1 00:19:42,160 --> 00:19:44,040 Let me just get the clicker real quick. 2 00:19:48,500 --> 00:19:50,460 Sweet, let's see if these slides show up first. 3 00:19:52,080 --> 00:19:52,740 All right. 4 00:19:53,380 --> 00:19:58,360 Hey everyone, I'm Jason from the developer experience team at OpenAI, and I'm super excited 5 00:19:58,360 --> 00:19:59,880 to kick off this track on 6 00:20:00,000 --> 00:20:04,460 raising your ambitions. You know, I joined in March and the Codex app wasn't 7 00:20:04,460 --> 00:20:09,780 really a thing for a couple weeks. There was no Astra yet. And when these things launched, 8 00:20:09,900 --> 00:20:13,140 I kind of realized that I started becoming someone that worked on the developer experience 9 00:20:13,140 --> 00:20:17,960 team and I worked on the everything experience team now that AI can be a tool that allows 10 00:20:17,960 --> 00:20:22,360 everyone to make things. And so I want this talk to not only to be addressable to the people 11 00:20:22,360 --> 00:20:25,600 here in this room as developers, but for everyone online as well. 12 00:20:26,220 --> 00:20:28,580 You know, in this year we've done a lot of different things, right? 13 00:20:28,620 --> 00:20:32,160 We wanted a better coding app, we built the Codex app, we've been contributing to things 14 00:20:32,160 --> 00:20:36,900 like biology, folks have wanted to make personal applications, so we launched sites, cybersecurity, 15 00:20:37,400 --> 00:20:41,600 voice, computer use, and in this past couple of months, we've done some really ambitious 16 00:20:41,600 --> 00:20:46,160 things like our contributions to math and physics and a lot of new physical engineering 17 00:20:46,160 --> 00:20:46,680 capabilities. 18 00:20:47,420 --> 00:20:50,440 It's kind of crazy to think that a couple years ago we were doing grade school math, 19 00:20:50,440 --> 00:20:53,080 and now we'll be able to like tackle Navier Stokes. 20 00:20:54,520 --> 00:20:56,700 So when I got asked to talk a little bit more 21 00:20:56,700 --> 00:20:57,860 about what ambition meant, 22 00:20:58,040 --> 00:21:00,160 I think the question I keep going back to is like, 23 00:21:00,880 --> 00:21:04,500 ambition isn't sort of how you do things, 24 00:21:04,600 --> 00:21:06,020 but like how you approach these things, right? 25 00:21:06,060 --> 00:21:06,740 What do you want to make? 26 00:21:06,940 --> 00:21:08,520 And how ambitious are you in thinking 27 00:21:08,520 --> 00:21:09,440 about the things that you make? 28 00:21:10,660 --> 00:21:12,540 I know this is a coding conference after all, 29 00:21:12,600 --> 00:21:14,740 and many of the talks here will talk about things 30 00:21:14,740 --> 00:21:17,020 like how to run like longer running tasks, 31 00:21:17,020 --> 00:21:20,080 or I think talk about dark factories and multi-agents. 32 00:21:20,500 --> 00:21:22,300 But I think from someone like me, 33 00:21:22,380 --> 00:21:24,000 it might not be as relatable to just say 34 00:21:24,000 --> 00:21:26,760 that I have thousands of agents running in the background. 35 00:21:26,980 --> 00:21:28,280 I have unlimited tokens. 36 00:21:28,520 --> 00:21:30,300 And so really I want to give a talk 37 00:21:30,300 --> 00:21:31,560 that talks a little bit more about 38 00:21:32,060 --> 00:21:35,280 what I think is the role and responsibility 39 00:21:35,280 --> 00:21:37,020 for developers in the future. 40 00:21:38,240 --> 00:21:40,700 And so using more tokens isn't what I mean 41 00:21:40,700 --> 00:21:42,000 by being more ambitious. 42 00:21:42,780 --> 00:21:45,800 To me, in a world where execution is now free, 43 00:21:45,800 --> 00:21:49,000 I think ambition is summed up by these three principles, right? 44 00:21:49,200 --> 00:21:51,340 Do you actually believe that you've created this work 45 00:21:51,340 --> 00:21:53,120 and that you have ownership of this work? 46 00:21:53,720 --> 00:21:56,240 Can you describe what your tastes are 47 00:21:56,240 --> 00:21:58,880 and understand how you can tell the model to make things better? 48 00:21:59,000 --> 00:22:00,500 And can you describe what good looks like? 49 00:22:00,980 --> 00:22:03,200 And then lastly, are you actually curious enough 50 00:22:03,200 --> 00:22:05,980 to use AI to pursue other fields, right? 51 00:22:06,000 --> 00:22:07,500 Are you actually able to expand your horizon 52 00:22:07,500 --> 00:22:09,120 rather than thinking that you're going to be replaced 53 00:22:09,120 --> 00:22:09,800 or something like that? 54 00:22:10,940 --> 00:22:14,300 And this mostly comes from three different fears 55 00:22:14,300 --> 00:22:17,320 that people have when it comes to like using AI more. 56 00:22:17,960 --> 00:22:19,900 Are we just gonna create more slop in the world? 57 00:22:20,480 --> 00:22:23,260 You know, is AI gonna destroy my ability to think critically? 58 00:22:23,540 --> 00:22:25,000 And is AI gonna replace us? 59 00:22:25,780 --> 00:22:27,800 And so I wanna talk about these things one at a time, 60 00:22:27,940 --> 00:22:30,760 and use that as a way of framing this conversation about ambition. 61 00:22:31,460 --> 00:22:33,740 And so the first thing is you really wanna think about 62 00:22:33,740 --> 00:22:35,280 owning the work that you ship. 63 00:22:36,420 --> 00:22:38,460 So the first question is like, are we gonna ship more slop? 64 00:22:38,840 --> 00:22:41,640 Right? And probably the answer is yes, 65 00:22:41,820 --> 00:22:43,660 but I try to think about this the same way that we thought about 66 00:22:43,660 --> 00:22:45,360 the camera when it first came out. 67 00:22:46,100 --> 00:22:47,920 As a raise of, like, show of hands, 68 00:22:48,020 --> 00:22:50,520 like, how many people here feel like the technical work 69 00:22:50,520 --> 00:22:51,960 they do is a creative pursuit? 70 00:22:53,660 --> 00:22:55,820 Yeah. You know, me too. 71 00:22:57,220 --> 00:22:59,060 These days, I don't think of myself as a developer. 72 00:22:59,480 --> 00:23:01,280 I mostly am thinking about myself as someone 73 00:23:01,280 --> 00:23:03,780 who's trying to will things into existence, right? 74 00:23:03,860 --> 00:23:05,300 And I have to, again, feel this ownership 75 00:23:05,300 --> 00:23:06,340 and describe what I like. 76 00:23:07,380 --> 00:23:09,520 In a past life, not many people know this, 77 00:23:09,740 --> 00:23:11,140 I used to be a machine learning engineer, 78 00:23:11,140 --> 00:23:13,200 I used to be a physicist, and I used to be an artist. 79 00:23:13,860 --> 00:23:16,600 And the story of the camera is something I keep thinking about 80 00:23:16,600 --> 00:23:19,420 as we introduce more ways of doing this kind of replication, 81 00:23:19,600 --> 00:23:21,040 as this way of creating things. 82 00:23:22,200 --> 00:23:23,840 When the camera first came out, 83 00:23:24,060 --> 00:23:27,920 painters who were busy focusing on their ability to recreate images, 84 00:23:28,080 --> 00:23:29,040 yeah, they were a little stressed out. 85 00:23:29,620 --> 00:23:31,820 But at that time, people weren't even thinking about 86 00:23:31,820 --> 00:23:34,520 whether or not photography could change the art world. 87 00:23:34,620 --> 00:23:36,200 We weren't thinking about what a movie meant. 88 00:23:36,280 --> 00:23:38,040 You see a Charlie Chaplin film when it comes out. 89 00:23:38,040 --> 00:23:41,320 It's really easy to dismiss that as some, you know, silly thing. 90 00:23:42,520 --> 00:23:48,800 But at some point you realize that not only is photography art, but it has changed the 91 00:23:48,800 --> 00:23:49,360 art world, right? 92 00:23:49,460 --> 00:23:53,400 And now we have film and all these other mediums and there's amazing things that you can explore. 93 00:23:54,240 --> 00:23:58,140 Walter Benjamin says this in a book, The Work of Art in the Age of Mechanical Reproduction, 94 00:23:58,180 --> 00:24:00,540 a book that came out right when the camera became popular. 95 00:24:00,660 --> 00:24:04,720 He says, photography freed the hand of the most important functions and brought them 96 00:24:05,120 --> 00:24:07,180 to the eyes and to the voice. 97 00:24:08,280 --> 00:24:10,280 This is kind of the same with AI now, 98 00:24:10,300 --> 00:24:12,500 so I just get to use voice mode on chat 99 00:24:12,500 --> 00:24:14,480 and just say, you know, do this and do that, 100 00:24:14,540 --> 00:24:16,140 in the same way that at some point, 101 00:24:16,620 --> 00:24:18,020 if you're a photographer or director, 102 00:24:18,040 --> 00:24:20,120 you can just say, you know, move that camera over there. 103 00:24:20,160 --> 00:24:21,120 I want a wider shot. 104 00:24:21,280 --> 00:24:22,760 I want to capture some kind of feeling, 105 00:24:22,880 --> 00:24:24,380 and maybe have a team around me 106 00:24:24,380 --> 00:24:25,520 that helps me capture that vision. 107 00:24:28,040 --> 00:24:30,420 And now we live in a world where everyone has a camera. 108 00:24:30,560 --> 00:24:33,760 And sure, maybe I have a thousand photos in my background, 109 00:24:33,900 --> 00:24:35,740 in my camera roll that I never look at. 110 00:24:35,740 --> 00:24:36,720 maybe that is a slot. 111 00:24:37,580 --> 00:24:39,380 But it only matters whether or not I have captured 112 00:24:39,380 --> 00:24:41,360 a couple of really important moments in my life. 113 00:24:41,940 --> 00:24:44,480 Right? And having ownership over those moments 114 00:24:44,480 --> 00:24:45,240 is the thing that's important. 115 00:24:45,360 --> 00:24:46,860 Like for example, like 30 years ago, like, 116 00:24:46,980 --> 00:24:48,500 I didn't really have childhood photos, 117 00:24:48,620 --> 00:24:50,380 because we just grew up in a place where we couldn't afford a camera. 118 00:24:50,880 --> 00:24:53,000 Right? And now technology has allowed everyone 119 00:24:53,000 --> 00:24:54,400 the ability to have that kind of access. 120 00:24:54,560 --> 00:24:56,260 And I sure would have liked to have a camera back then, 121 00:24:56,340 --> 00:24:58,200 even if it meant a little bit more slot. 122 00:24:59,860 --> 00:25:02,220 Right? And so if you go back to this idea of this painter, 123 00:25:02,620 --> 00:25:04,100 I think the question you have to ask yourself is, 124 00:25:04,100 --> 00:25:06,280 do you identify yourself as the painter in this example 125 00:25:06,280 --> 00:25:06,920 or as the artist? 126 00:25:07,700 --> 00:25:08,140 Right? 127 00:25:08,560 --> 00:25:10,840 Maybe it means that painting photo realistically 128 00:25:10,840 --> 00:25:12,360 isn't the skill that's really valuable, 129 00:25:12,620 --> 00:25:14,840 but if you think about yourself as the artist, 130 00:25:15,700 --> 00:25:17,720 right, whether it is the paintbrush or the camera, 131 00:25:17,900 --> 00:25:19,960 whether it is code or clay or anything else, 132 00:25:20,260 --> 00:25:22,700 these are not just mediums that allow you 133 00:25:22,700 --> 00:25:24,020 to express yourself and to make things. 134 00:25:25,320 --> 00:25:25,760 Right? 135 00:25:26,040 --> 00:25:27,740 I don't make a table and I say, like, 136 00:25:27,800 --> 00:25:29,020 DeWalt made this table. 137 00:25:29,400 --> 00:25:29,840 Right? 138 00:25:29,960 --> 00:25:32,500 Just because a director has thousands of people on set, 139 00:25:32,520 --> 00:25:33,560 it's still the director's movie, 140 00:25:33,560 --> 00:25:37,940 and even in open source, we might have thousands of contributors or even agents now running the system, 141 00:25:38,340 --> 00:25:42,900 but it's the core maintainers that help hold the responsibility for the quality of the work. 142 00:25:44,320 --> 00:25:48,240 And I don't think the danger these days is whether or not AI is going to create more slop, 143 00:25:49,120 --> 00:25:50,800 but instead it's about self-erasure. 144 00:25:52,060 --> 00:25:56,820 Are you going to be a passive observer of the work that's being created, which results in the slop, 145 00:25:56,920 --> 00:25:59,620 or are you going to be someone who is an active participant in the act of creation? 146 00:26:00,820 --> 00:26:04,240 I think many people have been sort of moving themselves away from this. 147 00:26:04,840 --> 00:26:08,820 And the only difference between a person who's making 1,000 PRs versus another is whether 148 00:26:08,820 --> 00:26:11,860 or not they actually care about the work and whether they are an active participant. 149 00:26:13,560 --> 00:26:17,640 And so as an engineer, what this means is you just have to use your products more, notice 150 00:26:17,640 --> 00:26:20,440 when things break, and decide whether things need another pass. 151 00:26:21,460 --> 00:26:25,160 A lot of the great examples you see online are not one-shot examples. 152 00:26:25,660 --> 00:26:29,180 These are things that you iterate over time because you feel proud to create this kind 153 00:26:29,180 --> 00:26:29,560 of work. 154 00:26:30,220 --> 00:26:34,480 You know, internally we have a lot of amazing work with, like, our ownership over, like, 155 00:26:34,540 --> 00:26:35,300 the Codex app, right? 156 00:26:36,060 --> 00:26:39,580 Most people here have an incredible sense of ownership over the product, and as things 157 00:26:39,580 --> 00:26:42,160 slip as we ship more, we want to bring them together. 158 00:26:42,560 --> 00:26:44,720 And again, this is mostly because of the fact that we care. 159 00:26:46,780 --> 00:26:52,780 So in terms of worrying about slop, I think that the medicine for slop is really just 160 00:26:52,780 --> 00:26:54,960 the fact that you really need to believe that you own this work. 161 00:26:56,840 --> 00:27:02,520 I think the second thing to address is whether or not you feel like you can actually describe 162 00:27:02,520 --> 00:27:05,620 what good looks like. And for that, you need to go out and be inspired. 163 00:27:07,780 --> 00:27:11,460 So the second question people are asking now is, like, do you think AI is going to destroy 164 00:27:11,460 --> 00:27:15,560 our ability to think critically, right? If I own the work, how do I know if something's 165 00:27:15,560 --> 00:27:20,360 good enough? If I just hand you a camera in this analogy, like, does it mean that it takes 166 00:27:20,360 --> 00:27:24,580 away all the effort of creation? I don't think the answer is yes. Right? 167 00:27:24,580 --> 00:27:27,000 Instead, it's now about having language 168 00:27:27,000 --> 00:27:28,200 to describe what's going on. 169 00:27:28,320 --> 00:27:29,280 So when I first started art school, 170 00:27:29,780 --> 00:27:31,500 maybe I don't know how to develop film, 171 00:27:31,620 --> 00:27:32,520 I just use my iPhone, 172 00:27:32,860 --> 00:27:35,700 but I still need to understand the elements of design 173 00:27:35,700 --> 00:27:37,240 and what does beautiful look like. 174 00:27:38,020 --> 00:27:39,520 Space, value, shape, texture. 175 00:27:40,480 --> 00:27:42,120 I would just spend a lot of time thinking, 176 00:27:42,240 --> 00:27:43,320 you know what, I don't think I understand 177 00:27:43,320 --> 00:27:44,280 what negative space means. 178 00:27:44,360 --> 00:27:46,840 Let me go do a lot of sketches and explore a little bit. 179 00:27:47,260 --> 00:27:49,420 It's the same thing with development now. 180 00:27:49,540 --> 00:27:51,500 I can make multiple apps and I have to think about 181 00:27:51,500 --> 00:27:53,060 why I like something and why I don't. 182 00:27:54,140 --> 00:27:56,180 And most of my time in art school was just thinking, 183 00:27:56,340 --> 00:27:58,880 man, like, I can make things and I can look at them 184 00:27:58,880 --> 00:28:00,800 and I just say, I don't like how this looks. 185 00:28:01,360 --> 00:28:03,760 And what I was noticing was that my taste 186 00:28:03,760 --> 00:28:05,440 and my judgment exceeded my skill. 187 00:28:05,600 --> 00:28:07,180 I could not do the thing I liked. 188 00:28:07,700 --> 00:28:09,520 And I can't just look in the mirror and tell myself, 189 00:28:09,600 --> 00:28:11,020 like, make it pop a hundred times 190 00:28:11,020 --> 00:28:12,400 and know that it's gonna be better. 191 00:28:12,580 --> 00:28:14,440 I have to go out and develop that language. 192 00:28:15,360 --> 00:28:17,000 People say, like, taste is the moat, 193 00:28:17,100 --> 00:28:19,120 but I think I generally hate that line 194 00:28:19,120 --> 00:28:21,460 because it doesn't tell you what taste really means. 195 00:28:22,320 --> 00:28:24,780 To me, taste is the ability to describe what you like 196 00:28:24,780 --> 00:28:26,360 and predict what others will like. 197 00:28:26,780 --> 00:28:29,920 And building this model of the world and of other people, 198 00:28:30,360 --> 00:28:31,380 that's the new thinking. 199 00:28:33,860 --> 00:28:36,560 Winstein's, the limits of my language means the limits of my world. 200 00:28:36,660 --> 00:28:40,440 And I think that's very much the case on pursuing ambitious projects. 201 00:28:41,620 --> 00:28:44,000 You know, there's a funny example of just like the use of language. 202 00:28:44,140 --> 00:28:46,440 I don't know if Steve was in the room, but he made a tweet a couple days ago 203 00:28:46,440 --> 00:28:49,680 that just says, find all the implicit state machines in my app. 204 00:28:49,680 --> 00:28:51,760 I had to go Google what that meant, right? 205 00:28:52,000 --> 00:28:54,720 And all of a sudden, all of the comments are just how, like, everyone's apps are, like, 206 00:28:54,820 --> 00:28:55,620 20% faster. 207 00:28:56,020 --> 00:28:59,780 These are now, like, really simple ways of just developing some language to understand 208 00:28:59,780 --> 00:29:01,880 what good looks like and how to make things better. 209 00:29:03,060 --> 00:29:06,580 If you think of, like, model progress, in machine learning, there's a really helpful 210 00:29:06,580 --> 00:29:10,260 concept called the generator-discriminator gap, right? 211 00:29:10,460 --> 00:29:11,420 You'll have two models. 212 00:29:11,580 --> 00:29:14,360 One grades the model, and then one actually creates the outputs. 213 00:29:14,780 --> 00:29:18,200 And what happens is, is if the grader is too good, then the model can't learn. 214 00:29:18,200 --> 00:29:19,640 And if the generator's too good, 215 00:29:19,760 --> 00:29:21,300 the discriminator can't help the model improve. 216 00:29:22,540 --> 00:29:24,560 We are now that discriminator, right? 217 00:29:24,840 --> 00:29:27,220 As model capabilities go up, at some point, 218 00:29:27,360 --> 00:29:30,000 maybe the model's output hits your own taste levels. 219 00:29:30,460 --> 00:29:32,320 Then it's really hard to actually improve. 220 00:29:32,600 --> 00:29:33,960 And then to make things that are better, 221 00:29:34,100 --> 00:29:36,300 you have to go out, develop that language, 222 00:29:36,380 --> 00:29:37,420 and know what good looks like. 223 00:29:37,880 --> 00:29:39,580 And the failure mode is here, 224 00:29:39,780 --> 00:29:41,920 just not being able to discern 225 00:29:42,620 --> 00:29:44,080 whether or not a model can improve. 226 00:29:44,940 --> 00:29:45,340 Right? 227 00:29:45,820 --> 00:29:47,620 And again, what this means is we have to go out 228 00:29:47,620 --> 00:29:48,940 and learn again. 229 00:29:49,100 --> 00:29:49,700 You have to think again. 230 00:29:50,800 --> 00:29:52,200 This is an example of a bunch of things 231 00:29:52,200 --> 00:29:54,600 I spent my past year learning, like in just building software. 232 00:29:55,020 --> 00:29:56,760 Or like, as I build these applications, 233 00:29:56,840 --> 00:29:58,800 I realize, oh man, these websites look really ugly. 234 00:29:58,920 --> 00:30:00,000 Like, I need to learn about typography. 235 00:30:00,340 --> 00:30:02,160 Like, why is everything italic? Like, I don't understand. 236 00:30:03,000 --> 00:30:06,880 And I spent a lot of time learning about typography. Then it was about, like, how the information 237 00:30:06,880 --> 00:30:11,280 was presented to avoid looking like slop. And then it was about, like, motion design, and 238 00:30:11,280 --> 00:30:14,640 how can I make things move in a beautiful way, and I got to spend so much time learning about 239 00:30:15,700 --> 00:30:19,080 animation, and then learning about, like, interaction patterns, and what makes an app feel beautiful. 240 00:30:19,780 --> 00:30:24,000 And this is really fun, right? And this is, again, spending more time learning in order to 241 00:30:24,000 --> 00:30:24,880 make great things. 242 00:30:26,220 --> 00:30:30,300 And most importantly, having the ability to describe things is what allows you to differentiate, 243 00:30:30,620 --> 00:30:30,780 right? 244 00:30:31,960 --> 00:30:35,500 If you think about these purple apps, the reason these existed was because the people making 245 00:30:35,500 --> 00:30:38,060 these applications, like, didn't have the site. 246 00:30:38,200 --> 00:30:41,720 They couldn't actually describe what was wrong, and they thought this was okay, right? 247 00:30:41,940 --> 00:30:44,760 And this limiting in language is sort of the problem that we face. 248 00:30:46,580 --> 00:30:49,580 And then so the question now is, okay, how do we develop this new language? 249 00:30:49,580 --> 00:30:55,360 And honestly, the line that I've sort of been like punished to now is just like, if you want good taste, you have to eat. 250 00:30:55,620 --> 00:30:57,900 You have to go out and explore things, right? 251 00:30:58,000 --> 00:30:59,620 People are like vibe coding video games. 252 00:30:59,820 --> 00:31:09,780 Maybe the answer isn't just to like tell the model to do better, but like play some video games and understand what feels good, what animations feel good, what stories matter, what stories do I want to tell, right? 253 00:31:09,800 --> 00:31:13,640 If you're building tools, try other people's tools and applications and understand what you like. 254 00:31:14,000 --> 00:31:15,120 Look at beautiful things. 255 00:31:15,120 --> 00:31:17,960 I've been doing a lot of like 3D printing with AI, 256 00:31:18,460 --> 00:31:20,800 and everything just looks like rectangles glued together. 257 00:31:21,020 --> 00:31:22,700 And I just kept saying like, make it more beautiful. 258 00:31:22,940 --> 00:31:24,500 And that's a really frustrating experience. 259 00:31:25,440 --> 00:31:27,140 And then I started looking at like sports cars, 260 00:31:27,300 --> 00:31:28,140 like industrial design. 261 00:31:28,280 --> 00:31:29,880 I started going to stores and looking at things 262 00:31:29,880 --> 00:31:31,620 and going like, let me describe and understand 263 00:31:31,620 --> 00:31:33,560 what like good curves look like. 264 00:31:33,840 --> 00:31:35,680 And again, it's like a really fun experience. 265 00:31:35,860 --> 00:31:38,080 And now more of my designs are like incredibly beautiful, 266 00:31:38,500 --> 00:31:40,160 for what I think is beautiful, right? 267 00:31:40,360 --> 00:31:41,640 I've been helping with like marketing, 268 00:31:41,840 --> 00:31:43,700 and I realized that like I tell the marketing team 269 00:31:43,700 --> 00:31:45,400 that they should use the Codex app more, 270 00:31:45,500 --> 00:31:46,620 but I'm not watching more ads. 271 00:31:47,220 --> 00:31:48,700 And as I'm watching more movies, 272 00:31:48,740 --> 00:31:50,100 I realize there's great stories to tell, 273 00:31:50,220 --> 00:31:51,480 and everything becomes better again. 274 00:31:51,980 --> 00:31:54,840 And this is kind of the really exciting part 275 00:31:54,840 --> 00:31:55,920 of using these tools. 276 00:31:57,440 --> 00:31:59,480 And thinking about why I like things 277 00:31:59,480 --> 00:32:00,520 and why things are amazing, 278 00:32:00,640 --> 00:32:02,720 why things feel good now is how I spend my time. 279 00:32:03,140 --> 00:32:05,480 So I'm not thinking less, I'm thinking more, 280 00:32:05,640 --> 00:32:07,180 and it's really enriching my world 281 00:32:07,180 --> 00:32:08,920 and making me see things differently. 282 00:32:09,680 --> 00:32:10,800 And if you're running a team, 283 00:32:10,920 --> 00:32:12,620 I think the one takeaway here is that 284 00:32:12,620 --> 00:32:16,140 Like, if you really believe that AI is making you more productive, 285 00:32:16,780 --> 00:32:19,100 I think you owe it to yourself and the rest of the team 286 00:32:19,100 --> 00:32:21,240 to make space to go out and explore things 287 00:32:21,240 --> 00:32:22,180 and try new applications 288 00:32:22,180 --> 00:32:26,420 and just actually go out and, like, experience the world, right? 289 00:32:26,660 --> 00:32:28,680 And if you are someone that makes really great work, 290 00:32:28,960 --> 00:32:30,480 inspiring others as well, right? 291 00:32:30,660 --> 00:32:32,500 And that's how you can come back with a point of view. 292 00:32:34,020 --> 00:32:35,560 So go out and be inspired. 293 00:32:35,840 --> 00:32:36,880 Learn to be a better critic 294 00:32:36,880 --> 00:32:39,340 and recognize that you're not actually thinking less, 295 00:32:39,440 --> 00:32:39,940 you're thinking more. 296 00:32:41,620 --> 00:32:43,820 And the last thing is actually just being curious, 297 00:32:43,980 --> 00:32:46,340 really, really curious to pursue new ideas. 298 00:32:46,880 --> 00:32:48,540 Like one question I want you guys to think about 299 00:32:48,540 --> 00:32:50,980 as you are thinking about your role as software engineers 300 00:32:50,980 --> 00:32:52,940 is like what were the things you gave up 301 00:32:52,940 --> 00:32:56,660 or put aside to pursue software, right? 302 00:32:56,820 --> 00:32:59,000 Where else could use your judgment and your experience 303 00:32:59,000 --> 00:33:00,320 you've developed as a software engineer? 304 00:33:00,740 --> 00:33:02,700 Because I think people worry about being replaced, 305 00:33:02,920 --> 00:33:05,360 but really I think people are needed in way more places. 306 00:33:06,500 --> 00:33:08,800 If you think about when the, you know, 307 00:33:08,800 --> 00:33:13,840 invention of like the pencil or sculpting became more easy, all these different industries 308 00:33:13,840 --> 00:33:19,620 in the creative world became really very much multidisciplinary, right? 309 00:33:19,800 --> 00:33:23,560 It completely made sense that if you were a furniture designer, you could be an architect 310 00:33:23,560 --> 00:33:28,140 and you could like make a movie and design kids' toys and playgrounds, right? 311 00:33:28,200 --> 00:33:31,960 Like Michelangelo is a sculptor and a painter and an architect and a poet, you know? 312 00:33:32,500 --> 00:33:35,220 It completely made sense that you could do all these things, right? 313 00:33:35,220 --> 00:33:36,060 And this is a creative world. 314 00:33:36,180 --> 00:33:38,280 Like you invent the pencil, you can do a lot of really amazing things. 315 00:33:38,960 --> 00:33:42,020 But with technology, it's become even more complicated to do so. 316 00:33:42,260 --> 00:33:44,760 I just can't get, like, four PhDs. 317 00:33:44,840 --> 00:33:46,700 And if I could, like, maybe I'll just be, like, 50. 318 00:33:47,000 --> 00:33:47,480 Right? 319 00:33:47,720 --> 00:33:50,420 And so being technical has become incredibly difficult. 320 00:33:51,780 --> 00:33:55,420 But when you do cross-pollinate these different industries, amazing things happen. 321 00:33:55,680 --> 00:33:55,920 Right? 322 00:33:55,920 --> 00:34:00,640 The physicist was, sorry, the founder of the molecular biology was a physicist. 323 00:34:00,960 --> 00:34:01,080 Right? 324 00:34:01,080 --> 00:34:04,200 A mathematician started leading the human genome project. 325 00:34:04,480 --> 00:34:04,820 Right? 326 00:34:04,820 --> 00:34:09,080 And the mathematics around magnets is what is used to help understand neural networks. 327 00:34:09,600 --> 00:34:13,300 But this is now, like, super hard to do, because you need so much information. 328 00:34:14,020 --> 00:34:14,420 Right? 329 00:34:14,620 --> 00:34:18,580 Even in our own world, like, we have a bunch of AI helping us design, like, chips and writing software. 330 00:34:18,780 --> 00:34:21,540 And these, again, are just very multidisciplinary pursuits. 331 00:34:22,940 --> 00:34:26,440 If you look at what it means to be ambitious outside in the real world, 332 00:34:26,600 --> 00:34:33,560 there's a machine learning consultant using AI and molecular biology to, like, help treat their dog's cancer. 333 00:34:33,560 --> 00:34:39,040 There was this public servant in Hokkaido who used, like, chat and, like, computer vision 334 00:34:39,040 --> 00:34:42,640 models to build, like, robots to automate their farm, right? 335 00:34:42,860 --> 00:34:48,660 This cross-pollination and this will to create these things and learn and cross-pollinate 336 00:34:48,660 --> 00:34:51,340 across these different paths is what's really inspiring, right? 337 00:34:51,780 --> 00:34:56,720 I think earlier today I saw folks who are using AI to, like, make more robots and also, 338 00:34:56,740 --> 00:34:58,820 like, discover exoplanets, right? 339 00:34:58,880 --> 00:35:02,700 And it's that curiosity and that enablement of what AI allows you to do in lowering that 340 00:35:02,700 --> 00:35:07,120 that barrier is what allows you to be ambitious in those fields, right? 341 00:35:07,340 --> 00:35:10,880 And so when it comes to thinking about whether or not you're gonna be replaced, a lot of 342 00:35:10,880 --> 00:35:15,640 it really is the fact that you now have the skills that everyone needs in every field. 343 00:35:15,740 --> 00:35:19,860 It's not just about like writing more apps, but actually thinking about, you know, what 344 00:35:19,860 --> 00:35:23,240 were the things I didn't pursue in college to pursue software engineering? 345 00:35:23,380 --> 00:35:24,740 How can I go back to those worlds? 346 00:35:24,900 --> 00:35:27,380 And how can AI make those things easier, right? 347 00:35:27,580 --> 00:35:32,060 You can, again, develop new language and understand what makes things amazing and then go pursue 348 00:35:32,060 --> 00:35:32,580 do those things. 349 00:35:33,520 --> 00:35:38,660 And the great joy of this, and the reason why these multidisciplinary artists are so successful, 350 00:35:39,040 --> 00:35:44,660 is as they pursue new mediums, the act of being a kid again means that you have the 351 00:35:44,660 --> 00:35:47,320 naivety to ask for what might be impossible. 352 00:35:48,460 --> 00:35:52,480 The reason this cross-pollination works is if you have the judgment you've built in software 353 00:35:52,480 --> 00:35:56,180 and you pursue another field, it means that you can ask a question that people might not 354 00:35:56,180 --> 00:35:56,880 think is possible. 355 00:35:57,060 --> 00:35:59,220 And you have the courage to use AI to help you get there. 356 00:35:59,780 --> 00:36:03,220 And so there's great benefits in being able to feel new again in the field 357 00:36:03,220 --> 00:36:05,740 while having all this experience in AI as well. 358 00:36:07,020 --> 00:36:09,280 In my personal life, I've been learning the drums, 359 00:36:09,420 --> 00:36:11,060 I've been building custom drumming software, 360 00:36:11,600 --> 00:36:13,600 AI is helping me learn how to use Ableton, 361 00:36:13,740 --> 00:36:16,300 I'm connecting my drums to MIDI and learning there. 362 00:36:16,800 --> 00:36:19,620 I've been using computer use to teach me new software 363 00:36:19,620 --> 00:36:21,320 that I never really had time to learn, 364 00:36:21,480 --> 00:36:25,960 and I've even been using AI to help me design physical engineering. 365 00:36:25,960 --> 00:36:30,720 So I'm 3D printing parts, building chips, and again, these are things I was always curious 366 00:36:30,720 --> 00:36:32,060 about but never had the time. 367 00:36:32,500 --> 00:36:37,160 And again, part of having AI is not only getting my time back, everything else is accelerated. 368 00:36:37,620 --> 00:36:38,020 Right? 369 00:36:38,080 --> 00:36:42,220 And so being ambitious is about exploring those pursuits, not just spending more tokens 370 00:36:42,220 --> 00:36:43,220 like building an app. 371 00:36:45,480 --> 00:36:47,520 And so this is also what I want you guys to think about, right? 372 00:36:48,320 --> 00:36:49,980 Again, what fields did you put aside? 373 00:36:50,060 --> 00:36:52,440 How can you pursue them now that you have more time with AI? 374 00:36:54,620 --> 00:36:56,860 And so when I'm talking about, like, increasing ambition, 375 00:36:57,020 --> 00:36:58,900 I'm not just saying, like, become a biologist, right? 376 00:36:59,140 --> 00:37:01,840 What I'm really saying is that, like, part of your job now, 377 00:37:01,980 --> 00:37:05,020 to become more ambitious, is to be really inspired. 378 00:37:05,380 --> 00:37:06,920 Pursue these things. Learn that language. 379 00:37:07,160 --> 00:37:09,420 And also for that next generation, to be inspiring. 380 00:37:10,580 --> 00:37:12,920 Right? When people used to ask me how to, like, 381 00:37:12,980 --> 00:37:15,600 learn how to code, it was never a case where I said, 382 00:37:15,640 --> 00:37:17,880 you know what, like, you should go read a textbook on Python. 383 00:37:18,080 --> 00:37:19,680 It was always build something, right? 384 00:37:19,820 --> 00:37:21,380 It was always go build a weather app. 385 00:37:21,380 --> 00:37:25,160 go build a note-taking app, have a real problem and solve it. 386 00:37:25,320 --> 00:37:27,360 And now the scope of these problems are much bigger. 387 00:37:27,580 --> 00:37:30,240 And so as software developers, I think our job now 388 00:37:30,240 --> 00:37:32,520 is to not only build amazing things and share them, 389 00:37:32,600 --> 00:37:35,700 but also inspire the next generation to have more hope, too. 390 00:37:36,040 --> 00:37:37,620 Right? Because maybe software is changing, 391 00:37:37,800 --> 00:37:39,720 but really we're going to be able to contribute 392 00:37:39,720 --> 00:37:41,660 to so much more things. Right? 393 00:37:42,000 --> 00:37:43,520 And like, we see this everywhere. 394 00:37:43,780 --> 00:37:45,560 Like, for example, the Codex app and Sites, 395 00:37:45,660 --> 00:37:47,200 these are all things that were acts of love. 396 00:37:47,620 --> 00:37:49,360 Codex app, the Sites product, 397 00:37:49,360 --> 00:37:51,620 we're all built by people who really cared about these things, 398 00:37:51,720 --> 00:37:54,040 or talking to people online, and working through these things. 399 00:37:54,300 --> 00:37:56,240 Right, and now our PMs can do engineering, 400 00:37:56,360 --> 00:37:57,420 our engineering can do design, 401 00:37:57,740 --> 00:38:00,240 and this cross-pollination is inspiring others to do the same, 402 00:38:00,360 --> 00:38:01,480 and take ownership of that work. 403 00:38:02,460 --> 00:38:05,560 Right, and most of the things here that exist are here 404 00:38:05,560 --> 00:38:07,800 because people are tenacious, and are trying, 405 00:38:08,000 --> 00:38:09,220 and want things to be beautiful. 406 00:38:10,660 --> 00:38:14,440 You know, and ultimately, our mission of OpenAI is to benefit all, 407 00:38:14,760 --> 00:38:17,440 and ultimately what that means isn't just having OpenAI 408 00:38:17,440 --> 00:38:19,640 I have all these AI to control and, you know, 409 00:38:20,180 --> 00:38:21,140 use agent swarms or whatever, 410 00:38:21,460 --> 00:38:23,460 but also making sure that everyone has this technology 411 00:38:23,460 --> 00:38:25,200 so they can contribute in their own ways. 412 00:38:27,360 --> 00:38:28,960 And so that's basically it, you know? 413 00:38:29,460 --> 00:38:30,800 Think about the things that you own. 414 00:38:31,260 --> 00:38:33,720 Go out, get inspired, develop the language you need 415 00:38:33,720 --> 00:38:35,540 to describe what beautiful things look like, 416 00:38:35,680 --> 00:38:38,160 and just make sure you're actually curious enough 417 00:38:38,160 --> 00:38:40,600 and take the time to pursue these new interests. 418 00:38:41,300 --> 00:38:43,280 And then that's how you become more ambitious. 419 00:38:43,640 --> 00:38:45,400 So I'm really excited to see what you guys are building 420 00:38:45,400 --> 00:38:47,960 and come by to the booth to chat. 421 00:38:48,280 --> 00:38:48,980 Thank you.