1 00:00:00,000 --> 00:00:01,740 Ready to dive into the future. 2 00:00:01,740 --> 00:00:04,000 Today we're looking at Dario Amodei's 3 00:00:04,000 --> 00:00:05,840 Machines of Loving Grace. 4 00:00:05,840 --> 00:00:07,780 He paints a pretty wild picture 5 00:00:07,780 --> 00:00:10,880 of what powerful AI could do, like really soon. 6 00:00:10,880 --> 00:00:13,080 Right, it's not just like little steps. 7 00:00:13,080 --> 00:00:15,760 He's talking about totally changing what's possible, 8 00:00:15,760 --> 00:00:17,600 maybe even while we're still around to see it. 9 00:00:17,600 --> 00:00:21,480 He throws around this idea of a compressed 21st century, 10 00:00:21,480 --> 00:00:24,780 like curing things like cancer, Alzheimer's, 11 00:00:24,780 --> 00:00:26,760 not in 100 years, but 10. 12 00:00:26,760 --> 00:00:28,440 Sounds kinda sci-fi, honestly. 13 00:00:28,440 --> 00:00:29,480 How does that even work? 14 00:00:29,480 --> 00:00:31,960 Does AI magically make scientists smarter? 15 00:00:31,960 --> 00:00:34,520 It's more about how discoveries happen, he thinks. 16 00:00:34,520 --> 00:00:37,400 Like a lot of the big leaps think CRISPR, 17 00:00:37,400 --> 00:00:40,840 or those mRNA vaccines, they totally change the game, right? 18 00:00:40,840 --> 00:00:43,760 Yeah, those aha moments where everything clicks. 19 00:00:43,760 --> 00:00:45,740 Exactly, and Amadeus' point is, 20 00:00:45,740 --> 00:00:47,520 those breakthroughs, looking back, 21 00:00:47,520 --> 00:00:49,480 we could have had them sooner if we connected the dots. 22 00:00:49,480 --> 00:00:51,760 You're saying we had the pieces, just not the big picture. 23 00:00:51,760 --> 00:00:54,900 It's like he says, imagine a country of geniuses, 24 00:00:54,900 --> 00:00:58,000 all in a data center, not one AI, but tons of them, 25 00:00:58,000 --> 00:00:59,880 each one a specialist, working together, 26 00:00:59,880 --> 00:01:04,120 analyzing stuff crazy fast, way beyond what we can do. 27 00:01:04,120 --> 00:01:06,180 So instead of waiting for some lone genius 28 00:01:06,180 --> 00:01:09,600 to put it all together, we get AI to do the heavy lifting. 29 00:01:09,600 --> 00:01:11,280 Sifting through all the research, 30 00:01:11,280 --> 00:01:13,440 making the connections no human could. 31 00:01:13,440 --> 00:01:15,340 Exactly, picture this. 32 00:01:15,340 --> 00:01:17,720 AI reading every single research paper, 33 00:01:17,720 --> 00:01:20,720 finding patterns, links, stuff we'd miss completely. 34 00:01:20,720 --> 00:01:22,480 He thinks this could speed up discoveries 35 00:01:22,480 --> 00:01:25,080 like tenfold, maybe even more. 36 00:01:25,080 --> 00:01:26,860 Warp speed for science. 37 00:01:26,860 --> 00:01:29,760 Okay, but even if we discover new stuff super fast, 38 00:01:29,760 --> 00:01:33,520 what about actually using it, like new treatments, trials, 39 00:01:33,520 --> 00:01:34,360 that all takes time. 40 00:01:34,360 --> 00:01:36,960 Well, for sure, Amodei’s not ignoring the real world, 41 00:01:36,960 --> 00:01:39,240 but even there, he sees AI helping. 42 00:01:39,240 --> 00:01:41,720 Imagine AI that designs clinical trials, 43 00:01:41,720 --> 00:01:45,080 figures out the best options fast, avoids dead ends. 44 00:01:45,080 --> 00:01:46,400 Working smarter, not harder. 45 00:01:46,400 --> 00:01:49,160 Exactly, or think about developing drugs. 46 00:01:49,160 --> 00:01:52,280 AI could simulate how they work at a cellular level, 47 00:01:52,280 --> 00:01:53,800 no more lengthy animal testing. 48 00:01:53,800 --> 00:01:55,080 Okay, I'm seeing the potential here. 49 00:01:55,080 --> 00:01:55,920 And the brain. 50 00:01:55,920 --> 00:01:59,320 Amodei’s really into how AI can change healthcare, biology. 51 00:01:59,320 --> 00:02:02,560 He even suggests we could double our lifespan, seriously. 52 00:02:02,560 --> 00:02:04,200 It's a bold claim, yeah. 53 00:02:04,200 --> 00:02:06,760 But he points to stuff like genomics, gene editing, 54 00:02:06,760 --> 00:02:07,960 all that's advancing. 55 00:02:07,960 --> 00:02:09,720 He thinks AI could supercharge it, 56 00:02:09,720 --> 00:02:12,880 not just cure diseases, but reverse aging itself. 57 00:02:12,880 --> 00:02:15,920 Hold on, reverse aging, that's movie stuff. 58 00:02:15,920 --> 00:02:17,600 What are we talking, immortality now? 59 00:02:17,600 --> 00:02:20,880 It's about figuring out how aging works at the cell level. 60 00:02:20,880 --> 00:02:23,560 Some research suggests it's not inevitable, 61 00:02:23,560 --> 00:02:26,600 more like a program in our cells we could change, 62 00:02:26,600 --> 00:02:27,880 maybe even reverse. 63 00:02:27,880 --> 00:02:29,920 Instead of treating wrinkles, we're going after the source. 64 00:02:29,920 --> 00:02:32,740 Right, he sees a future where we could reprogram ourselves 65 00:02:32,740 --> 00:02:36,380 to fix damage, regrow tissues, basically slow down, 66 00:02:36,380 --> 00:02:38,040 maybe even reverse the aging clock. 67 00:02:38,040 --> 00:02:39,800 Mind blowing, really. 68 00:02:39,800 --> 00:02:42,980 But aging aside, there are other big problems, like cancer. 69 00:02:42,980 --> 00:02:45,400 It mutates so fast, can we even keep up? 70 00:02:45,400 --> 00:02:48,480 That's where this country of geniuses idea shines. 71 00:02:48,480 --> 00:02:51,520 Imagine AI studying cancer cells nonstop, 72 00:02:51,520 --> 00:02:53,720 predicting how they'll mutate and making treatments 73 00:02:53,720 --> 00:02:54,840 that adapt along with it. 74 00:02:54,840 --> 00:02:57,220 Like a never ending arms race, but this time, 75 00:02:57,220 --> 00:02:59,800 we've got AI on our side calling the shots. 76 00:02:59,800 --> 00:03:01,520 Right, and it's not just reacting. 77 00:03:01,520 --> 00:03:05,120 Amodai says AI could spot disease way earlier 78 00:03:05,120 --> 00:03:06,760 before we even notice symptoms. 79 00:03:06,760 --> 00:03:10,200 World's best doctors, but it's AI working 247 80 00:03:10,200 --> 00:03:11,060 behind the scenes. 81 00:03:11,060 --> 00:03:12,800 And because they analyze so much data, 82 00:03:12,800 --> 00:03:14,920 they could catch things human doctors wouldn't. 83 00:03:14,920 --> 00:03:17,760 He thinks we could treat diseases like cancer super early, 84 00:03:17,760 --> 00:03:19,720 way better survival rates. 85 00:03:19,720 --> 00:03:22,680 It all sounds amazing, but is Amodai worried 86 00:03:22,680 --> 00:03:24,080 about any downsides? 87 00:03:24,080 --> 00:03:27,600 All this AI in health care, what about privacy data security? 88 00:03:27,600 --> 00:03:28,760 Totally valid concerns. 89 00:03:28,760 --> 00:03:30,940 We'll definitely get into the ethical stuff later on. 90 00:03:30,940 --> 00:03:32,080 Good to know. 91 00:03:32,080 --> 00:03:36,160 OK, so we've got potential for crazy health care breakthroughs, 92 00:03:36,160 --> 00:03:38,280 but Amodei doesn't stop there. 93 00:03:38,280 --> 00:03:40,400 He's also thinking about the economy, especially 94 00:03:40,400 --> 00:03:41,640 for developing countries. 95 00:03:41,640 --> 00:03:43,640 And that's where it gets even more interesting. 96 00:03:43,640 --> 00:03:47,340 Can AI actually help lift people out of poverty? 97 00:03:47,340 --> 00:03:51,600 He's cautiously optimistic, but there are definitely challenges. 98 00:03:51,600 --> 00:03:55,480 Yeah, he sees it as like AI could be this advisor, 99 00:03:55,480 --> 00:03:58,560 not replacing leaders, but giving them way better info 100 00:03:58,560 --> 00:03:59,680 to make decisions with. 101 00:03:59,680 --> 00:04:01,600 Empowering, not taking over. 102 00:04:01,600 --> 00:04:02,200 I like it. 103 00:04:02,200 --> 00:04:03,880 But what does that actually look like? 104 00:04:03,880 --> 00:04:04,720 Give me an example. 105 00:04:04,720 --> 00:04:06,720 He talks about things like resource allocation, 106 00:04:06,720 --> 00:04:08,100 building infrastructure. 107 00:04:08,100 --> 00:04:10,920 Imagine AI looking at a country's whole economy, 108 00:04:10,920 --> 00:04:12,640 figuring out the best investments, 109 00:04:12,640 --> 00:04:14,920 even predicting problems before they happen. 110 00:04:14,920 --> 00:04:17,240 A crystal ball, but with data to back it up. 111 00:04:17,240 --> 00:04:17,960 Right. 112 00:04:17,960 --> 00:04:19,320 And it's not just numbers. 113 00:04:19,320 --> 00:04:22,320 He says AI could help design those safety net programs, 114 00:04:22,320 --> 00:04:24,600 make sure food and medicine get where they're needed, 115 00:04:24,600 --> 00:04:27,160 even personalize education so everyone 116 00:04:27,160 --> 00:04:28,240 has a chance to succeed. 117 00:04:28,240 --> 00:04:30,800 OK, now we're talking about solving some seriously deep 118 00:04:30,800 --> 00:04:31,840 rooted problems. 119 00:04:31,840 --> 00:04:34,080 Poverty, inequality. 120 00:04:34,080 --> 00:04:35,920 Can AI really make a difference when 121 00:04:35,920 --> 00:04:37,360 there's so much else going on? 122 00:04:37,360 --> 00:04:40,480 Even Amodei admits tech alone isn't enough. 123 00:04:40,480 --> 00:04:43,800 He knows corruption, unstable governments, all that stuff 124 00:04:43,800 --> 00:04:44,960 still gets in the way. 125 00:04:44,960 --> 00:04:48,160 So AI isn't a magic wand. 126 00:04:48,160 --> 00:04:49,240 Then what's the point? 127 00:04:49,240 --> 00:04:50,400 Where does it fit in? 128 00:04:50,400 --> 00:04:53,320 Think of it like it amplifies what's already happening. 129 00:04:53,320 --> 00:04:54,920 Better data, better predictions. 130 00:04:54,920 --> 00:04:57,360 It helps those efforts actually work. 131 00:04:57,360 --> 00:04:59,600 He thinks the developing countries could skip ahead, 132 00:04:59,600 --> 00:05:01,800 do in years what used to take decades. 133 00:05:01,800 --> 00:05:03,640 Fast forwarding progress. 134 00:05:03,640 --> 00:05:05,960 But what are the people who get left behind? 135 00:05:05,960 --> 00:05:08,880 Seems like AI could help those who are already doing OK, 136 00:05:08,880 --> 00:05:10,680 and everyone else falls further behind. 137 00:05:10,680 --> 00:05:12,680 That's the opt out problem, he calls it. 138 00:05:12,680 --> 00:05:14,360 Even if the AI solutions are there, 139 00:05:14,360 --> 00:05:15,880 not everyone benefits equally. 140 00:05:15,880 --> 00:05:18,160 Some might not even have access in the first place. 141 00:05:18,160 --> 00:05:20,440 So how do we avoid this AI divide? 142 00:05:20,440 --> 00:05:22,440 It feels like that could define the whole century. 143 00:05:22,440 --> 00:05:24,760 Amodei is big on countries working together, 144 00:05:24,760 --> 00:05:26,280 especially democracies. 145 00:05:26,280 --> 00:05:29,760 He wants this entente strategy, basically. 146 00:05:29,760 --> 00:05:32,800 Democratic countries setting the rules for AI, 147 00:05:32,800 --> 00:05:35,960 sharing what they know, making sure it's good for everyone. 148 00:05:35,960 --> 00:05:38,240 Makes you wonder about the whole governance side of things. 149 00:05:38,240 --> 00:05:38,740 Yeah. 150 00:05:38,740 --> 00:05:40,720 If AI is going to shape the future, 151 00:05:40,720 --> 00:05:43,080 how do we make sure it's the future we actually want? 152 00:05:43,080 --> 00:05:44,880 And that's where it gets really interesting. 153 00:05:44,880 --> 00:05:47,320 He thinks AI could actually make democracy stronger, 154 00:05:47,320 --> 00:05:49,680 make the world more peaceful. 155 00:05:49,680 --> 00:05:51,240 But we got to be smart about it. 156 00:05:51,240 --> 00:05:51,560 OK. 157 00:05:51,560 --> 00:05:52,760 Now I'm really curious. 158 00:05:52,760 --> 00:05:54,000 How's that supposed to work? 159 00:05:54,000 --> 00:05:55,320 Any specific examples? 160 00:05:55,320 --> 00:05:57,360 He talks about fighting misinformation, 161 00:05:57,360 --> 00:05:59,280 all that propaganda online. 162 00:05:59,280 --> 00:06:03,280 Imagine AI that spots deep fakes, those bot accounts, 163 00:06:03,280 --> 00:06:06,160 all that manipulation stuff right as it's happening. 164 00:06:06,160 --> 00:06:08,800 That's scary relevant right now, with social media blowing up 165 00:06:08,800 --> 00:06:10,560 and everyone believing everything they read. 166 00:06:10,560 --> 00:06:11,240 For sure. 167 00:06:11,240 --> 00:06:14,400 And Amodei says AI could keep governments honest too. 168 00:06:14,400 --> 00:06:17,640 Imagine it analyzing their data, seeing where the money's going, 169 00:06:17,640 --> 00:06:19,040 flagging corruption. 170 00:06:19,040 --> 00:06:20,720 Digital watchdogs, making sure everyone's 171 00:06:20,720 --> 00:06:21,920 playing by the rules. 172 00:06:21,920 --> 00:06:22,840 Kind of, yeah. 173 00:06:22,840 --> 00:06:26,800 But he's also clear AI should help human institutions not 174 00:06:26,800 --> 00:06:28,280 replace them. 175 00:06:28,280 --> 00:06:31,600 We decide what matters, what the AI should prioritize. 176 00:06:31,600 --> 00:06:33,560 It's a tool for us to use. 177 00:06:33,560 --> 00:06:35,160 It's a lot to think about. 178 00:06:35,160 --> 00:06:38,080 Every benefit seems to come with its own set of problems. 179 00:06:38,080 --> 00:06:40,360 But even if we figure all that out, 180 00:06:40,360 --> 00:06:41,800 there's still the big one. 181 00:06:41,800 --> 00:06:44,320 If AI can do almost anything better than us, 182 00:06:44,320 --> 00:06:45,880 what does that make us? 183 00:06:45,880 --> 00:06:46,960 What's our purpose then? 184 00:06:46,960 --> 00:06:48,240 That's the big question, right? 185 00:06:48,240 --> 00:06:52,280 If AI has got this covered, cures, solutions, even 186 00:06:52,280 --> 00:06:55,040 composing music, what are we here for? 187 00:06:55,040 --> 00:06:56,640 Just sit back and enjoy the show. 188 00:06:56,640 --> 00:06:57,960 Could be one way to see it. 189 00:06:57,960 --> 00:06:59,560 But Amodei, he's got a different take. 190 00:06:59,560 --> 00:07:02,080 He says being human isn't just about being productive 191 00:07:02,080 --> 00:07:03,760 or better than machines at stuff. 192 00:07:03,760 --> 00:07:06,800 So what we focus on the stuff AI can't do yet, 193 00:07:06,800 --> 00:07:07,920 like the feeling stuff. 194 00:07:07,920 --> 00:07:09,560 It's about what makes us us. 195 00:07:09,560 --> 00:07:12,800 Creativity, compassion, love, connecting with each other, 196 00:07:12,800 --> 00:07:14,920 even just wanting to learn things because it's cool, 197 00:07:14,920 --> 00:07:16,400 not because it's useful. 198 00:07:16,400 --> 00:07:19,160 Amodei thinks that's where we'll thrive, even more so 199 00:07:19,160 --> 00:07:20,480 when AI takes care of the rest. 200 00:07:20,480 --> 00:07:21,000 Interesting. 201 00:07:21,000 --> 00:07:25,200 Like, AI helps us remember what being human is all about. 202 00:07:25,200 --> 00:07:26,040 But let's be real. 203 00:07:26,040 --> 00:07:27,520 Even if we've got art and feelings, 204 00:07:27,520 --> 00:07:29,160 what about paying the bills? 205 00:07:29,160 --> 00:07:32,760 If AI has got jobs on lock, how is that fair to everyone else? 206 00:07:32,760 --> 00:07:34,400 Big issue, for sure. 207 00:07:34,400 --> 00:07:37,440 Amodei says we might need a whole new economic system. 208 00:07:37,440 --> 00:07:40,680 He even talks about that universal basic income idea. 209 00:07:40,680 --> 00:07:42,600 Everyone gets paid no matter what, 210 00:07:42,600 --> 00:07:44,520 so they can follow their passions, help out, 211 00:07:44,520 --> 00:07:46,920 or just enjoy life more. 212 00:07:46,920 --> 00:07:47,640 Radical, yeah. 213 00:07:47,640 --> 00:07:51,000 But with AI taking jobs, more people are thinking that way. 214 00:07:51,000 --> 00:07:53,280 But even if the money part works out, 215 00:07:53,280 --> 00:07:55,920 can society handle that biggest shift? 216 00:07:55,920 --> 00:07:57,200 It's got to be scary. 217 00:07:57,200 --> 00:07:58,880 This much change, this fast. 218 00:07:58,880 --> 00:08:01,280 Amodei’s big on education, talking about this stuff 219 00:08:01,280 --> 00:08:03,480 openly, not just the techie parts of AI, 220 00:08:03,480 --> 00:08:06,280 but the ethics, the impact on society, all of it. 221 00:08:06,280 --> 00:08:08,600 We've got to teach the next generation to think critically, 222 00:08:08,600 --> 00:08:11,000 make good choices with this powerful tech. 223 00:08:11,000 --> 00:08:14,280 So not just coding skills, but how to be a decent human 224 00:08:14,280 --> 00:08:15,240 in an AI world. 225 00:08:15,240 --> 00:08:15,960 Exactly. 226 00:08:15,960 --> 00:08:19,360 The future isn't set in stone, he says, is what we make it, 227 00:08:19,360 --> 00:08:21,680 through the choices we make, what we value, 228 00:08:21,680 --> 00:08:23,560 being open to what's possible. 229 00:08:23,560 --> 00:08:24,480 I like that. 230 00:08:24,480 --> 00:08:26,080 It's easy to get freaked out about AI, 231 00:08:26,080 --> 00:08:29,560 but Amodei reminds us we have a say in all this. 232 00:08:29,560 --> 00:08:32,960 He compares it to when we invented nukes, kind of, 233 00:08:32,960 --> 00:08:34,280 huge tech leap. 234 00:08:34,280 --> 00:08:36,360 But then we had to choose, destroy ourselves, 235 00:08:36,360 --> 00:08:37,840 or use it for good. 236 00:08:37,840 --> 00:08:40,920 And AI is like that, but on a way bigger scale. 237 00:08:40,920 --> 00:08:42,640 It's not just countries or weapons. 238 00:08:42,640 --> 00:08:45,200 It's what, the future of intelligence itself. 239 00:08:45,200 --> 00:08:46,680 He doesn't pretend to know it all, 240 00:08:46,680 --> 00:08:48,640 but he leaves us with this. 241 00:08:48,640 --> 00:08:50,520 If a brighter future is possible, 242 00:08:50,520 --> 00:08:52,720 what are you going to do to make it happen? 243 00:08:52,720 --> 00:08:53,760 That's a good one to end on. 244 00:08:53,760 --> 00:08:55,320 It's not about predicting the future. 245 00:08:55,320 --> 00:08:56,160 It's making it. 246 00:08:56,160 --> 00:08:59,600 And that's our deep dive into machines of loving grace. 247 00:08:59,600 --> 00:09:02,080 Hopefully, you've got a better sense of Amodei’s vision 248 00:09:02,080 --> 00:09:04,000 and the role AI could play in it. 249 00:09:04,000 --> 00:09:05,360 Definitely got me thinking. 250 00:09:05,360 --> 00:09:06,880 I'm sure we'll be talking about this more 251 00:09:06,880 --> 00:09:09,040 as AI keeps shaking things up. 252 00:09:09,040 --> 00:09:12,000 Until then, keep asking the big questions, 253 00:09:12,000 --> 00:09:35,000 and don't be afraid to explore the possibilities.