WEBVTT 00:00:00.000 --> 00:00:05.960 Imagine a voice AI, so natural, it actually detects 00:00:05.960 --> 00:00:08.320 your laughter, adjusts its tone. Or think about 00:00:08.320 --> 00:00:11.839 an AI predicting flu strains, maybe even better, 00:00:11.880 --> 00:00:14.480 faster than global health organizations. This 00:00:14.480 --> 00:00:16.579 isn't like some distant sci -fi idea, is it? 00:00:16.739 --> 00:00:19.300 Well, it's happening right now. Welcome to this 00:00:19.300 --> 00:00:22.109 deep dive, everyone. Our mission today really 00:00:22.109 --> 00:00:25.089 is to cut through all the noise. We want to pull 00:00:25.089 --> 00:00:27.769 out the crucial insights from the, well, the 00:00:27.769 --> 00:00:30.030 absolute latest in AI. We're going to kick things 00:00:30.030 --> 00:00:34.070 off by unpacking a breakthrough in human -like 00:00:34.070 --> 00:00:37.130 voice AI. Then we'll skim across some other really 00:00:37.130 --> 00:00:39.229 exciting stuff, creative tools, coding assistance, 00:00:39.609 --> 00:00:41.729 things changing how we work. And finally, we'll 00:00:41.729 --> 00:00:44.049 land on something truly groundbreaking, a medical 00:00:44.049 --> 00:00:46.469 application that could honestly reshape public 00:00:46.469 --> 00:00:48.960 health globally. It's going to be a fast ride, 00:00:49.219 --> 00:00:50.859 but yeah, it's all pretty key to understanding 00:00:50.859 --> 00:00:53.219 where AI is today. Okay. Let's dive in then. 00:00:53.259 --> 00:00:56.659 Our first big story. It feels like a really profound 00:00:56.659 --> 00:00:59.640 shift in how we interact with machines. This 00:00:59.640 --> 00:01:04.280 new era of voice AI, specifically open AI's GPT 00:01:04.280 --> 00:01:06.920 real time. For years, we've heard these promises, 00:01:07.140 --> 00:01:09.780 you know, genuinely human -like voice agents. 00:01:09.859 --> 00:01:12.060 Always just around the corner. Exactly. Always 00:01:12.060 --> 00:01:15.159 just around the corner. But our sources are saying 00:01:15.159 --> 00:01:17.680 that corner has finally been turned. This thing 00:01:17.680 --> 00:01:19.840 we've been waiting for, the truly natural voice 00:01:19.840 --> 00:01:21.900 interface. It seems like it's actually arrived. 00:01:22.180 --> 00:01:24.359 It really feels like it. And what hits you immediately 00:01:24.359 --> 00:01:27.480 is just how natural it sounds. OpenAI has got 00:01:27.480 --> 00:01:30.359 these two new voices, Marin and Cedar. They're 00:01:30.359 --> 00:01:32.400 described as just incredibly smooth, really nuanced. 00:01:32.760 --> 00:01:35.920 And for you, the user, it gets even more specific. 00:01:35.980 --> 00:01:37.959 You can tell it talk faster, talk softer, sound 00:01:37.959 --> 00:01:40.579 warm, even sound cold. You can even ask for a 00:01:40.579 --> 00:01:43.500 specific accent. All 10 of their voice options 00:01:43.500 --> 00:01:47.329 apparently sound way more natural now. leap more 00:01:47.329 --> 00:01:50.010 than just clear pronunciation right it's beyond 00:01:50.010 --> 00:01:52.810 just clarity and the clever part isn't only the 00:01:52.810 --> 00:01:55.769 voice quality it's it's the understanding behind 00:01:55.769 --> 00:01:58.810 it exactly this gpt real time it can pick up 00:01:58.810 --> 00:02:02.049 on non -verbal stuff like actual laughter or 00:02:02.049 --> 00:02:05.310 sighs or pauses yeah yeah and crucially it handles 00:02:05.310 --> 00:02:08.129 code switching like a champ Yes. Switching languages 00:02:08.129 --> 00:02:10.449 mid -sentence. It just flows with you. Doesn't 00:02:10.449 --> 00:02:12.110 get tripped up. Doesn't force you to pick one 00:02:12.110 --> 00:02:14.370 language. That's pretty advanced. Yeah. It's 00:02:14.370 --> 00:02:17.169 not just mimicking them. Yeah. It's closer to 00:02:17.169 --> 00:02:19.930 real comprehension, real responsiveness. Which 00:02:19.930 --> 00:02:21.810 leads to the accuracy question, right? How much 00:02:21.810 --> 00:02:24.830 better is it? And the numbers are pretty striking. 00:02:25.050 --> 00:02:27.389 Okay. On Big Bench Audio Reasoning, that's a 00:02:27.389 --> 00:02:30.090 test for audio understanding accuracy, jumped 00:02:30.090 --> 00:02:33.930 to 82 .8%. Wow. What was it before? It was 65 00:02:33.930 --> 00:02:38.409 .6%. So big jump. Then instruction following 00:02:38.409 --> 00:02:41.210 on multi -challenge, that went up to 30 .5 % 00:02:41.210 --> 00:02:44.370 from 20 .6%. Also significant. And maybe most 00:02:44.370 --> 00:02:46.949 important for like actually using it day to day. 00:02:47.449 --> 00:02:50.210 function calling on complex funkbench it hits 00:02:50.210 --> 00:02:54.990 66 .5 up from 49 .7 okay hang on function calling 00:02:54.990 --> 00:02:57.389 just define that simply for sure it just means 00:02:57.389 --> 00:02:59.629 the ai actually doing something for you performing 00:02:59.629 --> 00:03:02.289 an action or task like book me a table exactly 00:03:02.289 --> 00:03:04.409 like that it goes and does the booking that's 00:03:04.409 --> 00:03:06.430 function calling got it and that improvement 00:03:06.430 --> 00:03:08.789 makes it way more useful obviously definitely 00:03:08.789 --> 00:03:11.030 and here's another detail i thought was really 00:03:11.030 --> 00:03:13.870 smart really speaks to making it feel seamless 00:03:13.870 --> 00:03:17.710 oh yeah if there's a long api call Like if the 00:03:17.710 --> 00:03:20.930 AI needs a moment to fetch some info, GPT real 00:03:20.930 --> 00:03:23.210 time keeps chatting. It doesn't just go quiet. 00:03:23.310 --> 00:03:25.250 Nope. Doesn't leave you hanging, wondering if 00:03:25.250 --> 00:03:27.949 it crashed. It's more like how a person might 00:03:27.949 --> 00:03:31.169 pause, maybe say, hmm, let me look that up. Right. 00:03:31.270 --> 00:03:34.090 It stays engaged. Yeah. That small thing, maintaining 00:03:34.090 --> 00:03:36.389 the flow, it really changes the feel of the interaction, 00:03:36.610 --> 00:03:40.710 makes it less. Tense somehow. Yeah, I can see 00:03:40.710 --> 00:03:42.610 that. And this isn't just a small update, right? 00:03:42.710 --> 00:03:44.810 No, not at all. Our sources are really emphasizing 00:03:44.810 --> 00:03:48.210 this. GPT Realtime is the first open AI model 00:03:48.210 --> 00:03:51.629 built as a voice -native AI agent right from 00:03:51.629 --> 00:03:53.870 the start. Meaning? Meaning it integrates audio, 00:03:54.129 --> 00:03:56.409 images, live function calling right into its 00:03:56.409 --> 00:03:59.490 core design. It's multimodal, built for conversation 00:03:59.490 --> 00:04:02.840 first. Okay. A multi -mobile beast designed for 00:04:02.840 --> 00:04:04.900 conversation, and the impact is pretty clear 00:04:04.900 --> 00:04:07.180 then. Oh, yeah. Sources say it's kind of making 00:04:07.180 --> 00:04:09.939 competitors like Eleven Labs maybe sound a bit 00:04:09.939 --> 00:04:13.280 dated already. And it's definitely a direct serious 00:04:13.280 --> 00:04:15.460 challenger to Google Gemini Voice. It marks a 00:04:15.460 --> 00:04:18.500 real transition point from voice AI being mostly 00:04:18.500 --> 00:04:20.439 a demo gimmick, cool for a minute, but not that 00:04:20.439 --> 00:04:23.199 useful to being a genuinely deployable product, 00:04:23.319 --> 00:04:25.000 something you can actually build things with. 00:04:25.160 --> 00:04:28.339 Okay. So boiling it down. What's the biggest 00:04:28.339 --> 00:04:32.459 shift this new voice AI brings to our daily interactions? 00:04:32.819 --> 00:04:36.759 It makes natural, real -time voice conversations 00:04:36.759 --> 00:04:39.800 with AI truly practical, a tool you can actually 00:04:39.800 --> 00:04:42.100 rely on. Right, practical and reliable, not just 00:04:42.100 --> 00:04:44.980 a novelty anymore. That leap from demo gimmick 00:04:44.980 --> 00:04:48.560 to deployable product for voice, it really is 00:04:48.560 --> 00:04:50.079 something. It makes you wonder how that's playing 00:04:50.079 --> 00:04:55.019 out elsewhere. Let's talk about AI in... creativity 00:04:55.019 --> 00:04:56.800 and maybe more practical workflows. Absolutely. 00:04:56.980 --> 00:04:58.959 So on the creative front, Google's got this new 00:04:58.959 --> 00:05:01.699 image model, Nano Banana. Nano Banana. Yeah. 00:05:01.819 --> 00:05:04.300 Sounds fun, right? But it's becoming a real fan 00:05:04.300 --> 00:05:06.339 favorite. People are using it for all sorts of 00:05:06.339 --> 00:05:08.540 diverse things, unique visuals, experimental 00:05:08.540 --> 00:05:10.879 art. Seems like it's sparking a lot of creativity. 00:05:11.319 --> 00:05:13.459 Interesting. And sticking with visuals, I saw 00:05:13.459 --> 00:05:16.910 something about UCLA researchers. An AI image 00:05:16.910 --> 00:05:19.269 generator using light instead of electricity. 00:05:19.490 --> 00:05:21.750 Oh, yeah. I saw that, too. Wild stuff. Almost 00:05:21.750 --> 00:05:24.370 zero power consumption, apparently. Those results 00:05:24.370 --> 00:05:27.149 are good. Supposedly really impressive artistically. 00:05:27.189 --> 00:05:28.610 It's just a totally different way of thinking 00:05:28.610 --> 00:05:31.550 about computation using light itself. Very cool. 00:05:31.689 --> 00:05:34.850 Okay, so switching gears to more practical stuff. 00:05:35.189 --> 00:05:39.149 Gemini 2 .5 Pro, that's getting buzzed for workflow 00:05:39.149 --> 00:05:42.509 automation, right? Huge potential there for streamlining 00:05:42.509 --> 00:05:45.670 tasks, yeah. Okay, here's a vulnerable admission. 00:05:46.050 --> 00:05:49.350 Beat. I still wrestle with prompt drift myself 00:05:49.350 --> 00:05:52.350 sometimes. Oh, tell me about it. Trying to get 00:05:52.350 --> 00:05:54.649 the most out of these big models, you know, where 00:05:54.649 --> 00:05:58.230 the AI starts kind of wandering away from your 00:05:58.230 --> 00:06:00.129 original instructions after a while. Yeah, losing 00:06:00.129 --> 00:06:02.410 the plot a bit. Happens to everyone. It's definitely 00:06:02.410 --> 00:06:04.689 a learning curve. Even for us, right? It shows 00:06:04.689 --> 00:06:06.290 that, you know, human guidance is still pretty 00:06:06.290 --> 00:06:08.350 key. Absolutely. It's not just plug and play 00:06:08.350 --> 00:06:10.029 magic yet. And we're seeing other specialized 00:06:10.029 --> 00:06:14.250 tools pop up, too, like XAI's GRUC code fast 00:06:14.250 --> 00:06:16.649 one. For coding. Yeah. People are calling it 00:06:16.649 --> 00:06:19.350 good for vibe coding, like capturing the feel 00:06:19.350 --> 00:06:21.250 or intent of the code quickly without getting 00:06:21.250 --> 00:06:23.490 totally bogged down. Vibe coding. I like that. 00:06:23.589 --> 00:06:27.629 And then there was that weird, brief, AI -enhanced 00:06:27.629 --> 00:06:30.920 Wizard of Oz thing. At the Las Vegas Sphere. 00:06:31.040 --> 00:06:33.759 Oh, right. With the faces of James Dolan and 00:06:33.759 --> 00:06:36.360 David Zaslav popping up for a second. Yeah. Super 00:06:36.360 --> 00:06:39.980 random. Super brief. But kind of shows AI bleeding 00:06:39.980 --> 00:06:42.980 into entertainment and spectacle in unexpected 00:06:42.980 --> 00:06:45.699 ways. Totally unexpected. Now, all this innovation, 00:06:45.800 --> 00:06:48.689 this deployment. Yeah. It takes serious money, 00:06:48.730 --> 00:06:50.050 right? Right, for sure. And the investment is 00:06:50.050 --> 00:06:53.449 pouring in. Like Accelera AI seeking, what, $150 00:06:53.449 --> 00:06:55.889 million? Yep, backed by Samsung, too, for their 00:06:55.889 --> 00:06:58.370 energy -efficient AI chips. And Google put in 00:06:58.370 --> 00:07:02.069 a massive $9 billion. Plus, Cohere raised $500 00:07:02.069 --> 00:07:04.930 million for enterprise AI. Exactly. This isn't 00:07:04.930 --> 00:07:07.569 just garage tinkering anymore. It's big players, 00:07:07.670 --> 00:07:10.110 big money, signals huge confidence across the 00:07:10.110 --> 00:07:12.990 whole sector, a real industrial shift. So when 00:07:12.990 --> 00:07:14.370 you look at all these different things, creative 00:07:14.370 --> 00:07:17.209 tools, workflow helpers, coding aids, massive 00:07:17.209 --> 00:07:19.970 investments, how do these varied innovations 00:07:19.970 --> 00:07:22.250 reflect the current state of AI development? 00:07:22.649 --> 00:07:25.209 AI is getting both more specialized for specific 00:07:25.209 --> 00:07:28.310 jobs and more deeply woven into our basic tools 00:07:28.310 --> 00:07:31.009 and infrastructure. Right. Specialized and integrated. 00:07:31.490 --> 00:07:34.250 Building that foundation. Okay. That makes sense. 00:07:34.389 --> 00:07:37.829 Let's zoom out a bit now. Broader AI trends, 00:07:37.970 --> 00:07:40.069 maybe where things are heading next. Feels like 00:07:40.069 --> 00:07:42.709 development has been just... It's incredibly 00:07:42.709 --> 00:07:44.990 intense lately. Intense is the word. Almost a 00:07:44.990 --> 00:07:47.509 Cambrian explosion, right? New models everywhere. 00:07:48.050 --> 00:07:50.269 Yeah. What really caught my eye was that one 00:07:50.269 --> 00:07:56.500 week, GBC 5, Clog 4 .1, new Gemini models. all 00:07:56.500 --> 00:07:58.959 landed practically at once. I know. My social 00:07:58.959 --> 00:08:01.079 feed just exploded. It was like trying to drink 00:08:01.079 --> 00:08:03.519 from a fire hose, just keeping up. Exactly. It 00:08:03.519 --> 00:08:05.500 really hammers home how fast we have to relearn 00:08:05.500 --> 00:08:07.959 what's even possible. Definitely. And beyond 00:08:07.959 --> 00:08:10.339 those huge foundational models, there were other 00:08:10.339 --> 00:08:13.160 interesting little bits. Like MathGPT expanding. 00:08:13.740 --> 00:08:16.339 It's positioning itself as a cheat -proof tutor, 00:08:16.480 --> 00:08:18.639 apparently rolling out to over 50 organizations. 00:08:18.860 --> 00:08:21.860 Interesting angle. Cheat -proof. And even Microsoft's 00:08:21.860 --> 00:08:24.899 CEO, Satya Nadella, shared his top five favorite 00:08:24.899 --> 00:08:27.670 prompts for... GPT -5 and Copilot. Shows how 00:08:27.670 --> 00:08:30.069 leaders are actually using these tools day to 00:08:30.069 --> 00:08:32.269 day and, you know, that everyone's still figuring 00:08:32.269 --> 00:08:34.529 out the best ways to talk to these things. That's 00:08:34.529 --> 00:08:36.470 a good point. We're all still learning. What 00:08:36.470 --> 00:08:39.110 about bigger strategic moves? Well, China's plans 00:08:39.110 --> 00:08:41.629 to triple its AI output stand out. Triple it. 00:08:41.710 --> 00:08:44.649 Wow. Yeah, it's a clear strategic push. Bolster 00:08:44.649 --> 00:08:46.809 their own companies, rely less on foreign tech. 00:08:47.210 --> 00:08:49.590 Makes sense from their perspective. And on the 00:08:49.590 --> 00:08:53.429 flip side, open AI. They seem focused on user 00:08:53.429 --> 00:08:56.100 safety. Talking about parental controls, emergency 00:08:56.100 --> 00:09:00.240 contacts for GPT. So innovation, but also trying 00:09:00.240 --> 00:09:03.000 to manage the risks. Exactly. That constant balancing 00:09:03.000 --> 00:09:06.600 act. Push forward, but do it responsibly. And 00:09:06.600 --> 00:09:08.480 looking further out, what glimpses are we getting 00:09:08.480 --> 00:09:10.679 of the future? Well, Microsoft Copilot being 00:09:10.679 --> 00:09:15.679 embedded into Samsung's 2025 TVs is pretty telling. 00:09:15.799 --> 00:09:19.600 AI in the TV. As standard. Seems like it. Think 00:09:19.600 --> 00:09:22.039 about that. AI isn't just an app on your phone 00:09:22.039 --> 00:09:24.460 anymore. It's becoming a default feature in your 00:09:24.460 --> 00:09:26.940 living room. Your TV is an actual intelligent 00:09:26.940 --> 00:09:30.299 companion, not just a screen. Right. That fundamentally 00:09:30.299 --> 00:09:32.639 changes how we interact with our home tech, doesn't 00:09:32.639 --> 00:09:35.399 it? Much more pervasive. It really does. Okay, 00:09:35.480 --> 00:09:38.299 so looking at all these trends, the rapid model 00:09:38.299 --> 00:09:40.659 launches, the strategic plays, the safety features, 00:09:40.779 --> 00:09:42.960 the deep integration, what's the overarching 00:09:42.960 --> 00:09:46.039 theme these suggest for AI's immediate future? 00:09:46.429 --> 00:09:50.610 AI is shifting from niche tools to being everywhere, 00:09:50.909 --> 00:09:53.610 often invisible, acting like cognitive backup 00:09:53.610 --> 00:09:56.190 for daily life. Ubiquitous, invisible, cognitive 00:09:56.190 --> 00:09:59.149 backup. Got it. Okay. Now, for our final segment, 00:09:59.190 --> 00:10:01.690 let's turn to something, well, truly profound. 00:10:02.649 --> 00:10:06.730 AI revolutionizing healthcare. MIT researchers 00:10:06.730 --> 00:10:09.690 have introduced Vaxir, an AI system that predicts 00:10:09.690 --> 00:10:12.830 flu strains. And get this, apparently outperforms 00:10:12.830 --> 00:10:14.830 the World Health Organization. Yeah, this one 00:10:14.830 --> 00:10:17.090 really made me pause. It's fascinating when you 00:10:17.090 --> 00:10:18.710 compare it to the traditional way. Which is? 00:10:18.830 --> 00:10:20.929 Well, picking flea vaccine strains is tough. 00:10:21.110 --> 00:10:23.269 They have to make the call six months, sometimes 00:10:23.269 --> 00:10:25.750 more, in advance, just to get production going 00:10:25.750 --> 00:10:28.389 globally. Right, it's a long lead time. So Vaxir. 00:10:28.529 --> 00:10:30.629 It takes a totally different approach. Proactive, 00:10:30.669 --> 00:10:32.669 data -driven. It uses something called protein 00:10:32.669 --> 00:10:34.990 language models. Protein language models. Explain 00:10:34.990 --> 00:10:37.629 that. Okay, think of it like AI learning the 00:10:37.629 --> 00:10:40.429 language or grammar of proteins. The grammar 00:10:40.429 --> 00:10:43.690 of life itself. Kind of. It learns how the building 00:10:43.690 --> 00:10:46.750 blocks, the amino acids, arrange themselves and 00:10:46.750 --> 00:10:49.950 interact. This helps it predict how a virus's 00:10:49.950 --> 00:10:52.250 surface proteins, the parts the vaccine targets, 00:10:52.429 --> 00:10:54.830 might change or mutate over time. Okay, so it's 00:10:54.830 --> 00:10:56.789 understanding the virus at a fundamental level. 00:10:56.909 --> 00:10:59.269 Exactly. Then it combines that understanding 00:10:59.269 --> 00:11:02.809 with real -world lab data, plus simulations of 00:11:02.809 --> 00:11:04.960 how diseases spread. And what does it do with 00:11:04.960 --> 00:11:07.559 all that? It doesn't just predict which flu strains 00:11:07.559 --> 00:11:09.940 might be dominant next season. It goes further. 00:11:10.100 --> 00:11:13.200 It identifies the best specific vaccine strains 00:11:13.200 --> 00:11:16.320 to actually neutralize those predicted dominant 00:11:16.320 --> 00:11:18.620 strains. So it's predictive and prescriptive, 00:11:18.620 --> 00:11:20.820 finding the best countermeasure. Precisely. It's 00:11:20.820 --> 00:11:23.519 a much more holistic, proactive way to tackle 00:11:23.519 --> 00:11:26.240 an old problem. And the evidence. Does it work? 00:11:26.620 --> 00:11:28.259 This is the really incredible part. They did 00:11:28.259 --> 00:11:30.580 a 10 -year retrospective study. Looking back, 00:11:31.139 --> 00:11:34.899 Vaxier beat the WHO's selections in nine out 00:11:34.899 --> 00:11:37.080 of those 10 seasons. Nine out of 10. Seriously. 00:11:37.419 --> 00:11:39.919 Seriously. And it matched or outperformed the 00:11:39.919 --> 00:11:42.659 WHO in six seasons overall. Pretty consistent. 00:11:42.779 --> 00:11:45.639 And there was one specific case back in 2016. 00:11:45.860 --> 00:11:48.639 It picked the right strain a full year before 00:11:48.639 --> 00:11:52.860 the WHO did. Whoa. Hold on. A year ahead. A full 00:11:52.860 --> 00:11:55.679 year. Imagine that. AI predicting the flu, not 00:11:55.679 --> 00:11:57.779 just, you know, slightly better, but potentially 00:11:57.779 --> 00:12:01.159 years ahead. That feels like a whole new frontier 00:12:01.159 --> 00:12:04.360 for proactive health. Game changing. That is 00:12:04.360 --> 00:12:07.399 genuinely mind blowing. A year's warning. And 00:12:07.399 --> 00:12:09.600 this wasn't just a simulation, right? It matched 00:12:09.600 --> 00:12:12.259 real world data. Completely. The predictions 00:12:12.259 --> 00:12:15.000 lined up with actual effectiveness data from 00:12:15.000 --> 00:12:17.840 the CDC in the U .S., Canada's Sentinel program, 00:12:18.000 --> 00:12:20.919 and Europe's iMove initiative. So it's validated. 00:12:20.980 --> 00:12:23.610 This isn't just a lab curiosity. Nope. It's real 00:12:23.610 --> 00:12:26.750 science. And the implication is huge. It suggests 00:12:26.750 --> 00:12:28.730 a whole new way of thinking about fighting disease, 00:12:28.950 --> 00:12:31.629 moving from just reacting to outbreaks. To predicting 00:12:31.629 --> 00:12:33.509 and preventing them on a global scale. Exactly. 00:12:33.730 --> 00:12:36.289 Proactive, predictive, preventative. Okay. So 00:12:36.289 --> 00:12:39.350 stepping back from just the flu. Yeah. What does 00:12:39.350 --> 00:12:42.389 Vaxier's success suggest for AI in broader scientific 00:12:42.389 --> 00:12:44.830 prediction, like other diseases or even other 00:12:44.830 --> 00:12:47.529 fields? It shows AI can seriously speed up and 00:12:47.529 --> 00:12:49.610 improve scientific discovery, moving us from 00:12:49.610 --> 00:12:51.809 reaction to proactive problem solving everywhere. 00:12:52.480 --> 00:12:55.220 From reaction to proaction. A fundamental shift 00:12:55.220 --> 00:13:01.679 enabled by AI. Wow. Okay. Sponsor. So let's try 00:13:01.679 --> 00:13:03.500 and wrap this up to recap our deep dive today. 00:13:04.360 --> 00:13:07.019 The core takeaway, it feels pretty clear, AI 00:13:07.019 --> 00:13:10.679 has really crossed the chasm, hasn't it? Yeah, 00:13:10.679 --> 00:13:12.840 definitely. From being experimental tech to being, 00:13:12.899 --> 00:13:15.600 well, deployable, genuinely impactful products. 00:13:15.960 --> 00:13:18.139 We've seen it in voice interactions becoming... 00:13:18.269 --> 00:13:20.830 Incredibly natural, almost human. And creative 00:13:20.830 --> 00:13:24.230 tools, smart coding, help making workflows better. 00:13:24.370 --> 00:13:26.950 Right. All the way to potentially life -saving 00:13:26.950 --> 00:13:30.090 medical predictions like Vaxir. The impact just 00:13:30.090 --> 00:13:32.049 feels undeniable now, and it's growing fast. 00:13:32.269 --> 00:13:34.250 And what gets me is not just the depth of it, 00:13:34.269 --> 00:13:36.769 but the sheer speed and breadth. Yeah, the pace 00:13:36.769 --> 00:13:38.330 is something else. These breakthroughs are hitting 00:13:38.330 --> 00:13:40.549 everything. Our daily work, how we create things, 00:13:40.649 --> 00:13:42.250 even like we just discussed, the foundations 00:13:42.250 --> 00:13:45.169 of global public health. It's exciting, maybe 00:13:45.169 --> 00:13:47.669 a little dizzying sometimes, but you can't ignore 00:13:47.669 --> 00:13:50.649 it. No, you really can't. So maybe a final thought 00:13:50.649 --> 00:13:53.230 for you, our listener. As you encounter these 00:13:53.230 --> 00:13:56.490 evolving AI capabilities, think about how these 00:13:56.490 --> 00:13:58.730 advancements are changing things. From a voice 00:13:58.730 --> 00:14:01.509 that seems to really understand you to medicine 00:14:01.509 --> 00:14:04.870 guided by AI. How is this reshaping our trust 00:14:04.870 --> 00:14:07.750 in technology? And maybe even, you know, reshaping 00:14:07.750 --> 00:14:09.889 our own human potential. Lots to think about 00:14:09.889 --> 00:14:11.669 there. Thank you for joining us on this deep 00:14:11.669 --> 00:14:14.230 dive. Until next time, keep exploring. Out to 00:14:14.230 --> 00:14:14.690 your own music.