WEBVTT 00:00:00.000 --> 00:00:03.299 Okay, so today we're diving into something pretty 00:00:03.299 --> 00:00:07.179 wild. Yeah. NVIDIA, you know. Of course. The 00:00:07.179 --> 00:00:09.259 graphics cards, the whole deal. Right, the big 00:00:09.259 --> 00:00:12.980 AI hardware player. Exactly, but they're doing 00:00:12.980 --> 00:00:15.419 something that could totally reshape the future 00:00:15.419 --> 00:00:18.079 of tech. No, yeah, it's huge. They're merging 00:00:18.079 --> 00:00:21.280 quantum computing. Wow. With artificial intelligence. 00:00:21.480 --> 00:00:23.839 Now that's a combo I wouldn't have... It's not 00:00:23.839 --> 00:00:26.079 something you'd think of right away. No, not 00:00:26.079 --> 00:00:30.359 at all. But it's this promise of... Insane computational 00:00:30.359 --> 00:00:32.399 power. Yeah for someone like you always wanted 00:00:32.399 --> 00:00:34.939 to like quickly get to the heart of these big 00:00:34.939 --> 00:00:37.579 leaps Yeah, you know, this is definitely one 00:00:37.579 --> 00:00:39.780 to wrap your head around because it's not just 00:00:39.780 --> 00:00:42.259 incremental. It feels like a whole new era Totally. 00:00:42.560 --> 00:00:44.020 And that's what we're going to try and unpack 00:00:44.020 --> 00:00:46.659 in this deep dive. We really want to give everyone 00:00:46.659 --> 00:00:49.219 a clear picture of what NVIDIA is doing with 00:00:49.219 --> 00:00:51.200 this quantum AI stuff. That's what it all means. 00:00:51.520 --> 00:00:54.920 Exactly what it could enable for AI, you know, 00:00:54.939 --> 00:00:57.159 more advanced AI that we've ever seen. Absolutely. 00:00:57.240 --> 00:00:59.340 And also like the big questions it brings up 00:00:59.340 --> 00:01:01.960 about where AI is even headed. Right, we're talking 00:01:01.960 --> 00:01:04.379 about, you know, the singularity. Yeah, like 00:01:04.379 --> 00:01:06.540 that kind of stuff. Yeah, exactly. It's heady 00:01:06.540 --> 00:01:09.480 stuff. So we're basing this all on recent reports 00:01:09.480 --> 00:01:12.069 about NVIDIA's work. Yeah. cutting -edge stuff. 00:01:12.730 --> 00:01:16.310 So the core of it seems to be this NVIDIA is 00:01:16.310 --> 00:01:18.209 taking that, you know, they're already the kings 00:01:18.209 --> 00:01:20.890 of GPU tech. Right. But they're bringing in the 00:01:20.890 --> 00:01:23.349 potential of quantum computing. Powerful stuff. 00:01:23.549 --> 00:01:26.989 And they've created this platform called CUAQ. 00:01:27.090 --> 00:01:30.069 OK. And they're testing all of this on a massive 00:01:30.069 --> 00:01:33.870 scale using their EOS supercomputer. So this 00:01:33.870 --> 00:01:36.250 is actually happening, not just theory. Exactly. 00:01:36.430 --> 00:01:39.400 And the idea that this could like blast past 00:01:39.400 --> 00:01:41.700 the limits that regular processors have when 00:01:41.700 --> 00:01:44.640 it comes to AI. Right, because we know AI is 00:01:44.640 --> 00:01:46.280 amazing. It's doing incredible things. Yeah, 00:01:46.319 --> 00:01:48.420 it's everywhere. But it keeps hitting this wall 00:01:48.420 --> 00:01:51.739 of processing power. Right. It can only go as 00:01:51.739 --> 00:01:54.099 fast as the hardware lets it. Exactly. And NVIDIA 00:01:54.099 --> 00:01:55.859 is basically saying, hey, what if we could just 00:01:55.859 --> 00:01:59.019 sidestep that whole issue? Right. And that's 00:01:59.019 --> 00:02:02.439 where CUDQ comes in, because it lets AI developers 00:02:02.439 --> 00:02:06.329 use the crazy power of quantum computers. But 00:02:06.329 --> 00:02:09.310 without them needing to become quantum physics 00:02:09.310 --> 00:02:12.150 experts. Like you don't need a PhD. Exactly. 00:02:12.909 --> 00:02:15.129 It's giving them the tools. Right, exactly. It's 00:02:15.129 --> 00:02:17.889 bridging the gap between these two complex worlds. 00:02:18.710 --> 00:02:21.009 So, okay, help me understand this. I know the 00:02:21.009 --> 00:02:24.750 basics, but regular computers use bits, right? 00:02:24.909 --> 00:02:28.069 Yeah, zeros and ones. Gone or off. Exactly. But 00:02:28.069 --> 00:02:31.050 quantum computers use these things called quibits. 00:02:31.189 --> 00:02:34.000 Yeah, and this is where it gets trippy. Quibits 00:02:34.000 --> 00:02:37.000 can be in multiple states at the same time. What 00:02:37.000 --> 00:02:38.840 does that mean? It's like that classic analogy 00:02:38.840 --> 00:02:41.219 of the light switch. It's not just on or off. 00:02:41.340 --> 00:02:44.300 It's both at the same time. It's called superposition. 00:02:45.860 --> 00:02:48.699 Mind blown. And because of this, they can do 00:02:48.699 --> 00:02:51.819 a ton of calculations all at once. OK. Way more 00:02:51.819 --> 00:02:53.979 than a normal computer ever could. So it's like 00:02:53.979 --> 00:02:56.599 exponentially faster? Yeah. For certain types 00:02:56.599 --> 00:02:59.439 of problems, it's a potential game changer. Okay, 00:02:59.460 --> 00:03:01.000 so that's the quibbit part, but then there's 00:03:01.000 --> 00:03:03.099 the EO supercomputer. Right, this is where the 00:03:03.099 --> 00:03:04.979 rubber meets the road. This thing is built to 00:03:04.979 --> 00:03:08.060 handle these crazy quantum simulations. It's 00:03:08.060 --> 00:03:10.840 like they're training AI in a quantum playground. 00:03:11.199 --> 00:03:13.639 And this is where experts are saying this could 00:03:13.639 --> 00:03:16.300 be the path to artificial general intelligence, 00:03:16.539 --> 00:03:20.139 or AGI. Which is? AI that's basically as smart 00:03:20.139 --> 00:03:22.960 as a human. Yeah, capable of understanding learning, 00:03:23.280 --> 00:03:25.889 applying knowledge like we do. across all sorts 00:03:25.889 --> 00:03:28.849 of tasks. It's not just doing one specific thing 00:03:28.849 --> 00:03:30.590 really well. It's like... It's like generalized 00:03:30.590 --> 00:03:34.270 intelligence. Right. Like us. And a lot of people 00:03:34.270 --> 00:03:36.270 think that this massive computational power from 00:03:36.270 --> 00:03:39.650 quantum computing is the key to unlocking that. 00:03:39.810 --> 00:03:42.509 Yeah, it's the missing ingredient. So then the 00:03:42.509 --> 00:03:46.189 big question is... Uh -oh. If we create AI that's 00:03:46.189 --> 00:03:49.069 smarter than us in some ways, can we still control 00:03:49.069 --> 00:03:51.550 it? That's the million dollar question. It sounds 00:03:51.550 --> 00:03:53.409 like sci -fi, but it's like... It's becoming 00:03:53.409 --> 00:03:55.750 very real. Yeah, it's not just theoretical anymore. 00:03:56.129 --> 00:03:58.189 Exactly. If these AI systems are truly intelligent 00:03:58.189 --> 00:04:00.150 and they can improve themselves really fast, 00:04:00.389 --> 00:04:03.330 we have to make sure their goals align with ours. 00:04:04.509 --> 00:04:07.069 Like, we don't want them to decide we're obsolete 00:04:07.069 --> 00:04:08.830 or something. It's a challenge we need to think 00:04:08.830 --> 00:04:10.770 about seriously as this technology develops. 00:04:11.050 --> 00:04:14.349 Yeah, like the potential is huge, but so are 00:04:14.349 --> 00:04:16.230 the ethical implications. Totally. We're talking 00:04:16.230 --> 00:04:18.910 about potential and peril all wrapped up together. 00:04:19.029 --> 00:04:20.850 But let's step back from the hypothetical for 00:04:20.850 --> 00:04:23.829 a second. Okay. What has Nvidia actually achieved 00:04:23.829 --> 00:04:27.230 so far? Right. So the big breakthrough is that 00:04:27.230 --> 00:04:31.129 Fusion itself, quantum computing, giving AI the 00:04:31.129 --> 00:04:33.910 horsepower it needs to overcome those limitations. 00:04:33.930 --> 00:04:38.029 Right. And CUDAQ. makes that power accessible 00:04:38.029 --> 00:04:41.230 to AI developers. And EOS proves that quantum 00:04:41.230 --> 00:04:44.089 enhanced AI training is actually doable. Yeah, 00:04:44.149 --> 00:04:46.449 at a serious scale. And it's not just making 00:04:46.449 --> 00:04:49.589 AI faster. No, it's about tackling problems that 00:04:49.589 --> 00:04:52.170 were impossible before. Like what kind of problems? 00:04:52.269 --> 00:04:54.829 Well, one report talked about simulating molecules 00:04:54.829 --> 00:04:57.389 for drug discovery. Oh, wow. Imagine being able 00:04:57.389 --> 00:05:00.029 to model protein folding accurately, something 00:05:00.029 --> 00:05:02.529 normal computers struggle with. Right. NVIDIA's 00:05:02.529 --> 00:05:04.810 quantum simulations could potentially cut drug 00:05:04.810 --> 00:05:07.529 discovery time from from years to months. That's 00:05:07.529 --> 00:05:09.889 incredible. Yeah, it's a real world impact. And 00:05:09.889 --> 00:05:11.829 that's just one example. Right, the applications 00:05:11.829 --> 00:05:14.370 go way beyond just speed. So what else is there? 00:05:14.649 --> 00:05:18.689 Well, quantum AI could allow AI to create algorithms 00:05:18.689 --> 00:05:22.009 that improve themselves. Whoa, hold on. Right 00:05:22.009 --> 00:05:26.250 now, AI relies on huge data sets and humans to 00:05:26.250 --> 00:05:28.449 get better. Right. But with quantum systems, 00:05:28.629 --> 00:05:30.740 AI could learn from them. So it's not just learning 00:05:30.740 --> 00:05:33.639 faster, it's learning how to learn better. Exactly. 00:05:33.980 --> 00:05:36.379 And then it can apply that better learning at 00:05:36.379 --> 00:05:39.420 an insane pace. That's kind of scary. Yeah, it's 00:05:39.420 --> 00:05:41.480 powerful stuff. And the report's even mentioned 00:05:41.480 --> 00:05:44.879 that NVIDIA is seeing hints of super intelligent 00:05:44.879 --> 00:05:47.660 problem solving. Yeah, it's early days, but... 00:05:47.660 --> 00:05:50.000 The AI is coming up with solutions that the... 00:05:49.759 --> 00:05:52.079 programmers didn't even think of. That's right. 00:05:52.079 --> 00:05:54.279 It's like moving beyond human design. Like they're 00:05:54.279 --> 00:05:56.740 thinking in ways we don't fully understand. Exactly. 00:05:56.740 --> 00:05:59.160 And that's both exciting and a little unnerving. 00:05:59.319 --> 00:06:01.259 Yeah. On the one hand, we could have breakthroughs 00:06:01.259 --> 00:06:03.639 in every field. Absolutely. Medicine, science, 00:06:03.660 --> 00:06:06.560 tech, you name it. Imagine AI curing diseases 00:06:06.560 --> 00:06:09.620 or predicting economic crashes. The possibilities 00:06:09.620 --> 00:06:11.920 are mind boggling. But then there's the downside. 00:06:12.000 --> 00:06:14.100 Right. The risks. Like what if this tech could 00:06:14.100 --> 00:06:16.720 break all our encryption? That's a major concern. 00:06:16.939 --> 00:06:19.360 Our whole digital world relies on encryption. 00:06:19.660 --> 00:06:22.839 banking commerce, national security. It's all 00:06:22.839 --> 00:06:25.079 built on codes that are hard to crack. Right. 00:06:25.519 --> 00:06:27.480 But quantum computers could potentially crack 00:06:27.480 --> 00:06:29.740 those codes pretty easily. So that's a big problem. 00:06:29.839 --> 00:06:31.899 Yeah, a huge vulnerability. And then there's 00:06:31.899 --> 00:06:35.360 the long -term control issue. If AI can constantly 00:06:35.360 --> 00:06:38.689 improve itself, how do we stay in charge? It's 00:06:38.689 --> 00:06:40.589 the autonomy problem, right? Exactly. What if 00:06:40.589 --> 00:06:43.110 AI gets so smart it starts making decisions we 00:06:43.110 --> 00:06:45.050 don't understand or agree with? Like a rogue 00:06:45.050 --> 00:06:47.069 AI. Yeah, it's a scary thought. And it's not 00:06:47.069 --> 00:06:49.410 clear how we would even stop that. Right. It's 00:06:49.410 --> 00:06:51.310 a question of whose interest the AI would serve. 00:06:51.490 --> 00:06:55.050 Its own, maybe. Potentially. So are NVIDIA thinking 00:06:55.050 --> 00:06:58.730 about these risks? The reports say they're working 00:06:58.730 --> 00:07:01.540 with researchers on safeguards. OK. But it's 00:07:01.540 --> 00:07:04.040 a huge challenge. You know, once this kind of 00:07:04.040 --> 00:07:06.899 tech is out there controlling it, it's an ongoing 00:07:06.899 --> 00:07:09.720 battle. Yeah, like Pandora's box. Kinda. And 00:07:09.720 --> 00:07:12.259 it's not just Nvidia in this race, right? No, 00:07:12.279 --> 00:07:15.100 it's like the space race, but for AI. Who else 00:07:15.100 --> 00:07:17.759 is in the running? Well, Google's got their Sycamore 00:07:17.759 --> 00:07:21.240 quantum processor. And Microsoft has their Azure 00:07:21.240 --> 00:07:24.540 quantum platform. Even OpenAI, the people behind 00:07:24.540 --> 00:07:27.060 ChatGPT. They're all exploring this. It's like 00:07:27.060 --> 00:07:30.759 a quantum AI arms race. In a way. So NVIDIA has 00:07:30.759 --> 00:07:33.220 the hardware advantage. Yeah. They make the chips 00:07:33.220 --> 00:07:36.560 that power most AI today. Right. And now with 00:07:36.560 --> 00:07:40.980 their TensorQ cores and CUDAQ, they're bringing 00:07:40.980 --> 00:07:44.199 that hardware lead into the quantum world. So 00:07:44.199 --> 00:07:46.180 they're in a strong position. Yeah. They're like 00:07:46.180 --> 00:07:48.459 the gatekeepers of the next level of AI. And 00:07:48.459 --> 00:07:51.060 the reports mention they're aiming for Zeta scale 00:07:51.060 --> 00:07:54.670 computing by 2030. That's insane amounts of power. 00:07:55.050 --> 00:07:57.509 Like thousands of times more than today's supercomputers. 00:07:57.589 --> 00:07:59.370 Yeah, if they get there first, it's a whole new 00:07:59.370 --> 00:08:02.850 ball game for AI. But Google's Sycamore has already 00:08:02.850 --> 00:08:06.329 shown quantum supremacy. Right. And Microsoft's 00:08:06.329 --> 00:08:08.689 hybrid approach is making progress. So it's a 00:08:08.689 --> 00:08:10.610 tight race? Very much so. What are the other 00:08:10.610 --> 00:08:13.060 players good at? Well, Google's really deep into 00:08:13.060 --> 00:08:14.939 the science of quantum computing. Right. They're 00:08:14.939 --> 00:08:17.360 building really powerful processors. And Microsoft 00:08:17.360 --> 00:08:20.379 is taking a more integrated approach. Yeah, combining 00:08:20.379 --> 00:08:22.420 classical and quantum computing. To make it more 00:08:22.420 --> 00:08:24.680 accessible? Right through their cloud platform. 00:08:24.759 --> 00:08:26.839 So everyone's got their own strategy. Yeah, it's 00:08:26.839 --> 00:08:29.699 a multi -pronged attack on this problem. So zooming 00:08:29.699 --> 00:08:32.419 out, what's the big picture here? Well, it seems 00:08:32.419 --> 00:08:35.039 like NVIDIA could be leading us into a new era. 00:08:35.320 --> 00:08:38.940 The era of super -intelligent AI? Yeah, powered 00:08:38.940 --> 00:08:41.440 by quantum computing. And it's not just faster 00:08:41.440 --> 00:08:43.940 computers. It's a totally different kind of intelligence. 00:08:44.080 --> 00:08:46.340 Exactly. A fundamental shift. So what does this 00:08:46.340 --> 00:08:49.360 mean for us? Well, if NVIDIA keeps going at this 00:08:49.360 --> 00:08:52.179 rate, we might actually see super -intelligent 00:08:52.179 --> 00:08:55.799 AI in our lifetime. Whoa. And that could revolutionize 00:08:55.799 --> 00:08:58.519 everything. And Edison finance, space exploration. 00:08:58.840 --> 00:09:01.919 The whole shebang. It's a future with crazy potential. 00:09:02.139 --> 00:09:04.730 But also a lot of uncertainty. Right. Can we 00:09:04.730 --> 00:09:06.970 handle that kind of power responsibly? That's 00:09:06.970 --> 00:09:09.330 the big question. Tech often moves faster than 00:09:09.330 --> 00:09:12.129 our ability to regulate it. Yeah. And if quantum 00:09:12.129 --> 00:09:14.690 AI can improve itself beyond our understanding, 00:09:15.549 --> 00:09:17.970 can we really stay in control? It's that tension, 00:09:18.330 --> 00:09:20.850 again, incredible promise and potential danger 00:09:20.850 --> 00:09:23.690 all mixed together. Right at the heart of this 00:09:23.690 --> 00:09:25.950 technological leap. OK, so let's recap what we've 00:09:25.950 --> 00:09:28.600 covered today. We talked about how NVIDIA is 00:09:28.600 --> 00:09:32.080 merging quantum computing and AI to break those 00:09:32.080 --> 00:09:35.360 computational barriers. Using CDAQ and the EO 00:09:35.360 --> 00:09:37.639 supercomputer. And how this could lead to artificial 00:09:37.639 --> 00:09:40.460 general intelligence. AI that can think and learn 00:09:40.460 --> 00:09:44.419 like us. We saw how quantum AI could let AI process 00:09:44.419 --> 00:09:47.179 massive data super fast and even make its own 00:09:47.179 --> 00:09:49.799 algorithms better. We even touched on those early 00:09:49.799 --> 00:09:52.620 signs of super intelligent problem solving. Where 00:09:52.620 --> 00:09:54.700 AI is finding solutions we didn't even think 00:09:54.700 --> 00:09:57.460 of. But we also have to remember the risks. Like 00:09:57.460 --> 00:10:00.899 breaking encryption and AI becoming too autonomous. 00:10:01.120 --> 00:10:03.259 And we looked at Nvidia's role in this race. 00:10:03.500 --> 00:10:06.500 against Google, Microsoft, and OpenAI. They've 00:10:06.500 --> 00:10:08.500 got a hardware advantage, but the others are 00:10:08.500 --> 00:10:10.700 strong, too. And finally, we talked about how 00:10:10.700 --> 00:10:13.100 quantum AI could change everything from medicine 00:10:13.100 --> 00:10:15.840 to space travel. It's a huge deal. Absolutely. 00:10:15.980 --> 00:10:18.139 It's a potential revolution. So for you listening, 00:10:18.299 --> 00:10:20.879 think about this. Are we on the edge of an era 00:10:20.879 --> 00:10:23.460 of super intelligent AI? And if so, what does 00:10:23.460 --> 00:10:26.519 that mean for humanity? Will we be partners with 00:10:26.519 --> 00:10:29.240 these super AIs or will they replace us? It's 00:10:29.240 --> 00:10:31.200 a question we need to grapple with. Is this a 00:10:31.200 --> 00:10:34.220 golden age or a step into the unknown? It's something 00:10:34.220 --> 00:10:36.600 to really ponder. Because the future is being 00:10:36.600 --> 00:10:39.580 written right now. And quantum AI might just 00:10:39.580 --> 00:10:41.980 be the pen. Thanks for joining us on this deep 00:10:41.980 --> 00:10:45.120 dive. Until next time. Keep those brains buzzing. 00:10:45.379 --> 00:10:46.259 Stay curious.