1 00:00:00,080 --> 00:00:02,637 part of the IIT Delhi community. Let's 2 00:00:02,720 --> 00:00:09,240 have another huge round of applause uh for our guest guys. Come on. 3 00:00:13,520 --> 00:00:15,197 They say the best way to predict the 4 00:00:15,280 --> 00:00:19,197 future is by inventing it. And today we 5 00:00:19,280 --> 00:00:20,957 have two stalwarts from different 6 00:00:21,040 --> 00:00:23,437 generations who have each foreseen the 7 00:00:23,520 --> 00:00:26,237 future for the world by building it 8 00:00:26,320 --> 00:00:28,717 themselves. Our first guest is a name 9 00:00:28,800 --> 00:00:30,637 that resonates deeply within these 10 00:00:30,720 --> 00:00:33,277 walls. He's not just a global icon, but 11 00:00:33,360 --> 00:00:38,397 he's also our very own alumni. In the early8s, he co-founded Sun Micros 12 00:00:38,480 --> 00:00:41,997 Systemystems, a company that fundamentally revamped the structure of 13 00:00:42,080 --> 00:00:46,237 Silicon Valley at a time when computing was closed and proprietary. He 14 00:00:46,320 --> 00:00:50,957 championed open systems. His vision led to workstations that powered the early 15 00:00:51,040 --> 00:00:53,117 internet and the creation of Java. 16 00:00:53,200 --> 00:00:55,597 Later, he founded Kosla Ventures. He has 17 00:00:55,680 --> 00:00:57,677 consistently backed ideas that aim to 18 00:00:57,760 --> 00:01:00,317 solve the society's largest problems 19 00:01:00,400 --> 00:01:02,957 from clean energy to digital tech and as 20 00:01:03,040 --> 00:01:07,677 the first institutional investor in open AI. Ladies and gentlemen, let's have 21 00:01:07,760 --> 00:01:09,517 another huge round of applause for Mr. 22 00:01:09,600 --> 00:01:13,080 Vinod Kosla. 23 00:01:15,520 --> 00:01:18,317 With him, we have arguably the most 24 00:01:18,400 --> 00:01:20,397 influential person in technology right 25 00:01:20,480 --> 00:01:22,797 now. Steering the development of the 26 00:01:22,880 --> 00:01:25,917 most powerful AI models in history. As 27 00:01:26,000 --> 00:01:28,637 the CEO of Open AI, he has kicked off 28 00:01:28,720 --> 00:01:33,037 the generative AI era, changing how the world inter interacts with the 29 00:01:33,120 --> 00:01:35,277 information, creativity, and 30 00:01:35,360 --> 00:01:42,200 intelligence itself. A huge round of applause for Mr. Samman 31 00:01:44,400 --> 00:01:47,837 to formally welcome our distinguished guests and set the context for this 32 00:01:47,920 --> 00:01:50,077 historic event. I would now like to call 33 00:01:50,160 --> 00:01:56,200 upon the stage the director of ID Delhi, Professor Rangan Banerjee. 34 00:01:58,080 --> 00:01:59,997 Good morning. 35 00:02:00,080 --> 00:02:03,597 So nice to see so many of you here. 36 00:02:03,680 --> 00:02:07,437 And when it was known that Mr. Sam Alman 37 00:02:07,520 --> 00:02:09,757 and Vinod Kosla are going to come here. 38 00:02:09,840 --> 00:02:12,637 We quickly realized that the Dogra hall 39 00:02:12,720 --> 00:02:16,557 is not going to be enough and all of you 40 00:02:16,640 --> 00:02:19,277 are actually privileged. We could have 41 00:02:19,360 --> 00:02:23,037 two Dogra halls here if we wanted to fit 42 00:02:23,120 --> 00:02:25,437 in all those who had registered from IIT 43 00:02:25,520 --> 00:02:29,517 Delhi itself. Uh we have the minor exams 44 00:02:29,600 --> 00:02:31,677 going on. I don't know how many of you 45 00:02:31,760 --> 00:02:33,677 have skipped the minor exams to come 46 00:02:33,760 --> 00:02:37,117 here. And I was told some people have 47 00:02:37,200 --> 00:02:40,157 finished the 2-hour exam in an hour and 48 00:02:40,240 --> 00:02:43,197 come down here. How many of you are from 49 00:02:43,280 --> 00:02:45,997 outside IIT? Please raise your hands. 50 00:02:46,080 --> 00:02:49,277 Excellent. So we hope that all of you 51 00:02:49,360 --> 00:02:51,997 would also come for our student tech 52 00:02:52,080 --> 00:02:57,480 event on next week. 53 00:02:58,320 --> 00:03:01,917 I'd like to just take this opportunity 54 00:03:02,000 --> 00:03:05,357 to say that as a higher educational 55 00:03:05,440 --> 00:03:09,997 institute IIIT Delhi AI 56 00:03:10,080 --> 00:03:12,877 provides an opportunity but also a 57 00:03:12,960 --> 00:03:17,037 challenge and we need to figure out how 58 00:03:17,120 --> 00:03:20,637 we train the future generation of people 59 00:03:20,720 --> 00:03:23,437 who are going into the workforce. Uh 60 00:03:23,520 --> 00:03:27,037 towards this we created a committee of 61 00:03:27,120 --> 00:03:29,357 faculty members, students and staff to 62 00:03:29,440 --> 00:03:34,877 look at how do we integrate responsible 63 00:03:34,960 --> 00:03:38,717 AI and what is what is ethical use of AI 64 00:03:38,800 --> 00:03:42,477 by faculty by students 65 00:03:42,560 --> 00:03:45,837 and this is an ongoing conversation 66 00:03:45,920 --> 00:03:51,197 we have again through uh we have a 67 00:03:51,280 --> 00:03:55,277 dedicated school of AI. We have a MTech 68 00:03:55,360 --> 00:03:57,437 program in machine intelligence and data 69 00:03:57,520 --> 00:04:01,037 sciences. We have a PhD program. We've 70 00:04:01,120 --> 00:04:04,477 also setting up this year. We'll be 71 00:04:04,560 --> 00:04:07,117 starting the MTech program in machine 72 00:04:07,200 --> 00:04:09,917 intelligence and data sciences in our 73 00:04:10,000 --> 00:04:12,557 Abu Dhabi campus. We did a curriculum 74 00:04:12,640 --> 00:04:14,717 review last year and in that curriculum 75 00:04:14,800 --> 00:04:19,517 review we we ensured that every student 76 00:04:19,600 --> 00:04:22,717 of the institute has 77 00:04:22,800 --> 00:04:27,597 exposure to AI in terms of courses AI and sustainability. This was one of the 78 00:04:27,680 --> 00:04:29,517 features of the curriculum review. I'm 79 00:04:29,600 --> 00:04:32,237 not going to go on about this because I 80 00:04:32,320 --> 00:04:35,117 we want to hear Mr. Sam Alman and Mr. 81 00:04:35,200 --> 00:04:38,237 Vinod Kosla. I just like to say that 82 00:04:38,320 --> 00:04:42,557 just give you one example. Uh we la we 83 00:04:42,640 --> 00:04:45,517 with the ministry of tribal affairs we 84 00:04:45,600 --> 00:04:48,397 developed a tool called Adibani 85 00:04:48,480 --> 00:04:50,637 which are in 86 00:04:50,720 --> 00:04:53,757 four tribal languages Santali, Bhili, 87 00:04:53,840 --> 00:04:56,477 Mundari and others where we are looking 88 00:04:56,560 --> 00:04:59,837 at AI for the community in communities 89 00:04:59,920 --> 00:05:03,677 which currently do not have access to 90 00:05:03,760 --> 00:05:06,237 this kind of content. 91 00:05:06,320 --> 00:05:10,157 We are looking at ways and we we are 92 00:05:10,240 --> 00:05:12,717 delighted that we are entering into a 93 00:05:12,800 --> 00:05:16,797 partnership with open AI. We are 94 00:05:16,880 --> 00:05:19,997 looking at ways where our students and 95 00:05:20,080 --> 00:05:24,317 our faculty will engage with both 96 00:05:24,400 --> 00:05:26,797 companies like Open AI as well as with 97 00:05:26,880 --> 00:05:29,837 sovereign AI models so that we can 98 00:05:29,920 --> 00:05:33,997 co-develop and create responsible AI 99 00:05:34,080 --> 00:05:37,917 which makes a difference in people's lives. 100 00:05:38,000 --> 00:05:43,037 I'd like to say that we are delighted to have Mr. Sam Alman here. She's a person 101 00:05:43,120 --> 00:05:44,877 who's been ahead of the curve in terms 102 00:05:44,960 --> 00:05:47,757 of thinking on AI and generative AI. 103 00:05:47,840 --> 00:05:50,317 We're also delighted to have with us one 104 00:05:50,400 --> 00:05:54,477 of our own Mr. Binot Kla who has always 105 00:05:54,560 --> 00:05:57,597 been ahead of the curve and as a student 106 00:05:57,680 --> 00:06:02,797 here I think 1971 to 1976. 107 00:06:02,880 --> 00:06:07,117 During that time he set up one of the 108 00:06:07,200 --> 00:06:10,637 first computer laboratories 109 00:06:10,720 --> 00:06:15,117 in the IIT system and uh so he was ahead 110 00:06:15,200 --> 00:06:17,437 of the curve at that time. He was also 111 00:06:17,520 --> 00:06:21,197 ahead of the curve when he in 2004 I 112 00:06:21,280 --> 00:06:23,917 think he donated 113 00:06:24,000 --> 00:06:28,237 to our institute for setting up the school of IT which is named after his 114 00:06:28,320 --> 00:06:32,920 parents and 115 00:06:34,080 --> 00:06:36,557 We at IIT Delhi 116 00:06:36,640 --> 00:06:39,997 are now a 65 year old institution 117 00:06:40,080 --> 00:06:43,837 and we are in the process of reimagining 118 00:06:43,920 --> 00:06:46,477 our future. We want to be ahead of the 119 00:06:46,560 --> 00:06:48,557 curve in education. 120 00:06:48,640 --> 00:06:52,077 We are looking at IIT Delhi 2035. 121 00:06:52,160 --> 00:06:56,477 We are looking at changing our infrastructure. We are 122 00:06:56,560 --> 00:06:58,797 going vertical and we would we are 123 00:06:58,880 --> 00:07:01,597 hoping to have Mr. Kosla's support in 124 00:07:01,680 --> 00:07:04,477 form of ideas as we do this 125 00:07:04,560 --> 00:07:09,197 reimagination. So with these few words I'd like to just say that like all of 126 00:07:09,280 --> 00:07:14,317 you I'm looking forward to this conversation. Thank you so much. 127 00:07:14,400 --> 00:07:20,840 Thank you. I'd like to call upon the stage Mr. Vinard Kosla and Mr. San 128 00:07:28,080 --> 00:07:29,677 where should we start? Hopefully 129 00:07:29,760 --> 00:07:33,277 somebody will keep us honest on time. 130 00:07:33,360 --> 00:07:35,997 Sam maybe uh first I'm always excited to 131 00:07:36,080 --> 00:07:39,037 see so many young people. So maybe you 132 00:07:39,120 --> 00:07:41,997 can talk about your observations here. 133 00:07:42,080 --> 00:07:46,717 >> Yeah. Uh this has been a wonderful trip uh to 134 00:07:46,800 --> 00:07:49,277 India for me and I have a lot more exciting stuff today but this is the 135 00:07:49,360 --> 00:07:53,757 thing I've been most looking forward to of the entire trip. uh watching what's 136 00:07:53,840 --> 00:07:55,277 happening with young people around the 137 00:07:55,360 --> 00:07:58,077 world uh especially in India though 138 00:07:58,160 --> 00:08:01,037 building with incredible energy with AI 139 00:08:01,120 --> 00:08:05,117 uh the new startups that are going to happen are already happening the the new 140 00:08:05,200 --> 00:08:06,557 products and services that are going to 141 00:08:06,640 --> 00:08:09,837 get created the just total embrace of AI 142 00:08:09,920 --> 00:08:13,117 and the the excitement and the desire to 143 00:08:13,200 --> 00:08:16,317 bring all of this agency and energy and 144 00:08:16,400 --> 00:08:20,477 you know speed to the future is awesome. 145 00:08:20,560 --> 00:08:24,797 Um I I don't know about you have a note but I think this is the biggest shift 146 00:08:24,880 --> 00:08:29,437 that has happened in the landscape of what a young person can accomplish in my 147 00:08:29,520 --> 00:08:33,197 entire life. >> Uh it's totally amazing to see and 148 00:08:33,280 --> 00:08:35,357 watching what people are doing is just incredible. 149 00:08:35,440 --> 00:08:38,157 >> India has its 100th anniversary of 150 00:08:38,240 --> 00:08:41,197 independence in 2047 151 00:08:41,280 --> 00:08:45,037 and I can't imagine a larger shift by 152 00:08:45,120 --> 00:08:48,637 that day than anybody 153 00:08:48,720 --> 00:08:53,037 would imagine. So imagine 2047 and what 154 00:08:53,120 --> 00:08:57,997 India might be like by then. We haven't prepped our question. We just wanted to 155 00:08:58,080 --> 00:09:01,437 have a conversation. >> I mean it's going to depend what all all 156 00:09:01,520 --> 00:09:04,637 of these people do. Uh you all will 157 00:09:04,720 --> 00:09:07,997 enter adulthood um with super 158 00:09:08,080 --> 00:09:12,317 intelligence and that's going to change things a lot. 159 00:09:12,400 --> 00:09:14,797 But that is an incredible opportunity. 160 00:09:14,880 --> 00:09:16,477 Uh, 161 00:09:16,560 --> 00:09:20,477 I have a one-year-old kid. Uh, and he is 162 00:09:20,560 --> 00:09:23,037 never going to know a world where he was 163 00:09:23,120 --> 00:09:27,757 smarter than a computer ever. He was born just at the right time to never 164 00:09:27,840 --> 00:09:29,597 outrun a computer. Uh, in raw 165 00:09:29,680 --> 00:09:33,197 intelligence. And, you know, for him 166 00:09:33,280 --> 00:09:38,077 it'll be totally normal life. For us, it's a big adaptation. For you all, it's 167 00:09:38,160 --> 00:09:39,917 like a little strange, but you know, 168 00:09:40,000 --> 00:09:42,237 you're excited about it. what people can 169 00:09:42,320 --> 00:09:46,877 do with GPUs now um will be the way that 170 00:09:46,960 --> 00:09:49,997 new value gets created in the world. Uh 171 00:09:50,080 --> 00:09:54,317 so I don't know what that's going to look like. Um the node doesn't know what 172 00:09:54,400 --> 00:09:57,837 that's going to look like. We got a bunch of highle guesses, but we're not 173 00:09:57,920 --> 00:10:02,077 going to be the ones that build that into uh all of these incredible new 174 00:10:02,160 --> 00:10:04,317 companies that people will do with that. 175 00:10:04,400 --> 00:10:08,637 Um examples that we both like are education is clearly going to totally 176 00:10:08,720 --> 00:10:11,997 change. Healthcare is clearly totally going to change. What it means to do 177 00:10:12,080 --> 00:10:13,837 work is going to totally change. 178 00:10:13,920 --> 00:10:16,637 Companies may be a few people and a 179 00:10:16,720 --> 00:10:19,117 giant data center and build incredible 180 00:10:19,200 --> 00:10:23,757 new kinds of value. Um what it means to do physical work will change as robots 181 00:10:23,840 --> 00:10:26,557 can automate more and more manufacturing. Certainly what it means 182 00:10:26,640 --> 00:10:28,637 to discover science is going to change. 183 00:10:28,720 --> 00:10:30,877 The growth rate of economic value in the 184 00:10:30,960 --> 00:10:36,157 world. Uh the increase in quality of life. A quote that someone from OpenAI 185 00:10:36,240 --> 00:10:37,437 said to me recently that has really 186 00:10:37,520 --> 00:10:41,517 stuck with me is we built up a lot of 187 00:10:41,600 --> 00:10:45,997 instincts and institutions, policies, uh 188 00:10:46,080 --> 00:10:50,477 structure to deal with a world of scarcity. But almost none of that 189 00:10:50,560 --> 00:10:52,477 applies well to a world of abundance. 190 00:10:52,560 --> 00:10:57,117 And in addition to building all these new kinds of companies uh and products 191 00:10:57,200 --> 00:11:00,557 and services, society is going to have to get reimagined for this world of 192 00:11:00,640 --> 00:11:04,237 incredible abundance. You've thought about this a lot. Yeah, I've spent a lot 193 00:11:04,320 --> 00:11:08,397 of time. Let's ask some practical questions. Do you think GDP rates on 194 00:11:08,480 --> 00:11:10,477 whatever is projected by economists 195 00:11:10,560 --> 00:11:13,677 today and every country has its 15 and 196 00:11:13,760 --> 00:11:18,877 25 year forecasts. Do you think they more double more than double? What 197 00:11:18,960 --> 00:11:21,757 happens? >> I think GDP is going to turn out to be a 198 00:11:21,840 --> 00:11:25,117 terrible metric um because AI is so 199 00:11:25,200 --> 00:11:29,277 deflationary. What we what we want to measure and I think what GDP is meant to 200 00:11:29,360 --> 00:11:31,437 approximate is sort of quality of life. 201 00:11:31,520 --> 00:11:35,117 Yeah. and quality of life. >> Technically, it's a definition of goods 202 00:11:35,200 --> 00:11:40,637 and services, but it's measured in dollars, which is money will be a whole 203 00:11:40,720 --> 00:11:44,317 different thing. >> Yeah. So, I think a proper metric of 204 00:11:44,400 --> 00:11:48,157 what GDP is trying to approximate will growth rate will more than double for 205 00:11:48,240 --> 00:11:50,477 sure. Um 206 00:11:50,560 --> 00:11:53,677 GDP measured in nominal dollars 207 00:11:53,760 --> 00:11:57,837 that probably gets kind of squirly actually to ask you a weird question. uh 208 00:11:57,920 --> 00:12:00,157 interest rates in five more years 209 00:12:00,240 --> 00:12:02,077 negative 2% or 20%. 210 00:12:02,160 --> 00:12:06,477 >> Um I think by 2035 for sure we will have 211 00:12:06,560 --> 00:12:09,837 a hugely deflationary economy that the 212 00:12:09,920 --> 00:12:12,317 world isn't thinking about. It's 213 00:12:12,400 --> 00:12:15,357 interesting because uh Google just 214 00:12:15,440 --> 00:12:19,757 issued 100-year bonds and I said by 2035 215 00:12:19,840 --> 00:12:22,557 you should own all those bonds because 216 00:12:22,640 --> 00:12:25,757 the world will be so deflationary that 217 00:12:25,840 --> 00:12:28,717 it's hard to imagine anybody pretending 218 00:12:28,800 --> 00:12:32,797 and this is sort of this notion of uh 219 00:12:32,880 --> 00:12:35,997 you know the world doesn't yet know 220 00:12:36,080 --> 00:12:38,157 how large the change is and they're 221 00:12:38,240 --> 00:12:41,997 going on as business as usual. Uh I 222 00:12:42,080 --> 00:12:44,317 think uh one of your engineers wrote a 223 00:12:44,400 --> 00:12:49,277 great manifesto. Um can you talk about that a little bit? 224 00:12:49,360 --> 00:12:51,837 >> Which manifesto? >> Uh people do a few of these 225 00:12:51,920 --> 00:12:53,437 >> forgetting the name. I just mentioned it 226 00:12:53,520 --> 00:12:57,757 earlier in the conversation. Um the uh 227 00:12:57,840 --> 00:13:02,077 essentially that the world is clueless about what's going on and the magnitude 228 00:13:02,160 --> 00:13:05,917 of the change. >> It'll come to me in just a second. I I 229 00:13:06,000 --> 00:13:08,157 think even people who work in AI, we 230 00:13:08,240 --> 00:13:10,317 feel a little bit clueless about the 231 00:13:10,400 --> 00:13:15,837 magnitude of the change. Codeex is an example. Uh by the way, Codex is growing 232 00:13:15,920 --> 00:13:19,037 in India faster than anywhere in the world. Uh and mostly with young people, 233 00:13:19,120 --> 00:13:23,757 but Codex is an example of where everyone's intuitions even though we 234 00:13:23,840 --> 00:13:25,917 knew this intellectually emotionally we 235 00:13:26,000 --> 00:13:28,957 were all hit by it. um we have like lots 236 00:13:29,040 --> 00:13:33,517 of codeax access inside of OpenAI and what it means to be an engineer uh what 237 00:13:33,600 --> 00:13:38,237 it means to be a researcher changed so much so quickly that I that is just 238 00:13:38,320 --> 00:13:39,517 going to happen again and again 239 00:13:39,600 --> 00:13:44,797 throughout the economy. Uh and 240 00:13:44,880 --> 00:13:48,397 if you believe which I do that the the 241 00:13:48,480 --> 00:13:51,197 top level driver of the economy and of 242 00:13:51,280 --> 00:13:55,997 life getting better and better every year is scientific discovery. And then 243 00:13:56,080 --> 00:14:00,157 the second most important thing is the economic engine around that that brings 244 00:14:00,240 --> 00:14:04,157 these new capabilities out into the world. And both of those are going to 245 00:14:04,240 --> 00:14:07,037 get automated at the same time. We're going to automate scientific progress. 246 00:14:07,120 --> 00:14:10,717 We're going to automate the whole economy. And you think about what that 247 00:14:10,800 --> 00:14:14,877 really would mean if these things can happen 10 times faster, 100 times faster 248 00:14:14,960 --> 00:14:17,677 than we've been able to do it on our own. 249 00:14:17,760 --> 00:14:21,437 We were not we did not evolve to have intuitions 250 00:14:21,520 --> 00:14:25,037 about that kind of a growth rate and that kind of compounding. So I don't 251 00:14:25,120 --> 00:14:29,997 know. But if you were back here as a student, what would you do tomorrow? 252 00:14:30,080 --> 00:14:31,837 >> No, I've I've thought about it a lot. 253 00:14:31,920 --> 00:14:35,837 The the way I frame the minimum level of 254 00:14:35,920 --> 00:14:38,157 change to imagine 255 00:14:38,240 --> 00:14:44,237 15 years from now. So imagine 2040. Uh 256 00:14:44,320 --> 00:14:48,397 I have to say I imagine life 257 00:14:48,480 --> 00:14:51,117 the delta being the same between now and 258 00:14:51,200 --> 00:14:55,037 2014 uh 20 240 as it is between today 259 00:14:55,120 --> 00:14:57,357 and 1970. 260 00:14:57,440 --> 00:15:01,197 So look back at technology in 1970. 261 00:15:01,280 --> 00:15:06,157 That's my assumption about Delta and I think that's too conservative a notion 262 00:15:06,240 --> 00:15:08,397 because >> I think that's too conservative. 263 00:15:08,480 --> 00:15:12,637 >> Yeah. So that's the magnitude of change 264 00:15:12,720 --> 00:15:15,517 and and people ask me so what should we 265 00:15:15,600 --> 00:15:18,477 do I say learn how to learn really 266 00:15:18,560 --> 00:15:22,797 really fast like your area of expertise 267 00:15:22,880 --> 00:15:25,597 doesn't matter whether you're electrical 268 00:15:25,680 --> 00:15:27,757 engineer or a mechanical engineer is 269 00:15:27,840 --> 00:15:31,037 largely going to be irrelevant in fact I 270 00:15:31,120 --> 00:15:35,437 scared the director a little bit saying nothing you learn will be relevant even 271 00:15:35,520 --> 00:15:39,757 in 2030 but when when the first year students 272 00:15:39,840 --> 00:15:41,277 graduate 273 00:15:41,360 --> 00:15:44,317 you won't have learned anything uh that 274 00:15:44,400 --> 00:15:46,957 the AI doesn't know better I know Sam if 275 00:15:47,040 --> 00:15:49,597 you agree with me or not 276 00:15:49,680 --> 00:15:52,237 you you won't know any technical field 277 00:15:52,320 --> 00:15:56,157 better uh you may understand better what 278 00:15:56,240 --> 00:15:59,037 people want uh you may understand much 279 00:15:59,120 --> 00:16:01,757 better how to interact with people 280 00:16:01,840 --> 00:16:06,957 motivate people uh how to come up with creative ideas how to sort of adapt 281 00:16:07,040 --> 00:16:11,837 quickly create value. Um but yeah, any any 282 00:16:11,920 --> 00:16:15,677 technical skill that you learn the AI will get better at. Uh it's still worth 283 00:16:15,760 --> 00:16:19,437 learning, you know, like someone asked me recently, do I regret learning to 284 00:16:19,520 --> 00:16:22,317 program? Um because, you know, clearly 285 00:16:22,400 --> 00:16:23,837 the AI is a better programmer than me. 286 00:16:23,920 --> 00:16:27,677 No. The and the answer is uh not at all. 287 00:16:27,760 --> 00:16:31,997 I also don't regret learning math. I never use math as an adult. Uh learning 288 00:16:32,080 --> 00:16:35,437 how to think, learning how to learn quickly. this was all a valuable skill, 289 00:16:35,520 --> 00:16:37,357 but it it was all of the meta- skills 290 00:16:37,440 --> 00:16:41,277 that matter. Um, I mean, the one the one 291 00:16:41,360 --> 00:16:45,437 thing that I think we could say with confidence of a specific skill to learn 292 00:16:45,520 --> 00:16:49,357 is when I was in school, learning to program was the obvious thing. Now, 293 00:16:49,440 --> 00:16:53,437 learning to use these tools is the obvious thing. And the fluency that 294 00:16:53,520 --> 00:16:58,157 people pick up with each new generation, if you sort of are pushing as much as 295 00:16:58,240 --> 00:17:02,077 possible on the edge, that that seems very valuable. If you're a college 296 00:17:02,160 --> 00:17:05,437 student, what do you think matters the most? 297 00:17:05,520 --> 00:17:09,837 What's what will matter when the technical knowledge they have doesn't 298 00:17:09,920 --> 00:17:12,397 matter? And an electrical engineer can 299 00:17:12,480 --> 00:17:14,877 do a structural engineer's job quite 300 00:17:14,960 --> 00:17:17,517 easily. And by the way, we building both 301 00:17:17,600 --> 00:17:20,237 structural engineers, chip designers, 302 00:17:20,320 --> 00:17:23,757 uh, and electrical engineers today. 303 00:17:23,840 --> 00:17:27,837 That'll be used pervasive everywhere. 304 00:17:27,920 --> 00:17:31,117 So, what do you think matters? 305 00:17:31,200 --> 00:17:35,517 >> Well, I I know I just said this, but I I can't emphasize enough how important I 306 00:17:35,600 --> 00:17:37,117 think it is to stay on the forefront of 307 00:17:37,200 --> 00:17:39,917 the tools. I expect that to be a 308 00:17:40,000 --> 00:17:43,437 compounding accelerant. Uh so, building 309 00:17:43,520 --> 00:17:47,757 projects all the time, just figure out what 310 00:17:47,840 --> 00:17:51,357 what's going to be possible and it's going to rapidly increase every couple 311 00:17:51,440 --> 00:17:53,997 of months from now now on that that 312 00:17:54,080 --> 00:17:57,997 really matters. Um the class that I wish 313 00:17:58,080 --> 00:18:01,597 my college had offered um 314 00:18:01,680 --> 00:18:03,517 the the motto of Y cominator is make 315 00:18:03,600 --> 00:18:06,397 something people want and it is I think 316 00:18:06,480 --> 00:18:09,677 for 20 years now been like the best 317 00:18:09,760 --> 00:18:13,437 motto of any investor. Um, and I wish my 318 00:18:13,520 --> 00:18:15,677 school had offered a class on that 319 00:18:15,760 --> 00:18:18,557 because I think that is a learnable 320 00:18:18,640 --> 00:18:22,957 skill figuring that out. And that's going to be ever more important in a 321 00:18:23,040 --> 00:18:26,557 world where in a world where you can sort of make 322 00:18:26,640 --> 00:18:31,117 anything happen. Figuring out how to get good ideas, which like anything else, 323 00:18:31,200 --> 00:18:33,277 deliberate practice seems to improve. 324 00:18:33,360 --> 00:18:36,157 Uh, that seems to matter a lot. 325 00:18:36,240 --> 00:18:39,357 >> Yeah. So my personal view what's most 326 00:18:39,440 --> 00:18:41,997 important to learn is curiosity and 327 00:18:42,080 --> 00:18:45,197 agency. Can you tell me if you agree, 328 00:18:45,280 --> 00:18:48,477 disagree, elaborate on that? 329 00:18:48,560 --> 00:18:51,517 I totally agree. Uh I think agency is 330 00:18:51,600 --> 00:18:54,717 also a learnable skill. Um the world 331 00:18:54,800 --> 00:18:58,957 kind of beats it out of you. So you have to you have to like resist and continue 332 00:18:59,040 --> 00:19:02,237 to believe that you can sort of figure anything and make anything happen. And 333 00:19:02,320 --> 00:19:04,157 the more times you do it, the more 334 00:19:04,240 --> 00:19:07,757 self-belief you develop. Um, many people 335 00:19:07,840 --> 00:19:09,997 learn this late in their careers. It is 336 00:19:10,080 --> 00:19:12,877 super valuable to learn early. Like 337 00:19:12,960 --> 00:19:16,077 other things, you just have to start and 338 00:19:16,160 --> 00:19:17,597 you have to keep doing it and you'll 339 00:19:17,680 --> 00:19:21,037 develop confidence in it. But the the 340 00:19:21,120 --> 00:19:25,277 returns on agency clearly have never been higher. You know, we've talked for 341 00:19:25,360 --> 00:19:28,317 a long time about what one person is going to be able to do with AI. We're 342 00:19:28,400 --> 00:19:30,717 finally seeing it now. um a single 343 00:19:30,800 --> 00:19:32,557 person with drive and an idea and 344 00:19:32,640 --> 00:19:35,357 willpower um can make incredible things 345 00:19:35,440 --> 00:19:38,717 happen. And in some sense, the most 346 00:19:38,800 --> 00:19:43,037 important learning of my entire career has been that if you are right on the 347 00:19:43,120 --> 00:19:46,557 idea, no matter how unpopular it is, in fact, it's often better to be unpopular 348 00:19:46,640 --> 00:19:51,277 because you have no one else you're competing with doing it and you stick 349 00:19:51,360 --> 00:19:55,437 with unwavering conviction over a long time and you just knock down obstacle 350 00:19:55,520 --> 00:19:59,997 after obstacle with no loss of enthusiasm between each setback, uh, you 351 00:20:00,080 --> 00:20:03,997 can accomplish amazing things. This was certainly the story of OpenAI. Um, now 352 00:20:04,080 --> 00:20:05,757 it obviously looks like a great thing, 353 00:20:05,840 --> 00:20:09,517 but when we started it seemed totally 354 00:20:09,600 --> 00:20:14,077 crazy and it was just like a bunch of us were in a room with force of will and 355 00:20:14,160 --> 00:20:17,677 sort of a attitude that we were going to do it no matter what. And then even the 356 00:20:17,760 --> 00:20:20,877 first like four or five years of opening eye, it was sort of it seemed like a 357 00:20:20,960 --> 00:20:22,557 very bad idea. Thank you for believing 358 00:20:22,640 --> 00:20:27,037 in it by the way. Um, and the agency 359 00:20:27,120 --> 00:20:30,477 that we had all developed earlier in our careers, which I wish we had developed 360 00:20:30,560 --> 00:20:33,917 even earlier, was critical to making it happen. 361 00:20:34,000 --> 00:20:37,037 Yeah, I I would say you heard a couple 362 00:20:37,120 --> 00:20:41,277 of quotes. Uh I think uh one said 363 00:20:41,360 --> 00:20:44,157 the the best way to predict the future 364 00:20:44,240 --> 00:20:47,917 is to invent it. I go on to add to it. 365 00:20:48,000 --> 00:20:50,157 That was Alan K who said that and I say 366 00:20:50,240 --> 00:20:53,517 invent the future you want because now 367 00:20:53,600 --> 00:20:56,557 more than ever that's possible. The key 368 00:20:56,640 --> 00:20:59,197 is agency and it's always been available 369 00:20:59,280 --> 00:21:04,077 where you can make the world you want happen. The director said uh I was part 370 00:21:04,160 --> 00:21:05,917 of four students who started the first 371 00:21:06,000 --> 00:21:08,877 computer programming club at IIT Delhi 372 00:21:08,960 --> 00:21:12,797 which was any programming in any IT 373 00:21:12,880 --> 00:21:15,677 had never existed. We just wanted it so 374 00:21:15,760 --> 00:21:18,637 we made it happen. Um I also with 375 00:21:18,720 --> 00:21:21,837 professor gua here in 1970s 376 00:21:21,920 --> 00:21:25,997 started the biomedical engineering program which is the other side which is 377 00:21:26,080 --> 00:21:28,717 I got curious as an electrical engineer 378 00:21:28,800 --> 00:21:31,757 about electrical phenomena in the body 379 00:21:31,840 --> 00:21:34,717 and convinced professor guad to start 380 00:21:34,800 --> 00:21:37,437 the first biomed program anywhere in 381 00:21:37,520 --> 00:21:40,637 India with the allindia institute of 382 00:21:40,720 --> 00:21:42,877 medical sciences. I give that as example 383 00:21:42,960 --> 00:21:47,597 of I was 16 or 17. You can make anything 384 00:21:47,680 --> 00:21:49,757 happen if you want to and that's what 385 00:21:49,840 --> 00:21:52,557 agency is about. And when technology and 386 00:21:52,640 --> 00:21:55,837 tools make almost anything possible 387 00:21:55,920 --> 00:21:58,317 agency is what matters. Curiosity for 388 00:21:58,400 --> 00:21:59,917 learning and always staying at the 389 00:22:00,000 --> 00:22:03,437 leading edge and agency to make the 390 00:22:03,520 --> 00:22:08,157 world happen. There's never been an opportunity to do startups. I'm very 391 00:22:08,240 --> 00:22:11,277 excited by this little company we invested in. They approached me a few 392 00:22:11,360 --> 00:22:14,157 months ago. Um, Emergent launched a 393 00:22:14,240 --> 00:22:16,397 product in July or August. Last week 394 00:22:16,480 --> 00:22:18,397 they called me and said they'd exceeded 395 00:22:18,480 --> 00:22:23,037 100 million ARR rate in 8 months, the 396 00:22:23,120 --> 00:22:25,677 fastest I've seen anybody grow, even 397 00:22:25,760 --> 00:22:28,157 Lovable or Replet or any of those 398 00:22:28,240 --> 00:22:31,517 companies. Um, so that opportunity 399 00:22:31,600 --> 00:22:36,397 exists for all of you. I don't know if you want to do a finishing comment. 400 00:22:36,480 --> 00:22:40,637 >> Uh, I can't really say anything better than that. uh this is like a special 401 00:22:40,720 --> 00:22:42,797 time like I I this is I think the best 402 00:22:42,880 --> 00:22:47,597 time ever to be starting off careers uh 403 00:22:47,680 --> 00:22:49,517 and yeah the opportunity is incredible. 404 00:22:49,600 --> 00:22:53,997 >> Yeah, I'll finish by a last quote I have 405 00:22:54,080 --> 00:22:56,797 uh most people especially in India are 406 00:22:56,880 --> 00:22:59,677 especially averse to risk. There's a 407 00:22:59,760 --> 00:23:02,877 Harvard Business School where the case 408 00:23:02,960 --> 00:23:04,797 that starts with a quote from me saying, 409 00:23:04,880 --> 00:23:08,237 "My willingness to fail allows me to 410 00:23:08,320 --> 00:23:10,877 succeed." So, I really encourage all of 411 00:23:10,960 --> 00:23:13,757 you to go be risktaking. 412 00:23:13,840 --> 00:23:18,157 There's so much opportunity to take advantage of it, take advantage of 413 00:23:18,240 --> 00:23:20,557 tools, curiosity, 414 00:23:20,640 --> 00:23:24,237 um, agency, risk takingaking, and then 415 00:23:24,320 --> 00:23:26,557 persistence because the path is never 416 00:23:26,640 --> 00:23:29,517 simple or or smooth. So you have to 417 00:23:29,600 --> 00:23:34,877 persist through the hard times. Thank you. I think there's we are at the end 418 00:23:34,960 --> 00:23:38,600 of our talk. >> Okay. 419 00:23:38,960 --> 00:23:41,037 >> Uh what a wonderful conversation guys. 420 00:23:41,120 --> 00:23:43,117 Uh we had Miss Anita Williams a month 421 00:23:43,200 --> 00:23:46,317 back and the idea that learning is such 422 00:23:46,400 --> 00:23:50,317 an important thing for for such huge personas for such huge stalwarts. I 423 00:23:50,400 --> 00:23:52,317 think that is one of the most humbling 424 00:23:52,400 --> 00:23:55,757 things we learn here today. Uh now we'd 425 00:23:55,840 --> 00:23:59,917 like to open up the floor uh for students to ask their own questions. For 426 00:24:00,000 --> 00:24:01,997 this I'd like to invite on stage Mr. 427 00:24:02,080 --> 00:24:07,117 Raghav Gupta the head of education for open AI in India. Uh he's currently 428 00:24:07,200 --> 00:24:11,197 leading leading partnerships with academic institutions ensuring that AI 429 00:24:11,280 --> 00:24:16,840 is integrated responsibly across all dimensions. 430 00:24:19,840 --> 00:24:21,997 >> Good morning. It's incredible seeing a 431 00:24:22,080 --> 00:24:25,437 packed hall. Um Sam, I know you have a 432 00:24:25,520 --> 00:24:30,477 packed uh few days in India and there are many senior meetings but uh to me it 433 00:24:30,560 --> 00:24:33,117 seems like this is going to be the most exciting session. Would you would you 434 00:24:33,200 --> 00:24:35,917 agree? I do. And I'd love to ask the 435 00:24:36,000 --> 00:24:37,517 students as well. Are you all excited 436 00:24:37,600 --> 00:24:42,317 for this conversation? Uh I mean come on 437 00:24:42,400 --> 00:24:45,357 now you we can do better. 438 00:24:45,440 --> 00:24:50,200 Are you all excited for this conversation with Sam? 439 00:24:50,880 --> 00:24:55,517 So, so we have some of India's uh bright young minds here and they're studying 440 00:24:55,600 --> 00:24:58,877 engineering, they're studying computer science, I think we have health 441 00:24:58,960 --> 00:25:02,317 sciences, management, design and so on. 442 00:25:02,400 --> 00:25:05,277 Um I think you all sent us like 800 plus 443 00:25:05,360 --> 00:25:07,517 questions. Uh we can't take all 800 but 444 00:25:07,600 --> 00:25:10,397 I've shortlisted the broad themes and 445 00:25:10,480 --> 00:25:13,917 we'll invite you to ask them live. Um 446 00:25:14,000 --> 00:25:18,317 I'll kick us off a little bit. U Sam you wrote this um article in Times of India 447 00:25:18,400 --> 00:25:21,997 earlier this week and the headline kind of captured your message very nicely 448 00:25:22,080 --> 00:25:24,797 where it said that um AI will define 449 00:25:24,880 --> 00:25:28,877 India's future and um India will define 450 00:25:28,960 --> 00:25:32,397 AI's future as well and I know at the AI summit you've seen a lot of what's 451 00:25:32,480 --> 00:25:35,837 happening around uh development in AI would you like to talk a little bit 452 00:25:35,920 --> 00:25:38,557 about your impressions u there 453 00:25:38,640 --> 00:25:40,317 >> India is already our second largest 454 00:25:40,400 --> 00:25:43,997 market there's 100 million chatbt users 455 00:25:44,080 --> 00:25:47,437 a third of them are students fastest growing codex market in the world. India 456 00:25:47,520 --> 00:25:49,597 should be our largest market over time. 457 00:25:49,680 --> 00:25:52,317 More than just usage of products like 458 00:25:52,400 --> 00:25:54,957 Chachi PT that the building energy in 459 00:25:55,040 --> 00:25:57,837 India um seems like it is growing faster 460 00:25:57,920 --> 00:26:02,397 than anywhere else. The desire to really create new kinds of products and 461 00:26:02,480 --> 00:26:07,037 services, brand new companies that are all about AI. Uh this is Yeah, it's just 462 00:26:07,120 --> 00:26:09,277 amazing and so it's been great to be here. 463 00:26:09,360 --> 00:26:11,277 >> Wonderful. Um so we're going to go to 464 00:26:11,360 --> 00:26:13,917 the students next and u we'll open it up 465 00:26:14,000 --> 00:26:16,957 live. I I think we have Mukund who's 466 00:26:17,040 --> 00:26:21,517 going to ask the first question and Mukund is from IIT Delhi. Uh let's get 467 00:26:21,600 --> 00:26:26,477 get started with Mukund then. >> Hi Sam, I'm Mukun from the department of 468 00:26:26,560 --> 00:26:30,557 mathematics IT Delhi. I'm in my final year pursuing a integrated master's 469 00:26:30,640 --> 00:26:34,237 degree. So my question is you have talked about intelligence becoming a 470 00:26:34,320 --> 00:26:38,877 cheap utility but we are seeing a specialized high reasoning models being 471 00:26:38,960 --> 00:26:43,117 gated by premium subscriptions. So how do you reconcile the concept of broad 472 00:26:43,200 --> 00:26:45,597 access uh with the economic viability of 473 00:26:45,680 --> 00:26:48,077 these frontier models? 474 00:26:48,160 --> 00:26:52,077 >> I think the most amazing thing well 475 00:26:52,160 --> 00:26:55,117 there's many amazing things I can't say the most. One of the amazing things 476 00:26:55,200 --> 00:26:57,517 happening is the rate at which AI is 477 00:26:57,600 --> 00:26:59,197 getting cheaper at a given level of 478 00:26:59,280 --> 00:27:03,197 intelligence. Um, we are, if you look at 479 00:27:03,280 --> 00:27:05,197 a hard reasoning problem, what it would 480 00:27:05,280 --> 00:27:06,957 have cost you to solve it using an 481 00:27:07,040 --> 00:27:10,637 OpenAI API um, a year and a half ago 482 00:27:10,720 --> 00:27:14,157 versus now, it's down about a,000x. 483 00:27:14,240 --> 00:27:19,037 A,000x in 18 months is just that doesn't happen very often. Um, I don't think 484 00:27:19,120 --> 00:27:23,197 we'll be able to sustain another thousandx over the next 18 months, but 485 00:27:23,280 --> 00:27:25,517 maybe we will, but there will be a 486 00:27:25,600 --> 00:27:30,477 significant decrease in in cost. So our 487 00:27:30,560 --> 00:27:32,797 goal here is to relentlessly drive the 488 00:27:32,880 --> 00:27:34,397 cost of intelligence down as close to 489 00:27:34,480 --> 00:27:37,997 zero as we can. Um and then make that 490 00:27:38,080 --> 00:27:40,717 available to the world as a very lowcost 491 00:27:40,800 --> 00:27:44,637 asset. Uh and for a bunch of reasons uh 492 00:27:44,720 --> 00:27:48,557 accessibility being one of them. That's that's a critical part of our mission. 493 00:27:48,640 --> 00:27:51,437 Um there are going to be a bunch of 494 00:27:51,520 --> 00:27:53,757 societal disruptions around that. Like 495 00:27:53,840 --> 00:27:58,317 part of society has maybe 496 00:27:58,400 --> 00:28:01,677 We talked earlier about this issue of um 497 00:28:01,760 --> 00:28:05,357 abundance versus scarcity, but society has been used to intelligence being a 498 00:28:05,440 --> 00:28:09,997 relatively scarce resource. Um also the ability to make things happen in the 499 00:28:10,080 --> 00:28:14,077 physical world, energy, robotics, whatever. If all of that changes, I 500 00:28:14,160 --> 00:28:16,557 don't think society yet has a consensus 501 00:28:16,640 --> 00:28:18,717 on how things are going to work and 502 00:28:18,800 --> 00:28:24,157 where the the limits, the natural limits are going to be. But I do think the cost 503 00:28:24,240 --> 00:28:28,637 of intelligence is going to trend down and down and in terms of empowering the 504 00:28:28,720 --> 00:28:30,797 entire world that's very important. 505 00:28:30,880 --> 00:28:34,397 >> Uh we'll go to uh Tjis next who's also 506 00:28:34,480 --> 00:28:38,557 from IIT Delhi and I know the director also spoke a little bit about uh safety 507 00:28:38,640 --> 00:28:41,357 and I think Tjis your question is related to that. So go ahead please. 508 00:28:41,440 --> 00:28:45,277 >> Yeah. Hi Sam, I'm the Raj Mangala. I'm currently a final year student here at 509 00:28:45,360 --> 00:28:47,597 IT Delhi. I'm pursuing my bachelor's in 510 00:28:47,680 --> 00:28:50,077 computational mechanics. Uh my question 511 00:28:50,160 --> 00:28:53,277 for you is how do we keep AI safe 512 00:28:53,360 --> 00:28:57,757 without blocking innovation from things like prompt injection that are happening 513 00:28:57,840 --> 00:28:59,437 recently? 514 00:28:59,520 --> 00:29:02,957 I I think there's three things here. Um 515 00:29:03,040 --> 00:29:05,757 one, we have to figure out how to 516 00:29:05,840 --> 00:29:11,197 technically align AI. Um we have made better progress there than I thought we 517 00:29:11,280 --> 00:29:14,157 might have, but there's still things to do as we get to these systems that are 518 00:29:14,240 --> 00:29:17,037 truly much smarter than we are. How we 519 00:29:17,120 --> 00:29:19,437 supervise those systems 520 00:29:19,520 --> 00:29:21,517 requires new ideas. As we get to a world 521 00:29:21,600 --> 00:29:25,037 where AI is writing so much source code 522 00:29:25,120 --> 00:29:28,717 and doing so much research, we can't review all of it. There's a bunch of 523 00:29:28,800 --> 00:29:30,237 things that used to feel like sci-fi 524 00:29:30,320 --> 00:29:32,957 that are now clear alignment challenges 525 00:29:33,040 --> 00:29:35,997 that we have to that we have to address. 526 00:29:36,080 --> 00:29:38,877 Um, we've been doing work here for a while. Other labs have been doing work 527 00:29:38,960 --> 00:29:41,357 here for a while. On the whole, 528 00:29:41,440 --> 00:29:44,317 alignment has looked easier than I 529 00:29:44,400 --> 00:29:46,637 feared, but we're clearly not done. So, 530 00:29:46,720 --> 00:29:49,517 bucket one is technical safety work 531 00:29:49,600 --> 00:29:52,957 around alignment. Two is there's a whole 532 00:29:53,040 --> 00:29:56,797 new kind of security infrastructure that needs to get built. You mentioned prompt 533 00:29:56,880 --> 00:29:58,157 injection. I think that's a great 534 00:29:58,240 --> 00:30:01,597 example. Open claw is an amazing thing. 535 00:30:01,680 --> 00:30:05,997 Codeex is an amazing thing. I thought that I was going to be a very 536 00:30:06,080 --> 00:30:10,957 disciplined person and not let them run on my computer with root access uh 537 00:30:11,040 --> 00:30:14,157 because I thought that was an insane thing to do and that lasted about two 538 00:30:14,240 --> 00:30:17,917 hours and then I was like gh it's really annoying to approve everything. It's so 539 00:30:18,000 --> 00:30:21,037 convenient. I want to just I'm going to just like let it rip. And if I'm willing 540 00:30:21,120 --> 00:30:23,677 to do that a lot of other people will be willing to do that too. I should be one 541 00:30:23,760 --> 00:30:28,397 of the more conservative ones there. So I think we need to develop a a new kind 542 00:30:28,480 --> 00:30:31,997 of security architecture different from the 543 00:30:32,080 --> 00:30:34,797 model technical alignment to let people 544 00:30:34,880 --> 00:30:39,357 be able to use these agents without insane things happening. Like it is 545 00:30:39,440 --> 00:30:43,037 awesome that you can run something on your computer and it can get to know 546 00:30:43,120 --> 00:30:46,237 everything about your life and record your screen and use the web for you and 547 00:30:46,320 --> 00:30:50,397 everything else. Um it will be significantly less awesome if it starts 548 00:30:50,480 --> 00:30:54,957 like randomly telling your secrets to people it shouldn't. Um and then third 549 00:30:55,040 --> 00:30:57,037 and most important 550 00:30:57,120 --> 00:30:58,637 you mentioned not blocking innovation. I 551 00:30:58,720 --> 00:31:04,317 super agree. Uh the the kind of like these are caricatures because everybody 552 00:31:04,400 --> 00:31:05,597 is a little bit in the middle but the 553 00:31:05,680 --> 00:31:07,277 kind of the two paths the world is 554 00:31:07,360 --> 00:31:11,437 looking at right now is either we accept 555 00:31:11,520 --> 00:31:15,437 you know an AI company or country or model having sort of effective 556 00:31:15,520 --> 00:31:18,077 totalitarian control over the world for 557 00:31:18,160 --> 00:31:21,117 no safety problems at all. You have one 558 00:31:21,200 --> 00:31:23,037 AI to rule them all. has been people's 559 00:31:23,120 --> 00:31:25,837 vision over time. Um the other is that 560 00:31:25,920 --> 00:31:29,277 we superdemocratize access and we put 561 00:31:29,360 --> 00:31:33,757 powerful AI in everyone's hands and we trust that if you you know out of a 100 562 00:31:33,840 --> 00:31:37,117 people 99 are going to be good one person bad the good people will stop the 563 00:31:37,200 --> 00:31:40,797 bad some things are going to go wrong in that world people are for sure going to 564 00:31:40,880 --> 00:31:43,437 misuse AI uh in small and probably in 565 00:31:43,520 --> 00:31:47,437 big ways um but for a long time the 566 00:31:47,520 --> 00:31:51,837 world I think has shown that if people are basically most people are good and 567 00:31:51,920 --> 00:31:54,397 the good people stop the bad people as long as there's not a huge power 568 00:31:54,480 --> 00:31:56,797 imbalance. We can get to a world where 569 00:31:56,880 --> 00:31:59,997 we democratize AI. Everybody lifts up 570 00:32:00,080 --> 00:32:02,317 and that kind of keeps the new powerful 571 00:32:02,400 --> 00:32:07,037 bad forces in check. This is the story of of 572 00:32:07,120 --> 00:32:09,277 society for a long time. It will be 573 00:32:09,360 --> 00:32:14,077 messy. Things will go wrong. Um but 574 00:32:14,160 --> 00:32:18,317 I think empowering everyone and having a wide distribution of power is the best 575 00:32:18,400 --> 00:32:20,957 fairest and safest path forward. We call 576 00:32:21,040 --> 00:32:25,357 this resilience. So instead of just there's a alignment, there's security 577 00:32:25,440 --> 00:32:29,757 and then there's this newer idea of resilience where I think it no longer 578 00:32:29,840 --> 00:32:33,117 works for open AI to just say we will block our models from telling people how 579 00:32:33,200 --> 00:32:36,797 to make a pathogen because some open source model is going to help some bad 580 00:32:36,880 --> 00:32:40,477 guy make a pathogen and we have to have society ready to defend against that 581 00:32:40,560 --> 00:32:43,037 with rapid response vaccines sort of new 582 00:32:43,120 --> 00:32:45,037 ways to block respiratory transmission 583 00:32:45,120 --> 00:32:48,237 whatever else. Um yeah that's what I 584 00:32:48,320 --> 00:32:50,317 would say. So the next question is from 585 00:32:50,400 --> 00:32:52,157 Yash and he's not here so I'll read it 586 00:32:52,240 --> 00:32:55,997 out on his behalf. Um I read that uh a 587 00:32:56,080 --> 00:32:57,837 lot of research behind study mode in 588 00:32:57,920 --> 00:33:01,037 chat GPT happened in India last year and 589 00:33:01,120 --> 00:33:05,517 so if OpenAI were to build a product specifically for India not adapted but 590 00:33:05,600 --> 00:33:10,237 designed for India first what problem would you want to it to solve? 591 00:33:10,320 --> 00:33:11,837 >> I would rather one of you figure out the 592 00:33:11,920 --> 00:33:15,997 answer to that. Um, we 593 00:33:16,080 --> 00:33:17,677 way more innovation is going to happen 594 00:33:17,760 --> 00:33:20,077 outside of OpenAI than inside of OpenAI. 595 00:33:20,160 --> 00:33:23,357 And our strategy is to build the core 596 00:33:23,440 --> 00:33:25,917 model and the basic way to use it, 597 00:33:26,000 --> 00:33:28,877 ChachiBT, the API agents like Codex and 598 00:33:28,960 --> 00:33:30,797 the general agent and really empower 599 00:33:30,880 --> 00:33:33,117 everybody else to build it. Um, one of 600 00:33:33,200 --> 00:33:36,077 my learnings from OpenAI is that most 601 00:33:36,160 --> 00:33:38,477 innovation in the world uh comes from 602 00:33:38,560 --> 00:33:42,797 startups and we want to empower them. 603 00:33:42,880 --> 00:33:46,317 Um I think we're going to shift gears a little bit to a health related question 604 00:33:46,400 --> 00:33:52,077 which is uh going to come from Aish from as you have the microphone. Yeah, go 605 00:33:52,160 --> 00:33:54,957 ahead. >> Hello. Uh my name is Arnesh. I'm a 606 00:33:55,040 --> 00:33:57,437 thirdyear medical student at as uh New 607 00:33:57,520 --> 00:34:00,557 Delhi and uh my question is about uh 608 00:34:00,640 --> 00:34:03,837 Chad GBD health. So uh we know that Chad 609 00:34:03,920 --> 00:34:06,557 GBT health connects to Apple health and 610 00:34:06,640 --> 00:34:11,357 medical records but uh in India people 611 00:34:11,440 --> 00:34:14,397 uh most people have neither they have uh 612 00:34:14,480 --> 00:34:17,517 mostly just a phone. So my question is 613 00:34:17,600 --> 00:34:21,037 uh how would you rethink uh health AI in 614 00:34:21,120 --> 00:34:24,877 a country like India? >> I love using chat for health. I don't 615 00:34:24,960 --> 00:34:29,517 have it connected to anything. I type in lots of notes. Every time I get like a 616 00:34:29,600 --> 00:34:33,037 blood test or something, I upload it. If every time I have symptoms, I type it 617 00:34:33,120 --> 00:34:35,597 in. It works really well without medical 618 00:34:35,680 --> 00:34:38,397 records. Um, we'd love to connect to any 619 00:34:38,480 --> 00:34:40,557 platforms that emerge in India as 620 00:34:40,640 --> 00:34:44,157 important platforms, but one of a huge 621 00:34:44,240 --> 00:34:46,957 volume of chat GBT is health related 622 00:34:47,040 --> 00:34:51,597 queries and with no connectors at all, we've worked to make it very good. 623 00:34:51,680 --> 00:34:54,637 >> Um, I think you'll like the next question uh which is going to come from 624 00:34:54,720 --> 00:34:57,997 Vijay uh who's from IIT Delhi. Uh Vijay 625 00:34:58,080 --> 00:35:02,477 you have the microphone. Uh hi Sam I'm Vijay uh I'm a third year student at IIT 626 00:35:02,560 --> 00:35:07,357 Delhi in computer science department. So I would like to know um how would I 627 00:35:07,440 --> 00:35:11,037 undergrads be able to join OpenAI directly and what are the current and 628 00:35:11,120 --> 00:35:13,917 future objectives of OpenAI's office in 629 00:35:14,000 --> 00:35:19,720 India? Uh we would love for you to join OpenAI. 630 00:35:22,480 --> 00:35:24,477 Um, we would love for you to join join 631 00:35:24,560 --> 00:35:27,437 OpenAI either in our office in India or 632 00:35:27,520 --> 00:35:30,477 in the US. I think these next couple of 633 00:35:30,560 --> 00:35:33,037 years are going to be the most exciting 634 00:35:33,120 --> 00:35:38,797 years yet in the development of AI. Um, and the coolest, well, hopefully there's 635 00:35:38,880 --> 00:35:41,837 a lot of cool parts about being an open eye, but I think the coolest part is you 636 00:35:41,920 --> 00:35:45,757 get to see the future and use the tools like six months before everybody else. 637 00:35:45,840 --> 00:35:49,757 And the work that that enables in the development process is awesome. So, we 638 00:35:49,840 --> 00:35:52,637 would love to hear from you. By the way, you can ask me like harder, more 639 00:35:52,720 --> 00:35:53,997 controversial questions. I won't be 640 00:35:54,080 --> 00:35:57,597 offended. Like go for it. 641 00:35:57,680 --> 00:36:01,597 I like those. And I think you might have seen uh you know, we have an office in 642 00:36:01,680 --> 00:36:05,837 Delhi. We've just announced offices in Bangalore and Mumbai as well. So the 643 00:36:05,920 --> 00:36:07,757 team is growing rapidly. So would love 644 00:36:07,840 --> 00:36:11,597 to engage on this conversation. Um okay, 645 00:36:11,680 --> 00:36:14,717 we're going to go to Sahed next. Uh 646 00:36:14,800 --> 00:36:19,437 Sahi, sorry. Um who's from Amiti University and you have the microphone. 647 00:36:19,520 --> 00:36:22,637 Hi Sam, I'm Sah. I'm from Amit University. I'm a 30-year student 648 00:36:22,720 --> 00:36:25,197 pursuing artificial intelligence in AI 649 00:36:25,280 --> 00:36:27,837 in Amit University. So my question is in 650 00:36:27,920 --> 00:36:30,157 today's market where 18 year olds are 651 00:36:30,240 --> 00:36:33,037 competing for jobs with AI, what unfair 652 00:36:33,120 --> 00:36:37,837 advantage do we humans have a lot uh 653 00:36:37,920 --> 00:36:43,037 watching the very best people now work with AI is really incredible and so 654 00:36:43,120 --> 00:36:46,237 different than a year ago. But you watch these people that have like six monitors 655 00:36:46,320 --> 00:36:47,997 on their desk, three agents per monitor 656 00:36:48,080 --> 00:36:53,277 like going session by session. One person can do what a team of 657 00:36:53,360 --> 00:36:54,797 um 658 00:36:54,880 --> 00:36:57,837 a lot of engineers used to take to do. 659 00:36:57,920 --> 00:37:01,357 And when you can keep a whole problem in 660 00:37:01,440 --> 00:37:06,317 one human head and manage a bunch of agents, you are able to bring things to 661 00:37:06,400 --> 00:37:11,197 life at a speed and quality level you just couldn't do before. Um, and there's 662 00:37:11,280 --> 00:37:14,717 still, this may change something in the future, but there's still something that 663 00:37:14,800 --> 00:37:18,797 humans are so good at, that AI is so bad at about thinking through the sort of 664 00:37:18,880 --> 00:37:21,277 like highle strategy of a system or a 665 00:37:21,360 --> 00:37:25,197 product or whatever. Um, that people who 666 00:37:25,280 --> 00:37:27,597 are proficient at using AI have a huge 667 00:37:27,680 --> 00:37:32,797 advantage uh, not just over AI, but over other people who are not doing that. uh 668 00:37:32,880 --> 00:37:34,557 in the long-term future, I don't know 669 00:37:34,640 --> 00:37:36,157 exactly what the advantage will look 670 00:37:36,240 --> 00:37:41,117 like will be, but um we seem so wired to 671 00:37:41,200 --> 00:37:43,117 care about other people and care so 672 00:37:43,200 --> 00:37:48,557 little about machines uh that 673 00:37:48,640 --> 00:37:52,797 somehow that feels like it's going to be part of whatever the answer is. 674 00:37:52,880 --> 00:37:55,757 So the next question is a little linked to what the prime minister said 675 00:37:55,840 --> 00:37:57,997 yesterday as well and you know he said 676 00:37:58,080 --> 00:37:59,437 there are two kinds of people in the 677 00:37:59,520 --> 00:38:03,277 world. one who think that AI is fearful, 678 00:38:03,360 --> 00:38:07,677 the others who are very you know excited about the future of what AI can bring 679 00:38:07,760 --> 00:38:11,517 and I think the next question is a little bit about jobs and you know that 680 00:38:11,600 --> 00:38:13,837 comes up often of course uh which is 681 00:38:13,920 --> 00:38:17,117 from uh Guru Dhanush. So go ahead Chris. 682 00:38:17,200 --> 00:38:20,877 >> Hi sir, my name is Guru Danush. I am a second year BK student in mathematics 683 00:38:20,960 --> 00:38:25,277 and computing. My question is what do you say to the workers who fear that 684 00:38:25,360 --> 00:38:27,357 their jobs will be outdated in next five 685 00:38:27,440 --> 00:38:29,277 years and what skills should they be 686 00:38:29,360 --> 00:38:32,557 focusing on now? Anyone who says they're 687 00:38:32,640 --> 00:38:35,997 only excited or only fearful about AI is 688 00:38:36,080 --> 00:38:39,597 I think not being very thoughtful. I don't think it's two camps. I think 689 00:38:39,680 --> 00:38:41,837 there's a continuum of people based on 690 00:38:41,920 --> 00:38:44,157 whether you're 90% excited and 10% 691 00:38:44,240 --> 00:38:47,677 fearful or 90% fearful and 10% excited. 692 00:38:47,760 --> 00:38:51,757 But as with any change of this magnitude 693 00:38:51,840 --> 00:38:54,077 happening this quickly, um, and with 694 00:38:54,160 --> 00:38:59,197 this much unknown, this much promise, this much obvious peril, um, it's it you 695 00:38:59,280 --> 00:39:02,397 you have to be somewhere on that spectrum. You can't it can't be sort of 696 00:39:02,480 --> 00:39:05,757 at at either end. Um, 697 00:39:05,840 --> 00:39:08,637 a big part of that is what the impact on 698 00:39:08,720 --> 00:39:11,997 jobs is going to be. And 699 00:39:12,080 --> 00:39:15,917 I will caveat this with lots of other things by AI by saying that no one 700 00:39:16,000 --> 00:39:18,077 really knows including us. There will be 701 00:39:18,160 --> 00:39:21,437 many surprises. Um, some jobs will 702 00:39:21,520 --> 00:39:25,917 totally go away. Many jobs will not 703 00:39:26,000 --> 00:39:31,357 change as much as it seems like they should with AI because we have, you 704 00:39:31,440 --> 00:39:35,517 know, we want someone else to be doing that. we still go want a human doctor to 705 00:39:35,600 --> 00:39:37,517 tell us things even if they we know the 706 00:39:37,600 --> 00:39:40,077 AI gives better advice or any many other 707 00:39:40,160 --> 00:39:42,717 set of examples. Uh or you know 708 00:39:42,800 --> 00:39:47,437 entrepreneurs will still start companies they'll just grow faster and they'll do 709 00:39:47,520 --> 00:39:50,637 more and it'll be more disruptive to existing jobs. 710 00:39:50,720 --> 00:39:54,477 And then most of all, 711 00:39:54,560 --> 00:39:58,317 every time in history that there's been a major technological shift, we have 712 00:39:58,400 --> 00:40:01,677 underestimated how quickly the new jobs are going to 713 00:40:01,760 --> 00:40:02,797 come and how different they were going 714 00:40:02,880 --> 00:40:07,037 to be. It'd be very hard to think about 715 00:40:07,120 --> 00:40:11,037 my job, for example, before computers were invented. Um, and impossible to 716 00:40:11,120 --> 00:40:12,557 think about my job before, say, the 717 00:40:12,640 --> 00:40:15,437 industrial revolution. Um, there are 718 00:40:15,520 --> 00:40:18,477 jobs now that are 719 00:40:18,560 --> 00:40:20,557 incredibly good jobs that didn't exist 720 00:40:20,640 --> 00:40:23,597 as a category 20 years ago, even if 721 00:40:23,680 --> 00:40:26,637 someone from 500 years ago could look at 722 00:40:26,720 --> 00:40:29,997 the jobs of the world today, um, 723 00:40:30,080 --> 00:40:32,397 generally speaking, they would say about 724 00:40:32,480 --> 00:40:35,597 many of the jobs, um, certainly my job, 725 00:40:35,680 --> 00:40:40,077 you know, death's not a real job. You're doing fine. You could, you don't have to 726 00:40:40,160 --> 00:40:43,517 be working. You're like, you've got plenty of food. you've got access to 727 00:40:43,600 --> 00:40:46,557 great shelter. Like this is this is not 728 00:40:46,640 --> 00:40:47,837 a job they could have imagined and it 729 00:40:47,920 --> 00:40:51,437 wouldn't look like real work. I hope 730 00:40:51,520 --> 00:40:55,357 100 years from now, not even 500, we would all look at the jobs and be like, 731 00:40:55,440 --> 00:40:59,357 "Those people are impossibly rich. They have everything they could ever want. 732 00:40:59,440 --> 00:41:03,677 They are making up ways to spend their time to entertain themselves. It's a, 733 00:41:03,760 --> 00:41:07,437 you know, it's a player of game style existence. They don't really need this." 734 00:41:07,520 --> 00:41:10,157 Um, and 735 00:41:10,240 --> 00:41:13,037 that is sort of the history of society 736 00:41:13,120 --> 00:41:17,837 building up more collective welfare for each other. 737 00:41:17,920 --> 00:41:21,437 Uh, and I think that's great, but we never 738 00:41:21,520 --> 00:41:24,957 run out of things to do. We have never gotten to the sort of like thing that 739 00:41:25,040 --> 00:41:28,477 people have theorized for for centuries now, which is technology is going to 740 00:41:28,560 --> 00:41:31,757 progress so much that we just sit around and do nothing. We find new things to 741 00:41:31,840 --> 00:41:35,437 do. We find new ways to express our creativity. Um, we find new ways to be 742 00:41:35,520 --> 00:41:36,957 useful to each other. you find new ways 743 00:41:37,040 --> 00:41:39,517 to, you know, gain status or whatever, 744 00:41:39,600 --> 00:41:42,477 whatever it is. Uh 745 00:41:42,560 --> 00:41:47,357 many of you will eventually do jobs that don't exist today even really as a 746 00:41:47,440 --> 00:41:50,797 concept. It doesn't mean that the transition is 747 00:41:50,880 --> 00:41:53,037 not painful. I expect for many people it 748 00:41:53,120 --> 00:41:56,077 will be very painful. Um 749 00:41:56,160 --> 00:41:59,357 as society gets richer quickly, I hope 750 00:41:59,440 --> 00:42:04,637 we find new socioeconomic models to make it less painful. But I'm less worried 751 00:42:04,720 --> 00:42:07,997 about the long-term future. And I think we'll find new things to do in in not 752 00:42:08,080 --> 00:42:10,317 very long. >> So we've announced a partnership with 753 00:42:10,400 --> 00:42:14,317 the Tata group yesterday and you know it's about compute, it's about data 754 00:42:14,400 --> 00:42:16,477 centers and so on. Now I think the next 755 00:42:16,560 --> 00:42:18,237 question relates to that a little bit. 756 00:42:18,320 --> 00:42:21,517 Uh and which is from Kub from IIT Delhi. 757 00:42:21,600 --> 00:42:25,037 Uh go ahead Gaussub. >> Hi Sam, my name is Kausub. I'm a second 758 00:42:25,120 --> 00:42:29,357 year PhD student here at the school of AI IT Delhi. I wanted to ask a question 759 00:42:29,440 --> 00:42:33,677 about what are the steps that open AAI is taking towards building green AI 760 00:42:33,760 --> 00:42:35,997 solutions for the world. >> To building what solutions? 761 00:42:36,080 --> 00:42:37,517 >> Green AI solutions. >> Real 762 00:42:37,600 --> 00:42:38,157 >> green green 763 00:42:38,240 --> 00:42:43,037 >> green. Oh. Uh I think 764 00:42:43,120 --> 00:42:46,957 the world is only a few years away from 765 00:42:47,040 --> 00:42:49,597 fusion and really reliable solar wind 766 00:42:49,680 --> 00:42:53,117 plus storage. And I further think that 767 00:42:53,200 --> 00:42:56,237 those will become uh the cheapest source 768 00:42:56,320 --> 00:43:00,877 of energy in the world. And when that happens, the economy will 769 00:43:00,960 --> 00:43:03,197 do its thing. And not just AI, but 770 00:43:03,280 --> 00:43:05,757 everybody else will use new forms of 771 00:43:05,840 --> 00:43:09,517 energy. Um the this is important for a 772 00:43:09,600 --> 00:43:11,997 ton of reasons. Uh but one that gets 773 00:43:12,080 --> 00:43:14,477 talked about less is as part of this if 774 00:43:14,560 --> 00:43:19,117 we can drive massive abundance of energy um in the same way that we can drive 775 00:43:19,200 --> 00:43:21,277 abundance of intelligence I think that 776 00:43:21,360 --> 00:43:23,997 will do a huge amount to elevate people 777 00:43:24,080 --> 00:43:27,437 around the world. Um so obviously we 778 00:43:27,520 --> 00:43:30,317 care about it environmentally uh and are 779 00:43:30,400 --> 00:43:33,277 committed to do that. We really care 780 00:43:33,360 --> 00:43:35,917 about it from a 781 00:43:36,000 --> 00:43:37,677 sort of duty to the world to make energy 782 00:43:37,760 --> 00:43:40,717 abundant and perspective. Uh we're going 783 00:43:40,800 --> 00:43:45,277 to go uh change gears a little bit and uh I was saying earlier we have uh 784 00:43:45,360 --> 00:43:50,477 students from creative uh industries as well. Um so the next question u is 785 00:43:50,560 --> 00:43:54,557 around that and I think it's from Parisha. Uh if you have the microphone 786 00:43:54,640 --> 00:43:58,717 go ahead please. >> Uh hi Sam I'm Parisha from Pearl 787 00:43:58,800 --> 00:44:00,957 Academy. I'm pursuing interior and 788 00:44:01,040 --> 00:44:04,077 architectural design. So my question is 789 00:44:04,160 --> 00:44:07,037 uh like AI became the tool that that can 790 00:44:07,120 --> 00:44:12,397 generate uh images, space and uh create 791 00:44:12,480 --> 00:44:16,477 entire concept. How designer can ensure 792 00:44:16,560 --> 00:44:20,397 and uh ensure their work still reflect 793 00:44:20,480 --> 00:44:24,397 the human emotions, cultural identity 794 00:44:24,480 --> 00:44:29,357 and originality um rather than becoming too machine 795 00:44:29,440 --> 00:44:30,877 driven. 796 00:44:30,960 --> 00:44:33,277 >> Great question. Um, I used to be more 797 00:44:33,360 --> 00:44:36,557 worried about this and you know I wasn't 798 00:44:36,640 --> 00:44:41,997 sure totally how it was going to go when models could sort of create anything. 799 00:44:42,080 --> 00:44:46,477 Watching what's happened in practice though and talking to creators, uh, not 800 00:44:46,560 --> 00:44:50,477 everyone, but generally speaking, people say this is the best thing ever to 801 00:44:50,560 --> 00:44:51,917 happen as a tool in their creative 802 00:44:52,000 --> 00:44:55,517 process if they can explore ideas and 803 00:44:55,600 --> 00:45:00,237 have a machine help them search the space and then pick at each 804 00:45:00,320 --> 00:45:02,157 layer which direction they want to go. 805 00:45:02,240 --> 00:45:06,157 Um, and if we can make their creative 806 00:45:06,240 --> 00:45:08,157 ideation loop faster and faster and 807 00:45:08,240 --> 00:45:11,517 faster, where the time from idea to then 808 00:45:11,600 --> 00:45:15,277 seen and results and then coming up with a new idea from that gets shorter and 809 00:45:15,360 --> 00:45:18,157 shorter and talking to creators, whether 810 00:45:18,240 --> 00:45:20,317 they're visual artists or people making 811 00:45:20,400 --> 00:45:23,757 videos or music. Um, they all care a 812 00:45:23,840 --> 00:45:27,357 lot, as do I, that the human spirit and the human direction and the human 813 00:45:27,440 --> 00:45:30,797 imprint is in this, but they all think they'll be able to create much better 814 00:45:30,880 --> 00:45:35,037 stuff than ever before. Another thing for optimism, back on this earlier point 815 00:45:35,120 --> 00:45:37,517 about people really are very hardwired 816 00:45:37,600 --> 00:45:39,997 to care about other people. We are 817 00:45:40,080 --> 00:45:44,477 seeing a I think predictable but still 818 00:45:44,560 --> 00:45:47,917 surprising to some people result which 819 00:45:48,000 --> 00:45:51,357 is if you show someone an image if you 820 00:45:51,440 --> 00:45:55,197 say I I'll sell you this piece of art and an AI made it and it's beautiful but 821 00:45:55,280 --> 00:45:57,837 you know an AI made it the the value 822 00:45:57,920 --> 00:45:59,837 that people put on that effectively 823 00:45:59,920 --> 00:46:04,957 rounds to zero. Um, if a human made it, even with AI 824 00:46:05,040 --> 00:46:07,917 tools and it's, you know, someone who you can connect to and has a story and 825 00:46:08,000 --> 00:46:10,237 put their creative energy into it and had something they were trying to 826 00:46:10,320 --> 00:46:15,597 express, uh, but AI helped them in their creative process, it's worth a lot. But 827 00:46:15,680 --> 00:46:18,317 without the person like effectively signing their name to it and saying 828 00:46:18,400 --> 00:46:23,037 like, "This is my thing." We seem to not care. And I think that's another very 829 00:46:23,120 --> 00:46:26,237 positive sign for what it's going to be like for human creators. 830 00:46:26,320 --> 00:46:29,837 >> And I know you're on a discussion with Kavan Johar later today, so some of this 831 00:46:29,920 --> 00:46:32,077 will come up a little bit more as well. 832 00:46:32,160 --> 00:46:35,917 U and he's very popular movie director as you know. So I know this is going to 833 00:46:36,000 --> 00:46:38,477 be a conversation uh further up as well. 834 00:46:38,560 --> 00:46:42,317 Uh I'm a little conscious we're uh maybe coming up a little bit on time. I'll 835 00:46:42,400 --> 00:46:46,077 take two more questions and then I want to do a little bit of rapid fire at the 836 00:46:46,160 --> 00:46:48,317 end as well Sam if you're open to that. 837 00:46:48,400 --> 00:46:52,877 Um I think we have Paridi next. Uh Paridi if you have the microphone. Yeah 838 00:46:52,960 --> 00:46:56,077 go ahead. >> Hi Sam I am Paridi from Amit University. 839 00:46:56,160 --> 00:46:58,877 I'm currently in third year of my 840 00:46:58,960 --> 00:47:03,677 artificial intelligence degree. My question to you is that uh recently 841 00:47:03,760 --> 00:47:07,677 there have been a lot of concerns regarding the use of water in cooling 842 00:47:07,760 --> 00:47:09,997 down the servers and that it's leading 843 00:47:10,080 --> 00:47:13,037 to water scarcity. So what is open AI 844 00:47:13,120 --> 00:47:17,517 doing to you know uh people help people 845 00:47:17,600 --> 00:47:19,357 uh dismiss the these concerns. 846 00:47:19,440 --> 00:47:21,517 >> So the the energy concern and the need 847 00:47:21,600 --> 00:47:24,637 to move to green energy I think is very 848 00:47:24,720 --> 00:47:27,677 real. uh that that's something that at 849 00:47:27,760 --> 00:47:30,477 the scale AI is developing can be a 850 00:47:30,560 --> 00:47:32,797 significant issue. The water concern I 851 00:47:32,880 --> 00:47:35,197 think is no longer real. Uh there was a 852 00:47:35,280 --> 00:47:40,157 time when data centers were you using evaporative cooling. Um that is not 853 00:47:40,240 --> 00:47:45,037 something that people are doing much anymore. Um the there is you know 854 00:47:45,120 --> 00:47:49,117 obviously some water consumption in producing semiconductors 855 00:47:49,200 --> 00:47:51,517 but if you look at AI data centers they 856 00:47:51,600 --> 00:47:53,917 are not significant consumers of water. 857 00:47:54,000 --> 00:47:56,397 This has been an incredibly sticky meme. 858 00:47:56,480 --> 00:47:59,117 I don't totally understand why. You know, there's people who say, "Well, I 859 00:47:59,200 --> 00:48:02,477 don't use ChachiBT because each time you ask it a question, it uses up seven 860 00:48:02,560 --> 00:48:06,077 gallons of water or whatever." It's totally not true. We've tried to explain 861 00:48:06,160 --> 00:48:07,757 it's not true. It's been very sticky for 862 00:48:07,840 --> 00:48:09,997 whatever reason. Um, but I think the 863 00:48:10,080 --> 00:48:13,917 focus on energy does make sense. 864 00:48:14,000 --> 00:48:17,197 >> All right. I I know we have a lot more questions. Uh, but I'm I'm conscious 865 00:48:17,280 --> 00:48:21,117 we're coming up on time. So, Ocean will come to you and, uh, please go ahead and 866 00:48:21,200 --> 00:48:23,917 ask your question. >> Yeah. Hi, Sam. My name is Oshini Adaral. 867 00:48:24,000 --> 00:48:27,197 I'm a PhD scholar at the transportation research and injury prevention center 868 00:48:27,280 --> 00:48:30,797 here. So my question is, what is the biggest mistake you see young people 869 00:48:30,880 --> 00:48:35,997 make right now when they apparently prepare for AI? 870 00:48:36,080 --> 00:48:37,437 Man, I'm going to answer this. I hope 871 00:48:37,520 --> 00:48:40,957 this talk is not recorded. Uh 872 00:48:41,040 --> 00:48:47,720 I think listening to old people is the biggest mistake young people make. 873 00:48:48,640 --> 00:48:52,637 I I think your your parents obviously love you more than anyone. They're 874 00:48:52,720 --> 00:48:55,357 extremely well-meaning. Um for advice 875 00:48:55,440 --> 00:48:59,997 about how to live a happy life or how to be a a good person, there is no one you 876 00:49:00,080 --> 00:49:01,917 should listen to 877 00:49:02,000 --> 00:49:06,157 more than your parents. Um but 878 00:49:06,240 --> 00:49:10,877 for a predictor of what the world is going to be like going forward, um I 879 00:49:10,960 --> 00:49:15,997 think you should trust me for having good intuition of the rate of change. Um 880 00:49:16,080 --> 00:49:19,437 it young people always figure this out the 881 00:49:19,520 --> 00:49:22,877 best and the world you are all 882 00:49:22,960 --> 00:49:25,437 inheriting is going to be very different 883 00:49:25,520 --> 00:49:27,677 and you'll have to quickly develop your 884 00:49:27,760 --> 00:49:30,957 own intuitions and trust them. Um but I 885 00:49:31,040 --> 00:49:35,997 think the traditional career advice is probably not going to work as well. 886 00:49:36,080 --> 00:49:39,357 >> Wonderful. Thank you for taking all those questions. Uh let's do a quick 887 00:49:39,440 --> 00:49:43,837 rapid fire and you can choose to answer with one word or one sentence but uh 888 00:49:43,920 --> 00:49:45,837 that's what you need to uh kind of keep 889 00:49:45,920 --> 00:49:48,797 to um and I have like four or five 890 00:49:48,880 --> 00:49:53,357 questions. U the first one so will the next trillion dollar AI company come 891 00:49:53,440 --> 00:49:56,637 from India? >> I hope so. 892 00:49:56,720 --> 00:50:03,000 Nice. Most overrated fear about AI 893 00:50:03,920 --> 00:50:07,757 like killer robots in the streets. 894 00:50:07,840 --> 00:50:11,517 Most underrated opportunity 895 00:50:11,600 --> 00:50:14,477 >> zero person companies 896 00:50:14,560 --> 00:50:19,720 bigger change in five years education or healthcare 897 00:50:21,680 --> 00:50:25,997 education. >> And last one skill every IT back 898 00:50:26,080 --> 00:50:27,757 healthare that's close though. 899 00:50:27,840 --> 00:50:31,517 >> Okay last question. One skill every IIT 900 00:50:31,600 --> 00:50:34,317 student should build this year. 901 00:50:34,400 --> 00:50:35,277 >> Resilience. 902 00:50:35,360 --> 00:50:37,757 >> Resilience. Wonderful. 903 00:50:37,840 --> 00:50:41,997 Thank you all. Thank you. 904 00:50:42,080 --> 00:50:43,677 >> Thank you so much for interacting with 905 00:50:43,760 --> 00:50:46,877 our students and uh this has indeed been 906 00:50:46,960 --> 00:50:49,197 a very memorable evening for all of us 907 00:50:49,280 --> 00:50:51,917 and in order to ensure that you also 908 00:50:52,000 --> 00:50:54,877 remember us uh once you go out of this 909 00:50:54,960 --> 00:50:59,357 institute. Uh we'd like to present our small token of gratitude. For that I'd 910 00:50:59,440 --> 00:51:01,677 like to call Mr. Vinod Koslaser uh back 911 00:51:01,760 --> 00:51:08,400 on stage and also the director uh to present uh our guests. 912 00:51:49,120 --> 00:51:51,757 I I want to finish given you gave me a 913 00:51:51,840 --> 00:51:53,597 baseball cap. Thank you. With one 914 00:51:53,680 --> 00:51:56,877 comment. I think in the next few four 915 00:51:56,960 --> 00:51:59,997 years this baseball cap will read my 916 00:52:00,080 --> 00:52:03,197 mind and input directly into computers. 917 00:52:03,280 --> 00:52:09,037 We may succeed fail at that but we are trying. 918 00:52:09,120 --> 00:52:14,237 >> Uh >> oh yes. >> I'd like I'd like to request Sam now to 919 00:52:14,320 --> 00:52:17,677 capture this moment with the audience with a selfie. I'd like to call upon the 920 00:52:17,760 --> 00:52:21,677 stage uh the associate dean of academic outreach and new initiatives uh 921 00:52:21,760 --> 00:52:25,997 professor Shelpi Sharma and professor Deepak Jan for this. 922 00:52:26,080 --> 00:52:28,557 >> So we're doing a selfie right with with everybody. 923 00:52:28,640 --> 00:52:31,800 >> Thank you.