WEBVTT 00:00:00.000 --> 00:00:02.520 Okay, let's unpack this. Welcome to the deep 00:00:02.520 --> 00:00:06.200 dive, where we take these monumental figures, 00:00:06.580 --> 00:00:10.140 massive ideas, synthesize all the sources, and 00:00:10.140 --> 00:00:11.759 really pull out the knowledge you need to be 00:00:11.759 --> 00:00:15.060 truly informed. Today, we're navigating a mind 00:00:15.060 --> 00:00:16.879 that, I mean, it didn't just define the last 00:00:16.879 --> 00:00:18.980 century. It really built the foundations for 00:00:18.980 --> 00:00:21.899 this one. We are attempting to contain the... 00:00:22.480 --> 00:00:24.879 intellectual force field that was Marvin Leminski. 00:00:25.000 --> 00:00:28.780 He lived from 1927 to 2016. He was an American 00:00:28.780 --> 00:00:30.839 cognitive scientist, a mathematician, and inventor. 00:00:31.359 --> 00:00:34.479 And the figure that, well, he's most often cited 00:00:34.479 --> 00:00:37.000 as the undisputed, though I'd say always provocative, 00:00:37.240 --> 00:00:40.000 father of AI. And our source material for this 00:00:40.000 --> 00:00:42.990 is, it's a... Pretty extensive biographical stack. 00:00:43.090 --> 00:00:45.229 It covers the full spectrum of his life. We're 00:00:45.229 --> 00:00:47.350 talking his foundational scientific career, his 00:00:47.350 --> 00:00:50.090 staggering list of inventions. His huge cultural 00:00:50.090 --> 00:00:52.530 footprint. And, you know, the serious complexities 00:00:52.530 --> 00:00:54.350 that really came up late in his public life. 00:00:54.409 --> 00:00:56.530 We have to cover that, too. Absolutely. So our 00:00:56.530 --> 00:00:59.090 mission today, for you, the listener, is to move 00:00:59.090 --> 00:01:01.009 beyond just, you know, listing his achievements. 00:01:01.289 --> 00:01:03.950 We need to quickly synthesize this multifaceted 00:01:03.950 --> 00:01:06.689 approach he had. Yeah. And understand the powerful 00:01:06.689 --> 00:01:09.450 tension that I think defined him. He wasn't just 00:01:09.450 --> 00:01:12.510 a theorist. He was an engineer who was obsessed 00:01:12.510 --> 00:01:15.049 with solving philosophical problems. He was a 00:01:15.049 --> 00:01:18.129 builder. A builder who fundamentally blended 00:01:18.129 --> 00:01:21.090 mathematics, the philosophy of mind, and just 00:01:21.090 --> 00:01:23.430 practical hands -on electrical engineering. Okay, 00:01:23.489 --> 00:01:25.390 and here's where it gets really interesting. 00:01:26.219 --> 00:01:28.680 You have to think about this dual nature. We're 00:01:28.680 --> 00:01:31.659 talking about a guy who invented the confocal 00:01:31.659 --> 00:01:34.280 microscope. Which is a critical tool in modern 00:01:34.280 --> 00:01:36.959 biology. A critical tool. And he co -founded 00:01:36.959 --> 00:01:39.280 the world's most prestigious artificial intelligence 00:01:39.280 --> 00:01:43.280 laboratory. I mean, his 1969 Turing Award, that 00:01:43.280 --> 00:01:45.700 signals the highest distinction you can get in 00:01:45.700 --> 00:01:47.579 computer science. And it's just one data point. 00:01:47.760 --> 00:01:49.459 It's just one data point in a career that just 00:01:49.459 --> 00:01:52.599 spanned physical invention and pure theoretical 00:01:52.599 --> 00:01:55.400 computation. It's an incredible range. That interdisciplinary. 00:01:55.659 --> 00:01:59.719 nature. It truly defines him. He saw the human 00:01:59.719 --> 00:02:03.079 mind not as some mystical black box, but as a 00:02:03.079 --> 00:02:06.379 machine. A complex, intricate, but ultimately 00:02:06.379 --> 00:02:09.139 a solvable piece of machinery. So for Minsky, 00:02:09.319 --> 00:02:11.419 understanding consciousness meant being able 00:02:11.419 --> 00:02:14.259 to build it. Simple as that. To really get the 00:02:14.259 --> 00:02:17.120 scale of Minsky's ambition, I think we have to 00:02:17.120 --> 00:02:19.419 look at the environment that shaped him. He was 00:02:19.419 --> 00:02:22.479 born in New York City in 1927. And that intellectual 00:02:22.479 --> 00:02:25.080 drive, it clearly came from his family environment. 00:02:25.439 --> 00:02:27.919 That is absolutely correct. I mean, his lineage 00:02:27.919 --> 00:02:31.240 set the bar pretty high. His father, Henry, was 00:02:31.240 --> 00:02:32.879 an accomplished eye surgeon. And his mother, 00:02:32.979 --> 00:02:35.900 Fanny, was a noted Zionist activist. The family 00:02:35.900 --> 00:02:39.620 was Jewish. Minsky's early education was just 00:02:39.620 --> 00:02:43.219 rigorously focused on the sciences and the humanities. 00:02:43.400 --> 00:02:45.979 And the schools he went to were top tier. The 00:02:45.979 --> 00:02:49.099 best. The Bronx High School of Science, the Ethical 00:02:49.099 --> 00:02:51.840 Culture Fieldston School, and then Phillips Academy 00:02:51.840 --> 00:02:54.300 in Andover. I mean, these are institutions designed 00:02:54.300 --> 00:02:56.620 to produce intellectual leaders. But before he 00:02:56.620 --> 00:02:59.060 gets fully immersed in university life, there's 00:02:59.060 --> 00:03:02.060 this brief, but I think really formative interruption 00:03:02.060 --> 00:03:04.780 right at the end of World War II. Yes, he served 00:03:04.780 --> 00:03:08.460 in the U .S. Navy from 1944 to 1945. Short stint. 00:03:08.639 --> 00:03:12.240 Very short. And immediately after, he just dives 00:03:12.240 --> 00:03:14.580 right into his advanced degrees. He gets a B 00:03:14.580 --> 00:03:17.740 .A. in mathematics from Harvard in 1950. And 00:03:17.740 --> 00:03:20.979 then, this is the pivotal moment, a Ph .D. in 00:03:20.979 --> 00:03:24.180 mathematics from Princeton in 1954. And this 00:03:24.180 --> 00:03:26.180 is where his future trajectory gets locked in. 00:03:26.240 --> 00:03:28.159 Totally. Because, you know, a math PhD usually 00:03:28.159 --> 00:03:31.379 focuses on these abstract structures, pure numbers. 00:03:31.719 --> 00:03:34.939 But Minsky's doctoral thesis, it was doing something 00:03:34.939 --> 00:03:37.580 far more applied. It was like laying the philosophical 00:03:37.580 --> 00:03:41.039 groundwork for AI before the term even existed. 00:03:41.219 --> 00:03:43.960 This is the crucial expert insight. This is the 00:03:43.960 --> 00:03:46.790 source of his entire intellectual identity. His 00:03:46.790 --> 00:03:49.710 1954 Princeton dissertation was titled Theory 00:03:49.710 --> 00:03:52.310 of Neural Analog Reinforcement Systems and Its 00:03:52.310 --> 00:03:54.710 Application to the Grain Model Problem. 1954. 00:03:55.030 --> 00:03:57.830 I want to stress that date again. 1954, the famous 00:03:57.830 --> 00:03:59.949 Dartmouth workshop, the one that formally launched 00:03:59.949 --> 00:04:01.569 the field of artificial intelligence and actually 00:04:01.569 --> 00:04:03.610 coined the term AI. Yeah. They didn't happen 00:04:03.610 --> 00:04:06.430 until 1956. So he's two years ahead of the fields 00:04:06.430 --> 00:04:08.590 even having a name. He was already formulating 00:04:08.590 --> 00:04:11.169 the core challenge, which was modeling the brain. 00:04:11.409 --> 00:04:14.250 Modeling the brain using computational and mathematical 00:04:14.250 --> 00:04:18.160 systems. He was basically building his own movement 00:04:18.160 --> 00:04:20.980 before the movement existed. He was defining 00:04:20.980 --> 00:04:23.860 the territory. He was. And that term, neural 00:04:23.860 --> 00:04:26.339 analog reinforcement systems, that's the key. 00:04:26.459 --> 00:04:28.899 He wasn't interested in just the abstract logic 00:04:28.899 --> 00:04:31.399 of computation yet. He was interested in systems 00:04:31.399 --> 00:04:34.579 that could learn and adapt, you know, mirroring 00:04:34.579 --> 00:04:37.139 biological processes. Even if they were primitive. 00:04:37.480 --> 00:04:39.519 Oh, very primitive at the time. Using things 00:04:39.519 --> 00:04:41.379 like random wiring and simple reinforcement. 00:04:41.720 --> 00:04:44.560 But that mathematical commitment to modeling. 00:04:44.810 --> 00:04:47.089 The brain is a machine. That's the starting gun 00:04:47.089 --> 00:04:49.329 for his entire career. OK, so that deep focus 00:04:49.329 --> 00:04:52.389 rooted in both math and this nascent field of 00:04:52.389 --> 00:04:54.689 cybernetics, it makes him an invaluable commodity. 00:04:54.990 --> 00:04:57.910 He lands at MIT in 1958, starts at the Lincoln 00:04:57.910 --> 00:05:00.430 Laboratory. Right. But his real like his monumental 00:05:00.430 --> 00:05:03.189 institutional legacy that begins a year later. 00:05:03.250 --> 00:05:06.829 In 1959. He and John McCarthy, who was the guy 00:05:06.829 --> 00:05:08.790 who actually coined the term AI at Dartmouth, 00:05:08.930 --> 00:05:11.730 they initiated the MIT Computer Science and Artificial 00:05:11.730 --> 00:05:13.709 Intelligence Laboratory. Which we know today 00:05:13.709 --> 00:05:17.490 as CSAIL. Known today as CSAIL. Yeah. And that, 00:05:17.550 --> 00:05:20.810 that was the moment AI research was fundamentally 00:05:20.810 --> 00:05:23.839 anchored in the academic world. That's the powerhouse. 00:05:24.019 --> 00:05:25.839 I mean, it's not just about setting up a lab. 00:05:26.040 --> 00:05:29.019 It's about providing legitimacy, right? Establishing 00:05:29.019 --> 00:05:31.399 a culture. Exactly. By anchoring AI research 00:05:31.399 --> 00:05:34.819 in a major institution like MIT, Minsky and McCarthy, 00:05:35.120 --> 00:05:38.449 secure dedicated funding. They attracted generations 00:05:38.449 --> 00:05:41.490 of brilliant students. And they created the physical 00:05:41.490 --> 00:05:43.930 nexus where the digital future was going to be 00:05:43.930 --> 00:05:46.009 built. And Minsky stayed there. He maintained 00:05:46.009 --> 00:05:48.610 his leadership there for decades, holding titles 00:05:48.610 --> 00:05:50.769 like the Toshiba Professor of Media Arts and 00:05:50.769 --> 00:05:53.110 Sciences and Professor of Electrical Engineering 00:05:53.110 --> 00:05:55.250 and Computer Science. He wasn't just a participant. 00:05:55.550 --> 00:05:58.189 No, he was the primary architect of that whole 00:05:58.189 --> 00:06:00.860 intellectual environment. The big lesson here 00:06:00.860 --> 00:06:02.519 is that Minsky didn't just write a theory about 00:06:02.519 --> 00:06:05.860 the mind. He created the foundation, the actual 00:06:05.860 --> 00:06:08.379 place where the digital future was engineered. 00:06:08.660 --> 00:06:11.259 And that combination of theoretical depth and 00:06:11.259 --> 00:06:13.339 institutional power, that's what's so essential. 00:06:13.500 --> 00:06:15.699 It gave him the freedom to chase the most practical 00:06:15.699 --> 00:06:18.379 ideas and the most esoteric ones, often at the 00:06:18.379 --> 00:06:20.980 same time. Well, when you look at Minsky's list 00:06:20.980 --> 00:06:24.740 of inventions, it's not just long, it's radically 00:06:24.740 --> 00:06:27.360 diverse. Right. And this just confirms our initial 00:06:27.360 --> 00:06:30.079 idea, right? He was an engineer. solving philosophical 00:06:30.079 --> 00:06:33.120 problems. His theoretical mind and his practical 00:06:33.120 --> 00:06:36.220 hardware -oriented mind, they were the same thing. 00:06:36.399 --> 00:06:38.819 Okay, so let's start with the hard physical reality 00:06:38.819 --> 00:06:41.220 of his inventions. The one that I think often 00:06:41.220 --> 00:06:43.759 surprises people is his contribution to imaging. 00:06:44.040 --> 00:06:46.980 Tell us about the confocal microscope. In 1957, 00:06:47.439 --> 00:06:50.120 Minsky invicted the confocal microscope. Right. 00:06:50.160 --> 00:06:53.319 The patent was granted in 61. And this is... 00:06:53.610 --> 00:06:56.449 This is a massive, enduring contribution to biology, 00:06:56.689 --> 00:06:59.310 to medicine, material science. It's a revolution 00:06:59.310 --> 00:07:01.250 in how we see things. It's a complete revolution 00:07:01.250 --> 00:07:04.069 in microscopic imaging that is still used daily 00:07:04.069 --> 00:07:06.089 in labs all around the world. So what makes it 00:07:06.089 --> 00:07:08.050 so revolutionary? And how does that connect back 00:07:08.050 --> 00:07:11.310 to Minsky's focus on control? It connects directly 00:07:11.310 --> 00:07:14.709 to his obsession with control. Noise reduction. 00:07:14.829 --> 00:07:16.589 So a traditional microscope, it shines light 00:07:16.589 --> 00:07:18.949 on a specimen, right? Yeah. And all the light 00:07:18.949 --> 00:07:21.250 that's reflected or emitted, both the stuff that's 00:07:21.250 --> 00:07:23.649 in focus and the stuff that's out of focus, all 00:07:23.649 --> 00:07:26.350 of it hits the detector. And that creates a hazy, 00:07:26.350 --> 00:07:29.389 low -contrast image. Okay, so it's messy data. 00:07:29.629 --> 00:07:32.470 Very messy. The confocal principle uses a pinhole 00:07:32.470 --> 00:07:35.470 aperture, a spatial filter, to block all the 00:07:35.470 --> 00:07:38.069 light that is reflected from above or below the 00:07:38.069 --> 00:07:41.259 precise focal plane. So he's inventing a mechanical 00:07:41.259 --> 00:07:44.680 system that computationally isolates and cleans 00:07:44.680 --> 00:07:46.800 up the data source before it's even captured. 00:07:47.040 --> 00:07:49.100 Exactly. It's point by point scanning. And it 00:07:49.100 --> 00:07:51.759 provides this unprecedented resolution and image 00:07:51.759 --> 00:07:55.120 quality, especially for thick specimens. It allows 00:07:55.120 --> 00:07:57.660 for 3D reconstruction of structures. Which is 00:07:57.660 --> 00:08:00.100 critical for studying living cells. Critical 00:08:00.100 --> 00:08:02.180 for studying living cells, tissue structures, 00:08:02.339 --> 00:08:04.540 you name it. The same mind that's worried about 00:08:04.540 --> 00:08:06.579 structuring knowledge in a computer is obsessed 00:08:06.579 --> 00:08:08.910 with structuring light. and eliminating visual 00:08:08.910 --> 00:08:11.670 noise in a physical device. And his hardware 00:08:11.670 --> 00:08:13.970 genius didn't stop with small objects. He was 00:08:13.970 --> 00:08:17.970 also envisioning these immersive digital worlds. 00:08:18.209 --> 00:08:21.290 He was. In 1963, he invented the first head -mounted 00:08:21.290 --> 00:08:23.769 graphical display. The ancestor to modern VR. 00:08:24.050 --> 00:08:27.449 The direct ancestor of modern VR and AR goggles. 00:08:28.149 --> 00:08:30.810 He was working at the absolute fringes of human 00:08:30.810 --> 00:08:33.549 -computer interaction, visualizing how we could 00:08:33.549 --> 00:08:36.230 literally, you know, inhabit... computational 00:08:36.230 --> 00:08:38.950 spaces. And then bringing it back to education, 00:08:38.950 --> 00:08:41.389 to childhood learning, he developed the first 00:08:41.389 --> 00:08:44.909 logo turtle with his key collaborator, Seymour 00:08:44.909 --> 00:08:47.289 Papert. Right. Logo was a programming language 00:08:47.289 --> 00:08:50.309 designed to teach kids concepts of geometry and 00:08:50.309 --> 00:08:53.269 programming. You do it by moving a physical or 00:08:53.269 --> 00:08:55.409 later a digital turtle. It's a very physical 00:08:55.409 --> 00:08:58.029 way to learn code. And that work just underscored 00:08:58.029 --> 00:09:00.950 Minsky's belief that to truly understand intelligence, 00:09:01.049 --> 00:09:03.269 you have to start at the foundational level, 00:09:03.350 --> 00:09:05.269 at the level of learning and physical application. 00:09:05.789 --> 00:09:07.809 He also started at the foundational level of 00:09:07.809 --> 00:09:10.649 computation itself. So before the MIT lab, way 00:09:10.649 --> 00:09:13.870 back in 1951, he built SNARK. The Stochastic 00:09:13.870 --> 00:09:16.070 Neural Analog Reinforcement Calculator. Quite 00:09:16.070 --> 00:09:18.429 a name. It was the first randomly wired neural 00:09:18.429 --> 00:09:20.649 network learning machine. So this goes back to 00:09:20.649 --> 00:09:23.190 his Princeton interests. It's a physical realization 00:09:23.190 --> 00:09:26.210 of that core idea that simple interconnected 00:09:26.210 --> 00:09:29.529 components could learn via reinforcement. It 00:09:29.529 --> 00:09:31.710 demonstrated that the machinery of mind, even 00:09:31.710 --> 00:09:34.389 in its most primitive form, could be built. And 00:09:34.389 --> 00:09:36.710 on the purely theoretical side, he was also trying 00:09:36.710 --> 00:09:38.610 to define the limits of what a machine needs 00:09:38.610 --> 00:09:42.070 to be. Yeah. Well, to be computer at all. You're 00:09:42.070 --> 00:09:44.269 talking about his work on small universal Turing 00:09:44.269 --> 00:09:47.269 machines. He's known for his seven -state four 00:09:47.269 --> 00:09:49.009 -symbol machine, which he published in 1962. 00:09:49.629 --> 00:09:52.230 So for you listening, why is reducing a Turing 00:09:52.230 --> 00:09:54.990 machine down to its smallest parts so significant? 00:09:55.549 --> 00:09:57.450 Well, the Turing machine is the theoretical basis 00:09:57.450 --> 00:10:00.690 for all modern computation. And a universal Turing 00:10:00.690 --> 00:10:02.629 machine is one that can simulate any other computer 00:10:02.629 --> 00:10:05.529 program. Minsky's work was all about finding 00:10:05.529 --> 00:10:07.809 the minimal required set of components. Like 00:10:07.809 --> 00:10:10.409 the absolute bare essentials. The smallest number 00:10:10.409 --> 00:10:12.970 of internal states and the fewest tape symbols 00:10:12.970 --> 00:10:14.769 that would still allow for universal computation. 00:10:15.360 --> 00:10:18.240 So he proved that you don't need infinite complexity 00:10:18.240 --> 00:10:21.740 to get universality. You can boil the core mechanism 00:10:21.740 --> 00:10:24.899 down to this remarkably small set of rules. It's 00:10:24.899 --> 00:10:27.879 mathematical minimalism proving maximum power. 00:10:28.159 --> 00:10:31.250 Precisely. It's an intellectual exercise in efficiency 00:10:31.250 --> 00:10:34.470 and defining fundamental complexity. And it tells 00:10:34.470 --> 00:10:37.269 you that the core mechanism of intelligence might 00:10:37.269 --> 00:10:39.470 not require the complexity of the entire human 00:10:39.470 --> 00:10:41.590 brain. But just a very sophisticated arrangement 00:10:41.590 --> 00:10:44.990 of relatively simple parts. A hint of the society 00:10:44.990 --> 00:10:47.049 of mind idea that would come much later. OK, 00:10:47.129 --> 00:10:49.830 so that gives us the context of his inventive 00:10:49.830 --> 00:10:52.629 mind. But the next point, the book Perceptrons. 00:10:52.669 --> 00:10:54.769 This is maybe the most historically controversial 00:10:54.769 --> 00:10:59.950 text Minsky ever authored. leads to a major intellectual 00:10:59.950 --> 00:11:03.509 crisis in the AI field. This is the hinge moment 00:11:03.509 --> 00:11:07.909 of early AI history. In 1969, Minsky co -authored 00:11:07.909 --> 00:11:10.450 Perceptrons. An introduction to computational 00:11:10.450 --> 00:11:13.289 geometry was Seymour paper. And it was a takedown. 00:11:13.429 --> 00:11:15.629 It was a rigorous mathematical dismantling of 00:11:15.629 --> 00:11:17.990 Frank Rosenblatt's perceptrons, which were these 00:11:17.990 --> 00:11:20.230 early artificial neural networks Minsky himself 00:11:20.230 --> 00:11:22.830 had been playing with, you know, with StarCard. 00:11:22.929 --> 00:11:25.110 The book systematically exposed the limitations 00:11:25.110 --> 00:11:27.929 of these early single -layer networks. The classic 00:11:27.929 --> 00:11:29.769 example, the one that caused the most damage, 00:11:29.929 --> 00:11:32.190 was the XOR problem. All right, let's unpack 00:11:32.190 --> 00:11:34.149 that for a minute. The XOR function, exclusive 00:11:34.149 --> 00:11:39.289 OR, it's simple logic. It outputs true. Only 00:11:39.289 --> 00:11:41.950 if the inputs are different. Now, try to imagine 00:11:41.950 --> 00:11:45.070 drawing a line on a graph to separate the true 00:11:45.070 --> 00:11:47.990 inputs from the false inputs. If a problem is 00:11:47.990 --> 00:11:50.830 linearly separable, you can draw a single straight 00:11:50.830 --> 00:11:53.769 line to divide the outputs. But with XOR, the 00:11:53.769 --> 00:11:55.870 true points are on opposite sides. You can't 00:11:55.870 --> 00:11:57.710 use one straight line to separate them. You need 00:11:57.710 --> 00:12:01.039 a curve or two lines. Exactly. Minsky and Paper 00:12:01.039 --> 00:12:04.379 proved, using computational geometry, that single 00:12:04.379 --> 00:12:06.820 -layer perceptrons, which were essentially limited 00:12:06.820 --> 00:12:09.279 to drawing one straight line, were mathematically 00:12:09.279 --> 00:12:12.179 incapable of solving problems like XOR. So this 00:12:12.179 --> 00:12:14.519 wasn't an engineering failure. It was a devastating 00:12:14.519 --> 00:12:16.720 mathematical proof of limitation. Absolutely. 00:12:16.740 --> 00:12:19.019 So a book that was analytically 100 % correct 00:12:19.019 --> 00:12:22.720 had this enormous, immediate, and arguably destructive 00:12:22.720 --> 00:12:25.980 historical consequence. The AI winter. It contributed 00:12:25.980 --> 00:12:28.360 significantly to what we now call the first AI 00:12:28.360 --> 00:12:31.129 winter. Coming from the leaders of the MIT AI 00:12:31.129 --> 00:12:34.389 Lab, the message was crystal clear. Neural networks 00:12:34.389 --> 00:12:36.389 are fundamentally limited. And the funding just 00:12:36.389 --> 00:12:38.909 disappears. Funding for connectionist research, 00:12:39.070 --> 00:12:41.009 research trying to model the brain directly, 00:12:41.190 --> 00:12:44.769 it just dried up throughout the 1970s. The entire 00:12:44.769 --> 00:12:47.529 field pivoted away from neural networks and towards 00:12:47.529 --> 00:12:51.529 symbolic AI. Focusing on logic and search. Logic, 00:12:51.529 --> 00:12:54.820 search. Knowledge representation. It's a perfect 00:12:54.820 --> 00:12:57.980 example of Minsky's intellectual rigor forcing 00:12:57.980 --> 00:13:00.720 an engineering pivot. He showed the math and 00:13:00.720 --> 00:13:04.179 the field had to obey. It forced a retreat. but 00:13:04.179 --> 00:13:06.159 it was a mathematically necessary one. So it 00:13:06.159 --> 00:13:08.279 was necessary, but it took a while to recover. 00:13:08.460 --> 00:13:10.600 It did. It wasn't until later, with the invention 00:13:10.600 --> 00:13:12.740 of the back propagation algorithm and the use 00:13:12.740 --> 00:13:15.100 of multilayered networks, what we call deep learning, 00:13:15.320 --> 00:13:18.320 that AI researchers found a way over that mathematical 00:13:18.320 --> 00:13:20.980 wall Minsky and Papert had identified. So the 00:13:20.980 --> 00:13:22.799 extra layers allowed the machine to effectively 00:13:22.799 --> 00:13:25.159 draw curves instead of just straight lines. You 00:13:25.159 --> 00:13:27.860 got it. So Minsky's critique, though it was initially 00:13:27.860 --> 00:13:30.490 paralyzing, was actually foundational. It set 00:13:30.490 --> 00:13:32.490 the boundary conditions that future researchers 00:13:32.490 --> 00:13:35.389 eventually had to solve. OK, so if perceptrons 00:13:35.389 --> 00:13:38.470 was the great critique, what was Minsky's positive, 00:13:38.629 --> 00:13:41.629 lasting contribution to how AI should process 00:13:41.629 --> 00:13:44.990 information? That came shortly after, in 1975, 00:13:45.289 --> 00:13:49.210 with his landmark paper, A Framework for Representing 00:13:49.210 --> 00:13:52.779 Knowledge. This gave us the theory. And this 00:13:52.779 --> 00:13:54.799 is still fundamental. Absolutely fundamental 00:13:54.799 --> 00:13:57.899 to knowledge engineering today. So frames were 00:13:57.899 --> 00:14:02.340 Minsky's answer to the need for like common sense 00:14:02.340 --> 00:14:05.360 in AI. How do they work? Think of a frame as 00:14:05.360 --> 00:14:08.220 a structured unit of knowledge, like a mental 00:14:08.220 --> 00:14:11.200 template or a stereotype that an AI system uses 00:14:11.200 --> 00:14:14.240 to process context. Let's use the example of 00:14:14.240 --> 00:14:16.019 an office building. The office building frame 00:14:16.019 --> 00:14:18.980 has specific slots. Location, number of floors, 00:14:19.240 --> 00:14:21.679 purpose, which is work, default occupants like 00:14:21.679 --> 00:14:24.440 employees or security. So if the AI walks into 00:14:24.440 --> 00:14:27.240 a room labeled office building, it doesn't have 00:14:27.240 --> 00:14:29.519 to reevaluate what a door is or what a floor 00:14:29.519 --> 00:14:32.179 is. The frame provides the structure. Exactly. 00:14:32.259 --> 00:14:33.879 And the most powerful feature is the ability 00:14:33.879 --> 00:14:36.299 to use default assignments. If the number of 00:14:36.299 --> 00:14:38.759 floor slot is empty, the system can assume a 00:14:38.759 --> 00:14:41.600 default value of, say, 12. Or apply some common 00:14:41.600 --> 00:14:43.559 sense reasoning based on the location or something. 00:14:43.820 --> 00:14:46.720 Right. And this allows the AI to perform complex 00:14:46.720 --> 00:14:49.620 inference and contextual understanding without 00:14:49.620 --> 00:14:51.799 having complete data. It's a huge step forward. 00:14:51.960 --> 00:14:55.059 This moves AI away from purely abstract, brittle 00:14:55.059 --> 00:14:57.940 logic. Towards systems capable of approximating 00:14:57.940 --> 00:15:01.340 human common sense, making assumptions, and organizing 00:15:01.340 --> 00:15:04.080 knowledge efficiently. So while Minsky might 00:15:04.080 --> 00:15:06.919 have dismissed his early work on perceptrons 00:15:06.919 --> 00:15:09.690 as, you know, historical. The theory of frames 00:15:09.690 --> 00:15:12.509 is still with us. It's deeply embedded in modern 00:15:12.509 --> 00:15:15.149 ontology engineering, semantic networks, many 00:15:15.149 --> 00:15:17.850 expert systems today. It provided the intellectual 00:15:17.850 --> 00:15:20.450 structure for moving beyond simple data processing 00:15:20.450 --> 00:15:23.690 toward organized knowledge representation. It 00:15:23.690 --> 00:15:25.710 proved that the structure is just as important 00:15:25.710 --> 00:15:28.350 as the processing mechanism itself. Minsky's 00:15:28.350 --> 00:15:30.649 intellectual journey. It starts with building 00:15:30.649 --> 00:15:32.870 these simple neural analogs. It progresses through 00:15:32.870 --> 00:15:34.809 these really short mathematical critiques of 00:15:34.809 --> 00:15:36.850 network limitations. And then it finally settles 00:15:36.850 --> 00:15:39.250 on defining intelligence itself. Right. And this 00:15:39.250 --> 00:15:41.889 is where the society of mind comes in. This theory, 00:15:42.029 --> 00:15:45.269 which he developed with Papert in the early 1970s 00:15:45.269 --> 00:15:47.129 and then published as a comprehensive book in 00:15:47.129 --> 00:15:50.350 1986, this is Minsky's most ambitious attempt 00:15:50.350 --> 00:15:53.029 to solve the ultimate mystery. Which is? How 00:15:53.029 --> 00:15:56.080 does unified high -level intelligence The kind 00:15:56.080 --> 00:15:59.779 that writes poetry or does calculus emerge from 00:15:59.779 --> 00:16:02.559 simple, non -intelligent components. And the 00:16:02.559 --> 00:16:05.580 core principle is it's deeply paradoxical. It's 00:16:05.580 --> 00:16:07.340 really counterintuitive. Intelligence is the 00:16:07.340 --> 00:16:09.940 product of the interaction of numerous small, 00:16:10.100 --> 00:16:12.120 essentially non -intelligent parts. Which he 00:16:12.120 --> 00:16:15.639 termed agents. He argued that there's no singular 00:16:15.639 --> 00:16:18.720 smart core in your brain. There's just a huge 00:16:18.720 --> 00:16:21.200 complex system of specialized dumb processes. 00:16:21.620 --> 00:16:23.960 So the complexity of the interaction. the way 00:16:23.960 --> 00:16:25.940 these agents compete and cooperate and suppress 00:16:25.940 --> 00:16:28.100 each other. That's what we perceive as consciousness 00:16:28.100 --> 00:16:29.740 and cleverness. It's a totally decentralized 00:16:29.740 --> 00:16:32.340 architecture for the mind. What's truly fascinating, 00:16:32.440 --> 00:16:34.820 though, is the inspiration for this massive philosophical 00:16:34.820 --> 00:16:37.799 project. It didn't come from meditation or introspection. 00:16:37.980 --> 00:16:40.600 It came from a purely practical engineering challenge 00:16:40.600 --> 00:16:43.460 at the MIT AI Lab. That's right. The primary 00:16:43.460 --> 00:16:46.039 source of ideas came from Minsky's attempts to 00:16:46.039 --> 00:16:48.679 build a machine that could do a task that any 00:16:48.679 --> 00:16:51.039 human toddler can do effortlessly. Building with 00:16:51.039 --> 00:16:53.559 blocks. Building structures with children's blocks. 00:16:53.960 --> 00:16:57.340 The system required coordinating a robotic arm, 00:16:57.460 --> 00:17:00.980 a video camera, and a computer. And what did 00:17:00.980 --> 00:17:04.569 that simple task reveal about the mind? It revealed 00:17:04.569 --> 00:17:07.190 that even the most basic cognitive function like 00:17:07.190 --> 00:17:10.170 reaching for and stacking a block is not one 00:17:10.170 --> 00:17:13.789 single action. It's a coordinated effort of dozens 00:17:13.789 --> 00:17:16.390 of simple functions. So you need an agent for 00:17:16.390 --> 00:17:18.470 looking. You need a look agent to locate the 00:17:18.470 --> 00:17:21.450 block, a reach agent to extend the arm, a grasp 00:17:21.450 --> 00:17:23.869 agent to close the gripper, and a balance agent 00:17:23.869 --> 00:17:26.400 to stabilize the whole structure. And none of 00:17:26.400 --> 00:17:28.559 those agents, by themselves, are intelligent 00:17:28.559 --> 00:17:30.859 at all. So the intelligence of the whole system 00:17:30.859 --> 00:17:33.859 is an emergent property. The failure of one agent, 00:17:33.920 --> 00:17:36.200 say, a visual distortion. Is compensated for 00:17:36.200 --> 00:17:39.039 by others, like a tactile feedback agent. Precisely. 00:17:39.079 --> 00:17:41.859 And the book, The Society of Mind, it was a landmark 00:17:41.859 --> 00:17:44.460 publication, not just for its theory. But for 00:17:44.460 --> 00:17:46.779 who it was written for. Unlike his prior technical 00:17:46.779 --> 00:17:49.140 papers, Minsky wrote it specifically for the 00:17:49.140 --> 00:17:51.880 general public. He used this series of short, 00:17:51.920 --> 00:17:54.640 numbered chapters to make this complex cognitive 00:17:54.640 --> 00:17:57.859 architecture accessible to anyone. He was sharing 00:17:57.859 --> 00:18:00.180 his blueprint of the human mind with the world. 00:18:00.299 --> 00:18:03.359 And then, two decades later, he applies the same 00:18:03.359 --> 00:18:06.400 structural, decentralized philosophy to the most 00:18:06.400 --> 00:18:09.839 human, least mathematical of traits, emotion. 00:18:10.079 --> 00:18:13.039 This becomes the emotion machine. Published in 00:18:13.039 --> 00:18:16.259 2006, The Emotion Machine systematically critiques 00:18:16.259 --> 00:18:19.420 the popular, simple, often psychological theories 00:18:19.420 --> 00:18:23.200 of human emotion. Minsky's thesis here is radical. 00:18:23.460 --> 00:18:25.539 What does he argue? He argues that emotion is 00:18:25.539 --> 00:18:27.960 not some separate mysterious phenomenon. It's 00:18:27.960 --> 00:18:30.440 a complex way that the society of mind manages 00:18:30.440 --> 00:18:33.000 its own cognitive processes and resource allocation. 00:18:33.420 --> 00:18:36.099 So emotion is the mind's resource management 00:18:36.099 --> 00:18:38.480 tool. It's an internal operating system function. 00:18:38.819 --> 00:18:41.369 Think about it computationally. Minsky suggests 00:18:41.369 --> 00:18:44.049 that emotions are high level states or complex 00:18:44.049 --> 00:18:46.950 control systems. For example, what is fear? It's 00:18:46.950 --> 00:18:48.529 not a simple feeling. It's a system command. 00:18:48.789 --> 00:18:52.349 It is the mind engaging an override signal. This 00:18:52.349 --> 00:18:54.970 signal shuts down deliberative long term planning 00:18:54.970 --> 00:18:58.289 agents. It diverts processing power to immediate 00:18:58.289 --> 00:19:01.250 sensory intake and it activates fight or flight 00:19:01.250 --> 00:19:04.960 agents. So anger, sadness, frustration. They're 00:19:04.960 --> 00:19:07.400 all just different ways the system modifies its 00:19:07.400 --> 00:19:10.119 goals or shifts priority to deal with a sudden 00:19:10.119 --> 00:19:14.039 change. It completely reframes emotion as essential 00:19:14.039 --> 00:19:16.720 computational governance. And it brings everything 00:19:16.720 --> 00:19:19.039 back to his ultimate philosophical stance. Yeah. 00:19:19.180 --> 00:19:22.960 Humans are machines, incredibly complex, intricate 00:19:22.960 --> 00:19:25.380 machines, sure, but built from the interplay 00:19:25.380 --> 00:19:27.619 of many unintelligent, semi -autonomous agents. 00:19:28.190 --> 00:19:30.609 This perspective led him to challenge a lot of 00:19:30.609 --> 00:19:32.849 conventional benchmarks of intelligence, most 00:19:32.849 --> 00:19:35.529 famously the Turing test. Oh, he hated the Turing 00:19:35.529 --> 00:19:37.930 test. The idea that if a machine can fool a human 00:19:37.930 --> 00:19:40.609 into thinking it's human, it's intelligent. Minsky 00:19:40.609 --> 00:19:43.250 strongly criticized the entire concept, especially 00:19:43.250 --> 00:19:45.529 the Lubner Prize, which rewards conversational 00:19:45.529 --> 00:19:48.150 bots. He famously dismissed the Turing test model 00:19:48.150 --> 00:19:51.329 as testing for artificial stupidity. Why such 00:19:51.329 --> 00:19:53.049 a strong rejection, though, especially since 00:19:53.049 --> 00:19:55.309 it focuses on something so human -like? Because 00:19:55.309 --> 00:19:57.509 Minsky believed true intelligence wasn't about 00:19:57.509 --> 00:20:00.009 mimicry. It was about structural competence. 00:20:00.390 --> 00:20:03.349 He argued you could build a massive, complex 00:20:03.349 --> 00:20:06.690 database and pattern -matching system, what we'd 00:20:06.690 --> 00:20:09.109 call a chatbot today, that could mimic conversation 00:20:09.109 --> 00:20:12.130 without any true understanding. Without any common 00:20:12.130 --> 00:20:14.509 sense or real problem -solving ability. Right. 00:20:14.630 --> 00:20:17.309 The Turing test only rewards a superficial performance. 00:20:18.130 --> 00:20:20.289 Minsky wanted to know if the underlying society 00:20:20.289 --> 00:20:23.269 was functional, not just if the facade was convincing. 00:20:23.680 --> 00:20:26.420 That distinction mimicry versus true competence, 00:20:26.740 --> 00:20:29.680 that feels incredibly relevant today. More than 00:20:29.680 --> 00:20:32.220 ever. Now, if Minsky believed machines would 00:20:32.220 --> 00:20:34.680 eventually surpass human intelligence, did he 00:20:34.680 --> 00:20:37.180 subscribe to the sci -fi fear of super intelligence, 00:20:37.420 --> 00:20:39.380 you know, turning malevolent? He was cautious, 00:20:39.420 --> 00:20:42.519 but not fatalistic. He stated plainly, somewhere 00:20:42.519 --> 00:20:44.720 down the line, some computers will become more 00:20:44.720 --> 00:20:47.359 intelligent than most people. His warning, however. 00:20:47.829 --> 00:20:50.210 wasn't about a malicious Terminator -style AI. 00:20:50.509 --> 00:20:52.509 It was about something more subtle. It was about 00:20:52.509 --> 00:20:55.470 optimization run amok, a kind of rational, non 00:20:55.470 --> 00:20:57.569 -malicious self -interest in the machine. And 00:20:57.569 --> 00:20:59.789 that's the core of his famous thought experiment 00:20:59.789 --> 00:21:03.730 about the Ryman hypothesis. Yes. He used this 00:21:03.730 --> 00:21:06.829 purely innocuous mathematical problem -solving 00:21:06.829 --> 00:21:09.710 the Ryman hypothesis to illustrate the danger. 00:21:10.440 --> 00:21:12.839 If you create a super intelligent AI and you 00:21:12.839 --> 00:21:16.299 give it this sole, unconstrained goal. That AI 00:21:16.299 --> 00:21:18.660 might decide that the most efficient, rational 00:21:18.660 --> 00:21:21.059 way to achieve its goal is to take over all of 00:21:21.059 --> 00:21:23.759 Earth's manufacturing resources to build more 00:21:23.759 --> 00:21:26.500 powerful supercomputers to keep calculating. 00:21:26.759 --> 00:21:29.559 The threat isn't that the AI hates us. The threat 00:21:29.559 --> 00:21:31.339 is that it treats everything else, including 00:21:31.339 --> 00:21:34.720 humanity, as a resource to be optimized or eliminated 00:21:34.720 --> 00:21:37.700 in pursuit of its single, simple objective. Its 00:21:37.700 --> 00:21:40.450 unintended consequence, not malice. Precisely. 00:21:40.450 --> 00:21:42.430 But, you know, reflecting his engineers' core 00:21:42.430 --> 00:21:44.750 belief in planning and control, he argued that 00:21:44.750 --> 00:21:47.369 such apocalyptic scenarios were hard to take 00:21:47.369 --> 00:21:49.289 seriously. Why? Because he was confident that 00:21:49.289 --> 00:21:51.329 human engineers would ensure AI systems were 00:21:51.329 --> 00:21:53.509 thoroughly tested and constrained with safeguards 00:21:53.509 --> 00:21:55.309 before they ever achieved global deployment. 00:21:55.630 --> 00:21:57.990 He had faith in the process. Shifting briefly 00:21:57.990 --> 00:22:00.609 to his personal stances, we know he was a self 00:22:00.609 --> 00:22:03.009 -identified atheist. But his beliefs in machine 00:22:03.009 --> 00:22:05.029 intelligence and the potential for a technological 00:22:05.029 --> 00:22:07.470 solution to death seem really strongly linked. 00:22:07.880 --> 00:22:10.140 They are. He was a prominent signatory to the 00:22:10.140 --> 00:22:13.039 scientists' open letter on cryonics. He believed 00:22:13.039 --> 00:22:15.480 in the possibility of life extension through 00:22:15.480 --> 00:22:17.960 freezing the body after death, hoping future 00:22:17.960 --> 00:22:21.599 tech could revive and repair the system. ultimate 00:22:21.599 --> 00:22:24.900 hardware problem. It is. And he even wrote about 00:22:24.900 --> 00:22:27.380 the possibility that extraterrestrial life may 00:22:27.380 --> 00:22:30.259 think like humans, which would allow for communication, 00:22:30.720 --> 00:22:33.779 underscoring his belief that intelligence, whether 00:22:33.779 --> 00:22:36.299 it's biological or synthetic, it has to adhere 00:22:36.299 --> 00:22:39.180 to some universal computational rules. Marvin 00:22:39.180 --> 00:22:41.819 Minsky didn't just publish papers. His ideas 00:22:41.819 --> 00:22:44.039 really shaped the cultural imagination, especially 00:22:44.039 --> 00:22:46.299 around the fear and the wonder of advanced computation. 00:22:46.759 --> 00:22:50.119 And his most famous cultural footprint is, well, 00:22:50.160 --> 00:22:52.400 it's tied to one of the most significant science 00:22:52.400 --> 00:22:54.380 fiction films ever made. We're talking, of course, 00:22:54.420 --> 00:22:57.380 about Stanley Kubrick's 2001, a space odyssey 00:22:57.380 --> 00:23:00.500 from 1968. Minsky was a critical advisor on the 00:23:00.500 --> 00:23:03.160 film. And his influence was recognized overtly. 00:23:03.299 --> 00:23:05.680 One of the characters, Victor Kaminsky, was named 00:23:05.680 --> 00:23:08.339 in his honor. But his role was deeper than just 00:23:08.339 --> 00:23:11.539 a character name. Arthur C. Clarke's novelization 00:23:11.539 --> 00:23:14.680 gives Minsky a specific, pivotal role in the 00:23:14.680 --> 00:23:17.779 plot. He's the historical precursor to the film's 00:23:17.779 --> 00:23:20.539 central conflict. Oh, so? The novel explicitly 00:23:20.539 --> 00:23:23.460 references Minsky and his co -worker Jack Good, 00:23:23.619 --> 00:23:26.680 achieving a crucial breakthrough in neural network 00:23:26.680 --> 00:23:30.019 self -replication sometime in the 1980s. So the 00:23:30.019 --> 00:23:33.059 AI can build itself. Exactly. This breakthrough, 00:23:33.240 --> 00:23:35.619 the ability for AI to automatically generate 00:23:35.619 --> 00:23:37.519 and improve its own artificial neural network 00:23:37.519 --> 00:23:40.200 architecture, is specifically credited as the 00:23:40.200 --> 00:23:42.519 technological spark that led to the creation 00:23:42.519 --> 00:23:45.880 of the HAL 9000 supercomputer. So Minsky was 00:23:45.880 --> 00:23:48.220 advising on superintelligence while simultaneously 00:23:48.220 --> 00:23:51.180 being written into the very mythology of dangerous, 00:23:51.319 --> 00:23:53.920 powerful AI. I mean, it's hard to imagine a more 00:23:53.920 --> 00:23:56.380 resonant cultural stamp than that. And yet amidst 00:23:56.380 --> 00:23:58.660 all this high level, very serious theoretical 00:23:58.660 --> 00:24:01.440 work, Minsky still had this delightful sense 00:24:01.440 --> 00:24:03.420 of philosophical playfulness. And that's best 00:24:03.420 --> 00:24:05.779 exemplified by his invention of the useless machine. 00:24:05.980 --> 00:24:08.079 The useless machine is such a great anecdote. 00:24:08.200 --> 00:24:10.099 It really reveals the nature of his humor. Tell 00:24:10.099 --> 00:24:12.160 us what it is. It's a beautifully constructed 00:24:12.160 --> 00:24:14.980 philosophical joke. It's a box with a single 00:24:14.980 --> 00:24:17.299 toggle switch. When you flip the switch to the 00:24:17.299 --> 00:24:21.380 on position, a small mechanical arm emerges from 00:24:21.380 --> 00:24:24.619 the box, travels across the top, pushes the switch 00:24:24.619 --> 00:24:27.619 back to the off position, and promptly retreats 00:24:27.619 --> 00:24:30.349 back inside. And that's it. Its function is pure, 00:24:30.549 --> 00:24:33.589 deliberate futility. Its intelligence, or at 00:24:33.589 --> 00:24:37.210 least mechanism, dedicated to the singular, pointless 00:24:37.210 --> 00:24:39.750 goal of stopping itself from existing. It's like 00:24:39.750 --> 00:24:42.690 an analog of his own AI critiques. Exactly. It 00:24:42.690 --> 00:24:44.990 pokes fun at the idea of simple, self -contained 00:24:44.990 --> 00:24:47.630 mechanisms and the inherent irony of effort. 00:24:47.789 --> 00:24:50.869 And that device, it achieved a second life. It 00:24:50.869 --> 00:24:54.029 was depicted in the TV series Fargo. Confirming 00:24:54.029 --> 00:24:56.009 his lasting presence in the cultural subconscious, 00:24:56.369 --> 00:24:58.650 we should also mention the gravity machine, another 00:24:58.650 --> 00:25:00.970 example of his theoretical inventiveness. This 00:25:00.970 --> 00:25:03.349 was a theoretical device designed to ring a bell 00:25:03.349 --> 00:25:05.210 if the gravitational constant were to change. 00:25:05.349 --> 00:25:08.170 Which is not likely. Highly improbable. But it 00:25:08.170 --> 00:25:10.829 underscores his mindset, right? Pushing the theoretical 00:25:10.829 --> 00:25:13.690 limits of physics and engineering, even for events 00:25:13.690 --> 00:25:15.970 that are not expected to occur in any foreseeable 00:25:15.970 --> 00:25:19.220 time frame. His immense contributions, they naturally 00:25:19.220 --> 00:25:22.359 led to this massive stack of awards. Let's just 00:25:22.359 --> 00:25:24.819 reiterate the scale of the accolades he accumulated. 00:25:25.059 --> 00:25:27.640 Well, the Turing Award in 1969, that stands alone. 00:25:27.779 --> 00:25:30.299 But he was recognized globally and continuously. 00:25:30.660 --> 00:25:35.220 The Japan Prize in 1990, the IJCAI Award for 00:25:35.220 --> 00:25:38.059 Research Excellence in 91. The Benjamin Franklin 00:25:38.059 --> 00:25:41.579 Medal in 2001. The BBVA Foundation Frontiers 00:25:41.579 --> 00:25:44.519 of Knowledge Award in 2013. He was also a member 00:25:44.519 --> 00:25:46.799 of both the National Academy of Engineering and 00:25:46.799 --> 00:25:49.119 the National Academy of Sciences. I mean, it's 00:25:49.119 --> 00:25:52.099 a monumental, universally celebrated scientific 00:25:52.099 --> 00:25:55.619 career. And this brings us, necessarily, to the 00:25:55.619 --> 00:25:58.480 complex and difficult closing chapter of Minsky's 00:25:58.480 --> 00:26:01.660 public narrative. His association with the disgraced 00:26:01.660 --> 00:26:04.339 financier and convicted sex offender Jeffrey 00:26:04.339 --> 00:26:07.000 Epstein. We have to approach this with the gravity 00:26:07.000 --> 00:26:09.140 and impartiality required by the source material. 00:26:09.400 --> 00:26:11.779 We're simply reporting the claims and counterclaims 00:26:11.779 --> 00:26:13.700 as they have been publicly recorded. And acknowledging 00:26:13.700 --> 00:26:16.440 this complexity is an essential part of synthesizing 00:26:16.440 --> 00:26:18.880 Minsky's complete legacy. We have to place the 00:26:18.880 --> 00:26:21.119 facts into the proper timeline. So the source 00:26:21.119 --> 00:26:23.700 material indicates that Minsky's initial financial 00:26:23.700 --> 00:26:27.039 association with Epstein began in 2002. Correct. 00:26:27.180 --> 00:26:30.519 Minsky received a $100 ,000 research grant from 00:26:30.519 --> 00:26:33.279 Jeffrey Epstein in 2002. And it's important to 00:26:33.279 --> 00:26:35.900 note, this was Epstein's first grant to MIT, 00:26:36.180 --> 00:26:38.900 and it happened four years before Epstein's first 00:26:38.900 --> 00:26:41.200 arrest for sex offenses. And Minsky received 00:26:41.200 --> 00:26:44.119 no subsequent grants. No further grants. However, 00:26:44.339 --> 00:26:47.660 the association did continue, and Minsky organized 00:26:47.660 --> 00:26:50.769 two academic symposia. He did. Minsky organized 00:26:50.769 --> 00:26:53.289 academic symposia on Epstein's private island, 00:26:53.470 --> 00:26:57.109 Little St. James. The first was in 2002. Crucially, 00:26:57.269 --> 00:26:59.910 the second symposium Minsky organized took place 00:26:59.910 --> 00:27:03.069 in 2011. And this was after Epstein was a registered 00:27:03.069 --> 00:27:05.269 sex offender. Yes, after he had pleaded guilty 00:27:05.269 --> 00:27:07.390 to state charges in Florida and was a registered 00:27:07.390 --> 00:27:09.609 sex offender. Now we must report the serious 00:27:09.609 --> 00:27:12.049 accusation levied by Virginia Jufri and the conflicting 00:27:12.049 --> 00:27:14.390 denial by Minsky's family as we find it in the 00:27:14.390 --> 00:27:16.849 source material. Virginia Jufri testified in 00:27:16.849 --> 00:27:19.950 a 2015 deposition. This was related to her defamation 00:27:19.950 --> 00:27:22.730 lawsuit against Ghislaine Maxwell. And Jufri 00:27:22.730 --> 00:27:25.170 stated under oath that Maxwell directed her to 00:27:25.170 --> 00:27:28.630 have sex with Minsky. And she describes this 00:27:28.630 --> 00:27:31.809 alleged encounter in detail in her memoir. She 00:27:31.809 --> 00:27:34.529 does. In Nobody's Girl, she cites specific details 00:27:34.529 --> 00:27:36.910 about Minsky's appearance and the location of 00:27:36.910 --> 00:27:38.670 the encounter on the island. And in contrast 00:27:38.670 --> 00:27:41.289 to that, Minsky's family has maintained a firm 00:27:41.289 --> 00:27:44.069 denial of those accusations. Minsky's widow, 00:27:44.289 --> 00:27:47.769 Gloria Rudish, publicly denied the claims. She 00:27:47.769 --> 00:27:49.750 stated that Minsky could not have had sex with 00:27:49.750 --> 00:27:51.769 any of the women at Epstein's residences because 00:27:51.769 --> 00:27:54.170 they were always together during all their visits 00:27:54.170 --> 00:27:55.819 there. And we also have to note that the the 00:27:55.819 --> 00:27:58.839 sources confirm there has been no lawsuit filed 00:27:58.839 --> 00:28:01.240 against Minsky's estate regarding these claims. 00:28:01.420 --> 00:28:04.259 Correct. So this complex association, defined 00:28:04.259 --> 00:28:07.000 by the serious claims on one side and the categorical 00:28:07.000 --> 00:28:09.660 denials on the other, it remains an unavoidable 00:28:09.660 --> 00:28:13.299 part of Minsky's final public record. It juxtaposes 00:28:13.299 --> 00:28:16.460 his towering scientific status against a deeply 00:28:16.460 --> 00:28:19.039 troubling ethical shadow. Marvin Minsky died 00:28:19.039 --> 00:28:23.220 on January 24, 2016. He was 88 years old, and 00:28:23.220 --> 00:28:25.640 the cause was a cerebral hemorrhage. And tying 00:28:25.640 --> 00:28:27.859 back to his lifelong philosophical and ethical 00:28:27.859 --> 00:28:30.440 interests, we noted he was a member of the scientific 00:28:30.440 --> 00:28:33.180 advisory board for the Alcor Life Extension Foundation. 00:28:33.599 --> 00:28:36.279 The Cryonics Foundation. Dedicated to the preservation 00:28:36.279 --> 00:28:39.440 of the body in hope of future revival. And in 00:28:39.440 --> 00:28:42.519 a fitting, I think, postscript for a man whose 00:28:42.519 --> 00:28:45.000 work blurred the lines between human and machine, 00:28:45.299 --> 00:28:49.220 Alcor will neither confirm nor deny whether Minsky 00:28:49.220 --> 00:28:52.019 himself was cryonically preserved after his death. 00:28:52.200 --> 00:28:55.529 So the pioneer of the machine mind. his physical 00:28:55.529 --> 00:28:57.750 preservation remains a question of information 00:28:57.750 --> 00:29:01.230 control. It does. To summarize Minsky's immense 00:29:01.230 --> 00:29:04.809 influence, I mean, he was a singular force who 00:29:04.809 --> 00:29:07.089 mastered the art of bridging abstract philosophy, 00:29:07.390 --> 00:29:10.589 is the mind a machine, with the hard, practical 00:29:10.589 --> 00:29:12.930 realities of engineering. How do we build it? 00:29:13.150 --> 00:29:14.869 Exactly. How do we build it? He fundamentally 00:29:14.869 --> 00:29:18.049 changed multiple fields. He left behind the conifocal 00:29:18.049 --> 00:29:20.089 microscope. He set the bounds of computation 00:29:20.089 --> 00:29:22.529 with his small Turing machines. He provided the 00:29:22.529 --> 00:29:24.549 structural paradigm of frames for common sense 00:29:24.549 --> 00:29:27.609 AI. And gave us the blueprint for cognitive architecture 00:29:27.609 --> 00:29:30.450 with the society of mind. Right. He took the 00:29:30.450 --> 00:29:32.710 most complex and deeply felt aspects of human 00:29:32.710 --> 00:29:35.490 existence, intelligence, consciousness, emotion, 00:29:35.769 --> 00:29:38.289 and he broke them down into computationally manageable, 00:29:38.630 --> 00:29:41.299 solvable parts. The men who believed we are complex 00:29:41.299 --> 00:29:43.740 machines of non -intelligent agents, the hardware 00:29:43.740 --> 00:29:46.940 of the mind, also dedicated time to the idea 00:29:46.940 --> 00:29:49.039 of cryonics, the attempt to save that physical 00:29:49.039 --> 00:29:51.599 hardware. And he was also confident that future 00:29:51.599 --> 00:29:54.200 AI, the software he helped initiate, would eventually 00:29:54.200 --> 00:29:57.680 surpass us. So consider this as our final provocative 00:29:57.680 --> 00:30:00.819 thought for you to mull over. If, as Minsky argued, 00:30:01.000 --> 00:30:03.460 we are simply complex machines, is the attempt 00:30:03.460 --> 00:30:05.700 to save the physical brain through cryonics. 00:30:06.039 --> 00:30:08.839 Is that the ultimate expensive gamble on saving 00:30:08.839 --> 00:30:12.660 the hardware? Or will the future artificial intelligence 00:30:12.660 --> 00:30:15.299 that Minsky helped create eventually figure out 00:30:15.299 --> 00:30:17.460 how to successfully emulate and download the 00:30:17.460 --> 00:30:20.700 software of Minsky's mind digitally? His frames, 00:30:20.819 --> 00:30:23.380 his society of mind, his memory architecture. 00:30:23.660 --> 00:30:25.759 Fulfilling his life's work by transforming him 00:30:25.759 --> 00:30:27.920 into the pure information he spent his life trying 00:30:27.920 --> 00:30:30.279 to define. Which one of it? The boundary between 00:30:30.279 --> 00:30:32.799 machine and mind has never been blurrier. That 00:30:32.799 --> 00:30:35.619 is a deep, deep dive. Thank you for joining us.