1 00:00:00,000 --> 00:00:01,520 Welcome back to the deep dive. 2 00:00:01,520 --> 00:00:03,600 Looks like we're diving into AI today, 3 00:00:03,600 --> 00:00:07,440 a whole stack of excerpts here from the Aesel Tana podcast. 4 00:00:07,440 --> 00:00:09,200 And you know what really stood out to me? 5 00:00:09,200 --> 00:00:10,040 What's that? 6 00:00:10,040 --> 00:00:11,920 All the stuff about the potential risks, 7 00:00:12,800 --> 00:00:16,280 especially what people on the inside are saying. 8 00:00:16,280 --> 00:00:17,160 Yeah. 9 00:00:17,160 --> 00:00:19,240 The ones actually working on this stuff. 10 00:00:19,240 --> 00:00:21,160 Yeah, it definitely hits different 11 00:00:21,160 --> 00:00:22,160 when it's coming from people 12 00:00:22,160 --> 00:00:23,780 who've been in the trenches, right? 13 00:00:23,780 --> 00:00:24,720 Not just theoretical. 14 00:00:24,720 --> 00:00:26,880 We've got folks here who worked at OpenAI, 15 00:00:26,880 --> 00:00:28,080 Meta, those kinds of places. 16 00:00:28,080 --> 00:00:29,080 Exactly. 17 00:00:29,080 --> 00:00:32,440 So we've got some juicy bits from a Senate hearing, right? 18 00:00:32,440 --> 00:00:35,800 Former AI developers, spilling the tea. 19 00:00:35,800 --> 00:00:38,640 And then some follow-up chats from the podcast, 20 00:00:38,640 --> 00:00:39,780 kind of unpacking it all. 21 00:00:39,780 --> 00:00:41,780 Where do we even begin? 22 00:00:41,780 --> 00:00:45,760 Well, there's this one story, more like a metaphor, I guess, 23 00:00:45,760 --> 00:00:47,120 that really captures the whole issue. 24 00:00:47,120 --> 00:00:49,400 They called it the metaphor of the bear. 25 00:00:49,400 --> 00:00:50,240 Oh, okay. 26 00:00:50,240 --> 00:00:51,060 I'm all ears. 27 00:00:51,060 --> 00:00:51,900 Hit me with this bear metaphor. 28 00:00:51,900 --> 00:00:52,740 So picture this. 29 00:00:52,740 --> 00:00:53,760 You're in a forest, right? 30 00:00:53,760 --> 00:00:55,680 Suddenly a bear shows up. 31 00:00:55,680 --> 00:00:57,320 Everyone starts running like crazy. 32 00:00:57,320 --> 00:00:58,160 Makes sense. 33 00:00:58,160 --> 00:00:59,160 Don't want to be bear food. 34 00:00:59,160 --> 00:01:00,440 The thing is, the only thing that matters 35 00:01:00,440 --> 00:01:01,760 is not being the slowest. 36 00:01:01,760 --> 00:01:03,680 Slowest one gets caught. 37 00:01:03,680 --> 00:01:05,480 Okay, I'm starting to see where you're going with this. 38 00:01:05,480 --> 00:01:06,600 Kind of a race to the bottom. 39 00:01:06,600 --> 00:01:07,440 Exactly. 40 00:01:07,440 --> 00:01:08,880 That's how some developers describe 41 00:01:08,880 --> 00:01:09,920 AI development right now. 42 00:01:09,920 --> 00:01:11,080 It's not about being the best. 43 00:01:11,080 --> 00:01:13,220 It's about not being the worst. 44 00:01:13,220 --> 00:01:15,960 Not being the one who makes the biggest mistake. 45 00:01:15,960 --> 00:01:18,960 So not building the safest AI, 46 00:01:18,960 --> 00:01:21,800 just making sure you're not the one 47 00:01:21,800 --> 00:01:24,540 unleashing the digital bear on the world. 48 00:01:24,540 --> 00:01:26,840 That's kind of scary. 49 00:01:26,840 --> 00:01:28,240 Is that really how it works out there? 50 00:01:28,240 --> 00:01:30,840 Or is that an exaggeration? 51 00:01:30,840 --> 00:01:32,200 I think it's closer to the truth 52 00:01:32,200 --> 00:01:33,200 than we'd like to admit. 53 00:01:33,200 --> 00:01:35,240 The pressure is immense. 54 00:01:35,240 --> 00:01:37,280 Everyone wants to be first. 55 00:01:37,280 --> 00:01:39,480 First to market, next big breakthrough. 56 00:01:39,480 --> 00:01:41,760 And sometimes safety takes a backseat. 57 00:01:41,760 --> 00:01:42,780 And that's what these whistleblowers 58 00:01:42,780 --> 00:01:43,920 are trying to tell us, right? 59 00:01:43,920 --> 00:01:45,880 This rush, this race. 60 00:01:45,880 --> 00:01:47,560 It's creating a blind spot. 61 00:01:47,560 --> 00:01:48,400 Absolutely. 62 00:01:48,400 --> 00:01:49,760 And they've got concrete examples. 63 00:01:49,760 --> 00:01:52,840 One guy, William Saunders, used to work at OpenAI, 64 00:01:52,840 --> 00:01:54,120 said he could have just walked out 65 00:01:54,120 --> 00:01:56,240 with their most advanced AI system. 66 00:01:56,240 --> 00:01:57,160 Just taken it. 67 00:01:57,160 --> 00:01:58,120 Wait, seriously? 68 00:01:58,120 --> 00:02:00,160 Just like pocketed it and walked away. 69 00:02:00,160 --> 00:02:01,000 That's insane. 70 00:02:01,000 --> 00:02:02,640 Security was that lax. 71 00:02:02,640 --> 00:02:04,960 Makes you wonder what else they might be missing, right? 72 00:02:04,960 --> 00:02:06,120 No kidding. 73 00:02:06,120 --> 00:02:08,000 So how real is this threat then? 74 00:02:08,000 --> 00:02:11,320 How soon could this bear actually catch up to us? 75 00:02:11,320 --> 00:02:13,700 Any guesses on the timeline for AGI, 76 00:02:13,700 --> 00:02:15,900 that artificial general intelligence thing? 77 00:02:15,900 --> 00:02:17,580 That's the million dollar question. 78 00:02:17,580 --> 00:02:19,440 And there's no easy answer. 79 00:02:19,440 --> 00:02:21,400 Some of these experts, they're saying AGI, 80 00:02:21,400 --> 00:02:23,080 it could be just a few years away. 81 00:02:23,080 --> 00:02:24,560 One to three even. 82 00:02:24,560 --> 00:02:25,920 Others think it's further out. 83 00:02:25,920 --> 00:02:27,080 Maybe a decade or two. 84 00:02:27,080 --> 00:02:30,800 OK, so we're looking at a pretty wide range of possibilities. 85 00:02:30,800 --> 00:02:32,520 But even if it's on the longer end, 86 00:02:32,520 --> 00:02:34,640 the potential consequences are huge, right? 87 00:02:34,640 --> 00:02:36,360 We've got to start thinking about this now. 88 00:02:36,360 --> 00:02:36,880 You got it. 89 00:02:36,880 --> 00:02:39,120 We're talking cyber attacks, even 90 00:02:39,120 --> 00:02:40,920 bioweapons powered by AGI. 91 00:02:40,920 --> 00:02:42,480 This isn't science fiction anymore. 92 00:02:42,480 --> 00:02:44,000 It's something we need to be ready for. 93 00:02:44,000 --> 00:02:46,120 Makes you wonder about the motives of the companies 94 00:02:46,120 --> 00:02:47,120 building this stuff. 95 00:02:47,120 --> 00:02:49,200 But let's talk about the whistleblowers themselves 96 00:02:49,200 --> 00:02:50,200 for a minute. 97 00:02:50,200 --> 00:02:52,240 They're really sticking their necks out here. 98 00:02:52,240 --> 00:02:54,760 What kind of challenges do they face speaking up? 99 00:02:54,760 --> 00:02:57,120 It's not an easy road, that's for sure. 100 00:02:57,120 --> 00:02:58,760 A lot of pressure to keep quiet, even 101 00:02:58,760 --> 00:03:00,120 if they see something wrong. 102 00:03:00,120 --> 00:03:03,080 Many of them sign these non-dispairagement agreements, 103 00:03:03,080 --> 00:03:05,360 you know, part of their contracts. 104 00:03:05,360 --> 00:03:07,200 So even if they see something that 105 00:03:07,200 --> 00:03:10,640 could be a danger to the public, they can't blow anything. 106 00:03:10,640 --> 00:03:11,640 Pretty much. 107 00:03:11,640 --> 00:03:13,520 It's really tough for them to speak freely. 108 00:03:13,520 --> 00:03:15,520 And on top of that, there aren't a lot 109 00:03:15,520 --> 00:03:19,560 of legal protections for whistleblowers in AI 110 00:03:19,560 --> 00:03:20,600 specifically. 111 00:03:20,600 --> 00:03:23,080 You'd think someone warning about dangerous AI 112 00:03:23,080 --> 00:03:24,400 would be protected, right? 113 00:03:24,400 --> 00:03:26,240 I mean, that seems like common sense. 114 00:03:26,240 --> 00:03:28,320 You'd think so, but the law hasn't caught up 115 00:03:28,320 --> 00:03:31,160 to how fast this technology is moving. 116 00:03:31,160 --> 00:03:34,400 A lot of the potential harm these folks are talking about, 117 00:03:34,400 --> 00:03:36,620 it's not even technically illegal yet. 118 00:03:36,620 --> 00:03:38,080 It's this gray area. 119 00:03:38,080 --> 00:03:40,480 They're risking their careers to speak out 120 00:03:40,480 --> 00:03:42,960 about something that might not even be against the rules. 121 00:03:42,960 --> 00:03:44,360 Wow. 122 00:03:44,360 --> 00:03:46,040 That's a tough position to be in. 123 00:03:46,040 --> 00:03:48,600 I bet there are way more people seeing this stuff 124 00:03:48,600 --> 00:03:50,120 but are too afraid to say anything. 125 00:03:50,120 --> 00:03:50,880 Probably. 126 00:03:50,880 --> 00:03:53,280 And then there's this whole thing about regulating AI, 127 00:03:53,280 --> 00:03:55,440 people saying it'll slow us down, especially compared 128 00:03:55,440 --> 00:03:56,080 to China. 129 00:03:56,080 --> 00:03:57,160 What do you think about that? 130 00:03:57,160 --> 00:03:58,700 Yeah, that's the big argument, right? 131 00:03:58,700 --> 00:03:59,880 We can't fall behind. 132 00:03:59,880 --> 00:04:02,040 But one of the experts, Helen Toner, 133 00:04:02,040 --> 00:04:03,600 she brought up an interesting point. 134 00:04:03,600 --> 00:04:06,760 China, they're already regulating their own AI sector. 135 00:04:06,760 --> 00:04:08,380 Pretty actively, actually. 136 00:04:08,380 --> 00:04:11,800 So it's not this free for all everyone thinks it is. 137 00:04:11,800 --> 00:04:12,440 Interesting. 138 00:04:12,440 --> 00:04:15,680 And Toner, she was saying that good regulation, 139 00:04:15,680 --> 00:04:19,880 smart regulation, it can actually boost innovation. 140 00:04:19,880 --> 00:04:22,000 It builds trust. 141 00:04:22,000 --> 00:04:23,720 People more likely to use something 142 00:04:23,720 --> 00:04:25,480 if they feel like it's safe and reliable. 143 00:04:25,480 --> 00:04:26,740 Yeah, that makes sense. 144 00:04:26,740 --> 00:04:28,440 It's like you don't want to build a house 145 00:04:28,440 --> 00:04:29,600 on a shaky foundation. 146 00:04:29,600 --> 00:04:30,160 Exactly. 147 00:04:30,160 --> 00:04:31,960 And it goes back to that bear metaphor, right? 148 00:04:31,960 --> 00:04:34,360 Everyone's so focused on outrunning each other, 149 00:04:34,360 --> 00:04:36,600 who's checking the ground beneath their feet. 150 00:04:36,600 --> 00:04:37,440 Good point. 151 00:04:37,440 --> 00:04:39,720 Someone needs to make sure we're not building this whole AI 152 00:04:39,720 --> 00:04:40,960 thing on shaky ground. 153 00:04:40,960 --> 00:04:41,720 Exactly. 154 00:04:41,720 --> 00:04:43,260 So we need those regulations, right, 155 00:04:43,260 --> 00:04:45,100 to make sure we're building on solid ground. 156 00:04:45,100 --> 00:04:45,800 Definitely. 157 00:04:45,800 --> 00:04:47,200 Ethical and safe. 158 00:04:47,200 --> 00:04:48,960 That's the foundation we need. 159 00:04:48,960 --> 00:04:51,880 OK, now, open source AI. 160 00:04:51,880 --> 00:04:53,480 That's a whole other can of worms, right? 161 00:04:53,480 --> 00:04:54,160 No. 162 00:04:54,160 --> 00:04:56,280 People say it's a way to make AI more democratic, 163 00:04:56,280 --> 00:04:58,400 but it's not that simple, is it? 164 00:04:58,400 --> 00:04:59,280 Not at all. 165 00:04:59,280 --> 00:05:02,040 The thing about open source, once those weights are out 166 00:05:02,040 --> 00:05:03,600 there, they're out there. 167 00:05:03,600 --> 00:05:05,800 No controlling who uses them or how. 168 00:05:05,800 --> 00:05:06,760 Like a recipe. 169 00:05:06,760 --> 00:05:08,960 Once it's shared, anyone can make the dish, 170 00:05:08,960 --> 00:05:10,880 even if they have no idea what they're doing. 171 00:05:10,880 --> 00:05:11,600 Exactly. 172 00:05:11,600 --> 00:05:14,680 Good intentions don't guarantee responsible use. 173 00:05:14,680 --> 00:05:17,880 And then there's jailbreaking. 174 00:05:17,880 --> 00:05:18,560 Jailbreaking. 175 00:05:18,560 --> 00:05:20,260 Yeah, I've heard that term thrown around. 176 00:05:20,260 --> 00:05:21,400 What's that all about? 177 00:05:21,400 --> 00:05:25,880 Basically, people finding ways to bypass the safety 178 00:05:25,880 --> 00:05:28,320 restrictions built into the AI. 179 00:05:28,320 --> 00:05:30,000 Oh, like picking a lock on a door that 180 00:05:30,000 --> 00:05:31,280 was supposed to be secure. 181 00:05:31,280 --> 00:05:32,080 Exactly. 182 00:05:32,080 --> 00:05:34,240 People will push the limits, especially 183 00:05:34,240 --> 00:05:36,320 with tech this powerful. 184 00:05:36,320 --> 00:05:38,360 You build in safeguards, but there's 185 00:05:38,360 --> 00:05:39,840 no guarantee they'll hold. 186 00:05:39,840 --> 00:05:42,080 It's like you're trying to contain something incredibly 187 00:05:42,080 --> 00:05:45,040 powerful, but the container might have holes. 188 00:05:45,040 --> 00:05:46,200 That's a good way to put it. 189 00:05:46,200 --> 00:05:48,680 So open source, it's got his pros and cons. 190 00:05:48,680 --> 00:05:50,600 It can make AI more accessible. 191 00:05:50,600 --> 00:05:53,320 But it also means potentially putting powerful tools 192 00:05:53,320 --> 00:05:54,240 in the wrong hands. 193 00:05:54,240 --> 00:05:55,280 Right, that's the dilemma. 194 00:05:55,280 --> 00:05:56,640 And then there's the whole ethical question, 195 00:05:56,640 --> 00:05:58,960 like who's responsible when something goes wrong. 196 00:05:58,960 --> 00:06:00,960 Yeah, it gets complicated fast. 197 00:06:00,960 --> 00:06:02,000 Absolutely. 198 00:06:02,000 --> 00:06:04,440 We're still figuring out how to use this powerful tool 199 00:06:04,440 --> 00:06:05,520 responsibly. 200 00:06:05,520 --> 00:06:07,440 OK, data privacy. 201 00:06:07,440 --> 00:06:08,400 That's a big one. 202 00:06:08,400 --> 00:06:13,120 Companies like Meta, using our data to train their AI, 203 00:06:13,120 --> 00:06:15,880 especially here in the US, where we've got weaker protections 204 00:06:15,880 --> 00:06:17,560 than, say, the EU. 205 00:06:17,560 --> 00:06:19,120 What were people saying about that? 206 00:06:19,120 --> 00:06:22,120 A lot of concern about transparency or the lack of it. 207 00:06:22,120 --> 00:06:23,600 And control. 208 00:06:23,600 --> 00:06:26,640 Do we really have a say in how our data is used? 209 00:06:26,640 --> 00:06:28,240 There was this example from the UK, 210 00:06:28,240 --> 00:06:29,760 really highlights the issue. 211 00:06:29,760 --> 00:06:33,080 Meta announced they were going to use public Facebook 212 00:06:33,080 --> 00:06:34,240 and Instagram data, right? 213 00:06:34,240 --> 00:06:35,520 Yeah, to do what? 214 00:06:35,520 --> 00:06:38,720 Train their AI to better reflect British culture. 215 00:06:38,720 --> 00:06:40,320 It caused a huge uproar. 216 00:06:40,320 --> 00:06:41,160 I can imagine. 217 00:06:41,160 --> 00:06:42,880 OK, we know our data is being used, 218 00:06:42,880 --> 00:06:45,320 but tied to our nationality. 219 00:06:45,320 --> 00:06:46,240 That feels different. 220 00:06:46,240 --> 00:06:48,480 People felt like their privacy was violated. 221 00:06:48,480 --> 00:06:49,160 Big time. 222 00:06:49,160 --> 00:06:50,880 And they weren't happy about it, especially 223 00:06:50,880 --> 00:06:52,120 without their consent. 224 00:06:52,120 --> 00:06:54,280 It raises the question, do we have any control 225 00:06:54,280 --> 00:06:55,720 over our digital footprint? 226 00:06:55,720 --> 00:06:58,160 And how comfortable are we with this stuff being used to train 227 00:06:58,160 --> 00:06:58,800 AI? 228 00:06:58,800 --> 00:06:59,720 Yeah, it's a lot to think about. 229 00:06:59,720 --> 00:07:01,320 Honestly, listening to all of this, 230 00:07:01,320 --> 00:07:03,280 it can feel a little overwhelming. 231 00:07:03,280 --> 00:07:04,360 I hear you. 232 00:07:04,360 --> 00:07:05,600 It's a lot to process. 233 00:07:05,600 --> 00:07:08,440 The truth is, we're all still learning about this stuff, 234 00:07:08,440 --> 00:07:10,560 what it can do, what the risks are. 235 00:07:10,560 --> 00:07:12,240 But that's why these conversations are 236 00:07:12,240 --> 00:07:13,160 so important, right? 237 00:07:13,160 --> 00:07:13,960 Absolutely. 238 00:07:13,960 --> 00:07:16,160 The more we talk about it, the better prepared we'll be. 239 00:07:16,160 --> 00:07:16,920 Exactly. 240 00:07:16,920 --> 00:07:19,600 And luckily, we've got a whole lot more to unpack here. 241 00:07:19,600 --> 00:07:21,840 OK, so we've got the bare metaphor, 242 00:07:21,840 --> 00:07:24,840 met some interesting folks who are putting it all on the line, 243 00:07:24,840 --> 00:07:28,240 and scratched the surface of data privacy and regulation. 244 00:07:28,240 --> 00:07:29,360 Where do we go from here? 245 00:07:29,360 --> 00:07:32,320 Well, next up, let's dig into some potential solutions. 246 00:07:32,320 --> 00:07:34,840 It's not all doom and gloom. 247 00:07:34,840 --> 00:07:37,040 There are people working hard to figure out 248 00:07:37,040 --> 00:07:37,960 how to do this right. 249 00:07:37,960 --> 00:07:39,600 All right, I'm ready for some good news. 250 00:07:39,600 --> 00:07:40,360 Lead the way. 251 00:07:40,360 --> 00:07:44,560 So after that Senate hearing, the Asel Tana podcast host, 252 00:07:44,560 --> 00:07:46,480 they had Helen Toner on as a guest. 253 00:07:46,480 --> 00:07:48,920 Oh, yeah, she's like the expert on AI policy, right? 254 00:07:48,920 --> 00:07:49,440 Pretty much. 255 00:07:49,440 --> 00:07:51,280 And she had some really interesting thoughts 256 00:07:51,280 --> 00:07:53,120 on how we move forward. 257 00:07:53,120 --> 00:07:56,320 One thing she kept coming back to was transparency, 258 00:07:56,320 --> 00:07:57,800 in AI development, I mean. 259 00:07:57,800 --> 00:07:58,680 Yeah, makes sense. 260 00:07:58,680 --> 00:08:00,760 Like shine a light on what's going on behind the scenes. 261 00:08:00,760 --> 00:08:01,520 Yeah. 262 00:08:01,520 --> 00:08:03,720 Sunlight is the best disinfectant. 263 00:08:03,720 --> 00:08:04,880 That's the idea. 264 00:08:04,880 --> 00:08:06,000 Lift the curtain. 265 00:08:06,000 --> 00:08:08,840 Make companies be more open about their safety practices, 266 00:08:08,840 --> 00:08:11,360 the testing they're doing, what data they're using 267 00:08:11,360 --> 00:08:13,080 to train their AI, all of it. 268 00:08:13,080 --> 00:08:16,280 Kind of like those making of documentaries for movies. 269 00:08:16,280 --> 00:08:17,920 You see all the work that goes into it. 270 00:08:17,920 --> 00:08:20,720 Except, obviously, the stakes are a bit higher with AI 271 00:08:20,720 --> 00:08:21,720 than with a movie, right? 272 00:08:21,720 --> 00:08:22,720 Yeah, no kidding. 273 00:08:22,720 --> 00:08:25,000 And transparency, it builds trust too. 274 00:08:25,000 --> 00:08:28,240 If people understand how AI is being made, what's being done 275 00:08:28,240 --> 00:08:31,960 to keep it safe, they're less likely to be afraid of it. 276 00:08:31,960 --> 00:08:33,560 OK, so open those studio doors. 277 00:08:33,560 --> 00:08:35,480 Let's see what's happening on the AI set. 278 00:08:35,480 --> 00:08:38,560 But transparency is just one piece of the puzzle, right? 279 00:08:38,560 --> 00:08:39,840 What else did they talk about? 280 00:08:39,840 --> 00:08:42,040 Well, another idea that's gaining traction 281 00:08:42,040 --> 00:08:44,040 is independent oversight. 282 00:08:44,040 --> 00:08:46,520 Like separate organizations not connected 283 00:08:46,520 --> 00:08:50,160 to the companies who can review and audit AI systems 284 00:08:50,160 --> 00:08:51,400 before they're released. 285 00:08:51,400 --> 00:08:53,520 So kind of like a safety inspection for a car 286 00:08:53,520 --> 00:08:54,640 before it hits the road. 287 00:08:54,640 --> 00:08:55,880 Make sure it's roadworthy. 288 00:08:55,880 --> 00:08:57,040 Exactly. 289 00:08:57,040 --> 00:09:00,080 And just like car safety, you need standards, right? 290 00:09:00,080 --> 00:09:02,680 Tests to make sure these AI systems are meeting 291 00:09:02,680 --> 00:09:04,640 ethical and safety benchmarks. 292 00:09:04,640 --> 00:09:05,760 Interesting. 293 00:09:05,760 --> 00:09:07,400 What about licensing AI developers? 294 00:09:07,400 --> 00:09:08,860 Have you heard anything about that? 295 00:09:08,860 --> 00:09:11,320 Oh, yeah, that's another one that's being tossed around. 296 00:09:11,320 --> 00:09:13,680 Setting qualifications. 297 00:09:13,680 --> 00:09:16,120 Standards for who can actually build these advanced systems. 298 00:09:16,120 --> 00:09:19,400 Sort of like doctors or lawyers needing a license. 299 00:09:19,400 --> 00:09:22,720 So not just anyone could whip up some powerful AI 300 00:09:22,720 --> 00:09:23,520 in their basement. 301 00:09:23,520 --> 00:09:24,280 Exactly. 302 00:09:24,280 --> 00:09:26,000 You need to show you've got the expertise, 303 00:09:26,000 --> 00:09:27,320 understand the risks. 304 00:09:27,320 --> 00:09:29,280 It could help professionalize the field, 305 00:09:29,280 --> 00:09:31,480 build a culture of responsibility. 306 00:09:31,480 --> 00:09:35,240 It all sounds great in theory, but are policymakers actually 307 00:09:35,240 --> 00:09:37,640 taking any of this seriously? 308 00:09:37,640 --> 00:09:38,840 Or is it just talk? 309 00:09:38,840 --> 00:09:40,800 Well, there are some signs that things are moving. 310 00:09:40,800 --> 00:09:42,740 I mean, the Senate even having that hearing 311 00:09:42,740 --> 00:09:43,760 is a good sign, right? 312 00:09:43,760 --> 00:09:46,080 It means lawmakers are starting to pay attention. 313 00:09:46,080 --> 00:09:46,580 True. 314 00:09:46,580 --> 00:09:47,880 Anything concrete happening? 315 00:09:47,880 --> 00:09:49,920 The EU, they're actually further along. 316 00:09:49,920 --> 00:09:53,240 They've got this AI act that's supposed to be finalized soon. 317 00:09:53,240 --> 00:09:55,560 And it lays out some pretty strict rules, especially 318 00:09:55,560 --> 00:09:58,560 for AI systems that are considered high risk. 319 00:09:58,560 --> 00:10:00,760 So they're actually doing something about it. 320 00:10:00,760 --> 00:10:02,240 Not just talking. 321 00:10:02,240 --> 00:10:03,840 What about the US? 322 00:10:03,840 --> 00:10:05,400 Any movement on our end? 323 00:10:05,400 --> 00:10:07,600 A little slower here, but there are a few bills 324 00:10:07,600 --> 00:10:12,120 being considered focused on AI safety and ethics. 325 00:10:12,120 --> 00:10:14,360 And the White House even put out a blueprint 326 00:10:14,360 --> 00:10:17,280 for an AI Bill of Rights, which is 327 00:10:17,280 --> 00:10:19,200 like a set of guiding principles. 328 00:10:19,200 --> 00:10:19,700 OK. 329 00:10:19,700 --> 00:10:21,360 So at least there's some momentum building. 330 00:10:21,360 --> 00:10:23,960 It feels like we're at a turning point here. 331 00:10:23,960 --> 00:10:26,240 And maybe this is me being cynical. 332 00:10:26,240 --> 00:10:28,760 You always hear that too much regulation is 333 00:10:28,760 --> 00:10:31,160 going to stifle innovation, right? 334 00:10:31,160 --> 00:10:33,820 How do we balance protecting people 335 00:10:33,820 --> 00:10:36,240 with letting technology advance? 336 00:10:36,240 --> 00:10:37,520 That is the question, isn't it? 337 00:10:37,520 --> 00:10:39,720 It's a debate that's been going on forever. 338 00:10:39,720 --> 00:10:41,680 But I think Helen Toner made a good point. 339 00:10:41,680 --> 00:10:43,520 Sometimes a little bit of constraint 340 00:10:43,520 --> 00:10:45,160 can actually lead to more creativity. 341 00:10:45,160 --> 00:10:47,480 It's like, if you give people a clear set of rules, 342 00:10:47,480 --> 00:10:51,360 it can focus their thinking, lead to better solutions. 343 00:10:51,360 --> 00:10:51,880 I like that. 344 00:10:51,880 --> 00:10:53,680 Constraints breed creativity. 345 00:10:53,680 --> 00:10:55,360 So maybe it's not about stopping AI. 346 00:10:55,360 --> 00:10:58,760 It's about giving it the right direction, the right guardrail. 347 00:10:58,760 --> 00:10:59,440 Exactly. 348 00:10:59,440 --> 00:11:01,520 Steer it in a safe and beneficial way. 349 00:11:01,520 --> 00:11:04,000 That's what those whistleblowers are pushing for, right? 350 00:11:04,000 --> 00:11:06,320 Not to shut it all down, but to do it responsibly, 351 00:11:06,320 --> 00:11:08,800 with safety and ethics at the forefront. 352 00:11:08,800 --> 00:11:10,480 OK, so we've talked about solutions. 353 00:11:10,480 --> 00:11:12,960 Sounds like things are moving, even if it's slow. 354 00:11:12,960 --> 00:11:15,080 But let's zoom out for a second. 355 00:11:15,080 --> 00:11:19,280 Why should the average person even care about all of this? 356 00:11:19,280 --> 00:11:21,280 How does this impact everyday life? 357 00:11:21,280 --> 00:11:23,600 That's a really important question. 358 00:11:23,600 --> 00:11:26,880 And the answer is, AI is already impacting our lives 359 00:11:26,880 --> 00:11:30,080 in so many ways, even if we don't realize it. 360 00:11:30,080 --> 00:11:33,380 What we see online, the things we buy, even health care 361 00:11:33,380 --> 00:11:36,200 decisions, AI is playing a role. 362 00:11:36,200 --> 00:11:39,040 It's like this invisible force shaping our world. 363 00:11:39,040 --> 00:11:40,600 And it's only going to get more powerful, right? 364 00:11:40,600 --> 00:11:41,200 Absolutely. 365 00:11:41,200 --> 00:11:42,760 And that's why we need to be informed. 366 00:11:42,760 --> 00:11:44,680 We need to be part of this conversation. 367 00:11:44,680 --> 00:11:48,200 Because the choices we make now about how we develop and use AI, 368 00:11:48,200 --> 00:11:50,440 they're going to have a huge impact on all of us. 369 00:11:50,440 --> 00:11:51,640 So it's not just a tech issue. 370 00:11:51,640 --> 00:11:52,640 It's a societal issue. 371 00:11:52,640 --> 00:11:53,640 It's about all of us. 372 00:11:53,640 --> 00:11:54,720 Exactly. 373 00:11:54,720 --> 00:11:57,520 AI isn't something happening in some lab somewhere, 374 00:11:57,520 --> 00:11:58,640 separate from our lives. 375 00:11:58,640 --> 00:12:00,800 It's becoming a part of our everyday reality. 376 00:12:00,800 --> 00:12:03,600 And it's going to affect how we live, how we work, 377 00:12:03,600 --> 00:12:04,960 how we interact with the world. 378 00:12:04,960 --> 00:12:07,920 It makes those whistleblower stories even more powerful, 379 00:12:07,920 --> 00:12:08,600 doesn't it? 380 00:12:08,600 --> 00:12:10,680 They're not just talking about some hypothetical thing. 381 00:12:10,680 --> 00:12:13,280 They're talking about something that could affect all of us. 382 00:12:13,280 --> 00:12:14,440 Exactly. 383 00:12:14,440 --> 00:12:17,240 They're giving us a glimpse behind the curtain showing us 384 00:12:17,240 --> 00:12:18,760 what can go wrong. 385 00:12:18,760 --> 00:12:21,800 And they're urging us to be more thoughtful about how 386 00:12:21,800 --> 00:12:23,600 we approach this technology. 387 00:12:23,600 --> 00:12:25,760 And it's not just up to the experts to figure it out, 388 00:12:25,760 --> 00:12:26,000 right? 389 00:12:26,000 --> 00:12:27,360 We all have a role to play. 390 00:12:27,360 --> 00:12:28,600 Absolutely. 391 00:12:28,600 --> 00:12:31,760 The future of AI, it's not set in stone. 392 00:12:31,760 --> 00:12:34,160 It's something we can shape through our choices 393 00:12:34,160 --> 00:12:35,360 and our actions. 394 00:12:35,360 --> 00:12:37,160 OK, now I'm feeling a bit more optimistic. 395 00:12:37,160 --> 00:12:39,360 There's something we can actually do. 396 00:12:39,360 --> 00:12:40,520 But where do we even start? 397 00:12:40,520 --> 00:12:44,320 What can we actually do to make a difference? 398 00:12:44,320 --> 00:12:47,040 Well, first things first, we need to educate ourselves. 399 00:12:47,040 --> 00:12:49,720 The more we understand about AI, the better equipped 400 00:12:49,720 --> 00:12:52,560 we'll be to make smart decisions about how it's developed, 401 00:12:52,560 --> 00:12:53,560 how it's used. 402 00:12:53,560 --> 00:12:55,920 So read articles, listen to podcasts, 403 00:12:55,920 --> 00:12:57,680 have these kinds of conversations. 404 00:12:57,680 --> 00:12:58,360 Exactly. 405 00:12:58,360 --> 00:12:59,640 And ask questions. 406 00:12:59,640 --> 00:13:02,880 Don't be afraid to ask, even if they seem basic. 407 00:13:02,880 --> 00:13:07,120 The AI world is moving so fast, everyone's learning as we go. 408 00:13:07,120 --> 00:13:09,240 And it's not just about the technical stuff. 409 00:13:09,240 --> 00:13:11,320 It's about the ethical side, too. 410 00:13:11,320 --> 00:13:13,600 What are the implications for society? 411 00:13:13,600 --> 00:13:15,800 Because AI is just a tool, right? 412 00:13:15,800 --> 00:13:17,160 It can be used for good or bad. 413 00:13:17,160 --> 00:13:19,280 It's up to us to decide how we want to use it. 414 00:13:19,280 --> 00:13:20,120 Exactly. 415 00:13:20,120 --> 00:13:22,240 And we can also be more conscious consumers. 416 00:13:22,240 --> 00:13:23,800 Think about the companies you support. 417 00:13:23,800 --> 00:13:27,200 Are they thinking about AI safety, about ethics? 418 00:13:27,200 --> 00:13:30,720 And be aware of how you're using AI in your own life. 419 00:13:30,720 --> 00:13:33,960 Like, are we OK with how social media uses our data? 420 00:13:33,960 --> 00:13:36,480 Or facial recognition in public spaces? 421 00:13:36,480 --> 00:13:37,120 Right. 422 00:13:37,120 --> 00:13:38,400 Those are the kinds of questions we 423 00:13:38,400 --> 00:13:39,600 need to be asking ourselves. 424 00:13:39,600 --> 00:13:42,200 It's about being critical, about questioning the tech that's 425 00:13:42,200 --> 00:13:43,080 all around us. 426 00:13:43,080 --> 00:13:44,880 And letting our voices be heard. 427 00:13:44,880 --> 00:13:46,200 Contact our elected officials. 428 00:13:46,200 --> 00:13:49,480 Support organizations that are fighting for responsible AI. 429 00:13:49,480 --> 00:13:51,480 Talk to our friends and family about this stuff. 430 00:13:51,480 --> 00:13:52,120 Exactly. 431 00:13:52,120 --> 00:13:54,240 The more people who are aware of the problem, 432 00:13:54,240 --> 00:13:56,840 the more pressure there is to find solutions. 433 00:13:56,840 --> 00:13:59,360 It's a long process, but it starts with individual voices 434 00:13:59,360 --> 00:14:00,240 speaking up. 435 00:14:00,240 --> 00:14:00,840 OK. 436 00:14:00,840 --> 00:14:02,640 I'm starting to see the bigger picture here. 437 00:14:02,640 --> 00:14:04,120 We've got the whistleblowers. 438 00:14:04,120 --> 00:14:06,280 We've got policymakers trying to catch up. 439 00:14:06,280 --> 00:14:09,320 And then there's us, everyday people with more power 440 00:14:09,320 --> 00:14:10,200 than we realize. 441 00:14:10,200 --> 00:14:12,240 And don't forget about staying curious. 442 00:14:12,240 --> 00:14:14,200 AI is changing so fast. 443 00:14:14,200 --> 00:14:17,200 Even the experts are constantly learning new things. 444 00:14:17,200 --> 00:14:18,120 Read those articles. 445 00:14:18,120 --> 00:14:19,720 Listen to those podcasts. 446 00:14:19,720 --> 00:14:22,600 It helps us stay informed and make better decisions. 447 00:14:22,600 --> 00:14:25,240 Speaking of podcasts, any memorable stories 448 00:14:25,240 --> 00:14:27,920 from Aysul Tanna about individuals 449 00:14:27,920 --> 00:14:28,800 making a difference? 450 00:14:28,800 --> 00:14:29,560 Oh, absolutely. 451 00:14:29,560 --> 00:14:32,600 There's one about this group of high school students. 452 00:14:32,600 --> 00:14:33,400 High schoolers. 453 00:14:33,400 --> 00:14:33,880 Wow. 454 00:14:33,880 --> 00:14:38,280 They built an AI-powered app that helps people identify bias 455 00:14:38,280 --> 00:14:40,960 in online content and report it. 456 00:14:40,960 --> 00:14:41,600 That's amazing. 457 00:14:41,600 --> 00:14:43,520 They're already out there changing things. 458 00:14:43,520 --> 00:14:44,840 It just shows you, you don't have 459 00:14:44,840 --> 00:14:46,360 to be an expert to make an impact. 460 00:14:46,360 --> 00:14:47,240 They saw a problem. 461 00:14:47,240 --> 00:14:48,640 They learned about AI. 462 00:14:48,640 --> 00:14:50,320 And they used it to do something good. 463 00:14:50,320 --> 00:14:51,320 Gives me hope, honestly. 464 00:14:51,320 --> 00:14:52,880 Me too. 465 00:14:52,880 --> 00:14:56,160 It's a good reminder that even with all the uncertainty, 466 00:14:56,160 --> 00:14:58,920 there are people out there who are passionate about using 467 00:14:58,920 --> 00:15:00,480 AI for good. 468 00:15:00,480 --> 00:15:02,400 And they're finding creative ways to do it. 469 00:15:02,400 --> 00:15:04,920 OK, so lots of ideas on the table. 470 00:15:04,920 --> 00:15:09,040 Transparency, oversight, even licensing for AI developers. 471 00:15:09,040 --> 00:15:10,600 But where does that leave us? 472 00:15:10,600 --> 00:15:11,800 The everyday folks. 473 00:15:11,800 --> 00:15:14,440 Just using technology, not building it. 474 00:15:14,440 --> 00:15:14,760 Right. 475 00:15:14,760 --> 00:15:16,240 It can feel pretty big picture. 476 00:15:16,240 --> 00:15:17,760 Like, what can we actually do? 477 00:15:17,760 --> 00:15:19,360 But honestly, a lot of this comes down 478 00:15:19,360 --> 00:15:21,200 to individual choices we make. 479 00:15:21,200 --> 00:15:21,720 Really? 480 00:15:21,720 --> 00:15:22,320 How so? 481 00:15:22,320 --> 00:15:23,000 Think about it. 482 00:15:23,000 --> 00:15:25,320 Every time you're on social media, searching online, 483 00:15:25,320 --> 00:15:28,200 buying something, those sites recommending stuff, 484 00:15:28,200 --> 00:15:29,920 you're creating data. 485 00:15:29,920 --> 00:15:32,800 And that data, it feeds these AI systems. 486 00:15:32,800 --> 00:15:33,300 Right. 487 00:15:33,300 --> 00:15:34,080 Our data is valuable. 488 00:15:34,080 --> 00:15:34,840 That's not new. 489 00:15:34,840 --> 00:15:37,320 But I guess I didn't think about it like, 490 00:15:37,320 --> 00:15:38,520 we're not just using AI. 491 00:15:38,520 --> 00:15:40,000 We're kind of shaping it as we go. 492 00:15:40,000 --> 00:15:40,520 Exactly. 493 00:15:40,520 --> 00:15:42,320 So that means we have some power here. 494 00:15:42,320 --> 00:15:44,080 Are we cool with how our data is used? 495 00:15:44,080 --> 00:15:48,080 Do we want to support companies that are doing AI ethically? 496 00:15:48,080 --> 00:15:50,360 Those choices, millions of people making them, 497 00:15:50,360 --> 00:15:51,360 that sends a message. 498 00:15:51,360 --> 00:15:53,680 Voting with our wallets, but also voting with our data. 499 00:15:53,680 --> 00:15:54,200 Exactly. 500 00:15:54,200 --> 00:15:57,480 And beyond that, staying informed, speaking up, 501 00:15:57,480 --> 00:16:01,000 talking to friends, family, writing to our representatives. 502 00:16:01,000 --> 00:16:03,360 The more we bring up these AI safety issues, 503 00:16:03,360 --> 00:16:05,160 the harder it is to ignore them. 504 00:16:05,160 --> 00:16:08,760 More people aware, more eyes watching those 505 00:16:08,760 --> 00:16:10,240 who need to be held accountable. 506 00:16:10,240 --> 00:16:11,480 That's the idea. 507 00:16:11,480 --> 00:16:13,720 Push for solutions that work for everyone, 508 00:16:13,720 --> 00:16:15,400 not just the tech companies. 509 00:16:15,400 --> 00:16:16,920 It's a long game, no doubt. 510 00:16:16,920 --> 00:16:19,840 But change starts with individual voices, doesn't it? 511 00:16:19,840 --> 00:16:20,280 OK. 512 00:16:20,280 --> 00:16:22,200 Yeah, I'm seeing how it all connects now. 513 00:16:22,200 --> 00:16:23,780 The whistleblowers sounding the alarm, 514 00:16:23,780 --> 00:16:26,000 the policymakers trying to figure it out, 515 00:16:26,000 --> 00:16:29,280 and then us, regular people, actually having some influence. 516 00:16:29,280 --> 00:16:31,720 And don't underestimate staying curious. 517 00:16:31,720 --> 00:16:33,920 AI changes so fast, even the pros 518 00:16:33,920 --> 00:16:36,640 are always playing catch up, reading articles, 519 00:16:36,640 --> 00:16:39,160 listening to podcasts like Aislethana. 520 00:16:39,160 --> 00:16:41,120 It helps us make smarter choices. 521 00:16:41,120 --> 00:16:43,840 Speaking of Aislethana, any examples 522 00:16:43,840 --> 00:16:46,320 they gave about people actually making a difference? 523 00:16:46,320 --> 00:16:47,560 Any stories stick out? 524 00:16:47,560 --> 00:16:50,200 Oh, yeah, there was this one about a group 525 00:16:50,200 --> 00:16:51,200 of high school students. 526 00:16:51,200 --> 00:16:51,800 High school? 527 00:16:51,800 --> 00:16:52,320 Wow, yeah. 528 00:16:52,320 --> 00:16:55,960 They built an app, uses AI to help people spot bias 529 00:16:55,960 --> 00:16:57,240 in online content. 530 00:16:57,240 --> 00:16:59,160 And then you can report it through the app. 531 00:16:59,160 --> 00:17:00,080 That's incredible. 532 00:17:00,080 --> 00:17:01,840 They're already out there doing this stuff. 533 00:17:01,840 --> 00:17:03,880 It just shows you, you don't have to be a genius 534 00:17:03,880 --> 00:17:06,720 or a politician to have an impact. 535 00:17:06,720 --> 00:17:08,800 They saw a problem, learned about AI, 536 00:17:08,800 --> 00:17:10,680 and used it to make things better. 537 00:17:10,680 --> 00:17:11,680 That's pretty inspiring. 538 00:17:11,680 --> 00:17:12,640 Gives you hope. 539 00:17:12,640 --> 00:17:13,720 Big time. 540 00:17:13,720 --> 00:17:16,520 With all the craziness, all the uncertainty, 541 00:17:16,520 --> 00:17:18,720 it's good to remember there are people out there who 542 00:17:18,720 --> 00:17:21,680 want to use this tech for good. 543 00:17:21,680 --> 00:17:24,160 And they're finding really creative ways to do it. 544 00:17:24,160 --> 00:17:25,600 This whole deep dive. 545 00:17:25,600 --> 00:17:29,000 I went from feeling overwhelmed to now feeling like, OK, 546 00:17:29,000 --> 00:17:30,520 maybe I can actually do something. 547 00:17:30,520 --> 00:17:31,720 That's what we like to hear. 548 00:17:31,720 --> 00:17:33,400 Knowledge is power, right? 549 00:17:33,400 --> 00:17:35,720 Now you've got the tools to make sense of this stuff. 550 00:17:35,720 --> 00:17:37,040 And you know, it's funny. 551 00:17:37,040 --> 00:17:39,840 We talk about AI, this cutting edge tech, 552 00:17:39,840 --> 00:17:43,680 but it comes down to really human stuff. 553 00:17:43,680 --> 00:17:46,000 Our choices, what we care about, what we're 554 00:17:46,000 --> 00:17:48,560 willing to put up with, what kind of future we want. 555 00:17:48,560 --> 00:17:50,440 It's not just the tech, it's us. 556 00:17:50,440 --> 00:17:51,840 Couldn't have said it better myself. 557 00:17:51,840 --> 00:17:54,840 AI's future, it's a reflection of our own humanity. 558 00:17:54,840 --> 00:17:56,560 And on that note, I think it's time 559 00:17:56,560 --> 00:17:57,920 to wrap up this deep dive. 560 00:17:57,920 --> 00:18:00,440 Yeah, it's been a pleasure exploring all this with you. 561 00:18:00,440 --> 00:18:02,280 Thanks so much for all your insights. 562 00:18:02,280 --> 00:18:04,280 And to our listener, we hope you found this journey 563 00:18:04,280 --> 00:18:06,320 as fascinating as we have. 564 00:18:06,320 --> 00:18:07,160 Keep exploring. 565 00:18:07,160 --> 00:18:08,760 Keep asking those tough questions. 566 00:18:08,760 --> 00:18:12,360 And remember, you have a voice in all of this, 567 00:18:12,360 --> 00:18:14,320 in shaping the future of AI. 568 00:18:14,320 --> 00:18:17,300 Until next time, keep learning, keep sharing, 569 00:18:17,300 --> 00:18:29,480 and keep diving deep.