00:27:47,839
You mentioned the awesome Copilot repository. Our friend Burke Holland

545
00:27:48,119 --> 00:27:51,319
did an amazing job putting together what he called beast Mode,

546
00:27:51,720 --> 00:27:54,920
which was right, a series of system prompts that help

547
00:27:54,960 --> 00:27:59,359
out the GPT models to really get you towards your solution.

548
00:27:59,799 --> 00:28:02,960
But there's also agent descriptions out there so that you

549
00:28:03,000 --> 00:28:07,039
can bring down those markdown files and load in and

550
00:28:07,119 --> 00:28:08,440
get that persona.

551
00:28:08,559 --> 00:28:08,680
Speaker 3: Right.

552
00:28:08,720 --> 00:28:11,920
Speaker 4: When we talk about talking to an LLM, one of

553
00:28:11,920 --> 00:28:14,559
the first things we refer to before you even get

554
00:28:14,559 --> 00:28:17,559
into defining the context of the problem that it needs

555
00:28:17,559 --> 00:28:21,359
to solve, is defining that persona who are you going

556
00:28:21,400 --> 00:28:24,599
to behave as what are you going to do? And

557
00:28:24,759 --> 00:28:27,599
you mentioned the expert c sharp You are an expert

558
00:28:27,640 --> 00:28:30,680
c sharp developer and you know how to do this,

559
00:28:30,799 --> 00:28:34,400
that and the other. I saw one that was one

560
00:28:34,440 --> 00:28:37,279
of the agents that's in that repository was talking about

561
00:28:37,359 --> 00:28:39,960
you are a fantastic at writing unit tests, and it

562
00:28:40,039 --> 00:28:43,519
even referenced personalities in the tech space and said, you

563
00:28:44,160 --> 00:28:46,559
have the ability to write tests like this person, and

564
00:28:46,599 --> 00:28:50,400
you know object oriented programming like Uncle Bob Martin, and

565
00:28:50,440 --> 00:28:53,680
you're going to be able to define and refine and

566
00:28:53,720 --> 00:28:54,319
do these things.

567
00:28:54,400 --> 00:28:55,400
Speaker 1: Well would uncle Bob do?

568
00:28:55,839 --> 00:28:56,200
Speaker 4: I know?

569
00:28:56,319 --> 00:28:59,920
Speaker 1: So we talked about this and Code with AI episode

570
00:29:00,039 --> 00:29:03,599
ten which came out this week, that writing agents, and

571
00:29:03,759 --> 00:29:08,160
we just sort of tried to differentiate between an LM

572
00:29:08,720 --> 00:29:13,720
and an MCP, which is a model control protocol, model

573
00:29:13,759 --> 00:29:17,279
content context protocol. I always get that messed up and

574
00:29:17,839 --> 00:29:21,400
agent right, So the agent is at the highest level

575
00:29:21,759 --> 00:29:25,599
and the LM you would give the system prompt, right,

576
00:29:25,680 --> 00:29:28,880
but the agent has a prompt. That's just like Jeff

577
00:29:28,960 --> 00:29:31,920
was saying, what is your persona, what is your area

578
00:29:31,960 --> 00:29:34,319
of focus that you're going to do? Right? Are you

579
00:29:34,400 --> 00:29:36,119
only going to work on testing? Are you going to

580
00:29:36,160 --> 00:29:38,680
work on code? Are you working on VB code? You're

581
00:29:38,680 --> 00:29:41,519
working on assembler code? Right, those are the kinds of

582
00:29:41,559 --> 00:29:44,519
things that an agent can help with, but it's at

583
00:29:44,519 --> 00:29:46,920
a high a much higher level, and it will do

584
00:29:47,039 --> 00:29:51,720
things on your behalf, not just not just code, right,

585
00:29:52,680 --> 00:29:54,400
And you just have to be careful that you don't

586
00:29:54,440 --> 00:29:57,160
give it permission to go, you know, sell your house

587
00:29:57,240 --> 00:29:57,680
or something.

588
00:29:58,279 --> 00:30:00,519
Speaker 4: Don't don't give it that essay past word into your

589
00:30:00,559 --> 00:30:01,319
SQL surfer.

590
00:30:02,240 --> 00:30:07,200
Speaker 3: No, not the production one anyway. Oh, they're supposed to

591
00:30:07,240 --> 00:30:11,880
be different, that's right, in production. I just told you

592
00:30:11,880 --> 00:30:17,119
guys still know essay? Like who uses essay anymore? Right? Yeah?

593
00:30:17,480 --> 00:30:18,279
Speaker 4: Managed identity?

594
00:30:18,720 --> 00:30:21,519
Speaker 3: Managed identities now, yeah, and we should take a break

595
00:30:21,559 --> 00:30:22,799
for these very important message.

596
00:30:22,839 --> 00:30:25,160
Speaker 1: Yeah, we should. We'll be right back. Mission.

597
00:30:25,519 --> 00:30:27,240
Speaker 4: What a show, it's intermission.

598
00:30:27,880 --> 00:30:28,480
Speaker 1: What do you know?

599
00:30:28,880 --> 00:30:29,279
Speaker 3: Okay?

600
00:30:29,440 --> 00:30:30,640
Speaker 1: I think that was inquisition?

601
00:30:31,079 --> 00:30:33,240
Speaker 3: Yeah, oh, agreed to get popcorn?

602
00:30:34,160 --> 00:30:36,880
Speaker 1: Did you know? There's a dot net on aws community.

603
00:30:37,559 --> 00:30:41,559
Follow the social media blogs, YouTube influencers, and open source

604
00:30:41,599 --> 00:30:45,319
projects and add your own voice. Get plugged into the

605
00:30:45,359 --> 00:30:50,039
dot net on AWS community at aws dot Amazon dot com,

606
00:30:50,079 --> 00:30:51,000
slash dot net.

607
00:30:54,559 --> 00:30:56,640
Speaker 3: And we're back. It's don I Rocks Emberger Campbell. That's

608
00:30:56,640 --> 00:31:00,200
called Franklin. Hey, sitting with our friends Jeff Fritz and

609
00:31:00,279 --> 00:31:02,599
Bill Wolf. We're here at Philly dot net.

610
00:31:06,880 --> 00:31:10,160
Speaker 1: Ten thousand people in the audience on a very first

611
00:31:10,240 --> 00:31:12,160
live show in quite a while, and really fun to

612
00:31:12,160 --> 00:31:14,319
be in front of sitting with everybody and having a

613
00:31:14,319 --> 00:31:16,440
little fun making a show about I think a pretty

614
00:31:16,440 --> 00:31:18,359
serious topic coming into the end of the year here.

615
00:31:18,519 --> 00:31:21,319
Speaker 4: But in all seriousness, is this the first time you've

616
00:31:21,319 --> 00:31:24,079
done a show where there's a court downstairs in the building,

617
00:31:24,359 --> 00:31:26,920
just in case any of these folks get rowdy.

618
00:31:27,000 --> 00:31:27,920
Speaker 1: That's true.

619
00:31:28,400 --> 00:31:29,079
Speaker 3: Co birds.

620
00:31:29,359 --> 00:31:32,240
Speaker 4: Yeah, I'll tell you know where in Philly.

621
00:31:32,799 --> 00:31:36,119
Speaker 3: You know, I've been working with a couple of software

622
00:31:36,160 --> 00:31:42,119
development consulting companies now that are not only going all in,

623
00:31:42,720 --> 00:31:44,880
but they're trying to get all of their developers not

624
00:31:44,920 --> 00:31:47,960
only using the tools, but they're coming up with sets

625
00:31:48,000 --> 00:31:49,960
of standards for how they want to make software as

626
00:31:50,000 --> 00:31:52,640
a whole for their customers. And so they're building out

627
00:31:52,680 --> 00:31:57,759
templates for deployment, they're building templates for infrastructure, they're building

628
00:31:57,759 --> 00:32:04,079
out templates for UAX, all focused on educating the cogenerators.

629
00:32:04,240 --> 00:32:07,680
And that's all these things are as cogenerators right to

630
00:32:08,160 --> 00:32:11,640
stay within the lines, so that each developer doesn't have

631
00:32:11,799 --> 00:32:14,640
to get those prompts right, they're just pre configured for them.

632
00:32:14,640 --> 00:32:18,039
Of like, this is the way we do things. There's

633
00:32:18,079 --> 00:32:20,599
some really great things to be said about those agents

634
00:32:20,680 --> 00:32:23,680
and being able to supply to them, or being able

635
00:32:23,759 --> 00:32:26,640
to pass in in a prompt, a prestructured prompt that

636
00:32:26,720 --> 00:32:28,799
you can have written down and out there on disk.

637
00:32:29,240 --> 00:32:32,400
Have those templates built out and sitting in markedown format,

638
00:32:32,720 --> 00:32:35,119
so that even when you browse your repository and you

639
00:32:35,160 --> 00:32:37,519
look at the markdown, you can look at the template

640
00:32:37,559 --> 00:32:38,279
and you know.

641
00:32:38,240 --> 00:32:40,359
Speaker 4: That it looks the right way that you expect it to.

642
00:32:41,039 --> 00:32:45,279
And the ability for the AI the LM to generate

643
00:32:45,359 --> 00:32:48,720
and stick to that is really really good. But I

644
00:32:48,720 --> 00:32:51,079
think Carl made a really good point. Don't just stick

645
00:32:51,079 --> 00:32:53,039
to the template, but you got to tell it what

646
00:32:53,240 --> 00:32:55,079
not to do about the template as well.

647
00:32:55,160 --> 00:32:56,720
Speaker 3: Right to get outside of the line.

648
00:32:57,559 --> 00:32:59,640
Speaker 4: There's an example I like to show where I gave.

649
00:33:00,599 --> 00:33:03,599
I gave an LLLM the ability to summarize the weather

650
00:33:04,200 --> 00:33:07,599
scenario for a weather forecast, and I gave it four

651
00:33:07,640 --> 00:33:12,720
options sonny, cloudy, rainy, and snowy and I would send

652
00:33:12,759 --> 00:33:17,880
into the LLLM a bunch of different forecasts, and inevitably

653
00:33:17,920 --> 00:33:22,079
it would come back and say clear, nice, clear, Clear

654
00:33:22,119 --> 00:33:24,240
isn't one of them, right, So you have to tell

655
00:33:24,279 --> 00:33:26,640
it what to do and what not to do.

656
00:33:27,960 --> 00:33:30,119
Speaker 1: This is a topic that we've talked about a lot

657
00:33:30,279 --> 00:33:32,799
recently on dot net Rocks, probably much to the chagrin

658
00:33:32,839 --> 00:33:35,519
of many of our listeners. Can you guys, please let's

659
00:33:35,559 --> 00:33:39,039
talk about something else than AI pat The fact of

660
00:33:39,079 --> 00:33:42,519
the matter is it's fundamentally changed the way we write code.

661
00:33:43,039 --> 00:33:46,279
For me, certainly has, I know for a lot of

662
00:33:46,279 --> 00:33:49,000
our listeners it has. Black Web may be one of

663
00:33:49,000 --> 00:33:52,680
the exceptions, but it really has changed the way I

664
00:33:52,720 --> 00:33:57,200
write code. So here's a really good example. And I

665
00:33:57,319 --> 00:34:01,119
love using my personal little stupid projects. I'm sitting there.

666
00:34:01,160 --> 00:34:03,839
I have a forty nine inch screen, and my wife

667
00:34:03,839 --> 00:34:07,519
and I play this game we call Sherlock, and it's

668
00:34:07,559 --> 00:34:11,239
a logic puzzle game. It's from the nineties, I think,

669
00:34:11,599 --> 00:34:13,480
but it's been ported to a bunch of things anyway,

670
00:34:14,159 --> 00:34:16,639
So it has to live on one side of the

671
00:34:16,679 --> 00:34:18,880
screen or the other because my wife and I sit

672
00:34:18,960 --> 00:34:22,400
together and usually I'm watching something on a streaming service

673
00:34:22,679 --> 00:34:25,719
she's playing, but then when she goes out, I want

674
00:34:25,760 --> 00:34:27,679
to switch them around, right, So I got to switch

675
00:34:27,719 --> 00:34:31,039
them like this. So I basically had co pilot generated

676
00:34:31,079 --> 00:34:33,800
little Windows app that says, Hey, I want to put

677
00:34:33,800 --> 00:34:35,920
this window here, in this window here, and swap them

678
00:34:36,199 --> 00:34:38,960
just swap them. Yeah, And I literally did it in

679
00:34:39,079 --> 00:34:43,519
fifteen minutes, and I know the code. It's not to

680
00:34:43,599 --> 00:34:48,440
do it, yeah, I don't know, God no, but but

681
00:34:48,679 --> 00:34:49,920
you know I know how to do it. It's not

682
00:34:50,000 --> 00:34:53,480
a magerate matter of me being lazy. I just it

683
00:34:53,559 --> 00:34:56,000
was a fifteen minute thing and I was just like, wow,

684
00:34:56,079 --> 00:34:56,840
that's so cool.

685
00:34:57,000 --> 00:34:59,800
Speaker 4: But vibe coding is a very powerful thing that I

686
00:34:59,800 --> 00:35:04,800
think is valuable for project managers. Program managers when they

687
00:35:04,880 --> 00:35:07,679
get the idea for a user interface update, for a patch,

688
00:35:08,000 --> 00:35:10,599
and they're able to take some code, they're able to

689
00:35:10,639 --> 00:35:14,559
take some screenshots, some ideas, talk to the AI generate

690
00:35:14,679 --> 00:35:17,519
those concepts. And for those of us that are in

691
00:35:17,599 --> 00:35:20,599
the industry that are experts, like you dear listener with

692
00:35:20,639 --> 00:35:24,360
the headphones on, like, you can do this, and you

693
00:35:24,440 --> 00:35:26,719
know those edge cases that you want it to do,

694
00:35:27,000 --> 00:35:29,559
and we also know how to tell the AI, hey

695
00:35:29,800 --> 00:35:33,000
there's a plan here. You need to build because I

696
00:35:33,039 --> 00:35:36,159
am a project planner. I am a project manager, and

697
00:35:36,320 --> 00:35:39,119
I need you to build that project plan, that spec

698
00:35:39,800 --> 00:35:43,000
so that we can walk through it. And Copilot is

699
00:35:43,159 --> 00:35:48,559
very good at following a document and execute on those plans.

700
00:35:48,559 --> 00:35:50,360
Speaker 1: Well, have you done any vibe coding with AI?

701
00:35:50,559 --> 00:35:50,760
Speaker 3: Yeah?

702
00:35:50,840 --> 00:35:51,360
Speaker 1: What do you think?

703
00:35:53,159 --> 00:35:53,880
Speaker 3: It helps a lot?

704
00:35:54,320 --> 00:35:54,519
Speaker 1: Yeah?

705
00:35:54,559 --> 00:35:57,400
Speaker 5: I just you know the old days of you weren't

706
00:35:57,440 --> 00:35:59,519
sure how to do something. You go to the web,

707
00:35:59,599 --> 00:36:03,559
you do a you know, a stack, and you know

708
00:36:03,760 --> 00:36:06,119
and copy stuff and a lot of a lot of

709
00:36:06,159 --> 00:36:10,000
programmers just copy exactly what's in the page. That leads

710
00:36:10,000 --> 00:36:10,880
to all sorts of pan.

711
00:36:10,920 --> 00:36:13,400
Speaker 3: Take the error message and the copy that but to Google.

712
00:36:15,639 --> 00:36:18,960
Speaker 5: But having AI sort of guide me through things that

713
00:36:19,000 --> 00:36:22,400
I haven't done in a while, it really really a

714
00:36:22,400 --> 00:36:22,840
time saver.

715
00:36:23,039 --> 00:36:25,199
Speaker 1: And you know, to the to the point that I

716
00:36:25,239 --> 00:36:27,239
was making a black web. You can ask it to

717
00:36:27,320 --> 00:36:30,599
explain itself. I don't understand this, what's it doing? Comment

718
00:36:30,679 --> 00:36:32,960
it or give me a summary of what you just did,

719
00:36:33,800 --> 00:36:34,039
you know.

720
00:36:34,559 --> 00:36:40,400
Speaker 5: But we also use it for QA, generating tests, DevOps,

721
00:36:41,280 --> 00:36:44,559
you know, figuring out you know, sort of scripts and

722
00:36:44,599 --> 00:36:50,719
recipes for deployments and you know resources. There's a lot

723
00:36:50,760 --> 00:36:54,639
of places that it applies beyond just the C sharp coder.

724
00:36:55,280 --> 00:36:58,159
Speaker 3: You know, if you're billing by the hour as a developer,

725
00:36:59,199 --> 00:37:03,079
you really don't want to No, you.

726
00:37:03,039 --> 00:37:05,599
Speaker 1: Don't want to tell anybody that you even have them installed.

727
00:37:05,719 --> 00:37:07,800
Speaker 3: Yeah, I think or if you are users tools, you

728
00:37:07,800 --> 00:37:10,000
gotta lie about your hours or matter. It's still I

729
00:37:10,000 --> 00:37:11,960
think you have to restructure HIO, you, Billy.

730
00:37:13,320 --> 00:37:16,360
Speaker 5: The other side of that, Richard, is I think you

731
00:37:16,440 --> 00:37:21,920
have contractors that work remotely that really understand these tools,

732
00:37:22,639 --> 00:37:27,599
and they can actually juggle multiple clients and build them concurrently.

733
00:37:28,400 --> 00:37:30,840
Not that that's a good thing, but I think you're

734
00:37:30,840 --> 00:37:32,239
seeing stuff like that going on too.

735
00:37:32,480 --> 00:37:32,679
Speaker 3: Yeah.

736
00:37:32,920 --> 00:37:36,000
Speaker 5: And one of my favorite parts, how many of you

737
00:37:36,079 --> 00:37:40,119
ever have to interview people? Yeah, don't you love when

738
00:37:40,159 --> 00:37:43,320
you ask them a question and their eyes are going

739
00:37:43,360 --> 00:37:46,199
back and forth, and you know that they're talking to

740
00:37:46,280 --> 00:37:49,280
Claude or some other model and saying, give me a

741
00:37:49,320 --> 00:37:50,519
good come on, give me something.

742
00:37:51,679 --> 00:37:54,840
Speaker 1: Yeah, there's no such thing as a radio quiz anymore.

743
00:37:54,960 --> 00:37:55,000
Speaker 5: No.

744
00:37:55,920 --> 00:37:58,039
Speaker 3: I talked to one interviewer. He said, here's how I

745
00:37:58,039 --> 00:38:02,639
asked the question. Close your off, now answer this question.

746
00:38:03,039 --> 00:38:03,559
That's good.

747
00:38:03,800 --> 00:38:07,079
Speaker 4: Yeah, Well, you made a good point earlier. Richard about

748
00:38:07,119 --> 00:38:10,119
how project teams are able to get through a six

749
00:38:10,159 --> 00:38:12,400
week sprint in days.

750
00:38:13,559 --> 00:38:15,960
Speaker 3: What was a project team for six weeks was a

751
00:38:16,159 --> 00:38:21,639
person with six or seven agents, burning credits like crazy. Sure,

752
00:38:21,800 --> 00:38:24,840
you know, running hard, but the results were again astonishing.

753
00:38:25,440 --> 00:38:29,480
Speaker 4: What we really can can get into now is those

754
00:38:29,519 --> 00:38:32,719
things that were in the parking lot on the combine board, right,

755
00:38:32,840 --> 00:38:33,840
they're now in play.

756
00:38:33,920 --> 00:38:36,480
Speaker 3: Well, you're exactly right. I could keep telling you know,

757
00:38:36,960 --> 00:38:39,119
black web handed at this everybody's talking about it's like

758
00:38:39,159 --> 00:38:40,679
we're all going to lose their jobs, and I'm like,

759
00:38:41,079 --> 00:38:43,480
it just doesn't look like it. A. We need the Shepherd,

760
00:38:43,559 --> 00:38:46,440
but b how much it's not like any of us

761
00:38:46,440 --> 00:38:48,719
we're getting to the bottom or to do list right. Yeah,

762
00:38:48,880 --> 00:38:52,199
Ever there's always more And how many other projects don't

763
00:38:52,239 --> 00:38:55,480
even get on the board exactly because the backlog is

764
00:38:55,519 --> 00:38:58,320
so far back Exactly. You go back to the Luddites

765
00:38:58,840 --> 00:39:03,400
and the industrial clothing production. While it was disruptive at

766
00:39:03,480 --> 00:39:06,760
the time, it also lowered the cost of cloth enough

767
00:39:06,760 --> 00:39:09,400
that people started owning more than one set of clothes.

768
00:39:09,719 --> 00:39:14,760
Imagine right, how many software projects have just never been

769
00:39:14,760 --> 00:39:17,480
written because they couldn't even get to the table.

770
00:39:18,039 --> 00:39:21,719
Speaker 1: It's interesting that the Luddites were the technologists of the day.

771
00:39:22,039 --> 00:39:25,760
They had the knowledge of what they were doing, clothes

772
00:39:26,000 --> 00:39:28,320
out of weaves, and it was suddenly threatened by machines.

773
00:39:28,400 --> 00:39:31,280
But they were you know, if they were us now,

774
00:39:32,199 --> 00:39:35,360
like us, we have the knowledge of how to build software,

775
00:39:35,480 --> 00:39:39,159
so we can interact with the AI. They didn't have AI, right,

776
00:39:39,199 --> 00:39:42,559
but they could probably be more productive with the machines then.

777
00:39:42,880 --> 00:39:45,719
Speaker 3: Well, in many on the street could follow that story.

778
00:39:45,719 --> 00:39:49,119
Over time, they got retrained on the new machine. Not

779
00:39:49,239 --> 00:39:51,480
all of them. Lots of people didn't want to play,

780
00:39:51,599 --> 00:39:54,880
but a lot of them did and it did change

781
00:39:54,880 --> 00:39:58,440
things around. It's hard to be in the disruption, but

782
00:39:58,559 --> 00:39:59,480
we're in the disruption.

783
00:40:00,119 --> 00:40:03,199
Speaker 1: So it reminds me of the CAD revolution, you remember that.

784
00:40:03,320 --> 00:40:06,800
So my mother was actually a draftsman at Electric Boat.

785
00:40:06,800 --> 00:40:10,360
Electric Boat, Yeah, and she she did everything by hand

786
00:40:10,400 --> 00:40:12,360
and she was very good at it. And her friend

787
00:40:12,559 --> 00:40:14,639
got on the CAD system. She's like, I don't want

788
00:40:14,639 --> 00:40:17,679
to do that, right, But they left her alone, and

789
00:40:17,760 --> 00:40:19,679
you know, she listened to her music and did her

790
00:40:19,760 --> 00:40:23,199
drawings and stuff, and when anybody wanted any real detail

791
00:40:23,239 --> 00:40:25,159
work done by hand, they called her.

792
00:40:25,599 --> 00:40:29,719
Speaker 4: I was one of the last classes at my university

793
00:40:30,159 --> 00:40:34,960
that that had a draftsman class required in engineering. I

794
00:40:35,280 --> 00:40:36,599
had to learn how to use.

795
00:40:36,519 --> 00:40:38,039
Speaker 1: All the tools and slide roll.

796
00:40:38,559 --> 00:40:40,440
Speaker 4: Yeah, a little bit of slide rule.

797
00:40:41,119 --> 00:40:44,000
Speaker 5: So that really makes me remember my father was an

798
00:40:44,000 --> 00:40:48,119
electrical dressman. Yeah, and he did the routing for electrical

799
00:40:48,159 --> 00:40:54,159
circuits in nuclear power plants. He worked out three mile Island.

800
00:40:54,559 --> 00:40:58,239
You can imagine, well, you know, but it's it is

801
00:40:58,280 --> 00:41:01,320
something that happened, yes, and uh, you know, I don't

802
00:41:01,360 --> 00:41:06,159
I doubt if his circuit caused the problem, but you

803
00:41:06,159 --> 00:41:09,519
can imagine all the drawings. It takes every little wire

804
00:41:09,639 --> 00:41:12,559
back then because it wasn't there was no tc P

805
00:41:12,679 --> 00:41:12,800
I P.

806
00:41:13,039 --> 00:41:16,280
Speaker 1: Well, Wover Dam was built with slide rules. Yep, no calculators,

807
00:41:16,280 --> 00:41:17,960
no computers still standing.

808
00:41:18,320 --> 00:41:20,760
Speaker 4: So let can we can we pivot the discussion a little.

809
00:41:20,760 --> 00:41:23,480
We've been talking about copilot, Claude, some of those tools

810
00:41:23,719 --> 00:41:27,800
that help us as developers. But I like to get

811
00:41:27,800 --> 00:41:31,000
outside that box a little bit and and something that

812
00:41:31,000 --> 00:41:34,039
that I want to make sure that that our developers

813
00:41:34,039 --> 00:41:37,719
here in Philadelphia and the listeners I want to I

814
00:41:37,719 --> 00:41:39,920
want to make sure that we're thinking about is not

815
00:41:40,039 --> 00:41:42,039
just how to use the AI to get your job done.

816
00:41:42,079 --> 00:41:42,480
Speaker 3: To build a.

817
00:41:42,480 --> 00:41:46,239
Speaker 4: Website, but how can you help your customers use the

818
00:41:46,320 --> 00:41:53,079
AI to deliver their requirements, their needs to their customers.

819
00:41:53,480 --> 00:41:57,719
There's there's more to that than just slapping a textbox

820
00:41:58,519 --> 00:42:01,440
into the middle of your appplications so they can ask

821
00:42:01,559 --> 00:42:04,320
questions about their reports. There's a lot more that we

822
00:42:04,360 --> 00:42:06,199
can do with that that we need to get. That's

823
00:42:06,239 --> 00:42:08,960
a great feature, Oh, without a doubt, it's a great feature.

824
00:42:09,119 --> 00:42:13,920
I can recommend a podcast, yes, but there's there's there's

825
00:42:13,960 --> 00:42:17,320
things that you can do to return value to those folks,

826
00:42:17,599 --> 00:42:19,480
and I think that's someplace that we need to help

827
00:42:19,519 --> 00:42:24,039
the enterprises understand and use the AI better so that

828
00:42:24,320 --> 00:42:28,119
they can turn right. The developers can help their end users,

829
00:42:28,119 --> 00:42:30,920
whether it's somebody working in a call center, somebody who's

830
00:42:30,920 --> 00:42:35,079
working in accounting or doing financial analysis, help those folks

831
00:42:35,360 --> 00:42:38,480
get that same multiplier that we're realizing as developers.

832
00:42:38,960 --> 00:42:43,039
Speaker 5: Some of the work I'm doing is square there and

833
00:42:43,280 --> 00:42:46,159
it relates to CAD because I work for a company

834
00:42:46,159 --> 00:42:52,360
that manufactures construction components to build buildings, and part of

835
00:42:52,360 --> 00:42:56,119
what we're looking at is AI. How do you, as

836
00:42:56,159 --> 00:43:00,199
an architect, how do you design a building for the

837
00:43:00,239 --> 00:43:04,599
coast of Florida, so it can handle hurricanes. And if

838
00:43:04,639 --> 00:43:07,199
you have and if you ever see those pictures after

839
00:43:07,280 --> 00:43:10,159
this dorm, you know, a whole bunch of buildings are flat.

840
00:43:10,679 --> 00:43:15,199
There's one or two buildings center block that know that

841
00:43:15,280 --> 00:43:19,280
are standing straight. And the AI said, but that's probably

842
00:43:19,880 --> 00:43:22,320
probably the company I'm at. It has to do with

843
00:43:22,400 --> 00:43:24,280
how you tie all the pieces together with.

844
00:43:24,239 --> 00:43:25,440
Speaker 3: Metal, roof doesn't get pulled out.

845
00:43:25,559 --> 00:43:28,639
Speaker 5: Yes, yes, and and we do all those calculations, but

846
00:43:28,679 --> 00:43:31,000
we're trying to figure out how to get a I

847
00:43:31,199 --> 00:43:33,079
to help do that kind of work.

848
00:43:33,199 --> 00:43:34,039
Speaker 3: That's tricky.

849
00:43:34,119 --> 00:43:34,920
Speaker 1: That's that's cool.

850
00:43:35,159 --> 00:43:37,760
Speaker 3: Yeah, we've been followed. One of the show recurring shows

851
00:43:37,760 --> 00:43:41,159
for us has been Vishwaz's show where he's now left

852
00:43:41,400 --> 00:43:46,039
gone into a startup to build l M tools for

853
00:43:46,119 --> 00:43:51,079
generating RFP responses to government contracts. Oh and it's been

854
00:43:51,079 --> 00:43:53,599
fascinting to listen to him as he's as they've learned

855
00:43:53,639 --> 00:43:55,880
more and hit the challenges and sort of that progress.

856
00:43:56,079 --> 00:43:57,599
It's been a couple of years of this now, like

857
00:43:57,639 --> 00:43:59,960
we're just trying to pull a narrative together, Like he's

858
00:44:00,079 --> 00:44:02,519
taking on a tough problem and he's learning more about

859
00:44:02,519 --> 00:44:05,440
it getting on there, like these different cases are part

860
00:44:05,480 --> 00:44:08,480
of us trying to find stories about what's working what isn't.

861
00:44:08,480 --> 00:44:09,599
Speaker 1: But a lot of these What's.

862
00:44:09,440 --> 00:44:13,679
Speaker 5: The biggest problem I have with AI is dealing with

863
00:44:14,760 --> 00:44:17,840
friends and family, oh god, yes, and holiday parties and

864
00:44:17,880 --> 00:44:18,280
they're like.

865
00:44:18,719 --> 00:44:20,079
Speaker 1: HEYI is ruining everything.

866
00:44:20,119 --> 00:44:21,280
Speaker 3: I'm like, oh, please don't.

867
00:44:21,599 --> 00:44:23,599
Speaker 5: And they come up to me at the party and

868
00:44:23,719 --> 00:44:27,239
specifically like, you are responsible, You're doing this.

869
00:44:27,280 --> 00:44:28,320
Speaker 1: Right, it's your fault.

870
00:44:29,599 --> 00:44:32,360
Speaker 3: Well. AI seems to be the current scapegoat for whatever

871
00:44:32,519 --> 00:44:35,039
is happening. It is, but listen, a I didn't lay

872
00:44:35,079 --> 00:44:38,679
anybody off. That was people. People laid up folks off

873
00:44:38,760 --> 00:44:41,360
and they may have used AI as their current excuse,

874
00:44:42,039 --> 00:44:44,880
but we're also seeing lots of them walk that back afterwards,

875
00:44:44,880 --> 00:44:47,400
but it doesn't get the results they wanted, or that

876
00:44:47,440 --> 00:44:48,880
it actually wasn't what they were to do in the

877
00:44:48,920 --> 00:44:49,840
first place. Right.

878
00:44:50,039 --> 00:44:52,320
Speaker 1: I know that, you know, I have a young gen

879
00:44:52,440 --> 00:44:57,599
Z daughter and her generation is feels hopeless, you know

880
00:44:57,679 --> 00:45:00,400
about the future that you know, AI kind of take

881
00:45:00,440 --> 00:45:03,239
everybody's jobs and all that stuff that you were talking about, Bill,

882
00:45:04,119 --> 00:45:07,960
and it's I just have to constantly express to her

883
00:45:08,119 --> 00:45:11,119
that you know, this is no excuse to not do

884
00:45:11,239 --> 00:45:14,679
your best at what you want to do in life.

885
00:45:14,800 --> 00:45:16,599
You know, if you want to learn something for the

886
00:45:16,840 --> 00:45:19,840
for the sake of learning it, learn it doesn't matter

887
00:45:19,880 --> 00:45:22,400
if there's an AI that knows it too. You go

888
00:45:22,559 --> 00:45:25,199
do what you want to do and be the best

889
00:45:25,239 --> 00:45:26,480
whatever that you can.

890
00:45:26,480 --> 00:45:28,360
Speaker 3: Assibly help you.

891
00:45:29,480 --> 00:45:33,559
Speaker 4: And yeah, that's that's a great point Bill, where I've

892
00:45:33,639 --> 00:45:36,719
I've talked to some folks, some artists who really don't

893
00:45:36,800 --> 00:45:39,880
like and I I agree with him, really don't like

894
00:45:40,000 --> 00:45:45,840
using AI to generate cartoons, images, photos. You're generating those

895
00:45:45,880 --> 00:45:50,960
images because you can't get a photographer or the personalities

896
00:45:50,960 --> 00:45:54,000
that you want to appear in those In those pictures,

897
00:45:54,039 --> 00:45:57,239
you're you're assembling them for some reason because you can't

898
00:45:57,280 --> 00:46:00,800
get that to happen, or you can't an artist to

899
00:46:00,840 --> 00:46:03,280
draw a cartoon. But what you can use the AI

900
00:46:03,440 --> 00:46:07,400
for is to analyze something that you drew or something

901
00:46:07,440 --> 00:46:10,840
that you made. Give it a picture and say, you

902
00:46:10,840 --> 00:46:12,920
know what, take a look at this, review it and

903
00:46:13,159 --> 00:46:15,159
tell me how I can make this better. And then

904
00:46:15,519 --> 00:46:17,840
to your point, I can learn how to improve my

905
00:46:17,960 --> 00:46:19,719
skills and do more I do.

906
00:46:20,199 --> 00:46:22,360
Speaker 3: On the consulting side, I'm not talking to folks who

907
00:46:22,440 --> 00:46:24,719
are making money off of cleaning up by code of

908
00:46:24,760 --> 00:46:30,239
projects that don't work. Oh yes, well, and I feel

909
00:46:30,320 --> 00:46:33,519
like like we're in a stupid period, right, this is

910
00:46:33,519 --> 00:46:36,360
the early days of this tech. We're still in the

911
00:46:36,440 --> 00:46:41,079
AI bubble, which is a very stupid period, and people

912
00:46:41,199 --> 00:46:43,559
are not knowing how to use these tools, are getting

913
00:46:43,559 --> 00:46:46,320
in over their heads. Yeah, like I did this with

914
00:46:46,440 --> 00:46:48,559
access back in the day. Lots of people spent the

915
00:46:48,639 --> 00:46:52,719
weekend and got themselves over their heads, Like there's money

916
00:46:52,719 --> 00:46:56,239
to be made cleaning up messes and building out the

917
00:46:56,320 --> 00:46:58,480
kit to get good at cleaning up the message this

918
00:46:58,519 --> 00:47:00,280
will pass only a few years.

919
00:47:00,280 --> 00:47:02,480
Speaker 1: One of the best customer comments they ever got is

920
00:47:02,480 --> 00:47:04,239
why is it taking so long? My brother could do

921
00:47:04,280 --> 00:47:05,000
this in access?

922
00:47:05,119 --> 00:47:09,639
Speaker 3: Yeah, like, you know what, you should try have your

923
00:47:09,639 --> 00:47:10,199
brother do it.

924
00:47:10,239 --> 00:47:11,239
Speaker 1: I'm done, But.

925
00:47:11,159 --> 00:47:14,360
Speaker 4: Richard, you're you're hinting at using AI to fix the

926
00:47:14,400 --> 00:47:19,880
mistakes that some human implemented with AI. Like the solution

927
00:47:19,960 --> 00:47:25,679
to AI being more AI feels like like a we're

928
00:47:26,239 --> 00:47:28,039
kicking the can down the road here a little bit.

929
00:47:28,039 --> 00:47:31,760
Speaker 3: It's also a normal escalation, but they are missed. You know,

930
00:47:31,880 --> 00:47:35,880
we've all we've often misused tools and inexperienced you misused

931
00:47:35,880 --> 00:47:38,440
tools and experienced people can get results, and then you

932
00:47:38,440 --> 00:47:41,880
start getting real costs. You know, it is possible with

933
00:47:42,039 --> 00:47:44,559
minimal skill to get to a certain point in building

934
00:47:44,599 --> 00:47:47,199
an application with these tools, but finishing it it's hard.

935
00:47:47,519 --> 00:47:51,000
Speaker 4: How many of us sell a sp net MVC applications

936
00:47:51,360 --> 00:47:55,400
with SQL statements sitting in a view right to our

937
00:47:55,480 --> 00:47:56,280
dot Net listeners?

938
00:47:56,840 --> 00:47:57,880
Speaker 3: You put them in the parameter.

939
00:48:01,159 --> 00:48:06,119
Speaker 1: Oh, just to change tax here. Anybody a musician or

940
00:48:06,119 --> 00:48:10,320
an amateur musician, play an instrument, write songs, jand up

941
00:48:10,360 --> 00:48:13,920
if you write songs, a couple of you, Okay, I

942
00:48:14,000 --> 00:48:17,880
want to know what you think of Suno and Suno

943
00:48:18,199 --> 00:48:24,159
is essentially something that can build a completely professional sounding song,

944
00:48:24,639 --> 00:48:29,280
complete with vocals and solos, just from a prompt and

945
00:48:29,639 --> 00:48:34,559
it sounds amazing. Now I'm a musician, it's very hard

946
00:48:34,599 --> 00:48:37,159
for me not to take offense at this, but I don't.

947
00:48:37,559 --> 00:48:40,119
I kind of look at this as like the Cassio

948
00:48:40,239 --> 00:48:42,639
keyboard of twenty twenty five, you know what I mean?

949
00:48:43,280 --> 00:48:46,159
Who used to have organs in the home where people

950
00:48:46,199 --> 00:48:48,360
could play take me out to the ballgame and learn

951
00:48:48,400 --> 00:48:51,079
their things. But that doesn't mean they're going to take

952
00:48:51,119 --> 00:48:56,280
a musician's job. Right. This thing, however, it's pretty awesome.

953
00:48:56,360 --> 00:48:58,960
So who raised their hand? What do you think about Suno?

954
00:48:59,039 --> 00:49:01,920
Andy down? Thumbs down?

955
00:49:02,960 --> 00:49:03,280
Speaker 3: What's that?

956
00:49:04,679 --> 00:49:08,440
Speaker 1: I can hear you, but i'll replay your yeah?

957
00:49:08,599 --> 00:49:08,760
Speaker 3: Right?

958
00:49:08,800 --> 00:49:11,000
Speaker 1: Why I have a famous pain or painting something when

959
00:49:11,039 --> 00:49:11,280
you can?

960
00:49:11,400 --> 00:49:12,599
Speaker 3: Yeah? Right?

961
00:49:12,840 --> 00:49:15,920
Speaker 1: Have you listened to a Suno generated song? There? Actually?

962
00:49:16,039 --> 00:49:18,559
Rick Piatto did a video where he made one in

963
00:49:18,719 --> 00:49:20,920
five minutes and it sounds like it was done in

964
00:49:20,960 --> 00:49:24,639
a professional studio with an emotion, a voice full of

965
00:49:24,679 --> 00:49:28,360
emotion and everything. Who else raised their hand? Over here?

966
00:49:28,440 --> 00:49:30,519
You you play piano? Oh you never heard of Suno?

967
00:49:30,760 --> 00:49:31,119
Speaker 3: All right?

968
00:49:31,280 --> 00:49:31,679
Speaker 1: Somebody else?

969
00:49:31,719 --> 00:49:32,199
Speaker 3: Yeah? Go ahead?

970
00:49:32,239 --> 00:49:35,800
Speaker 1: It feels gross to you? Yeah? Yeah, synthetic feels grotesque.

971
00:49:35,840 --> 00:49:36,039
Speaker 3: Yeah.

972
00:49:36,079 --> 00:49:38,119
Speaker 1: As a tool to create and refine art, it's kind

973
00:49:38,159 --> 00:49:40,880
of grotesque. Do you do you think you could tell

974
00:49:40,920 --> 00:49:45,800
the difference between a Suno song and a professionally recorded song,

975
00:49:45,840 --> 00:49:49,440
Because I'm a musician and a producer and I record

976
00:49:49,519 --> 00:49:52,880
bands and everything and make albums. I couldn't tell the difference.

977
00:49:53,320 --> 00:49:56,920
It was really really good a step, Yeah, it's really

978
00:49:56,960 --> 00:49:59,760
quite a step. But here's my opinion. I think that

979
00:50:00,039 --> 00:50:02,039
this is only going to make people want to go

980
00:50:02,119 --> 00:50:05,559
out and see live music that much more and I

981
00:50:05,599 --> 00:50:08,400
think that this is a wave, you know, and after

982
00:50:08,519 --> 00:50:12,880
people have been bamboozled or whatever by a what that's

983
00:50:12,920 --> 00:50:15,840
AI whatever, you know, then they're going to want to

984
00:50:16,239 --> 00:50:18,599
see real people performing with real talent.

985
00:50:19,039 --> 00:50:22,199
Speaker 5: Are we approaching a point where all elevator music is

986
00:50:22,239 --> 00:50:23,519
going to be generated by AI?

987
00:50:23,719 --> 00:50:24,639
Speaker 1: Probably is already.

988
00:50:24,800 --> 00:50:28,400
Speaker 3: Yeah. Spotify is basically admitting they're headed down that path.

989
00:50:28,440 --> 00:50:30,440
You know how you you can turn on a Spotify

990
00:50:30,519 --> 00:50:33,119
playlist of songs, you know, and then it just starts

991
00:50:33,159 --> 00:50:36,079
adding stuff to it. Well, they don't even want to

992
00:50:36,079 --> 00:50:38,320
pay those royalties anymore, so they're going to start using

993
00:50:38,400 --> 00:50:42,079
they're using these tools. If you're not paying attention to playlist,

994
00:50:42,239 --> 00:50:44,239
it's just going to start then sizing music on you.

995
00:50:44,519 --> 00:50:48,800
And the question is will you notice, Yeah, okay, don't

996
00:50:48,800 --> 00:50:49,960
really perform their songs.

997
00:50:49,960 --> 00:50:52,840
Speaker 1: Barry Manilowe, who sang I write the songs, didn't write

998
00:50:57,960 --> 00:51:03,360
And I don't want to make you just to paying

999
00:51:03,440 --> 00:51:08,239
someone to write the lyrics for me or paying someone

1000
00:51:08,239 --> 00:51:11,440
to play the guitar in the track, because they're real people.

1001
00:51:12,039 --> 00:51:16,039
So that's the difference. Someone would make that argument.

1002
00:51:16,360 --> 00:51:18,800
Speaker 4: I think there's a point when we when we do

1003
00:51:18,880 --> 00:51:28,400
talk about generating art whether it's music, video, even generating stories,

1004
00:51:28,480 --> 00:51:34,480
generating write fiction, drama that isn't just coming from nowhere, right.

1005
00:51:34,559 --> 00:51:38,400
We are feeding it information. We are shaping the direction

1006
00:51:38,599 --> 00:51:41,639
of those prompts and sending them down a path, and.

1007
00:51:41,639 --> 00:51:43,960
Speaker 1: For some using pieces from real art.

1008
00:51:44,000 --> 00:51:47,880
Speaker 4: It is so right, there's copyright concerns there. But when

1009
00:51:47,880 --> 00:51:52,760
we think about that as as creators, when you're looking

1010
00:51:52,840 --> 00:51:56,039
to get that that ball rolling, and you're able to

1011
00:51:56,119 --> 00:51:59,119
have a conversation with an AI with a with a

1012
00:51:59,199 --> 00:52:03,480
language model and start to tease out these things, then

1013
00:52:04,159 --> 00:52:06,639
if you're if you're talking through I want to write

1014
00:52:06,679 --> 00:52:09,960
a song, right and I'm just writing the lyrics and

1015
00:52:09,960 --> 00:52:12,280
and then I don't know. I want to throw help

1016
00:52:12,320 --> 00:52:15,039
me out understand what the what what the instruments are

1017
00:52:15,079 --> 00:52:17,440
that I might want to put at this. It's still

1018
00:52:17,519 --> 00:52:19,760
me that's saying, hey, you know what, let me let

1019
00:52:19,800 --> 00:52:21,719
me hear what a what a jazz piano sounds like?

1020
00:52:21,719 --> 00:52:21,960
Speaker 3: With this?

1021
00:52:22,960 --> 00:52:26,119
Speaker 4: Personally, I want to hear a human play that. But

1022
00:52:26,159 --> 00:52:28,639
when I'm writing the song, I don't know how to

1023
00:52:28,639 --> 00:52:30,480
play jazz piano. I don't know how to compose.

1024
00:52:30,559 --> 00:52:31,840
Speaker 1: You might want to get some ideas.

1025
00:52:31,920 --> 00:52:33,559
Speaker 4: I want to get some ideas and to be able

1026
00:52:33,559 --> 00:52:37,000
to get that first level expert to give me those

1027
00:52:37,039 --> 00:52:39,280
ideas demo, to get a demo, to.

1028
00:52:39,280 --> 00:52:41,679
Speaker 1: Hire some real musicians to put their own stamp on it.

1029
00:52:41,800 --> 00:52:43,400
Speaker 4: Because to the to the point that I've heard a

1030
00:52:43,440 --> 00:52:48,039
number of folks say, there's a human feel and emotion

1031
00:52:48,360 --> 00:52:51,800
that you get when that you see not just when

1032
00:52:51,840 --> 00:52:53,679
you listen to music, but when you look at an

1033
00:52:53,760 --> 00:52:56,920
art piece and you see the paint strokes in the painting,

1034
00:52:57,280 --> 00:53:01,199
when you when you see how an illustration is put together,

1035
00:53:01,519 --> 00:53:04,440
you can see and feel how that was done that

1036
00:53:04,800 --> 00:53:07,360
by a human, even if it is a human that's

1037
00:53:07,480 --> 00:53:09,719
drawing in Adobe Photoshop.

1038
00:53:09,800 --> 00:53:12,679
Speaker 1: And conversely, when you read something that's been generated by

1039
00:53:12,719 --> 00:53:15,320
an AI, it has that certain, I don't know, a

1040
00:53:15,920 --> 00:53:18,400
sycophantic feel to it, doesn't it.

1041
00:53:18,199 --> 00:53:20,920
Speaker 4: It's got an extra emoji and mdash everything.

1042
00:53:20,920 --> 00:53:22,960
Speaker 1: It's just the way that it sounds like in your

1043
00:53:23,000 --> 00:53:26,960
mind when you read it. I had this experience, But yeah,

1044
00:53:27,000 --> 00:53:29,360
I can't. I don't know if this is true because

1045
00:53:29,360 --> 00:53:32,119
I haven't confronted my friend about it. But I did

1046
00:53:32,119 --> 00:53:36,000
a post on Facebook and it wasn't I don't remember

1047
00:53:36,039 --> 00:53:38,360
what it was about, but one of my friends one

1048
00:53:38,400 --> 00:53:42,679
of my Facebook friends commented, and I swear to god,

1049
00:53:42,920 --> 00:53:46,239
this person just took my post, put it into chat

1050
00:53:46,280 --> 00:53:49,719
GPT and said, give me a positive reply about this,

1051
00:53:49,920 --> 00:53:52,199
and then they pasted it in. That's what it sounded

1052
00:53:52,239 --> 00:53:55,760
like to me, and because it kind of summarized everything,

1053
00:53:56,079 --> 00:53:58,079
you know, and said, oh, it's so good that you

1054
00:53:58,159 --> 00:54:01,159
blah blah blah, you know, as if to prove that

1055
00:54:01,239 --> 00:54:04,920
they understood what I was saying. Your real friends don't

1056
00:54:04,960 --> 00:54:08,880
do that. No, right, No, they don't summarize and bullet

1057
00:54:08,880 --> 00:54:13,719
point everything you said. So yeah, and let's they want

1058
00:54:13,719 --> 00:54:14,360
to make fun of it.

1059
00:54:14,360 --> 00:54:17,920
Speaker 5: Right, So to move this in a different direction sort

1060
00:54:17,960 --> 00:54:21,559
of positive to me, I think a good area for

1061
00:54:21,719 --> 00:54:27,119
AI is healthcare. And if I have something going on,

1062
00:54:28,000 --> 00:54:31,079
and yes, I want a doctor to read the X

1063
00:54:31,199 --> 00:54:35,320
ray or the MRI, I sure as hell want Claude

1064
00:54:35,320 --> 00:54:38,360
to also look at it, because if they find something

1065
00:54:38,400 --> 00:54:39,920
that the doctor missed.

1066
00:54:41,519 --> 00:54:43,480
Speaker 1: Claude is so positive even if you were going to

1067
00:54:43,519 --> 00:54:47,199
die and say, oh, if I just take a couple

1068
00:54:47,199 --> 00:54:47,880
of shots and.

1069
00:54:49,039 --> 00:54:50,880
Speaker 3: That's one of the best lung tumors I've ever seen.

1070
00:54:52,679 --> 00:54:55,880
Speaker 4: Right, Claude starts every response with you're absolutely.

1071
00:54:55,360 --> 00:54:58,360
Speaker 1: Right, right, say when you tell it it's wrong, but

1072
00:54:58,400 --> 00:55:01,599
you're expected lifespan is lesson sick Bill.

1073
00:55:01,679 --> 00:55:03,960
Speaker 4: You actually make a good point there, because I'm one

1074
00:55:04,000 --> 00:55:06,679
of those people who I'll wait to the last minute

1075
00:55:06,679 --> 00:55:08,960
to go to the doctor. Right, I've got don't do

1076
00:55:09,000 --> 00:55:09,760
that symptoms.

1077
00:55:09,960 --> 00:55:10,320
Speaker 3: Thank you.

1078
00:55:10,760 --> 00:55:13,800
Speaker 4: My wife, missus c Sharp Fritz, has been keeping me

1079
00:55:14,000 --> 00:55:17,920
honest with that. So but but right, I've got symptoms

1080
00:55:18,039 --> 00:55:19,599
x Y and say, you know, I've got this weird

1081
00:55:19,639 --> 00:55:21,440
pain in my leg that happens at some time in

1082
00:55:21,480 --> 00:55:23,519
the afternoon. And she's like, you really should talk to

1083
00:55:23,519 --> 00:55:27,239
the doctor about that. But I was like, I don't know, but.

1084
00:55:27,400 --> 00:55:29,639
Speaker 1: Okay, honestly, wouldn't trust an AI to give me any

1085
00:55:29,719 --> 00:55:31,440
kind of so anything.

1086
00:55:31,679 --> 00:55:34,440
Speaker 4: I mentioned here's the medications that I'm taking, here's the

1087
00:55:34,480 --> 00:55:36,400
weird symptoms that I'm feeling. And this happened to me

1088
00:55:36,440 --> 00:55:38,800
while I was in Portugal for a conference we were

1089
00:55:38,840 --> 00:55:41,039
speaking at and it came back and said, no, you

1090
00:55:41,039 --> 00:55:42,679
you might want to talk to your doctor about that,

1091
00:55:42,719 --> 00:55:45,599
because you might have a wrong dose on this medication

1092
00:55:45,679 --> 00:55:48,760
you're taking. Sure enough, talk to the doctor and yeah,

1093
00:55:48,840 --> 00:55:52,239
let's dial that back. And I don't have the problem.

1094
00:55:52,840 --> 00:55:53,599
Speaker 1: But I wouldn't.

1095
00:55:53,960 --> 00:55:57,639
Speaker 4: I would have right, because I'm a middle aged guy

1096
00:55:58,280 --> 00:56:01,840
in America. I would have been like, when you drive

1097
00:56:02,000 --> 00:56:08,400
or anything, it could be supreme tuma, but probably a headache.

1098
00:56:09,639 --> 00:56:16,880
Speaker 3: The radiology story is an interesting one because there was

1099
00:56:16,920 --> 00:56:20,760
this whole point made by like Jeff Hinton ten years

1100
00:56:20,800 --> 00:56:24,760
ago saying, you know, generative AI is now figured out radiology.

1101
00:56:24,800 --> 00:56:29,079
It's better at analyzing images than humans are. Radiologists are

1102
00:56:29,119 --> 00:56:33,000
are done, They're totally unnecessary. He was one hundred percent incorrect.

1103
00:56:33,320 --> 00:56:36,119
The demand for radiologists has only gone up, and part

1104
00:56:36,159 --> 00:56:39,079
of that is that there was such a huge unmet

1105
00:56:39,119 --> 00:56:43,280
demand for imaging, and the software accelerated the ability for

1106
00:56:43,360 --> 00:56:47,679
radiologists to do a good job. And so now they're

1107
00:56:47,719 --> 00:56:50,519
doing three times as much imaging and most radiologists now

1108
00:56:50,679 --> 00:56:55,719
use it's like seven hundred models certified by the FDA

1109
00:56:56,599 --> 00:56:57,400
for imaging.

1110
00:56:57,519 --> 00:56:59,639
Speaker 1: So this is where I totally agree with Bill that

1111
00:56:59,679 --> 00:57:04,440
you have these specified models that are trained on their

1112
00:57:04,519 --> 00:57:09,480
particular like radiology data, right, that are so narrowly focused

1113
00:57:09,519 --> 00:57:13,239
that they're going to give you a better outcome diagnosis.

1114
00:57:13,280 --> 00:57:15,599
Then if you just you know, type in a chat GPT,

1115
00:57:15,960 --> 00:57:16,760
you know, I got it.

1116
00:57:16,800 --> 00:57:19,400
Speaker 3: Yeah, I wonder if we're going to move away. As

1117
00:57:19,480 --> 00:57:21,840
the dumb winds down on this. You know, if you

1118
00:57:21,880 --> 00:57:27,199
think how much better the web got after two what's

1119
00:57:27,239 --> 00:57:30,239
the level you know, after the dot com boom ended,

1120
00:57:30,760 --> 00:57:33,000
and you know, after a year or two, we built

1121
00:57:33,039 --> 00:57:35,800
better websites, like we stopped racing, started to think about

1122
00:57:35,800 --> 00:57:41,119
what actually made sense. Right now, the business is good

1123
00:57:41,599 --> 00:57:45,000
for making them models bigger, right Those companies are all

1124
00:57:45,000 --> 00:57:48,199
incentive to make models bigger because they're going for these

1125
00:57:48,199 --> 00:57:50,880
outrageous evaluations. That means they have to constantly show they

1126
00:57:50,880 --> 00:57:54,480
need to spend more money counter to what they actually needed,

1127
00:57:54,480 --> 00:57:58,679
what's actually benefit show the customer. So as this ends,

1128
00:57:58,880 --> 00:58:01,320
and it must, Yeah, I'm wondering if we're not going

1129
00:58:01,280 --> 00:58:04,039
to go a local model. Right. So software development is

1130
00:58:04,039 --> 00:58:07,960
actually a pretty narrow domain space. It's pretty tight. So

1131
00:58:08,239 --> 00:58:11,000
I wonder when the dumb wind's down and we are

1132
00:58:11,039 --> 00:58:13,079
focused on the most efficient ways to do things, is

1133
00:58:13,079 --> 00:58:15,320
that the tookys are gonna become irrelevant. You're gonna own

1134
00:58:15,320 --> 00:58:15,679
that gear.

1135
00:58:15,800 --> 00:58:19,840
Speaker 1: And also it becomes more attractive to your customers because

1136
00:58:19,880 --> 00:58:23,920
you're not sharing screenshots and code of their Yeah, you know.

1137
00:58:23,880 --> 00:58:26,480
Speaker 3: There's how you get rid of the whole sovereignty problem.

1138
00:58:26,559 --> 00:58:27,400
Never leaves the building.

1139
00:58:27,480 --> 00:58:30,719
Speaker 4: Yeah, foundry local is is absolutely a thing, O Lama.

1140
00:58:30,920 --> 00:58:33,559
Running locally is absolutely a thing that we see folks

1141
00:58:33,639 --> 00:58:36,119
using more and more. Right, we get the we get

1142
00:58:36,119 --> 00:58:38,719
the fee models that are running locally on Windows. You

1143
00:58:38,719 --> 00:58:41,159
can run Quinn on Windows. And if you're running with

1144
00:58:41,559 --> 00:58:44,400
that the surface laptop with the NPU on it, you

1145
00:58:44,440 --> 00:58:47,639
can do some amazing things. If you're playing Fortnite and

1146
00:58:47,639 --> 00:58:51,000
you've got a fantastic GPU on your system at home, well, yeah.

1147
00:58:50,800 --> 00:58:54,960
Speaker 3: And the radio implodes. GPUs are only going to get cheaper. Yeah,

1148
00:58:55,039 --> 00:58:56,960
and we're probably going to be you know, racking up

1149
00:58:56,960 --> 00:58:59,199
a few of those things in an office. Somebody's got

1150
00:58:59,280 --> 00:59:00,639
to running them on stop.

1151
00:59:01,400 --> 00:59:05,920
Speaker 4: We are still seeing that that growth of data centers, right,

1152
00:59:05,960 --> 00:59:08,320
there's been there's a number of properties here in the

1153
00:59:09,360 --> 00:59:13,760
southeast Pennsylvania region around Philadelphia where we are seeing old

1154
00:59:13,920 --> 00:59:17,119
old warehouses, old factories that are being bought out and.

1155
00:59:17,039 --> 00:59:19,599
Speaker 5: Being turned into nuclear power plight.

1156
00:59:19,760 --> 00:59:22,320
Speaker 4: Let's come back to that in a second. But we're

1157
00:59:22,320 --> 00:59:25,440
seeing these older, right, industrial facilities that are being bought

1158
00:59:25,480 --> 00:59:28,519
out and turned into data centers. And there's people protesting

1159
00:59:28,880 --> 00:59:33,960
because it's it's higher demand on electricity to bill you

1160
00:59:33,960 --> 00:59:36,000
made an excellent point. We brought up three mile.

1161
00:59:35,920 --> 00:59:36,880
Speaker 3: Hour in Philadelphia.

1162
00:59:36,920 --> 00:59:40,159
Speaker 1: When you say we Microsoft, yeah, correct, Yeah, his blue

1163
00:59:40,199 --> 00:59:43,360
badge is showing. Yeah.

1164
00:59:43,599 --> 00:59:47,800
Speaker 3: So Microsoft has kicked in something two billion dollars to Constellation,

1165
00:59:47,960 --> 00:59:50,679
which is the company that had been operating three Mile Island.

1166
00:59:50,960 --> 00:59:53,880
They turned it off in twenty nineteen because it costs

1167
00:59:53,960 --> 00:59:58,079
more to operate than the natural gas combined cycle plants

1168
00:59:58,119 --> 01:00:01,400
that have been built all over Pennsylvania now. And but

1169
01:00:01,480 --> 01:00:04,119
Microsoft doesn't care. Eight hundred megawatts of electricity is eight

1170
01:00:04,159 --> 01:00:07,800
hundred megawats electricity, and it was a way to try

1171
01:00:07,800 --> 01:00:11,000
and put that more power online. It's probably gonna take

1172
01:00:11,000 --> 01:00:12,880
a few more years for they get it done. Twenty

1173
01:00:12,880 --> 01:00:15,559
seven seven, that's what they're shooting for. They're probably it'll

1174
01:00:15,559 --> 01:00:18,719
probably take longer debt. There's a dozen plants like that

1175
01:00:18,760 --> 01:00:23,280
around the US. So now Google has approached some some

1176
01:00:23,400 --> 01:00:26,159
folks in Cedar Springs and Idaho, same kind of situation.

1177
01:00:26,320 --> 01:00:30,039
Five reactors shut down for five years. So I mean

1178
01:00:30,079 --> 01:00:32,920
it's going to help. We're in any more power than that.

1179
01:00:33,719 --> 01:00:36,920
It's interesting to think in terms rather dystopic terms of

1180
01:00:37,519 --> 01:00:41,840
tech companies now going to advance technologies on power generation

1181
01:00:42,119 --> 01:00:45,199
of course, I'm about to record the energy geek out,

1182
01:00:45,320 --> 01:00:48,960
So here you go, you get a preview. Amazon's now

1183
01:00:49,000 --> 01:00:52,239
committed to a new reactor design. Like all the tech

1184
01:00:52,280 --> 01:00:54,000
giants are looking for other power source and.

1185
01:00:54,039 --> 01:00:58,159
Speaker 5: Another place AI is got a factor in as grid management. Yeah,

1186
01:00:58,239 --> 01:00:59,760
without getting the electricity.

1187
01:01:00,480 --> 01:01:04,039
Speaker 3: But I also think we're overbuilding, right, We're in that.

1188
01:01:04,199 --> 01:01:06,440
We're again in the bubble situation. Like what was the

1189
01:01:06,480 --> 01:01:10,599
overbuild during the dot com boom? It was fiber optic cable, right,

1190
01:01:10,679 --> 01:01:12,519
there was a ton of fiber optic cable was late

1191
01:01:12,559 --> 01:01:15,320
and many of those companies went broke shortly thereafter, and

1192
01:01:15,400 --> 01:01:17,559
the fire and the and the cable was bought up

1193
01:01:17,599 --> 01:01:21,280
ten cents on the dollar. Yeah. The only thing I

1194
01:01:21,280 --> 01:01:23,840
would say that's safer this time around is the tech

1195
01:01:23,880 --> 01:01:27,320
giants for the most part, have been spending cash money

1196
01:01:27,360 --> 01:01:29,519
they already have that it would have used on stock

1197
01:01:29,559 --> 01:01:32,719
buybacks or something and instead or turning it into land.

1198
01:01:33,119 --> 01:01:37,199
Speaker 4: But having that extra electric power on the grid not

1199
01:01:37,239 --> 01:01:40,719
only is going to facilitate AI focused data centers, but

1200
01:01:41,320 --> 01:01:43,400
we've been trying to do electric cars for how long?

1201
01:01:43,800 --> 01:01:46,239
We are doing electric cars right, but I mean at

1202
01:01:46,320 --> 01:01:50,480
scale with significant delivery. Here the amount of draw that

1203
01:01:50,480 --> 01:01:52,519
that puts on the grid. When we start thinking about

1204
01:01:53,800 --> 01:01:57,239
higher percentages of vehicles on the road electric based.

1205
01:01:57,519 --> 01:01:59,679
Speaker 3: They're not even close to the same league of what

1206
01:01:59,719 --> 01:02:01,199
pot data centers are trying to consume.

1207
01:02:01,360 --> 01:02:04,159
Speaker 4: No, no, not at all. But having that extra power on.

1208
01:02:04,119 --> 01:02:06,039
Speaker 3: The it's not have more power on ARID is not

1209
01:02:06,079 --> 01:02:08,840
going to hurt. You could stand to have more electricity.

1210
01:02:09,360 --> 01:02:12,159
I think it's going to be an overbuild. I think

1211
01:02:12,199 --> 01:02:13,440
a lot of them are not going to get finished.

1212
01:02:13,480 --> 01:02:15,239
I think it's interesting to think in terms of Microsoft

1213
01:02:15,280 --> 01:02:18,519
not being a software company anymore. Yeah, they're a utility company.

1214
01:02:18,519 --> 01:02:19,760
Their utility.

1215
01:02:19,880 --> 01:02:23,079
Speaker 1: So should we call this the future of not software

1216
01:02:23,119 --> 01:02:24,239
development but the future?

1217
01:02:24,800 --> 01:02:27,960
Speaker 3: No, I've renamed it the role of AI and software

1218
01:02:27,960 --> 01:02:31,000
development because that's clearly clearly we ended up for better

1219
01:02:31,079 --> 01:02:34,679
or worse. It's only a future. I've been doing this

1220
01:02:34,760 --> 01:02:38,760
AI hype keynote for a while now, very much focused

1221
01:02:38,840 --> 01:02:41,559
on what the end of this bubble looks like because

1222
01:02:41,599 --> 01:02:44,400
we've been through this before. Lots of people are going

1223
01:02:44,440 --> 01:02:47,599
to be hurt. I only it's going to be us. No,

1224
01:02:48,639 --> 01:02:52,440
We're a scarce resource and our ability to learn these

1225
01:02:52,480 --> 01:02:55,400
new tools isn't going to go away. This tool is

1226
01:02:55,440 --> 01:02:56,920
not going to go away anymore than the Web went

1227
01:02:56,960 --> 01:03:01,039
away after the dot com boom ended. In fact, efficiency

1228
01:03:01,119 --> 01:03:04,800
is going to be the word, and most people in

1229
01:03:04,840 --> 01:03:08,840
software actually like being efficient. They don't like the rampant inefficiency.

1230
01:03:09,039 --> 01:03:12,559
And so some ways the knuckle downtimes of a downturn

1231
01:03:13,119 --> 01:03:18,119
is where we thrive because we do make our companies

1232
01:03:18,119 --> 01:03:20,159
more efficient with what we do, and these tools can

1233
01:03:20,239 --> 01:03:22,119
do that. It's just that right now there's a lot

1234
01:03:22,159 --> 01:03:26,320
of people incentive to not focus on that. That will end.

1235
01:03:26,920 --> 01:03:27,840
The tools won't.

1236
01:03:27,880 --> 01:03:30,760
Speaker 4: So if we're on AI one point zero and Web

1237
01:03:30,840 --> 01:03:33,360
one point oh, was everybody build your HTML? Web two

1238
01:03:33,400 --> 01:03:35,159
point zero is social media? Now?

1239
01:03:35,559 --> 01:03:38,840
Speaker 3: It was? It was mash up too, right, Like I

1240
01:03:38,840 --> 01:03:43,079
would argue that the post bubble AI will be hybrid,

1241
01:03:43,440 --> 01:03:47,719
a lot more client side and a lot more focused

1242
01:03:47,719 --> 01:03:50,519
model types. Like I wonder if you think about how

1243
01:03:50,559 --> 01:03:52,679
constrained you are in a version three of the piece

1244
01:03:52,679 --> 01:03:55,599
of software, right if adding something new to the stack

1245
01:03:55,760 --> 01:03:58,719
is anathema. Right, this is our data store, this is

1246
01:03:58,719 --> 01:04:00,840
our client libraries, like your kind of in a box.

1247
01:04:00,960 --> 01:04:03,880
Like that's a pretty tight little model. Like here's an

1248
01:04:03,920 --> 01:04:07,000
interesting thought for you. The more mature your software gets,

1249
01:04:07,159 --> 01:04:09,639
the less it's going to cost to maintain because the

1250
01:04:09,760 --> 01:04:12,760
simpler the model is, because the constraints are so clear,

1251
01:04:13,360 --> 01:04:16,320
and your most experienced developers who don't want to maintain

1252
01:04:16,360 --> 01:04:18,280
that software and move on to the new things. The

1253
01:04:18,400 --> 01:04:21,840
juniors come in and these tools would protect them from

1254
01:04:21,880 --> 01:04:25,800
making mistakes. Your code coverage betters, your your rules are

1255
01:04:25,840 --> 01:04:30,639
clearly delineated. Like weirdly, we might actually get more reliable

1256
01:04:30,679 --> 01:04:34,639
software out of this for relatively low costs based on

1257
01:04:34,719 --> 01:04:37,360
the way models behave when they're well constrained.

1258
01:04:37,920 --> 01:04:41,760
Speaker 4: The push for better models that are smaller and use

1259
01:04:41,840 --> 01:04:45,639
less power means that we don't need to deploy and

1260
01:04:45,719 --> 01:04:47,760
run as many data centers consume.

1261
01:04:47,480 --> 01:04:50,559
Speaker 3: As much energy. That stuff's going to be on my desk.

1262
01:04:50,800 --> 01:04:52,519
Yeah right, I don't want to go out in the

1263
01:04:52,519 --> 01:04:53,079
cloud at all.

1264
01:04:53,480 --> 01:04:55,920
Speaker 4: Yeah, I want my Raspberry Pie tower running.

1265
01:04:55,679 --> 01:04:58,840
Speaker 1: It as Hey, I think we got to wrap it up,

1266
01:04:58,840 --> 01:05:02,480
but let's give one more round of applause for Jeff Fritz.

1267
01:05:05,719 --> 01:05:08,000
Speaker 2: And our friend walking out the door beIN a wolf.

1268
01:05:10,639 --> 01:05:13,639
Speaker 1: Congratulations, what do I hear?

1269
01:05:14,000 --> 01:05:14,880
Speaker 3: One? Two, three?

1270
01:05:16,880 --> 01:05:18,559
Speaker 5: All right, that's our rally cry at Philly.

1271
01:05:19,079 --> 01:05:22,840
Speaker 1: Awesome and thank you all for coming to dot net.

1272
01:05:22,840 --> 01:05:45,599
Speaker 2: Rocks dot net.

1273
01:05:45,679 --> 01:05:48,599
Speaker 1: Rocks is brought to you by Franklin's Net and produced

1274
01:05:48,599 --> 01:05:52,440
by Pop Studios, a full service audio, video and post

1275
01:05:52,440 --> 01:05:56,639
production facilities located physically in New London, Connecticut, and of

1276
01:05:56,639 --> 01:06:00,519
course in the cloud online at pwop dot com.

1277
01:06:01,320 --> 01:06:03,480
Speaker 2: Visit our website at d O T N E t

1278
01:06:03,679 --> 01:06:03,960
R O.

1279
01:06:04,000 --> 01:06:09,280
Speaker 1: C k S dot com for RSS feeds, downloads, mobile apps, comments,

1280
01:06:09,599 --> 01:06:12,119
and access to the full archives going back to show

1281
01:06:12,199 --> 01:06:14,840
number one, recorded in September two.

1282
01:06:14,639 --> 01:06:15,280
Speaker 3: Thousand and two.

1283
01:06:15,880 --> 01:06:18,239
Speaker 1: And make sure you check out our sponsors. They keep

1284
01:06:18,280 --> 01:06:21,480
us in business. Now go write some code. See you

1285
01:06:21,519 --> 01:06:26,800
next time. You got jud Middle Vans and