1
08:37:57,520 --> 08:37:57,940
Amazing.

2
08:37:58,280 --> 08:37:59,420
That was such a cool animation.

3
08:38:00,100 --> 08:38:02,300
But, so it's not World of Warcraft.

4
08:38:02,620 --> 08:38:03,980
I'm not sure what I'm doing here.

5
08:38:05,060 --> 08:38:07,960
But hey guys, I'm Tharik, nice to meet you.

6
08:38:09,740 --> 08:38:14,700
Yeah, I wanted to talk to you about a couple things I've been thinking about since I left

7
08:38:14,700 --> 08:38:15,320
Anthropik.

8
08:38:15,840 --> 08:38:21,760
It's been a wild year and there are a bunch of reflections I've had.

9
08:38:23,160 --> 08:38:29,120
And my conclusion is about what I'm calling the sweet side of the bitter lesson and the

10
08:38:29,120 --> 08:38:29,860
ways I've been wrong.

11
08:38:29,860 --> 08:38:32,980
So I think I got a lot of things right this year.

12
08:38:34,180 --> 08:38:39,840
I think I was right about models being able to edit and generate videos.

13
08:38:40,040 --> 08:38:43,440
I was talking about this sometime in June last year.

14
08:38:44,040 --> 08:38:48,960
I was right about models being able to use HTML to understand their work.

15
08:38:49,680 --> 08:38:55,420
And I was right about models being able to help you understand and figure out your unknown,

16
08:38:55,660 --> 08:38:58,400
to help plan and learn about things.

17
08:38:58,400 --> 08:39:06,480
So these were all things I was like right about and I was happy and I think that like a lot of you might be incorporating this sort of stuff in your work

18
08:39:07,560 --> 08:39:11,760
But I've been wrong a lot too and I think this is one of those things in AI where you sort of just like

19
08:39:12,340 --> 08:39:17,860
Memory hold the things that you're wrong about and you're like oh like I was just always expecting you know AI to be great

20
08:39:19,200 --> 08:39:23,720
But here are some of the things I've been wrong about so first like bash is all you need

21
08:39:23,720 --> 08:39:28,700
I think there was this moment where the agents were really good at using Bash and Cloud Code

22
08:39:28,700 --> 08:39:30,000
was built around the Bash tool.

23
08:39:30,720 --> 08:39:34,640
And so people are like, well, let's turn everything into a CLI.

24
08:39:35,440 --> 08:39:39,920
There's just going to be Bash and CLIs and that's going to take over the world.

25
08:39:40,980 --> 08:39:43,040
That's actually kind of not how it turned out.

26
08:39:43,140 --> 08:39:48,720
I actually prefer MCP to CLIs now, but now there's also code mode and skills and progressive

27
08:39:48,720 --> 08:39:49,140
disclosure.

28
08:39:49,140 --> 08:39:55,480
There are so many other ways that agents can act on the world.

29
08:39:55,920 --> 08:39:59,800
And the fact that they were using Bash was just like this weird remnant

30
08:40:00,000 --> 08:40:03,220
of them being trained in a certain way at the start.

31
08:40:05,020 --> 08:40:11,240
Then there's React is all you need. I think for a moment it seemed like the models were really, really good at writing React.

32
08:40:11,600 --> 08:40:17,480
And so if you wanted to create an AI-enabled app, you should figure out how to use React.

33
08:40:17,480 --> 08:40:20,960
Like, you should use React Native, you should use, you know, like, you should just make

34
08:40:20,960 --> 08:40:21,840
everything on the web.

35
08:40:24,520 --> 08:40:26,120
And that also has not been true.

36
08:40:26,260 --> 08:40:30,400
I think, like, now, you know, models can write it in HTML, they can write it in React, they

37
08:40:30,400 --> 08:40:36,340
can write it in Swift, they can write it in the Figma proprietary JSON format, and they'll

38
08:40:36,340 --> 08:40:38,600
all look roughly the same, right?

39
08:40:38,720 --> 08:40:42,420
And so I was afraid that, like, you know, the models have been trained on React, and

40
08:40:42,420 --> 08:40:46,320
so we'd just be writing React forever, and React would be, like, the last language almost,

41
08:40:46,500 --> 08:40:46,600
right?

42
08:40:46,600 --> 08:40:48,640
And that's almost certainly not true, right?

43
08:40:48,720 --> 08:40:50,680
I think a lot of people are writing their own languages now.

44
08:40:53,920 --> 08:40:56,680
And I think one of the other things I was really concerned about

45
08:40:56,680 --> 08:40:58,660
was like diffusion models will kind of like eat art.

46
08:40:58,800 --> 08:40:59,980
Like it would just be like, you know,

47
08:41:00,200 --> 08:41:02,480
we wouldn't have any control over these models

48
08:41:02,480 --> 08:41:04,960
or the kind of art that these models are creating,

49
08:41:05,380 --> 08:41:07,580
but it would just like create the most beautiful stuff.

50
08:41:07,780 --> 08:41:09,760
And that hasn't been true at all.

51
08:41:09,940 --> 08:41:13,180
Like I think now LLMs can make art through code,

52
08:41:13,480 --> 08:41:15,820
through controlling your computer and using, you know,

53
08:41:15,820 --> 08:41:20,680
Blender or Photoshop or Figma or whatever you'd like.

54
08:41:20,840 --> 08:41:28,880
And so these were all fears I had about, I think, the bitter lesson,

55
08:41:29,540 --> 08:41:34,780
our alien intelligence, the shoggoth, these are all words we use, I think,

56
08:41:34,880 --> 08:41:41,060
to describe this kind of fear that the models will just have this way of doing things

57
08:41:41,060 --> 08:41:44,560
and it will be weird and alien to us and we just need to accept it.

58
08:41:44,560 --> 08:41:48,540
You just have to submit to this foreign intelligence.

59
08:41:49,580 --> 08:41:52,060
But that's just like, I don't think that's what happened.

60
08:41:54,220 --> 08:42:00,260
And so I think the question I come back to a lot is like, okay, how can we be AGI-pilled and human-centered?

61
08:42:00,380 --> 08:42:03,500
And what can we learn about this progression?

62
08:42:04,700 --> 08:42:06,520
What can I learn about what I was wrong about?

63
08:42:07,360 --> 08:42:12,380
So I think this is really, really important because software engineers are the leading edge of AI adoption.

64
08:42:12,380 --> 08:42:17,940
Right? Like anything that is happening to software will happen to knowledge work and other parts of the economy.

65
08:42:18,300 --> 08:42:25,840
And so I think how we deal and engage with AI is how other people will as well.

66
08:42:26,160 --> 08:42:30,340
And I really enjoyed the different talks here talking about things like, yeah, fighting swap.

67
08:42:30,540 --> 08:42:34,960
I think we're the first defense for that, right? Or like being more ambitious.

68
08:42:34,960 --> 08:42:41,540
this, what I'd like to think about it as AI is both at the same time an industrial revolution

69
08:42:42,080 --> 08:42:43,580
and a renaissance, right?

70
08:42:43,800 --> 08:42:48,760
And I think this year was really about the industrial revolution, right?

71
08:42:48,900 --> 08:42:51,680
Like we use the word software factories very often.

72
08:42:53,940 --> 08:43:01,140
And like just sort of like turning software from this like craft that was very unreliable

73
08:43:01,140 --> 08:43:06,520
and hard to execute on into something that is not the limiting factor.

74
08:43:08,360 --> 08:43:11,820
And this sort of industrialization is absolutely happening.

75
08:43:11,940 --> 08:43:15,160
This is definitely something that is happening with software.

76
08:43:15,360 --> 08:43:18,840
But at the same time, there's also kind of a renaissance happening, right?

77
08:43:19,040 --> 08:43:23,420
Where people want a higher and higher quality software.

78
08:43:23,420 --> 08:43:32,580
and every marginal bit of creativity, utility, spirit, soul that you put into it is useful

79
08:43:32,580 --> 08:43:34,260
and economically valuable.

80
08:43:34,740 --> 08:43:39,020
And so simultaneously, you've got both of these things that are very hard to juggle

81
08:43:39,020 --> 08:43:41,880
and sort of hold in your head at the same time.

82
08:43:42,220 --> 08:43:44,100
But I think it is very true.

83
08:43:44,680 --> 08:43:47,260
And yeah, we're doing both of them.

84
08:43:47,380 --> 08:43:52,200
So oftentimes when you talk about AI capabilities, we talk about the jagged frontier.

85
08:43:52,200 --> 08:43:59,060
And so maybe you go back two years and you're asking AI to write this function for you,

86
08:44:00,240 --> 08:44:04,560
or a year and a half ago or a year ago, it's like, build this feature, and you're like,

87
08:44:04,580 --> 08:44:09,080
hey, build this to-do app, build a kind of simple to-do app, and then clone Slack for

88
08:44:09,080 --> 08:44:09,560
me now.

89
08:44:09,700 --> 08:44:15,880
And now it's like, okay, build Mario Kart from scratch, and then port it to the Wii

90
08:44:15,880 --> 08:44:16,360
or something.

91
08:44:16,360 --> 08:44:21,180
There's like the capability increases are the things that you see, right?

92
08:44:21,340 --> 08:44:25,400
Where, you know, whenever you look at the evals and benchmarks and stuff,

93
08:44:25,600 --> 08:44:30,020
you see this like just this jagged frontier of like capabilities.

94
08:44:30,340 --> 08:44:37,520
But I also like to talk about like the smooth interior where the way,

95
08:44:37,840 --> 08:44:42,700
whoops, the way the models, is there an animation playing here?

96
08:44:44,200 --> 08:44:52,620
The smooth interior is like, the idea is that the models are, as they get better, the animation

97
08:44:52,620 --> 08:44:55,480
here is like this jagged front and this circle in between.

98
08:44:55,840 --> 08:45:02,100
And so the circle sort of represents all of the things that the models are smart enough

99
08:45:02,100 --> 08:45:02,520
to do.

100
08:45:02,720 --> 08:45:08,200
And so like examples being like writing React code, right, or like using bash.

101
08:45:08,200 --> 08:45:15,360
and what you see is that like it's not jagged below the frontier, right? Like if it can write

102
08:45:15,360 --> 08:45:20,620
React, it can write Figma, it can write JSON, it can do sort of like all of these capabilities

103
08:45:20,620 --> 08:45:24,980
all together. And there's like a breadth of task that the models are really, really good at.

104
08:45:25,240 --> 08:45:31,740
And this is something that's also I think really like important and for me even more impactful

105
08:45:31,740 --> 08:45:37,220
because I don't really care that they can build Mario Kart from scratch anymore, right? What I

106
08:45:37,220 --> 08:45:41,800
really care about is like, hey, is the software that we're building, is it high quality, is

107
08:45:41,800 --> 08:45:44,700
it worthy of like a renaissance, you know what I mean?

108
08:45:47,360 --> 08:45:51,160
So yeah, like I think that like, you know, you can do things the way you want.

109
08:45:51,280 --> 08:45:54,640
So like if you're building mock-up, you can be like, hey, can I get an HTML, can I get

110
08:45:54,640 --> 08:45:59,180
a React, can I get a Figma file, can I computer use my computer and use After Effects, can

111
08:45:59,180 --> 08:46:00,180
you make a video?

112
08:46:00,460 --> 08:46:04,660
There are all these different ways of getting the like final result.

113
08:46:05,080 --> 08:46:06,620
And I've been really happy with this.

114
08:46:06,620 --> 08:46:11,740
I really was afraid that there would just be like one thing that the models love doing and we'd only be doing that

115
08:46:13,420 --> 08:46:14,300
and you know like

116
08:46:14,940 --> 08:46:18,160
Taken to the limit kind of like, you know, we talked about this a lot

117
08:46:18,160 --> 08:46:24,240
but like, you know, everything we've talked about so far about like, you know, how you build like these harnesses and software factories and

118
08:46:25,080 --> 08:46:27,220
these like this scaffolding for the models

119
08:46:28,020 --> 08:46:33,780
even things like, you know, for a while we were also like, hey, the model you'll like

120
08:46:34,780 --> 08:46:38,100
your job is to QA the model, you know, or review the code.

121
08:46:38,260 --> 08:46:39,660
And that's clearly not the case, right?

122
08:46:39,740 --> 08:46:41,640
I think as the models get smarter and smarter,

123
08:46:42,720 --> 08:46:46,580
it will become more of this, like, smooth interior, right?

124
08:46:46,840 --> 08:46:49,780
Where they're just doing things well, right?

125
08:46:49,880 --> 08:46:51,640
And, like, the perfect model you can imagine

126
08:46:51,640 --> 08:46:54,120
won't need to, like, really verify its work

127
08:46:54,120 --> 08:46:55,660
or scan for security vulnerabilities.

128
08:46:56,040 --> 08:46:57,600
We'll just, like, write the code and be like,

129
08:46:57,620 --> 08:46:58,760
hey, I'm done now.

130
08:46:59,840 --> 08:47:01,920
And you will check, and you'll be like, oh, you're right.

131
08:47:01,920 --> 08:47:06,880
Like, you know, according to the spec and the goals that we did, you did it correctly.

132
08:47:08,820 --> 08:47:10,760
And I think that's actually very exciting.

133
08:47:11,720 --> 08:47:17,640
One of the ways I like to sort of think about visualizing how much the model knows of this

134
08:47:17,640 --> 08:47:20,100
eval, I think Kaparthi posted it.

135
08:47:20,680 --> 08:47:27,520
But basically, you ask any, I think these are Claude models, you ask a model, you give

136
08:47:27,520 --> 08:47:32,280
it a longitude latitude and you say, hey, is this point earth or water?

137
08:47:32,640 --> 08:47:32,740
Right?

138
08:47:33,160 --> 08:47:34,740
Ground, sorry, land or water.

139
08:47:35,000 --> 08:47:37,960
And from that you plot a map of the earth.

140
08:47:39,260 --> 08:47:45,140
And yeah, this is Opus 4.6 is somewhere over there.

141
08:47:45,260 --> 08:47:46,700
Quite a good software engineering model.

142
08:47:47,040 --> 08:47:48,300
Kind of like a blurry view.

143
08:47:48,480 --> 08:47:51,820
And now you look at Opus 5.5 and it's almost perfect.

144
08:47:51,820 --> 08:47:57,820
It knows every piece of these coordinates.

145
08:47:57,820 --> 08:48:00,040
It knows the entire world, basically.

146
08:48:00,320 --> 08:48:07,300
And so this, I think, just gives you an idea besides the benchmarks about the jagged frontier.

147
08:48:07,560 --> 08:48:13,260
I feel like that's less and less important to us, and more important is just the general knowledge that the models have.

148
08:48:14,520 --> 08:48:20,360
And what's exciting, I think, about this is that we can use intelligence to understand

149
08:48:20,360 --> 08:48:22,360
and shape how we use intelligence, right?

150
08:48:22,420 --> 08:48:26,520
And I think that will be, like, the theme of the next year, right?

151
08:48:26,620 --> 08:48:30,580
Where this year, I think, was really about the industrialization of software and creating

152
08:48:30,580 --> 08:48:32,420
this, like, reliable way to make software.

153
08:48:34,200 --> 08:48:35,480
But what does that look like?

154
08:48:35,580 --> 08:48:37,620
So I have been building a game.

155
08:48:38,600 --> 08:48:41,080
It's this thing I've been posting on Twitter about

156
08:48:41,760 --> 08:48:44,300
It's a 2D isometric game

157
08:48:44,880 --> 08:48:46,440
It's kind of like a fighting game

158
08:48:46,440 --> 08:48:49,180
Inspired by things like Avatar and Smash Brothers

159
08:48:51,340 --> 08:48:53,640
And I have kind of like

160
08:48:53,640 --> 08:48:55,820
It's fun, but it's also quite frustrating to do

161
08:48:55,820 --> 08:48:59,300
I'm really trying to be a great vibe coder here

162
08:48:59,300 --> 08:49:02,400
I don't really know what each character does sometimes

163
08:49:02,400 --> 08:49:06,320
Because I will say, hey, I want this character to throw a rock

164
08:49:06,320 --> 08:49:11,880
And they'll be like, oh, yeah, great, but like there are all these like different things that can happen when you're throwing a rock

165
08:49:11,880 --> 08:49:14,140
Like this character can also summon a wall

166
08:49:14,140 --> 08:49:22,000
So does the rock go through the wall or not and like how what speed or like what what about size and charge and all these questions?

167
08:49:23,720 --> 08:49:24,120
and

168
08:49:24,120 --> 08:49:27,800
I I find that like the model just makes a bunch of choices for me and then I try it out

169
08:49:27,800 --> 08:49:30,520
I'm like wow I have no idea what this character actually does I

170
08:49:31,420 --> 08:49:34,320
Get distracted pretty easily. I don't feel in the zone very much

171
08:49:34,320 --> 08:49:38,380
You know, I sort of send off a prompt and I start doing something else and I come back

172
08:49:40,060 --> 08:49:43,800
And I honestly make less progress because of these things right and so

173
08:49:44,880 --> 08:49:47,660
I think this is frustrating and I think for a while I was like man

174
08:49:47,660 --> 08:49:51,520
I guess this is just like how you work with AI. This is just like what it's gonna be like, you know

175
08:49:51,520 --> 08:49:54,720
But I think reflecting I'm like man. I don't think that's right

176
08:49:54,720 --> 08:49:59,980
I think that like, you know, we have this like the AI's can work in the way that we

177
08:50:00,000 --> 08:50:00,420
want.

178
08:50:00,740 --> 08:50:03,140
And so, yeah, I don't think it's acceptable.

179
08:50:03,320 --> 08:50:05,520
I don't think that this is what leads to the renaissance.

180
08:50:05,680 --> 08:50:08,000
I don't think this is what leads to incredible software.

181
08:50:09,240 --> 08:50:12,220
And part of it, going back to what Dex said,

182
08:50:12,300 --> 08:50:16,920
is great work is just about this messy process of you

183
08:50:16,920 --> 08:50:18,980
don't know exactly what you want.

184
08:50:18,980 --> 08:50:23,280
It's more about the choices you made and what wasn't involved

185
08:50:23,280 --> 08:50:26,460
or what wasn't included versus what was included.

186
08:50:27,300 --> 08:50:31,560
And so, yeah, I think there's like,

187
08:50:32,780 --> 08:50:34,580
you kind of just have to be an expert.

188
08:50:34,680 --> 08:50:36,380
You have to just really know your stuff

189
08:50:36,380 --> 08:50:38,080
in order to make something great, right?

190
08:50:38,140 --> 08:50:41,080
In order to make something worthy of a renaissance, right?

191
08:50:41,140 --> 08:50:43,500
And I just like, when I look at something I'm vibe coding,

192
08:50:43,640 --> 08:50:44,920
I'm like, dude, I have no idea.

193
08:50:45,280 --> 08:50:47,140
Like, I just don't really get it, you know?

194
08:50:48,400 --> 08:50:51,680
And so what am I doing?

195
08:50:51,840 --> 08:50:52,680
How am I fixing this?

196
08:50:54,360 --> 08:50:56,540
One of the ways I'm doing it, and I'm still quite early on.

197
08:50:56,620 --> 08:50:58,300
I'm not saying that I've figured it all out.

198
08:50:58,520 --> 08:51:00,120
But I think one of the ways I'm doing it

199
08:51:00,120 --> 08:51:01,780
is that I have two parts of my code base.

200
08:51:01,860 --> 08:51:04,080
I've got this engine part of the code base.

201
08:51:04,180 --> 08:51:05,200
And then I have a folder.

202
08:51:05,640 --> 08:51:07,760
I think I call it spec, but I'm thinking I'm just

203
08:51:07,760 --> 08:51:08,920
calling it human or something.

204
08:51:09,040 --> 08:51:11,880
And that folder is the code that I read.

205
08:51:13,520 --> 08:51:15,660
And it's this code that's linted.

206
08:51:15,980 --> 08:51:18,680
And there's all these rules for Claude

207
08:51:18,680 --> 08:51:20,620
about what that code looks like.

208
08:51:20,660 --> 08:51:21,940
It's supposed to feel like English.

209
08:51:22,780 --> 08:51:26,080
It's kind of like a spec, and I'll show you what it looks like.

210
08:51:26,920 --> 08:51:32,800
But it's at the density where if I just read this code, I know what's happening, and I

211
08:51:32,800 --> 08:51:40,320
can edit it, I can tweak it, but more importantly, I just can have a bit of a ground truth of

212
08:51:40,320 --> 08:51:41,960
what my game looks like.

213
08:51:43,620 --> 08:51:46,420
So let's do a quick demo here.

214
08:51:47,620 --> 08:51:51,800
By demo, I guess I mean I'm just going to, whoops, this is my eTalk.

215
08:51:52,120 --> 08:51:53,280
Okay, here we go.

216
08:51:53,380 --> 08:51:55,840
Okay, so this is my game.

217
08:51:56,040 --> 08:51:59,740
I've got in here this folder game, and then I've got here this code.

218
08:51:59,860 --> 08:52:01,400
This is the agent code, right?

219
08:52:01,480 --> 08:52:05,620
This is the code where I'm like, dude, I don't know how cameras work or whatever.

220
08:52:06,300 --> 08:52:08,140
There's a bunch of code here that I don't understand.

221
08:52:08,440 --> 08:52:12,660
But then in this code, I have all my characters, and I have their abilities.

222
08:52:13,060 --> 08:52:19,200
And so I can go in here and I can say, like, okay, this is the rock move, right?

223
08:52:20,240 --> 08:52:23,340
And inside of this move, there's a bunch of ways to tune it.

224
08:52:23,880 --> 08:52:24,940
It's well commented.

225
08:52:25,340 --> 08:52:31,800
It uses things like meters, you know what I mean, instead of, like, x, like, 0.5 or whatever.

226
08:52:34,180 --> 08:52:37,560
It tries to, like, create a bunch of different abstractions along the way.

227
08:52:37,560 --> 08:52:43,020
And so I can see that one of the things that is useful is that there are three ways for

228
08:52:43,660 --> 08:52:45,140
this character to throw a rock.

229
08:52:45,240 --> 08:52:49,440
It can grab it from the ground, it can grab it from the wall or something.

230
08:52:50,900 --> 08:52:57,700
And I can see sort of the true logic of the character.

231
08:52:57,900 --> 08:53:04,900
I don't really know how the engine works or how this is implemented underneath, but I've

232
08:53:04,900 --> 08:53:07,960
essentially created almost a custom scripting language for

233
08:53:07,960 --> 08:53:08,400
my game.

234
08:53:09,320 --> 08:53:13,120
And so that means when I'm playing this game, I can

235
08:53:13,120 --> 08:53:13,900
tweak little things.

236
08:53:14,100 --> 08:53:15,900
I'm still working on it.

237
08:53:16,020 --> 08:53:19,080
So I can say I want the rock to be a lot bigger.

238
08:53:19,080 --> 08:53:23,360
So I can say 1.5, but I'm using an actual IDE.

239
08:53:24,060 --> 08:53:29,700
And so if I go here to my game, I can see that I have a

240
08:53:29,700 --> 08:53:31,980
rock, it can charge, it gets really big, and I throw it.

241
08:53:31,980 --> 08:53:38,200
right? But I'm like kind of, it's really interesting because I'm also starting to use

242
08:53:38,200 --> 08:53:42,280
cursor more and I'm using the tab model more, you know what I mean? Because now I can be like,

243
08:53:42,560 --> 08:53:49,520
oh, hey, what if I just want to say like, you know, if you summon a wall in the middle,

244
08:53:49,940 --> 08:53:56,720
like cue it up as an action, and then cursor will sort of like, you know, do the tab complete here.

245
08:53:57,820 --> 08:54:02,060
And this is kind of like, it's kind of like spec-driven development, you know, but the

246
08:54:02,060 --> 08:54:05,800
spec is this, like, this scripting language, basically.

247
08:54:07,880 --> 08:54:11,760
And, yeah, it's like, is this the final way I'm going to write this game?

248
08:54:11,800 --> 08:54:12,300
I'm not sure.

249
08:54:12,460 --> 08:54:16,320
It's like, you know, in some ways, when I need to do something complicated, I still

250
08:54:16,320 --> 08:54:18,000
need to ask Claude to do it, right?

251
08:54:18,060 --> 08:54:20,680
And I need to ask Claude to, like, make the primitives.

252
08:54:20,940 --> 08:54:23,580
But it's actually really easy for me to understand what's happening.

253
08:54:23,580 --> 08:54:26,240
And I feel like much more in control.

254
08:54:26,480 --> 08:54:29,640
And while Cloud is working, I can also sort of review it and work with it.

255
08:54:31,180 --> 08:54:32,660
And that's been really exciting.

256
08:54:32,860 --> 08:54:37,720
And so I think the takeaway for me has been that, like,

257
08:54:39,920 --> 08:54:42,180
the takeaway for me has been that, like,

258
08:54:42,500 --> 08:54:44,100
I now have to think about the way,

259
08:54:44,900 --> 08:54:48,900
the best way for me to deliver high-quality software, right?

260
08:54:48,900 --> 08:54:54,800
And I'm not as constrained and limited by the model's capabilities and what it likes

261
08:54:54,800 --> 08:54:56,120
to write code in or something.

262
08:54:56,360 --> 08:55:00,300
I'm just not thinking about, I need to think about how can I deliver great software.

263
08:55:01,060 --> 08:55:07,300
And so I think if we really internalize that, I think then the next year will be about a

264
08:55:07,300 --> 08:55:08,280
renaissance of software.

265
08:55:09,000 --> 08:55:15,120
And I think many people have talked today and this conference about how to be more ambitious.

266
08:55:15,120 --> 08:55:19,780
And I don't think any of this is quite new, but I just think there's many things to do,

267
08:55:19,900 --> 08:55:20,640
like new languages.

268
08:55:23,040 --> 08:55:27,420
Obviously, we want things like Rust that are very efficient, but we also want things to

269
08:55:27,420 --> 08:55:30,940
be potentially modular and hot reloadable.

270
08:55:31,680 --> 08:55:34,200
We want it to be potentially readable.

271
08:55:35,240 --> 08:55:39,100
New primitives, there are probably new types of databases, new types of servers.

272
08:55:40,700 --> 08:55:42,440
New operating system interactions.

273
08:55:42,440 --> 08:55:46,520
I've been sort of like vibe coding a bunch of personal software that sort of messes with

274
08:55:46,520 --> 08:55:47,520
my operating system.

275
08:55:48,460 --> 08:55:50,960
There are new interaction paradigms possible, I think.

276
08:55:51,560 --> 08:55:55,740
A lot of like old HCI research is now like very interesting to revisit.

277
08:55:56,440 --> 08:56:03,040
New aesthetics, I think the way that we like use and design software can change quite a

278
08:56:03,040 --> 08:56:03,320
lot.

279
08:56:04,320 --> 08:56:09,180
And yeah, I think that like there's so much to do here that I think, I don't think that

280
08:56:09,180 --> 08:56:13,360
And this is, every speaker, I think, has talked about this.

281
08:56:13,440 --> 08:56:17,060
And I don't have too much more to add.

282
08:56:17,240 --> 08:56:23,140
But I think that what I really want us to say is that what we need to do is to decide

283
08:56:23,140 --> 08:56:24,800
what the future of work looks like, right?

284
08:56:25,000 --> 08:56:27,240
Decide what our ideal experience is.

285
08:56:27,380 --> 08:56:33,240
Because the models are not this alien intelligence that's trying to force us into a specific

286
08:56:33,240 --> 08:56:33,860
way of working.

287
08:56:34,360 --> 08:56:36,640
They can kind of do everything, and they will.

288
08:56:37,400 --> 08:56:39,860
but our job is to bring about the renaissance.

289
08:56:40,320 --> 08:56:42,080
So, thank you guys.

290
08:56:42,460 --> 08:56:42,880
That's my talk.