1 00:00:00,000 --> 00:00:01,080 Ready for a deep dive. 2 00:00:01,080 --> 00:00:01,960 Always. 3 00:00:01,960 --> 00:00:04,420 Well, today we're diving into the world of AI 4 00:00:04,420 --> 00:00:05,840 and specifically is potential 5 00:00:05,840 --> 00:00:08,320 to like revolutionize scientific discovery. 6 00:00:08,320 --> 00:00:09,840 Sounds ambitious. 7 00:00:09,840 --> 00:00:11,800 Yeah, and to help us kind of navigate 8 00:00:11,800 --> 00:00:14,000 this complex landscape, 9 00:00:14,000 --> 00:00:16,040 we're turning to the insights of Demis Hasabis. 10 00:00:16,040 --> 00:00:17,360 Oh, yes. 11 00:00:17,360 --> 00:00:19,040 The co-founder of DeepMind. 12 00:00:19,040 --> 00:00:19,880 Exactly. 13 00:00:19,880 --> 00:00:22,640 And we're pulling from a fascinating conversation 14 00:00:22,640 --> 00:00:25,520 that he had with Chris Anderson from TED. 15 00:00:25,520 --> 00:00:26,640 I'm familiar with that one. 16 00:00:26,640 --> 00:00:28,280 It's a good one. 17 00:00:28,280 --> 00:00:29,600 Lots of interesting insights. 18 00:00:29,600 --> 00:00:30,440 Yeah. 19 00:00:30,440 --> 00:00:34,000 So our mission for today, I guess you could say, 20 00:00:34,000 --> 00:00:37,120 is to kind of explore Hasabis' journey, 21 00:00:37,120 --> 00:00:38,680 how he got started, 22 00:00:38,680 --> 00:00:42,040 what led him to these groundbreaking AI breakthroughs. 23 00:00:42,040 --> 00:00:44,200 And maybe, just maybe, we'll uncover some clues 24 00:00:44,200 --> 00:00:47,000 about unlocking the secrets of the universe along the way. 25 00:00:47,000 --> 00:00:47,840 You never know, right? 26 00:00:47,840 --> 00:00:48,680 You never know. 27 00:00:48,680 --> 00:00:51,720 So Hasabis' background, it's really fascinating. 28 00:00:51,720 --> 00:00:54,880 Kind of all seems to stem from this childhood obsession he had. 29 00:00:54,880 --> 00:00:55,720 Oh, you mean chess. 30 00:00:55,720 --> 00:00:56,560 Yeah. 31 00:00:56,560 --> 00:00:57,400 Yeah. 32 00:00:57,400 --> 00:00:58,240 Chess. 33 00:00:58,240 --> 00:00:59,080 The guy was basically a prodigy. 34 00:00:59,080 --> 00:01:01,880 I'm captaining England's under-11 chess team 35 00:01:01,880 --> 00:01:03,960 at just nine years old. 36 00:01:03,960 --> 00:01:05,480 Nine years old? 37 00:01:05,480 --> 00:01:06,440 Nine years old. 38 00:01:06,440 --> 00:01:08,080 I mean, that's insane. 39 00:01:08,080 --> 00:01:10,720 Most kids at that age are just figuring out 40 00:01:10,720 --> 00:01:11,960 how the pieces move. 41 00:01:11,960 --> 00:01:12,800 Yeah. 42 00:01:12,800 --> 00:01:13,640 You know? 43 00:01:13,640 --> 00:01:14,480 Totally. 44 00:01:14,480 --> 00:01:15,560 But he's already leading a national team. 45 00:01:15,560 --> 00:01:16,400 Like, wow. 46 00:01:16,400 --> 00:01:17,240 Right. 47 00:01:17,240 --> 00:01:20,600 That early fascination, I think, with strategy, you know, 48 00:01:20,600 --> 00:01:22,440 with the intelligence that's programmed 49 00:01:22,440 --> 00:01:23,880 into those early chess computers. 50 00:01:23,880 --> 00:01:25,800 It's like, that's what really, I think, 51 00:01:25,800 --> 00:01:27,640 ignited this lifelong passion. 52 00:01:27,640 --> 00:01:28,560 Understand intelligence. 53 00:01:28,560 --> 00:01:29,800 Yeah, I understand intelligence. 54 00:01:29,800 --> 00:01:32,080 And ultimately, to recreate it. 55 00:01:32,080 --> 00:01:34,440 It looks like his brain was kind of wired for strategy 56 00:01:34,440 --> 00:01:35,160 from the get-go. 57 00:01:35,160 --> 00:01:36,440 It seems that way. 58 00:01:36,440 --> 00:01:40,200 So speaking of strategy, this kind of strategic thinking 59 00:01:40,200 --> 00:01:44,280 led DeepMind to some really interesting early work 60 00:01:44,280 --> 00:01:45,120 using games. 61 00:01:45,120 --> 00:01:46,640 Oh, you're talking about the Atari stuff. 62 00:01:46,640 --> 00:01:47,800 Yeah, like breakout. 63 00:01:47,800 --> 00:01:49,640 Ah, breakout of classic. 64 00:01:49,640 --> 00:01:50,920 Classic, right? 65 00:01:50,920 --> 00:01:53,400 So they trained in AI to play it. 66 00:01:53,400 --> 00:01:57,720 And the results were, well, they were surprising, 67 00:01:57,720 --> 00:01:58,760 to say the least. 68 00:01:58,760 --> 00:01:59,880 OK, I'm intrigued. 69 00:01:59,880 --> 00:02:00,400 What happened? 70 00:02:00,400 --> 00:02:03,280 Did the AI just become super fast at bouncing the ball? 71 00:02:03,280 --> 00:02:04,240 Well, yeah, it won. 72 00:02:04,240 --> 00:02:06,320 But the way it won, that's what was unexpected. 73 00:02:06,320 --> 00:02:08,440 So the AI actually discovered this strategy. 74 00:02:08,440 --> 00:02:09,560 Well, this is a strategy. 75 00:02:09,560 --> 00:02:14,320 This strategy that most human players wouldn't even think of, 76 00:02:14,320 --> 00:02:17,680 it learned to tunnel behind the wall. 77 00:02:17,680 --> 00:02:18,000 Wow. 78 00:02:18,000 --> 00:02:20,760 Creating this kind of shortcut to victory. 79 00:02:20,760 --> 00:02:23,680 So it wasn't just following pre-programmed rules. 80 00:02:23,680 --> 00:02:26,200 No, it was coming up with its own solution. 81 00:02:26,200 --> 00:02:27,000 It's learning. 82 00:02:27,000 --> 00:02:28,120 It was adapting. 83 00:02:28,120 --> 00:02:28,640 Wow. 84 00:02:28,640 --> 00:02:30,680 And it was coming up with solutions 85 00:02:30,680 --> 00:02:33,360 that even its creators hadn't considered. 86 00:02:33,360 --> 00:02:35,360 OK, so that's one of those aha moments. 87 00:02:35,360 --> 00:02:36,440 I love in AI research. 88 00:02:36,440 --> 00:02:37,840 It's like, wait a minute. 89 00:02:37,840 --> 00:02:39,720 The machine is outsmarting us already. 90 00:02:39,720 --> 00:02:40,120 Right. 91 00:02:40,120 --> 00:02:42,400 And this was just a glimpse of what was to come. 92 00:02:42,400 --> 00:02:42,960 A glimpse. 93 00:02:42,960 --> 00:02:43,560 Yeah. 94 00:02:43,560 --> 00:02:46,040 So they took these learnings from breakout, 95 00:02:46,040 --> 00:02:48,840 and they developed AlphaGo, which then 96 00:02:48,840 --> 00:02:51,880 went head to head with the world champion, Lisa Doll, 97 00:02:51,880 --> 00:02:53,040 in the game of Go. 98 00:02:53,040 --> 00:02:54,920 OK, now Go. 99 00:02:54,920 --> 00:02:56,120 This is where it gets really interesting. 100 00:02:56,120 --> 00:02:58,760 Go is incredibly complex. 101 00:02:58,760 --> 00:02:59,320 It is. 102 00:02:59,320 --> 00:03:02,000 I mean, there are more possible moves in Go 103 00:03:02,000 --> 00:03:03,760 than there are atoms in the universe. 104 00:03:03,760 --> 00:03:04,680 That's what they say. 105 00:03:04,680 --> 00:03:05,840 It's just mind-boggling. 106 00:03:05,840 --> 00:03:06,800 It really is. 107 00:03:06,800 --> 00:03:09,640 And I've heard it said that Go is a game of intuition 108 00:03:09,640 --> 00:03:12,760 as much as it is strategy. 109 00:03:12,760 --> 00:03:15,880 So when AlphaGo was first announced, 110 00:03:15,880 --> 00:03:18,400 I remember there was a lot of debate 111 00:03:18,400 --> 00:03:22,440 about whether an AI could truly grasp those nuances, 112 00:03:22,440 --> 00:03:23,880 those subtle aspects of the game. 113 00:03:23,880 --> 00:03:24,440 Yeah. 114 00:03:24,440 --> 00:03:27,600 I mean, Go was considered this grand challenge in AI, 115 00:03:27,600 --> 00:03:30,680 something that experts thought was decades away, 116 00:03:30,680 --> 00:03:32,440 before a machine could even come close. 117 00:03:32,440 --> 00:03:34,760 But AlphaGo didn't just win. 118 00:03:34,760 --> 00:03:35,280 It does. 119 00:03:35,280 --> 00:03:36,840 No, it didn't just win. 120 00:03:36,840 --> 00:03:39,600 It invented entirely new strategies. 121 00:03:39,600 --> 00:03:40,200 Really? 122 00:03:40,200 --> 00:03:43,440 Yeah, revealing weaknesses in traditional Go thinking 123 00:03:43,440 --> 00:03:46,680 that had gone unnoticed for centuries. 124 00:03:46,680 --> 00:03:47,360 For centuries. 125 00:03:47,360 --> 00:03:48,520 So it's not just playing the game. 126 00:03:48,520 --> 00:03:50,320 It's pushing the boundaries of the game itself. 127 00:03:50,320 --> 00:03:50,760 Right. 128 00:03:50,760 --> 00:03:53,560 It's like it made people realize that AI wasn't just 129 00:03:53,560 --> 00:03:55,320 about brute force calculation. 130 00:03:55,320 --> 00:03:56,640 It was showing signs of creativity. 131 00:03:56,640 --> 00:03:58,600 Yeah, creativity, strategic thinking. 132 00:03:58,600 --> 00:04:00,440 It was exhibiting a level of, I don't know, 133 00:04:00,440 --> 00:04:01,920 it seemed almost human. 134 00:04:01,920 --> 00:04:04,840 OK, so if you can master something as complex as Go, 135 00:04:04,840 --> 00:04:06,320 what else can AI achieve? 136 00:04:06,320 --> 00:04:07,440 Well, they didn't stop there. 137 00:04:07,440 --> 00:04:08,880 They created AlphaZero. 138 00:04:08,880 --> 00:04:10,040 Oh, AlphaZero, OK. 139 00:04:10,040 --> 00:04:12,400 Which took this whole concept of self-learning 140 00:04:12,400 --> 00:04:13,600 to a whole new level. 141 00:04:13,600 --> 00:04:16,520 OK, so if AlphaGo was like a prodigy, 142 00:04:16,520 --> 00:04:18,720 AlphaZero was like the ultimate grandmaster. 143 00:04:18,720 --> 00:04:19,640 Pretty much, yeah. 144 00:04:19,640 --> 00:04:22,920 AlphaZero learned to play not just Go, but also chess 145 00:04:22,920 --> 00:04:23,880 and shokie. 146 00:04:23,880 --> 00:04:24,640 All from scratch. 147 00:04:24,640 --> 00:04:26,040 Entirely from scratch. 148 00:04:26,040 --> 00:04:27,800 No pre-programmed strategies. 149 00:04:27,800 --> 00:04:30,040 No, no human guidance. 150 00:04:30,040 --> 00:04:31,600 Just pure learning. 151 00:04:31,600 --> 00:04:34,440 From the ground up, and here's the kicker, 152 00:04:34,440 --> 00:04:37,400 it reached superhuman level play in these games 153 00:04:37,400 --> 00:04:38,880 within just a few hours. 154 00:04:38,880 --> 00:04:40,480 A few hours? 155 00:04:40,480 --> 00:04:41,240 I mean, come on. 156 00:04:41,240 --> 00:04:44,880 These are skills that take humans years, decades to achieve. 157 00:04:44,880 --> 00:04:46,240 And it did it in hours. 158 00:04:46,240 --> 00:04:48,040 That's the power of AI learning. 159 00:04:48,040 --> 00:04:48,880 That's the power. 160 00:04:48,880 --> 00:04:51,560 It makes you reconsider the limitations we put on learning 161 00:04:51,560 --> 00:04:52,520 and skill development. 162 00:04:52,520 --> 00:04:53,880 It's like, what else could we crack 163 00:04:53,880 --> 00:04:55,440 with this kind of learning power? 164 00:04:55,440 --> 00:04:57,680 What other seemingly insurmountable problem? 165 00:04:57,680 --> 00:04:59,440 Well, one of those insurmountable problems 166 00:04:59,440 --> 00:05:01,680 might just be next on Deep Mind's list. 167 00:05:01,680 --> 00:05:02,360 What's that? 168 00:05:02,360 --> 00:05:04,080 Protein folding. 169 00:05:04,080 --> 00:05:05,080 Yes. 170 00:05:05,080 --> 00:05:07,960 This is where Deep Mind's work takes a giant leap, 171 00:05:07,960 --> 00:05:12,920 from the world of games into real-world scientific breakthroughs. 172 00:05:12,920 --> 00:05:16,360 Yeah, because proteins, those tiny building blocks of life, 173 00:05:16,360 --> 00:05:17,560 have been a bit of a mystery. 174 00:05:17,560 --> 00:05:18,920 Oh, for a long time. 175 00:05:18,920 --> 00:05:19,360 Yeah. 176 00:05:19,360 --> 00:05:22,080 We know their 3D shape determines their function. 177 00:05:22,080 --> 00:05:25,440 But predicting that shape from its amino acid sequence. 178 00:05:25,440 --> 00:05:28,200 That's been a challenge, a challenge for decades. 179 00:05:28,200 --> 00:05:30,560 Yeah, for over 50 years, researchers 180 00:05:30,560 --> 00:05:35,120 have been painstakingly unraveling protein structures one by one. 181 00:05:35,120 --> 00:05:38,800 And this process could take years for a single protein. 182 00:05:38,800 --> 00:05:39,400 Years. 183 00:05:39,400 --> 00:05:39,880 Yeah. 184 00:05:39,880 --> 00:05:43,080 Incredibly time-consuming, resource-intensive. 185 00:05:43,080 --> 00:05:45,440 And that's where Deep Mind's alpha fold comes in. 186 00:05:45,440 --> 00:05:49,640 The AI specifically designed to crack the protein folding code. 187 00:05:49,640 --> 00:05:50,840 And crack it, it did. 188 00:05:50,840 --> 00:05:51,680 It did. 189 00:05:51,680 --> 00:05:54,280 OK, I have to ask, how did alpha fold change the game? 190 00:05:54,280 --> 00:05:56,280 It'll imagine a billion years. 191 00:05:56,280 --> 00:05:56,960 A billion. 192 00:05:56,960 --> 00:06:00,040 A billion years of PhD research time, 193 00:06:00,040 --> 00:06:01,920 compressed into a single year. 194 00:06:01,920 --> 00:06:03,680 That's essentially what alpha fold achieved. 195 00:06:03,680 --> 00:06:05,480 That's essentially what alpha fold achieved. 196 00:06:05,480 --> 00:06:08,520 It predicted the structure of all 200 million. 197 00:06:08,520 --> 00:06:09,280 200 million. 198 00:06:09,280 --> 00:06:12,720 200 million known proteins. 199 00:06:12,720 --> 00:06:13,320 In a year. 200 00:06:13,320 --> 00:06:14,080 In a year. 201 00:06:14,080 --> 00:06:15,120 That's just mind-blowing. 202 00:06:15,120 --> 00:06:15,680 It is. 203 00:06:15,680 --> 00:06:16,520 It is. 204 00:06:16,520 --> 00:06:20,280 But we're talking about tiny, complex structures here. 205 00:06:20,280 --> 00:06:23,360 How do we even know if those predictions are accurate? 206 00:06:23,360 --> 00:06:24,680 That's a good question. 207 00:06:24,680 --> 00:06:26,840 They compared alpha fold's predictions 208 00:06:26,840 --> 00:06:28,840 to experimentally determined structures. 209 00:06:28,840 --> 00:06:29,280 And? 210 00:06:29,280 --> 00:06:30,280 The results. 211 00:06:30,280 --> 00:06:33,240 Astonishingly accurate, matching experimental data 212 00:06:33,240 --> 00:06:35,920 down to the atomic level. 213 00:06:35,920 --> 00:06:39,040 So alpha fold isn't just spitting out random shapes. 214 00:06:39,040 --> 00:06:39,520 Nope. 215 00:06:39,520 --> 00:06:41,880 It's generating these really precise models 216 00:06:41,880 --> 00:06:43,160 that reflect reality. 217 00:06:43,160 --> 00:06:43,920 Exactly. 218 00:06:43,920 --> 00:06:46,240 And to really grasp the magnitude of this, 219 00:06:46,240 --> 00:06:47,240 we need to visualize it. 220 00:06:47,240 --> 00:06:50,520 We've got some images of these protein structures, 221 00:06:50,520 --> 00:06:53,240 both as predicted by alpha fold and as determined 222 00:06:53,240 --> 00:06:54,480 through experiments. 223 00:06:54,480 --> 00:06:56,560 OK, I'm looking at these images now. 224 00:06:56,560 --> 00:06:59,080 And wow, they're beautiful. 225 00:06:59,080 --> 00:07:00,520 Like intricate works of art. 226 00:07:00,520 --> 00:07:03,720 I can see why it would take a billion years to figure this out 227 00:07:03,720 --> 00:07:04,920 using traditional methods. 228 00:07:04,920 --> 00:07:06,600 The complexity is breathtaking. 229 00:07:06,600 --> 00:07:08,960 And the fact that alpha fold can predict these structures 230 00:07:08,960 --> 00:07:11,080 so accurately, well, it's a game changer. 231 00:07:11,080 --> 00:07:11,680 For biology. 232 00:07:11,680 --> 00:07:12,640 For biology. 233 00:07:12,640 --> 00:07:13,480 And medicine. 234 00:07:13,480 --> 00:07:14,040 Yeah. 235 00:07:14,040 --> 00:07:16,120 I mean, this has to have huge implications for drug 236 00:07:16,120 --> 00:07:18,240 discovery through understanding diseases, 237 00:07:18,240 --> 00:07:20,440 and maybe even curing diseases that have plagued us 238 00:07:20,440 --> 00:07:21,280 for centuries. 239 00:07:21,280 --> 00:07:22,120 Absolutely. 240 00:07:22,120 --> 00:07:24,680 It's the perfect example of how AI 241 00:07:24,680 --> 00:07:28,240 can accelerate scientific progress in ways 242 00:07:28,240 --> 00:07:29,480 we never thought possible. 243 00:07:29,480 --> 00:07:32,480 And the crazy thing is this might just be the beginning. 244 00:07:32,480 --> 00:07:32,800 Yeah. 245 00:07:32,800 --> 00:07:35,000 Alpha fold could just be one example 246 00:07:35,000 --> 00:07:37,800 of what AI can unlock in the scientific world. 247 00:07:37,800 --> 00:07:40,600 So if you take a step back and look at the big picture, 248 00:07:40,600 --> 00:07:43,720 alpha fold is just one piece of the puzzle 249 00:07:43,720 --> 00:07:46,560 when it comes to AI's potential to solve what 250 00:07:46,560 --> 00:07:49,160 Hasabah calls root node problems. 251 00:07:49,160 --> 00:07:50,640 Root node problems. 252 00:07:50,640 --> 00:07:52,840 He uses this really interesting analogy 253 00:07:52,840 --> 00:07:54,760 of a tree of knowledge. 254 00:07:54,760 --> 00:07:56,000 I like that analogy. 255 00:07:56,000 --> 00:07:56,560 Yeah. 256 00:07:56,560 --> 00:07:58,840 So how does this tree of knowledge work? 257 00:07:58,840 --> 00:07:59,160 OK. 258 00:07:59,160 --> 00:08:01,680 So imagine all the knowledge in the universe 259 00:08:01,680 --> 00:08:05,080 represented as this vast sprawling tree. 260 00:08:05,080 --> 00:08:05,480 OK. 261 00:08:05,480 --> 00:08:07,880 With branches reaching out in countless directions, 262 00:08:07,880 --> 00:08:09,920 representing different fields of knowledge, 263 00:08:09,920 --> 00:08:13,040 the knowledge we have today, just a tiny fraction 264 00:08:13,040 --> 00:08:14,160 of this vast tree. 265 00:08:14,160 --> 00:08:16,160 So we've only explored a few branches, 266 00:08:16,160 --> 00:08:18,640 and it's a whole mass of trees still waiting to be discovered. 267 00:08:18,640 --> 00:08:19,520 Exactly. 268 00:08:19,520 --> 00:08:20,020 OK. 269 00:08:20,020 --> 00:08:22,520 And Hasabah sees AI as the tool that 270 00:08:22,520 --> 00:08:25,480 can help us explore that entire tree, 271 00:08:25,480 --> 00:08:27,320 uncovering branches of knowledge. 272 00:08:27,320 --> 00:08:28,440 That we haven't even imagined yet. 273 00:08:28,440 --> 00:08:30,160 That we haven't even imagined yet. 274 00:08:30,160 --> 00:08:32,480 And root node problems are the keys 275 00:08:32,480 --> 00:08:34,080 to unlocking those new branches. 276 00:08:34,080 --> 00:08:34,920 Exactly. 277 00:08:34,920 --> 00:08:36,840 Solving a root node problem opens up 278 00:08:36,840 --> 00:08:39,040 a whole new branch of that tree, leading 279 00:08:39,040 --> 00:08:42,280 to vast new areas of research and discovery. 280 00:08:42,280 --> 00:08:45,640 So like the protein folding problem, right? 281 00:08:45,640 --> 00:08:49,280 By solving that, AlphaFold essentially unlocked 282 00:08:49,280 --> 00:08:51,800 a whole new branch of biology and medicine. 283 00:08:51,800 --> 00:08:52,920 Precisely. 284 00:08:52,920 --> 00:08:55,880 And Hasabah believes there are countless other root node 285 00:08:55,880 --> 00:08:58,120 problems out there waiting to be solved. 286 00:08:58,120 --> 00:08:59,120 In all sorts of fields. 287 00:08:59,120 --> 00:09:02,560 In all sorts of fields, physics, cosmology, energy, 288 00:09:02,560 --> 00:09:03,720 materials, science. 289 00:09:03,720 --> 00:09:06,400 I mean, who knows what breakthroughs await us. 290 00:09:06,400 --> 00:09:08,880 This is getting me really excited about the possibilities. 291 00:09:08,880 --> 00:09:12,320 AI could revolutionize the way we approach scientific discovery. 292 00:09:12,320 --> 00:09:12,880 It could. 293 00:09:12,880 --> 00:09:15,840 But you know, with great power comes great responsibility, 294 00:09:15,840 --> 00:09:16,080 right? 295 00:09:16,080 --> 00:09:17,400 Of course, yeah. 296 00:09:17,400 --> 00:09:21,200 As we delve deeper into this world of AI-driven science, 297 00:09:21,200 --> 00:09:24,240 there are some ethical considerations we can't ignore. 298 00:09:24,240 --> 00:09:25,160 That's a good point. 299 00:09:25,160 --> 00:09:27,640 We can't just get caught up in the excitement of all this 300 00:09:27,640 --> 00:09:30,720 discovery and forget about the potential downsides. 301 00:09:30,720 --> 00:09:31,600 Exactly. 302 00:09:31,600 --> 00:09:35,040 And Hasabah himself highlights one of these concerns, 303 00:09:35,040 --> 00:09:37,040 something he calls the Mollok trap. 304 00:09:37,040 --> 00:09:38,760 The Mollok trap. 305 00:09:38,760 --> 00:09:40,320 OK, that sounds a little ominous. 306 00:09:40,320 --> 00:09:41,160 What is that? 307 00:09:41,160 --> 00:09:45,360 Well, it refers to the risk that companies driven 308 00:09:45,360 --> 00:09:49,200 by intense competition may be pushed to develop AI 309 00:09:49,200 --> 00:09:51,080 in potentially harmful ways. 310 00:09:51,080 --> 00:09:53,240 Even if it goes against their better judgment. 311 00:09:53,240 --> 00:09:55,560 Even if those actions go against the better judgment 312 00:09:55,560 --> 00:09:57,240 of the individuals involved. 313 00:09:57,240 --> 00:10:00,360 So it's like a runaway train where everyone is on board, 314 00:10:00,360 --> 00:10:03,680 but no one's quite sure how to apply the brakes. 315 00:10:03,680 --> 00:10:04,800 That's a great analogy. 316 00:10:04,800 --> 00:10:05,280 Thanks. 317 00:10:05,280 --> 00:10:09,080 And this is precisely why Hasabah emphasizes the need 318 00:10:09,080 --> 00:10:12,800 for collaboration, thoughtful development, ethical frameworks. 319 00:10:12,800 --> 00:10:16,560 To ensure that AI is developed and used safely for all of humanity. 320 00:10:16,560 --> 00:10:17,840 For all of humanity. 321 00:10:17,840 --> 00:10:19,200 Yeah, that makes a lot of sense. 322 00:10:19,200 --> 00:10:21,680 But there's another factor that worries me. 323 00:10:21,680 --> 00:10:26,000 The sheer scale of resources being poured into AI development, 324 00:10:26,000 --> 00:10:28,080 particularly by those tech giants we were talking about, 325 00:10:28,080 --> 00:10:30,760 we're talking billions, even hundreds of billions of dollars. 326 00:10:30,760 --> 00:10:31,040 I know. 327 00:10:31,040 --> 00:10:33,760 Massive data centers, supercomputers. 328 00:10:33,760 --> 00:10:35,520 It's just staggering. 329 00:10:35,520 --> 00:10:40,400 And I wonder if this intense competition might lead to someone 330 00:10:40,400 --> 00:10:42,080 taking a shortcut. 331 00:10:42,080 --> 00:10:44,840 Pushing an AI system too far, too fast, 332 00:10:44,840 --> 00:10:46,480 in an attempt to gain an edge. 333 00:10:46,480 --> 00:10:48,360 It's like a high stakes poker game, 334 00:10:48,360 --> 00:10:50,680 where the players are tempted to go all in, 335 00:10:50,680 --> 00:10:53,720 even if they're not sure what cards the other players are holding. 336 00:10:53,720 --> 00:10:54,960 That's a good analogy. 337 00:10:54,960 --> 00:10:57,120 And Hasabah recognizes this risk. 338 00:10:57,120 --> 00:10:57,720 He does. 339 00:10:57,720 --> 00:10:59,240 He does, yeah. 340 00:10:59,240 --> 00:11:02,360 He suggests we need to be proactive in establishing safeguards 341 00:11:02,360 --> 00:11:04,120 to prevent those kinds of scenarios. 342 00:11:04,120 --> 00:11:05,360 OK, so how do we do that? 343 00:11:05,360 --> 00:11:09,280 He proposes that instead of trying to bolt on safety measures 344 00:11:09,280 --> 00:11:13,920 after the fact, we should focus on developing safe and robust AI 345 00:11:13,920 --> 00:11:15,640 architectures from the outset. 346 00:11:15,640 --> 00:11:17,600 So building in safety from the ground up? 347 00:11:17,600 --> 00:11:18,680 From the ground up. 348 00:11:18,680 --> 00:11:20,800 Ensuring that these systems are inherently designed 349 00:11:20,800 --> 00:11:25,000 to operate within certain ethical and safety parameters. 350 00:11:25,000 --> 00:11:27,920 He even suggests it might be possible to develop AI 351 00:11:27,920 --> 00:11:30,920 architectures with mathematical or practical guarantees 352 00:11:30,920 --> 00:11:32,160 around their behavior. 353 00:11:32,160 --> 00:11:35,120 So almost like a guarantee that they'll operate as intended. 354 00:11:35,120 --> 00:11:35,720 Right. 355 00:11:35,720 --> 00:11:37,720 Providing a higher level of assurance? 356 00:11:37,720 --> 00:11:39,800 That would certainly alleviate some of the fears 357 00:11:39,800 --> 00:11:42,640 about AI going rogue or something like that. 358 00:11:42,640 --> 00:11:43,480 Right. 359 00:11:43,480 --> 00:11:45,120 It's a complex challenge, but one 360 00:11:45,120 --> 00:11:47,400 that Hasabah believes we can overcome. 361 00:11:47,400 --> 00:11:48,400 Through collaboration. 362 00:11:48,400 --> 00:11:50,960 Through collaboration, careful planning, 363 00:11:50,960 --> 00:11:55,280 and a constant focus on those ethical considerations. 364 00:11:55,280 --> 00:11:56,080 Yeah. 365 00:11:56,080 --> 00:11:59,080 He remains optimistic about the future of AI. 366 00:11:59,080 --> 00:12:01,560 Seeing it as this powerful tool that 367 00:12:01,560 --> 00:12:04,560 can help us unlock extraordinary possibilities. 368 00:12:04,560 --> 00:12:06,240 So what are some of those possibilities? 369 00:12:06,240 --> 00:12:09,760 What's the big picture vision that fuels his optimism? 370 00:12:09,760 --> 00:12:12,160 Well, remember that Tree of Knowledge analogy we talked about? 371 00:12:12,160 --> 00:12:12,660 Yeah. 372 00:12:12,660 --> 00:12:15,440 He sees AI as the ultimate explorer, 373 00:12:15,440 --> 00:12:18,240 charting these uncharted territories of knowledge, 374 00:12:18,240 --> 00:12:20,080 pushing the boundaries of human understanding. 375 00:12:20,080 --> 00:12:20,580 Right. 376 00:12:20,580 --> 00:12:24,240 The idea that AI can help us discover whole new branches 377 00:12:24,240 --> 00:12:26,000 of knowledge that we haven't even imagined yet. 378 00:12:26,000 --> 00:12:26,640 Exactly. 379 00:12:26,640 --> 00:12:29,920 And he believes AI can help us tackle those root node 380 00:12:29,920 --> 00:12:33,600 problems, those fundamental challenges that once solved 381 00:12:33,600 --> 00:12:36,800 open up whole new fields of research and discovery, 382 00:12:36,800 --> 00:12:39,400 like alpha fold solving the protein folding problem. 383 00:12:39,400 --> 00:12:41,960 Which has the potential to revolutionize medicine. 384 00:12:41,960 --> 00:12:43,080 Exactly. 385 00:12:43,080 --> 00:12:46,280 And Hasabah believes there are countless other root node 386 00:12:46,280 --> 00:12:49,200 problems out there in fields ranging from physics 387 00:12:49,200 --> 00:12:52,440 and cosmology to energy and material science. 388 00:12:52,440 --> 00:12:54,560 He even talks about using AI to tackle 389 00:12:54,560 --> 00:12:58,000 some of the biggest questions humanity has ever pondered. 390 00:12:58,000 --> 00:12:58,960 Like what kinds of questions? 391 00:12:58,960 --> 00:13:01,440 Questions about the fundamental nature of reality. 392 00:13:01,440 --> 00:13:01,920 Whoa. 393 00:13:01,920 --> 00:13:03,640 The origins of the universe. 394 00:13:03,640 --> 00:13:05,760 Maybe even the meaning of life itself. 395 00:13:05,760 --> 00:13:06,260 Wow. 396 00:13:06,260 --> 00:13:08,200 Those are some truly grand aspirations. 397 00:13:08,200 --> 00:13:09,720 It's inspiring. 398 00:13:09,720 --> 00:13:13,040 But also a little daunting to think about the scale 399 00:13:13,040 --> 00:13:14,040 of what he's envisioning. 400 00:13:14,040 --> 00:13:15,120 It is. 401 00:13:15,120 --> 00:13:18,240 But Hasabah truly believes that AI, if developed and used 402 00:13:18,240 --> 00:13:21,000 responsibly, can be the key to unlocking answers 403 00:13:21,000 --> 00:13:22,920 to these profound questions. 404 00:13:22,920 --> 00:13:24,960 He talks about a future of radical abundance, 405 00:13:24,960 --> 00:13:27,320 a future where disease is eradicated, 406 00:13:27,320 --> 00:13:29,560 consciousness expands beyond Earth, 407 00:13:29,560 --> 00:13:31,800 and human potential reaches new heights. 408 00:13:31,800 --> 00:13:34,800 That's a vision that sparks both awe and, I have to admit, 409 00:13:34,800 --> 00:13:36,280 a bit of healthy skepticism. 410 00:13:36,280 --> 00:13:36,880 Of course. 411 00:13:36,880 --> 00:13:39,520 It's important to be excited about the possibilities, 412 00:13:39,520 --> 00:13:41,640 but also to remain grounded and aware 413 00:13:41,640 --> 00:13:43,040 of the potential risks. 414 00:13:43,040 --> 00:13:46,960 So as we stand at this crossroads, 415 00:13:46,960 --> 00:13:50,520 with AI rapidly advancing and its potential impact 416 00:13:50,520 --> 00:13:53,600 on humanity becoming increasingly clear, 417 00:13:53,600 --> 00:13:55,200 what's the takeaway for our listeners? 418 00:13:55,200 --> 00:13:58,560 What should they be thinking about as they navigate 419 00:13:58,560 --> 00:14:01,200 this complex and ever-evolving landscape? 420 00:14:01,200 --> 00:14:02,480 That's a great question. 421 00:14:02,480 --> 00:14:05,320 I think the key is to approach AI with a sense 422 00:14:05,320 --> 00:14:07,000 of informed optimism. 423 00:14:07,000 --> 00:14:08,080 Informed optimism. 424 00:14:08,080 --> 00:14:08,580 Yeah. 425 00:14:08,580 --> 00:14:11,320 We need to be aware of the potential risks, 426 00:14:11,320 --> 00:14:14,640 but also open to the extraordinary possibilities 427 00:14:14,640 --> 00:14:15,560 that AI offers. 428 00:14:15,560 --> 00:14:18,320 So embrace the potential while also advocating 429 00:14:18,320 --> 00:14:21,480 for responsible development and those ethical considerations. 430 00:14:21,480 --> 00:14:22,040 Exactly. 431 00:14:22,040 --> 00:14:23,320 It's a balancing act. 432 00:14:23,320 --> 00:14:26,240 But a crucial one if we want to harness the power of AI 433 00:14:26,240 --> 00:14:28,120 for the benefit of all humankind. 434 00:14:28,120 --> 00:14:28,560 Yeah. 435 00:14:28,560 --> 00:14:31,480 It's like finding that balance between pushing 436 00:14:31,480 --> 00:14:34,520 the boundaries of AI, but also ensuring it's 437 00:14:34,520 --> 00:14:35,520 developed responsibly. 438 00:14:35,520 --> 00:14:37,360 That seems like it's going to be one of the big challenges 439 00:14:37,360 --> 00:14:37,920 moving forward. 440 00:14:37,920 --> 00:14:39,560 Oh, it's a tight rope walk, for sure. 441 00:14:39,560 --> 00:14:42,640 And the rapid release of these powerful AI models 442 00:14:42,640 --> 00:14:45,520 like ChatGPT, that really brought this whole tension 443 00:14:45,520 --> 00:14:46,800 into focus, I think. 444 00:14:46,800 --> 00:14:47,120 Yeah. 445 00:14:47,120 --> 00:14:48,920 It felt like it came out of nowhere. 446 00:14:48,920 --> 00:14:51,320 And all of a sudden, everyone's talking about AI, 447 00:14:51,320 --> 00:14:52,920 experimenting with it. 448 00:14:52,920 --> 00:14:55,160 Realizing its potential in a whole new way. 449 00:14:55,160 --> 00:14:57,000 It was a turning point, for sure. 450 00:14:57,000 --> 00:14:59,800 I mean, AI went from this kind of niche topic 451 00:14:59,800 --> 00:15:03,200 to a mainstream phenomenon almost overnight. 452 00:15:03,200 --> 00:15:05,360 And it definitely shook things up in the tech world. 453 00:15:05,360 --> 00:15:08,680 Like Microsoft, with its backing of open AI, 454 00:15:08,680 --> 00:15:11,760 was looking to, I don't know, challenge 455 00:15:11,760 --> 00:15:13,480 Google's dominance in search. 456 00:15:13,480 --> 00:15:13,960 Oh, yeah. 457 00:15:13,960 --> 00:15:16,000 And they certainly made Google scramble to catch up. 458 00:15:16,000 --> 00:15:20,360 Google definitely danced as Microsoft's CEO put it. 459 00:15:20,360 --> 00:15:20,600 Yeah. 460 00:15:20,600 --> 00:15:23,120 Suddenly, they're rushing to release their own AI products 461 00:15:23,120 --> 00:15:24,000 like Gemini. 462 00:15:24,000 --> 00:15:27,280 It felt like that MoLok trap in action. 463 00:15:27,280 --> 00:15:29,720 With these tech giants being driven by competition 464 00:15:29,720 --> 00:15:32,560 to push the boundaries of AI development, 465 00:15:32,560 --> 00:15:34,160 maybe faster than they would have otherwise. 466 00:15:34,160 --> 00:15:35,840 Yeah, it's an interesting dilemma, isn't it? 467 00:15:35,840 --> 00:15:38,200 This rapid progress, it's exciting. 468 00:15:38,200 --> 00:15:40,800 But it also raises some valid concerns. 469 00:15:40,800 --> 00:15:43,560 Even Hasada acknowledges that these systems, 470 00:15:43,560 --> 00:15:46,360 as impressive as they are, still have flaws. 471 00:15:46,360 --> 00:15:46,840 Like what? 472 00:15:46,840 --> 00:15:50,760 Like generating incorrect or misleading information, 473 00:15:50,760 --> 00:15:52,800 those what people call hallucinations. 474 00:15:52,800 --> 00:15:53,400 Right. 475 00:15:53,400 --> 00:15:55,600 And those hallucinations can be a real problem, 476 00:15:55,600 --> 00:15:57,840 especially if people start relying on these AI 477 00:15:57,840 --> 00:16:02,160 systems for critical information without understanding 478 00:16:02,160 --> 00:16:03,080 their limitations. 479 00:16:03,080 --> 00:16:04,200 Exactly. 480 00:16:04,200 --> 00:16:06,200 And then there's the whole open source debate, 481 00:16:06,200 --> 00:16:09,480 like Meta, for example, released its own large language model 482 00:16:09,480 --> 00:16:12,640 as open source, making the technology freely available 483 00:16:12,640 --> 00:16:14,560 for anyone to use and modify. 484 00:16:14,560 --> 00:16:17,880 That sounds both incredibly democratizing 485 00:16:17,880 --> 00:16:20,440 and potentially terrifying, all at the same time. 486 00:16:20,440 --> 00:16:23,080 Yeah, it's a double-edged sword, for sure. 487 00:16:23,080 --> 00:16:25,200 On one hand, you have the potential 488 00:16:25,200 --> 00:16:28,400 to accelerate innovation, make this powerful technology 489 00:16:28,400 --> 00:16:30,680 accessible to a wider range of people. 490 00:16:30,680 --> 00:16:34,160 But on the other hand, you have concerns about safety, 491 00:16:34,160 --> 00:16:36,760 control, the potential for misuse. 492 00:16:36,760 --> 00:16:38,160 It's a lot to grapple with. 493 00:16:38,160 --> 00:16:40,480 So given all these complexities, like how do we move forward? 494 00:16:40,480 --> 00:16:43,200 How do we ensure that AI is developed and used responsibly 495 00:16:43,200 --> 00:16:47,560 as we enter this new era of increasingly powerful systems? 496 00:16:47,560 --> 00:16:49,840 Well, Hussabus believes that collaboration is key. 497 00:16:49,840 --> 00:16:50,640 Collaboration. 498 00:16:50,640 --> 00:16:53,160 Especially as we get closer to developing 499 00:16:53,160 --> 00:16:55,720 artificial general intelligence, or AGI. 500 00:16:55,720 --> 00:16:59,000 AGI, that's the idea of AI that can perform 501 00:16:59,000 --> 00:17:01,560 any intellectual task a human can. 502 00:17:01,560 --> 00:17:04,520 So essentially, AI that's as smart as or even smarter 503 00:17:04,520 --> 00:17:05,560 than humans. 504 00:17:05,560 --> 00:17:06,640 Exactly. 505 00:17:06,640 --> 00:17:09,600 And Hussabus argues that as we get closer to that point, 506 00:17:09,600 --> 00:17:12,640 we need to shift our mindset from competition 507 00:17:12,640 --> 00:17:14,280 to cooperation. 508 00:17:14,280 --> 00:17:16,120 I could see why that would be important. 509 00:17:16,120 --> 00:17:17,680 But is it realistic? 510 00:17:17,680 --> 00:17:20,600 In a world that's so driven by technological advancement 511 00:17:20,600 --> 00:17:22,880 and economic competition, can we really 512 00:17:22,880 --> 00:17:25,960 expect companies, countries even, to just put aside 513 00:17:25,960 --> 00:17:27,440 their rivalries and work together? 514 00:17:27,440 --> 00:17:30,000 Well, Hussabus believes it's not just idealistic, 515 00:17:30,000 --> 00:17:34,360 it's essential for the safe and beneficial development of AGI. 516 00:17:34,360 --> 00:17:36,960 He points out that many of the leading AI scientists 517 00:17:36,960 --> 00:17:38,720 around the world, they already know each other. 518 00:17:38,720 --> 00:17:39,480 They communicate. 519 00:17:39,480 --> 00:17:41,160 They collaborate to some extent. 520 00:17:41,160 --> 00:17:43,320 So there's already a foundation of trust and shared 521 00:17:43,320 --> 00:17:44,360 understanding in place. 522 00:17:44,360 --> 00:17:45,080 Right. 523 00:17:45,080 --> 00:17:47,480 And Hussabus argues that that foundation needs 524 00:17:47,480 --> 00:17:50,600 to be strengthened and expanded to include governments, 525 00:17:50,600 --> 00:17:52,400 civil society, academia. 526 00:17:52,400 --> 00:17:53,120 On the stakeholders. 527 00:17:53,120 --> 00:17:54,360 All the stakeholders. 528 00:17:54,360 --> 00:17:57,600 He sees this as a crucial step towards establishing 529 00:17:57,600 --> 00:18:01,320 those ethical frameworks, guidelines for AI development. 530 00:18:01,320 --> 00:18:05,280 So it's like creating a global set of rules of the road for AI, 531 00:18:05,280 --> 00:18:08,400 ensuring that everyone's driving on the same side, 532 00:18:08,400 --> 00:18:10,000 following the same traffic signals. 533 00:18:10,000 --> 00:18:11,120 Exactly. 534 00:18:11,120 --> 00:18:13,560 We need to ensure that the development and deployment 535 00:18:13,560 --> 00:18:17,560 of these powerful AI systems align with human values 536 00:18:17,560 --> 00:18:20,720 and prioritize safety and well-being. 537 00:18:20,720 --> 00:18:21,720 That makes a lot of sense. 538 00:18:21,720 --> 00:18:24,720 But there's another factor that worries me. 539 00:18:24,720 --> 00:18:29,040 The sheer scale of resources being poured into AI development, 540 00:18:29,040 --> 00:18:31,160 particularly by those tech giants we were talking about. 541 00:18:31,160 --> 00:18:34,920 We're talking billions, even hundreds of billions of dollars 542 00:18:34,920 --> 00:18:38,840 being invested in massive data centers and supercomputers. 543 00:18:38,840 --> 00:18:41,120 Yeah, the scale of investment is staggering. 544 00:18:41,120 --> 00:18:42,560 And it's understandable to wonder 545 00:18:42,560 --> 00:18:45,360 if all this intense competition might lead to someone 546 00:18:45,360 --> 00:18:49,520 taking a shortcut, pushing an AI system too far too fast 547 00:18:49,520 --> 00:18:51,320 in an attempt to gain an edge. 548 00:18:51,320 --> 00:18:52,760 It's like a high stakes poker game 549 00:18:52,760 --> 00:18:54,880 where the players are tempted to go all in, 550 00:18:54,880 --> 00:18:56,960 even if they're not sure what cards the other players are 551 00:18:56,960 --> 00:18:57,640 holding. 552 00:18:57,640 --> 00:18:58,920 That's a good analogy. 553 00:18:58,920 --> 00:19:01,000 And Hussabus recognizes this risk. 554 00:19:01,000 --> 00:19:01,760 He does. 555 00:19:01,760 --> 00:19:04,600 He suggests that we need to be proactive in establishing 556 00:19:04,600 --> 00:19:07,680 safeguards to prevent those kinds of scenarios. 557 00:19:07,680 --> 00:19:09,160 Like what kinds of safeguards? 558 00:19:09,160 --> 00:19:11,440 Well, he proposes that instead of trying 559 00:19:11,440 --> 00:19:15,240 to bolt on saty measures after the fact, 560 00:19:15,240 --> 00:19:18,440 we should focus on developing safe and robust AI 561 00:19:18,440 --> 00:19:20,960 architectures from the outset. 562 00:19:20,960 --> 00:19:22,760 So building in safety from the ground up? 563 00:19:22,760 --> 00:19:23,840 From the ground up. 564 00:19:23,840 --> 00:19:26,160 Ensuring that these systems are inherently designed 565 00:19:26,160 --> 00:19:30,280 to operate within certain ethical and safety parameters. 566 00:19:30,280 --> 00:19:32,840 He even suggests that it might be possible to develop 567 00:19:32,840 --> 00:19:36,560 AI architectures with mathematical or practical 568 00:19:36,560 --> 00:19:38,000 guarantees around their behavior. 569 00:19:38,000 --> 00:19:39,480 So almost like a guarantee that they'll 570 00:19:39,480 --> 00:19:40,560 operate as intended. 571 00:19:40,560 --> 00:19:42,800 Providing a higher level of assurance. 572 00:19:42,800 --> 00:19:44,720 That would certainly alleviate some of the fears 573 00:19:44,720 --> 00:19:47,640 about AI going rogue or something like that. 574 00:19:47,640 --> 00:19:49,640 It's a complex challenge, but one 575 00:19:49,640 --> 00:19:52,720 that Hussabus believes we can overcome 576 00:19:52,720 --> 00:19:57,720 through collaboration, through careful planning, 577 00:19:57,720 --> 00:20:01,080 and a constant focus on those ethical considerations. 578 00:20:01,080 --> 00:20:04,040 He remains optimistic about the future of AI, 579 00:20:04,040 --> 00:20:06,840 seeing it as this powerful tool that can help us unlock 580 00:20:06,840 --> 00:20:08,640 these extraordinary possibilities. 581 00:20:08,640 --> 00:20:10,080 So what are some of those possibilities? 582 00:20:10,080 --> 00:20:12,880 What's the big picture vision that fuels his optimism? 583 00:20:12,880 --> 00:20:15,840 Well, remember that Tree of Knowledge analogy. 584 00:20:15,840 --> 00:20:19,080 He sees AI as the ultimate explorer, 585 00:20:19,080 --> 00:20:22,520 charting these uncharted territories of knowledge, 586 00:20:22,520 --> 00:20:24,800 pushing the boundaries of human understanding. 587 00:20:24,800 --> 00:20:25,300 Right. 588 00:20:25,300 --> 00:20:28,520 The idea that AI can help us discover whole new branches 589 00:20:28,520 --> 00:20:30,520 of knowledge that we haven't even imagined yet. 590 00:20:30,520 --> 00:20:31,840 Exactly. 591 00:20:31,840 --> 00:20:35,160 And he believes that AI can help us tackle those root node 592 00:20:35,160 --> 00:20:39,160 problems, those fundamental challenges that, once solved, 593 00:20:39,160 --> 00:20:42,400 open up whole new fields of research and discovery, 594 00:20:42,400 --> 00:20:45,440 like alpha fold, solving the protein folding problem. 595 00:20:45,440 --> 00:20:48,160 Which has the potential to revolutionize medicine. 596 00:20:48,160 --> 00:20:49,120 Precisely. 597 00:20:49,120 --> 00:20:52,400 And Hussabus believes there are countless other root node 598 00:20:52,400 --> 00:20:56,040 problems out there and feels ranging from physics and cosmology 599 00:20:56,040 --> 00:20:58,480 to energy and material science. 600 00:20:58,480 --> 00:21:00,320 He even talks about using AI to tackle 601 00:21:00,320 --> 00:21:03,240 some of the biggest questions humanity has ever pondered. 602 00:21:03,240 --> 00:21:04,520 Like what kinds of questions? 603 00:21:04,520 --> 00:21:06,920 Questions about the fundamental nature of reality, 604 00:21:06,920 --> 00:21:08,720 the origins of the universe, maybe even 605 00:21:08,720 --> 00:21:10,080 the meaning of life itself. 606 00:21:10,080 --> 00:21:12,400 Wow, those are some truly grand aspirations. 607 00:21:12,400 --> 00:21:14,400 It's inspiring, but also a little daunting 608 00:21:14,400 --> 00:21:16,760 to think about the scale of what he's envisioning. 609 00:21:16,760 --> 00:21:17,840 It is. 610 00:21:17,840 --> 00:21:21,520 But Hussabus truly believes that AI, if developed and used 611 00:21:21,520 --> 00:21:24,640 responsibly, can be the key to unlocking answers 612 00:21:24,640 --> 00:21:26,520 to these profound questions. 613 00:21:26,520 --> 00:21:29,720 He talks about a future of radical abundance, 614 00:21:29,720 --> 00:21:32,160 a future where disease is eradicated, 615 00:21:32,160 --> 00:21:34,360 consciousness expands beyond Earth, 616 00:21:34,360 --> 00:21:36,760 and human potential reaches new heights. 617 00:21:36,760 --> 00:21:38,800 It's a vision that sparks both awe 618 00:21:38,800 --> 00:21:40,880 and I have to admit a bit of healthy skepticism. 619 00:21:40,880 --> 00:21:41,720 Of course. 620 00:21:41,720 --> 00:21:43,840 It's important to be excited about the possibilities, 621 00:21:43,840 --> 00:21:46,040 but also to remain grounded and aware 622 00:21:46,040 --> 00:21:47,200 of the potential risks. 623 00:21:47,200 --> 00:21:49,360 So as we stand at this crossroads, 624 00:21:49,360 --> 00:21:52,160 with AI rapidly advancing and its potential impact 625 00:21:52,160 --> 00:21:54,520 on humanity becoming increasingly clear, 626 00:21:54,520 --> 00:21:56,160 what's the takeaway for our listeners? 627 00:21:56,160 --> 00:21:58,200 What should they be thinking about as they navigate 628 00:21:58,200 --> 00:22:01,160 this complex and ever-evolving landscape? 629 00:22:01,160 --> 00:22:02,080 That's a great question. 630 00:22:02,080 --> 00:22:04,080 I think the key is to approach AI 631 00:22:04,080 --> 00:22:06,280 with a sense of informed optimism. 632 00:22:06,280 --> 00:22:07,240 Informed optimism. 633 00:22:07,240 --> 00:22:08,080 Yeah. 634 00:22:08,080 --> 00:22:09,800 We need to be aware of the potential risks, 635 00:22:09,800 --> 00:22:12,160 but also open to the extraordinary possibilities 636 00:22:12,160 --> 00:22:13,240 that AI offers. 637 00:22:13,240 --> 00:22:16,040 So embrace the potential while also advocating 638 00:22:16,040 --> 00:22:19,840 for responsible development and those ethical considerations. 639 00:22:19,840 --> 00:22:20,680 Exactly. 640 00:22:20,680 --> 00:22:23,000 It's a balancing act, but a crucial one 641 00:22:23,000 --> 00:22:25,600 if we want to harness the power of AI 642 00:22:25,600 --> 00:22:27,600 for the benefit of all humankind. 643 00:22:27,600 --> 00:22:30,840 It's like we're standing on the edge of a whole new era 644 00:22:30,840 --> 00:22:32,600 of exploration. 645 00:22:32,600 --> 00:22:34,960 And AI is the ship that's going to take us there. 646 00:22:34,960 --> 00:22:35,480 I like that. 647 00:22:35,480 --> 00:22:36,920 It's a powerful image. 648 00:22:36,920 --> 00:22:40,600 But as we embark on this journey, it's important to remember, 649 00:22:40,600 --> 00:22:44,120 we all have a role to play in shaping the future of AI. 650 00:22:44,120 --> 00:22:44,440 Right. 651 00:22:44,440 --> 00:22:47,320 It's not just about the big tech companies or research labs. 652 00:22:47,320 --> 00:22:48,720 It's about all of us. 653 00:22:48,720 --> 00:22:50,360 You're engaging in this conversation, 654 00:22:50,360 --> 00:22:52,200 thinking critically about the implications. 655 00:22:52,200 --> 00:22:54,120 It's not just about passively observing. 656 00:22:54,120 --> 00:22:57,560 It's about actively shaping the direction that this all takes. 657 00:22:57,560 --> 00:23:00,720 We need to ask questions, challenge assumptions, 658 00:23:00,720 --> 00:23:02,600 demand transparency and accountability 659 00:23:02,600 --> 00:23:06,080 from those who are developing and deploying these systems. 660 00:23:06,080 --> 00:23:09,680 It's about being informed citizens, not just bystanders. 661 00:23:09,680 --> 00:23:11,400 In this technological revolution that's 662 00:23:11,400 --> 00:23:12,760 happening all around us. 663 00:23:12,760 --> 00:23:15,160 And I think sometimes it's easy to forget 664 00:23:15,160 --> 00:23:19,520 that AI isn't some far-off futuristic concept. 665 00:23:19,520 --> 00:23:20,240 It's already here. 666 00:23:20,240 --> 00:23:22,960 It's already woven into our lives in so many ways. 667 00:23:22,960 --> 00:23:24,920 From the recommendations we see online 668 00:23:24,920 --> 00:23:27,200 to the medical diagnosis we receive, 669 00:23:27,200 --> 00:23:29,960 it's shaping our world whether we realize it or not. 670 00:23:29,960 --> 00:23:34,120 And its influence, it's only going to grow in the years to come. 671 00:23:34,120 --> 00:23:35,600 So we better start paying attention. 672 00:23:35,600 --> 00:23:37,320 We better start paying attention. 673 00:23:37,320 --> 00:23:42,200 We need to ensure that AI is used to enhance human well-being, 674 00:23:42,200 --> 00:23:45,240 to promote fairness, justice, to create a more 675 00:23:45,240 --> 00:23:47,560 sustainable and equitable world. 676 00:23:47,560 --> 00:23:50,880 Those are some pretty lofty goals, but they're essential ones. 677 00:23:50,880 --> 00:23:53,680 If we want to create a future where AI benefits everyone. 678 00:23:53,680 --> 00:23:54,200 They are. 679 00:23:54,200 --> 00:23:56,480 And you know, Sasabas, despite all the challenges 680 00:23:56,480 --> 00:23:59,000 and all the complexities, he remains optimistic. 681 00:23:59,000 --> 00:24:01,800 He really believes that AI, if developed and used wisely, 682 00:24:01,800 --> 00:24:04,800 can be the key to solving some of humanity's greatest 683 00:24:04,800 --> 00:24:08,640 challenges and unlocking this future of extraordinary 684 00:24:08,640 --> 00:24:09,600 possibilities. 685 00:24:09,600 --> 00:24:12,000 So as we wrap up this deep dive into Dimus' 686 00:24:12,000 --> 00:24:14,600 Sasabas' world, his vision for the future of AI, 687 00:24:14,600 --> 00:24:16,600 I'm left with a question for our listeners. 688 00:24:16,600 --> 00:24:19,280 If AI has the potential to be this incredible tool 689 00:24:19,280 --> 00:24:22,320 for exploration and discovery, what root node problems 690 00:24:22,320 --> 00:24:23,320 might it help you solve? 691 00:24:23,320 --> 00:24:26,160 What breakthroughs are waiting to be unlocked in your field, 692 00:24:26,160 --> 00:24:27,960 in your life, in the world around you? 693 00:24:27,960 --> 00:24:30,000 Keep your eyes open because the future of AI 694 00:24:30,000 --> 00:24:31,720 is unfolding right now. 695 00:24:31,720 --> 00:24:57,680 And it's going to be an incredible journey.