WEBVTT 00:00:00.000 --> 00:00:03.220 When you really look at the fundamental architecture 00:00:03.220 --> 00:00:06.040 of our digital world, the code, the concepts, 00:00:06.219 --> 00:00:08.480 the whole infrastructure, it's kind of built 00:00:08.480 --> 00:00:10.759 on these pillars, you know. And we often hear 00:00:10.759 --> 00:00:13.140 the same few names, Alan Turing, maybe John von 00:00:13.140 --> 00:00:16.440 Neumann or Marvin Minsky. And they are absolutely 00:00:16.440 --> 00:00:18.640 crucial, yes. But if you were to point to just 00:00:18.640 --> 00:00:23.620 one person, one individual who... Simultaneously 00:00:23.620 --> 00:00:26.460 architected the field's philosophy, created its 00:00:26.460 --> 00:00:29.679 main early language, and predicted its modern 00:00:29.679 --> 00:00:31.940 business model. The whole multi -billion dollar 00:00:31.940 --> 00:00:34.539 business model. Exactly. That singular focus. 00:00:34.719 --> 00:00:37.280 Yeah. It really points to John McCarthy. Right. 00:00:37.359 --> 00:00:38.780 And that's our mission today. We're going to 00:00:38.780 --> 00:00:41.060 do a deep dive into the life and, I mean, the 00:00:41.060 --> 00:00:43.340 monumental contributions of John McCarthy, American 00:00:43.340 --> 00:00:45.619 computer scientist, cognitive scientist, and 00:00:45.619 --> 00:00:48.759 one of the true yet sometimes overlooked intellectual 00:00:48.759 --> 00:00:51.759 giants of the 20th century. He really was. We 00:00:51.759 --> 00:00:53.420 want to pull out the core insights that show 00:00:53.420 --> 00:00:56.200 how one person managed to define three, I mean, 00:00:56.219 --> 00:00:58.500 three radically different but equally foundational 00:00:58.500 --> 00:01:02.340 areas of computer science. It's an almost unbelievable 00:01:02.340 --> 00:01:05.560 trifecta of achievement. These are what we call 00:01:05.560 --> 00:01:09.920 his big three claims to fame. And they are, believe 00:01:09.920 --> 00:01:12.400 it or not, still shaping the technology you're 00:01:12.400 --> 00:01:15.219 using right now. So claim number one. Back in 00:01:15.219 --> 00:01:18.180 1956, he officially coined the term artificial 00:01:18.180 --> 00:01:21.659 intelligence. He basically established AI as 00:01:21.659 --> 00:01:24.200 a formal field of study. He gave the future a 00:01:24.200 --> 00:01:26.579 name. Then there's claim number two. He created 00:01:26.579 --> 00:01:29.609 the Lisp programming language. A technical masterpiece, 00:01:29.829 --> 00:01:32.170 really rooted in pure mathematics, which then 00:01:32.170 --> 00:01:35.569 went on to define the entire symbolic and functional 00:01:35.569 --> 00:01:37.730 programming paradigm. And claim number three, 00:01:37.750 --> 00:01:39.489 and this is the one that touches every part of 00:01:39.489 --> 00:01:41.890 the modern Internet. He was the main driver behind 00:01:41.890 --> 00:01:44.530 timesharing. He popularized it. And that is the 00:01:44.530 --> 00:01:47.590 direct conceptual ancestor of modern cloud computing. 00:01:48.480 --> 00:01:50.579 The utility model. The whole thing. That breadth 00:01:50.579 --> 00:01:52.980 is why he's so important. He didn't just contribute 00:01:52.980 --> 00:01:55.120 to a field that already existed. He was literally 00:01:55.120 --> 00:01:57.980 drawing the maps for multiple future fields. 00:01:58.159 --> 00:01:59.980 And to set the stage, you have to understand 00:01:59.980 --> 00:02:02.260 the academic world he was in. We're talking about 00:02:02.260 --> 00:02:04.299 high stakes environments. The absolute top tier. 00:02:04.700 --> 00:02:08.319 McCarthy, who lived from 1927 to 2011. He graduated 00:02:08.319 --> 00:02:11.139 from Caltech, got his Ph .D. in math from Princeton 00:02:11.139 --> 00:02:14.300 in 51. And then he just he proceeded to populate 00:02:14.300 --> 00:02:16.520 the most important intellectual hubs of computing 00:02:16.520 --> 00:02:18.740 history. It's like a grand tour of innovation. 00:02:18.939 --> 00:02:22.180 He went from Dartmouth in 1955, which is where 00:02:22.180 --> 00:02:26.300 the AI story really kicks off, to MIT as a research 00:02:26.300 --> 00:02:29.159 fellow, and then finally to Stanford. He was 00:02:29.159 --> 00:02:31.780 a full professor there from 1962 all the way 00:02:31.780 --> 00:02:34.159 until he retired in 2000. Right. And these are 00:02:34.159 --> 00:02:35.840 the intellectual battlegrounds where computer 00:02:35.840 --> 00:02:38.120 science was literally hammered into shape. Yeah. 00:02:38.240 --> 00:02:41.580 And his peers, his institutions, they all recognized 00:02:41.580 --> 00:02:44.479 the sheer scope of his influence. The awards 00:02:44.479 --> 00:02:46.900 he won really confirmed this duality. status 00:02:46.900 --> 00:02:49.780 he had as both a brilliant engineer and a profound 00:02:49.780 --> 00:02:51.159 scientific theorist. They were talking about 00:02:51.159 --> 00:02:53.439 the big ones. The biggest. He received the Turing 00:02:53.439 --> 00:02:56.479 Award in 1971 for his work in AI, the prestigious 00:02:56.479 --> 00:02:59.639 Kyoto Prize in 1988, and the United States National 00:02:59.639 --> 00:03:02.240 Medal of Science in 1990. I mean, these aren't 00:03:02.240 --> 00:03:04.819 just lifetime achievement awards. They're a recognition 00:03:04.819 --> 00:03:07.439 that he laid the structural bedrock for almost 00:03:07.439 --> 00:03:10.139 everything that came after. OK, so to understand 00:03:10.139 --> 00:03:12.300 the architect, we have to understand the environment 00:03:12.300 --> 00:03:15.360 that, you know, that shaped him and his early 00:03:15.360 --> 00:03:19.520 life was well, it was decidedly atypical for 00:03:19.520 --> 00:03:22.240 a figure who would later be seen as a pillar 00:03:22.240 --> 00:03:24.879 of Silicon Valley conservatism. Oh, very much 00:03:24.879 --> 00:03:27.979 so. John McCarthy was born in Boston in 1927, 00:03:28.180 --> 00:03:31.500 and his heritage was really defined by the melting 00:03:31.500 --> 00:03:34.500 pot of that era. His father was an Irish immigrant 00:03:34.500 --> 00:03:37.259 and his mother, a Lithuanian Jewish immigrant. 00:03:37.610 --> 00:03:39.789 So that dual background was just the start of 00:03:39.789 --> 00:03:42.810 a childhood that was full of, well, intellectual 00:03:42.810 --> 00:03:45.370 rigor and some pretty intense politics. It was. 00:03:45.409 --> 00:03:47.030 The family moved around a lot. They didn't really 00:03:47.030 --> 00:03:48.949 stay put. This was during the Great Depression, 00:03:49.069 --> 00:03:50.810 so they were constantly relocating until they 00:03:50.810 --> 00:03:53.110 finally settled in Los Angeles. Why Los Angeles? 00:03:53.169 --> 00:03:55.629 His father found stable work there as an organizer 00:03:55.629 --> 00:03:58.069 for the Amalgamated Clothing Workers Union. Ah, 00:03:58.289 --> 00:04:00.889 a union organizer. Exactly. And that move to 00:04:00.889 --> 00:04:03.750 LA coincided with a really deep political commitment. 00:04:04.189 --> 00:04:05.870 Both of his parents were active members of the 00:04:05.870 --> 00:04:08.710 Communist Party during the 1930s. Which, looking 00:04:08.710 --> 00:04:11.949 back from today, is. It's a very specific political 00:04:11.949 --> 00:04:15.150 affiliation. It is. But for the young McCarthy, 00:04:15.349 --> 00:04:17.389 what it really meant was growing up in an environment 00:04:17.389 --> 00:04:22.089 where debate, critical reading, rigorous analysis, 00:04:22.370 --> 00:04:25.170 questioning authority, all of that wasn't just 00:04:25.170 --> 00:04:27.529 allowed. It was actively encouraged. It was a 00:04:27.529 --> 00:04:29.870 fiercely intellectual upbringing, even if the 00:04:29.870 --> 00:04:32.389 ideology itself was controversial. So his home 00:04:32.389 --> 00:04:35.050 life was basically a constant seminar. But the 00:04:35.050 --> 00:04:38.230 actual spark for science, that came from a pretty 00:04:38.230 --> 00:04:40.970 unexpected source, didn't it? It did. His scientific 00:04:40.970 --> 00:04:43.509 curiosity wasn't sparked by, you know, an American 00:04:43.509 --> 00:04:46.189 textbook. It came from a Russian popular science 00:04:46.189 --> 00:04:49.189 book for children called I'm 100 ,000 Wise. The 00:04:49.189 --> 00:04:51.529 Russian kids book. Yeah. The book was a collection 00:04:51.529 --> 00:04:54.509 of scientific facts, explanations, all presented 00:04:54.509 --> 00:04:57.240 in a really accessible way. And it just. It completely 00:04:57.240 --> 00:04:59.439 captivated him. And this wasn't just some brief 00:04:59.439 --> 00:05:01.459 fascination. It inspired him to become fluent 00:05:01.459 --> 00:05:03.819 in Russian later in life. Which led him to talk 00:05:03.819 --> 00:05:06.980 to scientists in the Soviet Union. Exactly. He 00:05:06.980 --> 00:05:08.920 maintained these intellectual friendships even 00:05:08.920 --> 00:05:11.439 during the height of the Cold War. And that curiosity. 00:05:12.350 --> 00:05:14.629 It quickly turned into some serious academic 00:05:14.629 --> 00:05:16.649 talent. I mean, he didn't just graduate high 00:05:16.649 --> 00:05:18.550 school early. He basically skipped the first 00:05:18.550 --> 00:05:21.069 two years of college math. That's right. He taught 00:05:21.069 --> 00:05:24.290 himself advanced college -level mathematics during 00:05:24.290 --> 00:05:27.430 his teens. He just studied the textbooks that 00:05:27.430 --> 00:05:29.870 were used at Caltech. So by the time he actually 00:05:29.870 --> 00:05:32.250 arrived on campus, he could immediately test 00:05:32.250 --> 00:05:34.910 out of all the foundational math courses. That 00:05:34.910 --> 00:05:38.009 level of self -directed learning is... It's just 00:05:38.009 --> 00:05:40.009 incredible. It's remarkable. But for all that 00:05:40.009 --> 00:05:42.370 brilliance, he was still, you know, a teenager. 00:05:42.670 --> 00:05:46.610 And there's this deeply humanizing story. The 00:05:46.610 --> 00:05:49.350 future architect of AI was actually suspended 00:05:49.350 --> 00:05:51.870 from Caltech. It's the great quirk in his record. 00:05:52.620 --> 00:05:55.220 And it wasn't for academic reasons. He was suspended 00:05:55.220 --> 00:05:57.579 for failure to attend physical education classes. 00:05:57.720 --> 00:06:00.139 For skipping gym class. Basically, yeah. It was 00:06:00.139 --> 00:06:02.560 such a non -academic bureaucratic issue. He had 00:06:02.560 --> 00:06:04.480 to take a detour. He served briefly in the U 00:06:04.480 --> 00:06:07.040 .S. Army before being readmitted to finally get 00:06:07.040 --> 00:06:10.139 his Bachelor of Science in math in 1948. It's 00:06:10.139 --> 00:06:12.379 a great reminder that even geniuses have to, 00:06:12.379 --> 00:06:15.240 you know, deal with the mundane rules of life. 00:06:15.379 --> 00:06:18.120 So when he finished his undergrad work, he went 00:06:18.120 --> 00:06:21.329 straight to Princeton for his Ph .D. Was the 00:06:21.329 --> 00:06:23.829 path to computation already set in stone for 00:06:23.829 --> 00:06:26.670 him or was there another singular moment of inspiration? 00:06:27.129 --> 00:06:29.709 The path was definitely solidifying, but there 00:06:29.709 --> 00:06:32.290 was one event. While he was still at Caltech, 00:06:32.470 --> 00:06:35.810 he attended a lecture given by the, well, the 00:06:35.810 --> 00:06:37.910 towering figure of modern computing, John von 00:06:37.910 --> 00:06:39.790 Neumann. Oh, wow. And von Neumann's lecture, 00:06:39.870 --> 00:06:41.750 which was all about the logical structure of 00:06:41.750 --> 00:06:45.329 computation and self -reproducing automata. Yeah. 00:06:45.490 --> 00:06:48.240 It was, by all accounts, electrifying. That was 00:06:48.240 --> 00:06:50.459 the experience that really cemented McCarthy's 00:06:50.459 --> 00:06:53.980 decision to move beyond pure math and dedicate 00:06:53.980 --> 00:06:56.899 his life to formalizing thought through machines. 00:06:57.339 --> 00:06:59.399 OK, let's circle back to his political roots 00:06:59.399 --> 00:07:01.399 for a second. You mentioned he was raised in 00:07:01.399 --> 00:07:03.920 this committed communist household, but he died 00:07:03.920 --> 00:07:06.930 a staunch conservative Republican. That ideological 00:07:06.930 --> 00:07:09.689 shift is fascinating. It is, because it proves 00:07:09.689 --> 00:07:11.569 that the critical thinking skills his parents 00:07:11.569 --> 00:07:14.629 taught him ultimately trumped the specific ideology 00:07:14.629 --> 00:07:16.829 they held. So it wasn't a rejection of their 00:07:16.829 --> 00:07:19.889 method, just their conclusion. Exactly. His transformation 00:07:19.889 --> 00:07:22.310 was driven entirely by direct experience. I mean, 00:07:22.310 --> 00:07:24.329 despite his friendships and fluency in Russian, 00:07:24.509 --> 00:07:27.750 his visits to the Soviet bloc, particularly after 00:07:27.750 --> 00:07:30.009 witnessing the Soviet invasion of Czechoslovakia 00:07:30.009 --> 00:07:34.149 in 1968, that forced a profound reevaluation 00:07:34.149 --> 00:07:37.199 of his principles. What was it specifically about 00:07:37.199 --> 00:07:39.819 the invasion of Czechoslovakia that catalyzed 00:07:39.819 --> 00:07:41.620 that shift? Well, the invasion, which was known 00:07:41.620 --> 00:07:44.379 as the Prague Spring, was just this brutal suppression 00:07:44.379 --> 00:07:46.819 of liberal reform movement. For someone like 00:07:46.819 --> 00:07:49.639 McCarthy, who valued intellectual freedom and 00:07:49.639 --> 00:07:52.339 rigor above all else. To see that crushed. To 00:07:52.339 --> 00:07:54.899 witness that raw authoritarianism, the denial 00:07:54.899 --> 00:07:57.600 of free expression, the huge gap between Soviet 00:07:57.600 --> 00:08:00.180 propaganda and the reality on the ground. It 00:08:00.180 --> 00:08:03.040 was intolerable to him. His shift wasn't a rejection 00:08:03.040 --> 00:08:05.000 of his parents' value of intellectual independence. 00:08:05.259 --> 00:08:07.860 It was the application of that very independence 00:08:07.860 --> 00:08:10.819 to empirical data. So he looked at the evidence 00:08:10.819 --> 00:08:13.639 and changed his mind. He did. He became a conservative 00:08:13.639 --> 00:08:16.879 because he believed the structures he observed 00:08:16.879 --> 00:08:19.040 in the West, especially around free markets and 00:08:19.040 --> 00:08:22.199 limited government, offered a more logical and 00:08:22.199 --> 00:08:26.060 a more robust path to freedom than the systems 00:08:26.060 --> 00:08:27.980 he'd witnessed firsthand behind the Iron Curtain. 00:08:28.079 --> 00:08:30.680 And that commitment to logical rigor, you know, 00:08:30.699 --> 00:08:33.340 regardless of where it led him politically, that 00:08:33.340 --> 00:08:35.580 translated directly into his professional life. 00:08:35.700 --> 00:08:38.769 And that leads us to his first monumental achievement, 00:08:39.110 --> 00:08:42.049 actually defining the discipline of artificial 00:08:42.049 --> 00:08:44.590 intelligence. Right. The field needed a common 00:08:44.590 --> 00:08:48.389 banner. It needed a shared goal. Research into 00:08:48.389 --> 00:08:50.950 thinking machines had been floating around under 00:08:50.950 --> 00:08:53.190 different names, automotive theory, cybernetics. 00:08:53.269 --> 00:08:55.450 Kind of all over the place. Exactly. But McCarthy 00:08:55.450 --> 00:08:57.190 and his colleagues, they knew they were doing 00:08:57.190 --> 00:08:59.480 something fundamentally different. And that all 00:08:59.480 --> 00:09:03.820 culminated in the famous 1956 Dartmouth Summer 00:09:03.820 --> 00:09:06.019 Research Project on Artificial Intelligence. 00:09:06.399 --> 00:09:08.480 This wasn't just another conference. It was a 00:09:08.480 --> 00:09:10.779 watershed moment. It was the formal establishment 00:09:10.779 --> 00:09:13.299 of the field. McCarthy co -authored the proposal 00:09:13.299 --> 00:09:16.240 for that project alongside giants like Martin 00:09:16.240 --> 00:09:18.639 Minsky, Nathaniel Rochester and Claudia Shannon. 00:09:18.840 --> 00:09:21.320 And the objective they stated in that proposal. 00:09:22.139 --> 00:09:25.100 It was incredibly ambitious. It was. They wrote 00:09:25.100 --> 00:09:27.799 that the goal was to proceed on the basis of 00:09:27.799 --> 00:09:30.340 the conjecture that every aspect of learning 00:09:30.340 --> 00:09:33.580 or any other feature of intelligence can in principle 00:09:33.580 --> 00:09:36.980 be so precisely described that a machine can 00:09:36.980 --> 00:09:38.980 be made to simulate it. And it's right there 00:09:38.980 --> 00:09:41.100 in the proposal, the naming ceremony. They needed 00:09:41.100 --> 00:09:43.899 a term to capture that ambitious goal. And that 00:09:43.899 --> 00:09:47.600 term was artificial intelligence or AI. By coining 00:09:47.600 --> 00:09:50.700 it. McCarthy really did two things. First, he 00:09:50.700 --> 00:09:52.840 gave the field a distinct identity, you know, 00:09:52.860 --> 00:09:54.840 separating it from general computing or control 00:09:54.840 --> 00:09:57.840 theory. And second, he set a high but very clearly 00:09:57.840 --> 00:10:01.139 defined goal. the creation of intelligent artifacts. 00:10:01.500 --> 00:10:03.559 What's so important here is that McCarthy didn't 00:10:03.559 --> 00:10:06.179 just stop at naming the field. He immediately 00:10:06.179 --> 00:10:08.500 started to define the logical structure that 00:10:08.500 --> 00:10:11.039 would be necessary to actually realize that name. 00:10:11.240 --> 00:10:14.220 Absolutely. His 1958 proposal for the advice 00:10:14.220 --> 00:10:16.659 taker is a perfect example of this. Now, this 00:10:16.659 --> 00:10:19.139 wasn't a physical program yet. It was a conceptual 00:10:19.139 --> 00:10:21.919 design for an AI system based on logical principles. 00:10:22.159 --> 00:10:24.740 So what would the advice taker do? The idea was 00:10:24.740 --> 00:10:27.059 that it would process knowledge facts and rules 00:10:27.059 --> 00:10:30.000 given to it and then use formal logical deduction 00:10:30.000 --> 00:10:33.120 to derive conclusions. So essentially, it would 00:10:33.120 --> 00:10:35.779 answer questions and plan actions based on the 00:10:35.779 --> 00:10:38.139 advice it had stored. How was that different 00:10:38.139 --> 00:10:40.200 from how computers were being used at the time? 00:10:40.419 --> 00:10:42.320 Well, at that time, computers were basically 00:10:42.320 --> 00:10:44.879 just number crunchers. They were designed to 00:10:44.879 --> 00:10:48.259 execute very specific, predefined algorithms 00:10:48.259 --> 00:10:51.320 for calculations like missile projectories or 00:10:51.320 --> 00:10:53.590 accounting. Right. The advice taker, though, 00:10:53.669 --> 00:10:56.049 introduced this idea of representing knowledge 00:10:56.049 --> 00:10:59.649 symbolically and applying general logical inference 00:10:59.649 --> 00:11:02.509 rules to that knowledge. It suggested a machine 00:11:02.509 --> 00:11:05.169 that could reason, not just calculate. And that 00:11:05.169 --> 00:11:07.129 was the seminal concept that inspired all the 00:11:07.129 --> 00:11:09.490 later research in logic programming and sophisticated 00:11:09.490 --> 00:11:12.090 question answering systems. And an ambition like 00:11:12.090 --> 00:11:14.330 that immediately attracted the best minds. So 00:11:14.330 --> 00:11:16.350 you start to see collaboration, but also some 00:11:16.350 --> 00:11:18.870 intense intellectual competition shaping the 00:11:18.870 --> 00:11:20.649 field. Oh, definitely. The early AI community 00:11:20.649 --> 00:11:23.230 was tiny and that competition was formative. 00:11:23.409 --> 00:11:26.529 You had McCarthy, Minsky, Herbert Simon, Alan 00:11:26.529 --> 00:11:28.990 Newell. That was the core group. So when Minsky, 00:11:29.210 --> 00:11:31.269 who is a key collaborator on the Dartmouth proposal, 00:11:31.610 --> 00:11:35.190 joined McCarthy at MIT in 1959. The stage was 00:11:35.190 --> 00:11:37.690 set. It was set for some of the most intense, 00:11:37.789 --> 00:11:40.809 productive rivalry in academic history. They 00:11:40.809 --> 00:11:43.769 were racing, often against each other, to realize 00:11:43.769 --> 00:11:46.950 the promise of AI. And that competitive energy 00:11:46.950 --> 00:11:49.889 just fueled this exponential progress in the 00:11:49.889 --> 00:11:54.149 late 50s and 60s. So if naming AI was the theoretical 00:11:54.149 --> 00:11:57.230 foundation that his second monumental contribution. 00:11:57.820 --> 00:11:59.899 the creation of the Lisp programming language. 00:12:00.299 --> 00:12:02.799 That was the indispensable toolkit. And we need 00:12:02.799 --> 00:12:05.100 to do a genuine deep dive here because the significance 00:12:05.100 --> 00:12:08.000 of Lisp is often misunderstood. Okay, let's get 00:12:08.000 --> 00:12:10.320 technical. Lisp, which stands for list processing, 00:12:10.559 --> 00:12:13.340 was created in the late 1950s. Why did they need 00:12:13.340 --> 00:12:15.100 a completely new language? I mean, languages 00:12:15.100 --> 00:12:17.500 like Fortran were already emerging. Because Fortran 00:12:17.500 --> 00:12:20.320 was designed for numerical computation, for crunching 00:12:20.320 --> 00:12:22.940 floating point numbers really fast. But AI needed 00:12:22.940 --> 00:12:25.370 to manipulate knowledge. And that meant manipulating 00:12:25.370 --> 00:12:27.950 abstract symbols and relationships. Which Fortran 00:12:27.950 --> 00:12:31.029 just wasn't built for. Not at all. McCarthy realized 00:12:31.029 --> 00:12:33.409 he couldn't do this efficiently using standard 00:12:33.409 --> 00:12:36.669 procedural logic. So his solution was revolutionary. 00:12:37.610 --> 00:12:40.409 He created Lisp by extending the mathematical 00:12:40.409 --> 00:12:43.610 concept of primitive recursive functions to operate 00:12:43.610 --> 00:12:46.710 on symbolic expressions, specifically on lists. 00:12:47.029 --> 00:12:48.909 Okay, tell us more about primitive recursive 00:12:48.909 --> 00:12:50.610 functions. Why were they the jumping off point? 00:12:50.919 --> 00:12:52.779 Well, primitive recursive functions are at the 00:12:52.779 --> 00:12:55.480 core of computability theory. They're the simplest 00:12:55.480 --> 00:12:58.159 functions that computers can calculate. And McCarthy 00:12:58.159 --> 00:13:01.120 basically had this insight. He realized that 00:13:01.120 --> 00:13:02.879 if he could extend these fundamental functions 00:13:02.879 --> 00:13:05.519 to operate on lists of symbols, not just numbers, 00:13:05.700 --> 00:13:08.399 he could create a powerful, flexible structure 00:13:08.399 --> 00:13:11.139 for representing any kind of knowledge. So anything 00:13:11.139 --> 00:13:13.899 from facts about the world to the rules of logic 00:13:13.899 --> 00:13:16.179 itself. Exactly. And this brings us to that famous 00:13:16.179 --> 00:13:19.159 mathematical borrowing he introduced, lambda 00:13:19.159 --> 00:13:22.159 calculus. Now, that's a term that scares a lot 00:13:22.159 --> 00:13:24.759 of non -experts, but it's really the structural 00:13:24.759 --> 00:13:27.559 heart of Lisp. It is the conceptual core, isn't 00:13:27.559 --> 00:13:30.340 it? It is. Lambda calculus, which was developed 00:13:30.340 --> 00:13:33.320 by Alonzo Church, is a formal system in mathematical 00:13:33.320 --> 00:13:36.419 logic for expressing computation based on function 00:13:36.419 --> 00:13:39.259 abstraction and application. McCarthy borrowed 00:13:39.259 --> 00:13:42.139 the syntax of lambda calculus, specifically the 00:13:42.139 --> 00:13:44.620 lambda notation, to define how Lisp functions 00:13:44.620 --> 00:13:47.259 should work. How can we visualize that for the 00:13:47.259 --> 00:13:49.179 listener? I mean, what does that mean practically? 00:13:49.440 --> 00:13:52.299 Okay, think of it this way. In most traditional 00:13:52.299 --> 00:13:55.799 procedural languages, like early Fortran, a function 00:13:55.799 --> 00:13:58.779 is a rigid step -by -step recipe for turning 00:13:58.779 --> 00:14:01.659 inputs into an output. Right, a set of instructions. 00:14:01.940 --> 00:14:04.220 Exactly. In lambda calculus, and therefore in 00:14:04.220 --> 00:14:06.899 Lisp, functions are treated as first -class citizens. 00:14:07.200 --> 00:14:09.879 They're almost like data. A function can be an 00:14:09.879 --> 00:14:12.100 argument to another function, or a function can 00:14:12.100 --> 00:14:14.720 return a new function as its output. So it lets 00:14:14.720 --> 00:14:16.860 you build these incredibly flexible self -referential 00:14:16.860 --> 00:14:19.600 structures. It's like being able to use a recipe 00:14:19.600 --> 00:14:21.840 as an ingredient in a new recipe, and then the 00:14:21.840 --> 00:14:24.820 result is a third brand new recipe. That is a 00:14:24.820 --> 00:14:27.860 perfect analogy. It's metaprogramming at its 00:14:27.860 --> 00:14:31.019 earliest and purest. And this design fundamentally 00:14:31.019 --> 00:14:34.049 changed the nature of programming. It defined 00:14:34.049 --> 00:14:36.610 the symbolic programming paradigm and became 00:14:36.610 --> 00:14:39.309 the immediate, indispensable language for virtually 00:14:39.309 --> 00:14:42.149 all early AI applications after it was published 00:14:42.149 --> 00:14:45.350 in 1960. It's incredible. And beyond that, every 00:14:45.350 --> 00:14:48.490 modern functional language, from Haskell to Scheme 00:14:48.490 --> 00:14:51.690 to Clojure, owes its philosophical and structural 00:14:51.690 --> 00:14:54.830 heritage directly to Liszt's foundation in lambda 00:14:54.830 --> 00:14:57.220 notation. It's amazing to think that a language 00:14:57.220 --> 00:14:59.759 designed to solve these abstract problems of 00:14:59.759 --> 00:15:02.820 1960s logic programming is directly responsible 00:15:02.820 --> 00:15:05.919 for shaping the programming style of, you know, 00:15:05.940 --> 00:15:09.059 today. But the creation of Lisp, with its radical 00:15:09.059 --> 00:15:11.220 flexibility and heavy use of dynamic memory, 00:15:11.440 --> 00:15:14.039 immediately created a massive practical engineering 00:15:14.039 --> 00:15:16.360 problem that McCarthy also had to solve. And 00:15:16.360 --> 00:15:18.000 that was the invention of garbage collection. 00:15:18.320 --> 00:15:20.399 That's right. This is a perfect example of a 00:15:20.399 --> 00:15:22.779 theoretical breakthrough forcing an engineering 00:15:22.779 --> 00:15:25.080 necessity. So what was the problem that Lisp... 00:15:25.100 --> 00:15:28.259 created? Well, Lisp operates by constantly creating 00:15:28.259 --> 00:15:31.320 and destroying symbolic data structures in memory 00:15:31.320 --> 00:15:34.559 lots, trees, expressions. When a programmer is 00:15:34.559 --> 00:15:36.700 constantly creating new objects, the computer's 00:15:36.700 --> 00:15:39.360 memory quickly gets fragmented and cluttered 00:15:39.360 --> 00:15:41.519 with pieces of data that are no longer being 00:15:41.519 --> 00:15:44.240 referenced or used. The garbage. The garbage. 00:15:44.320 --> 00:15:46.679 And if you don't clean it up, the program runs 00:15:46.679 --> 00:15:49.139 out of memory and it crashes. And this was traditionally 00:15:49.139 --> 00:15:51.899 managed manually by programmers, right? Which 00:15:51.899 --> 00:15:54.840 sounds tedious and very error prone. Precisely. 00:15:55.159 --> 00:15:57.179 Manual memory management is one of the biggest 00:15:57.179 --> 00:16:00.019 sources of bugs and instability in early programs. 00:16:00.460 --> 00:16:03.159 So around 1959, McCarthy invented... methods 00:16:03.159 --> 00:16:06.580 for garbage collection, an automated memory management 00:16:06.580 --> 00:16:08.899 technique. So instead of the programmer tracking 00:16:08.899 --> 00:16:11.539 everything. The system does it for you. The system 00:16:11.539 --> 00:16:14.320 itself would periodically pause, identify all 00:16:14.320 --> 00:16:16.820 the unreferenced data structures, and then free 00:16:16.820 --> 00:16:19.379 up that memory for new use. Wait, so the need 00:16:19.379 --> 00:16:22.100 to manage Lisp's complex structures led to the 00:16:22.100 --> 00:16:23.919 invention of a technique that is now fundamental 00:16:23.919 --> 00:16:26.659 to almost every high -level language we use today. 00:16:26.840 --> 00:16:30.379 Java, Python, C -Shark. It is entirely foundational. 00:16:30.940 --> 00:16:33.120 Garbage collection allows millions of developers 00:16:33.120 --> 00:16:35.919 today to focus purely on the application's logic, 00:16:36.120 --> 00:16:38.779 freeing them from the low -level, error -prone 00:16:38.779 --> 00:16:41.639 task of manual memory management. So the pursuit 00:16:41.639 --> 00:16:43.899 of pure symbolic thought led to the solution 00:16:43.899 --> 00:16:46.580 for a fundamental infrastructure problem. The 00:16:46.580 --> 00:16:49.029 irony is wonderful, isn't it? And beyond Lisp, 00:16:49.129 --> 00:16:51.110 we really shouldn't overlook his influence on 00:16:51.110 --> 00:16:53.690 general purpose programming standards. He played 00:16:53.690 --> 00:16:57.330 a significant role in defining LGOL 60, which 00:16:57.330 --> 00:16:59.649 was one of the most important early high -level 00:16:59.649 --> 00:17:01.450 languages. He was on the committee for that, 00:17:01.470 --> 00:17:03.950 right? A key member of the ACM committee. And 00:17:03.950 --> 00:17:06.369 his contributions there were absolutely transformative 00:17:06.369 --> 00:17:09.509 for general computing. Specifically, in August 00:17:09.509 --> 00:17:12.950 of 1959, he proposed incorporating two essential 00:17:12.950 --> 00:17:15.930 concepts, recursion and conditional expressions. 00:17:16.559 --> 00:17:18.839 And why were recursion and conditional expressions 00:17:18.839 --> 00:17:22.539 so critical for LGL? Well, recursion, the ability 00:17:22.539 --> 00:17:24.619 for a function to call itself, is absolutely 00:17:24.619 --> 00:17:27.420 necessary for elegantly solving problems that 00:17:27.420 --> 00:17:29.380 have naturally recursive structures, like file 00:17:29.380 --> 00:17:31.920 systems or trees or certain mathematical patterns. 00:17:32.039 --> 00:17:34.380 Before that, it was just messy loops. Convoluted 00:17:34.380 --> 00:17:38.359 loops. By embedding recursion into LGL, McCarthy 00:17:38.359 --> 00:17:40.819 helped elevate the structural elegance and power 00:17:40.819 --> 00:17:43.460 of general -purpose languages. and conditional 00:17:43.460 --> 00:17:45.940 expressions using if -then -else statements that 00:17:45.940 --> 00:17:49.160 return a value also streamline code. moving away 00:17:49.160 --> 00:17:51.319 from simple jump statements to more functional 00:17:51.319 --> 00:17:54.319 logical constructs. His participation here and 00:17:54.319 --> 00:17:56.900 later as a member of IFIP Working Group 2 .1 00:17:56.900 --> 00:17:59.720 ensured that his rigorous logical approach to 00:17:59.720 --> 00:18:01.960 language design got baked into international 00:18:01.960 --> 00:18:04.500 standards. So we have the philosopher who named 00:18:04.500 --> 00:18:07.400 AI and the logician who invented functional programming 00:18:07.400 --> 00:18:10.220 and modern memory management. Now we shift from 00:18:10.220 --> 00:18:13.259 software and logic to physical infrastructure 00:18:13.259 --> 00:18:16.000 and economics. Part four covers his third monumental 00:18:16.000 --> 00:18:18.450 contribution, timesharing, which leads... directly 00:18:18.450 --> 00:18:21.190 to the modern cloud. And this is where McCarthy's 00:18:21.190 --> 00:18:23.950 foresight becomes almost unsettlingly prophetic. 00:18:24.289 --> 00:18:26.269 He wasn't just thinking about how machines think. 00:18:26.349 --> 00:18:28.230 He was thinking about how people would access 00:18:28.230 --> 00:18:30.410 machines and how that access would be paid for. 00:18:30.720 --> 00:18:33.720 Right, because in the early 1960s, a computer 00:18:33.720 --> 00:18:36.799 was this massive single -user machine. It cost 00:18:36.799 --> 00:18:39.460 millions of dollars. If you were a researcher, 00:18:39.700 --> 00:18:42.400 you had to book time, submit your punch cards, 00:18:42.579 --> 00:18:44.859 and then wait hours for the result. Computing 00:18:44.859 --> 00:18:47.720 was a scarce, exclusive resource. And timesharing 00:18:47.720 --> 00:18:50.299 was the radical answer to this scarcity. It was. 00:18:50.380 --> 00:18:53.250 The idea was simple. but technically very complex. 00:18:53.670 --> 00:18:56.190 Instead of dedicating an entire mainframe to 00:18:56.190 --> 00:18:58.670 one user, the operating system would rapidly 00:18:58.670 --> 00:19:01.289 switch between processes, giving the illusion 00:19:01.289 --> 00:19:04.450 of simultaneous access to multiple users at separate 00:19:04.450 --> 00:19:06.970 terminals. And McCarthy was instrumental in making 00:19:06.970 --> 00:19:09.029 this happen, practically. He helped create three 00:19:09.029 --> 00:19:11.509 of the earliest systems, the Compatible Timesharing 00:19:11.509 --> 00:19:15.390 System, or CTSS at MIT, the BBN Timesharing System, 00:19:15.470 --> 00:19:17.910 and the Dartmouth Timesharing System. And timesharing 00:19:17.910 --> 00:19:20.210 wasn't just, you know, an efficiency hack. It 00:19:20.210 --> 00:19:23.089 was a democratization of computing power. It 00:19:23.089 --> 00:19:25.390 allowed dozens of students or researchers to 00:19:25.390 --> 00:19:28.009 interact directly with the machine in real time. 00:19:28.150 --> 00:19:30.349 It completely changed the workflow from batch 00:19:30.349 --> 00:19:33.049 processing to interactive computing. Yeah. And 00:19:33.049 --> 00:19:35.890 this technological leap, it led directly to his 00:19:35.890 --> 00:19:37.970 economic prophecy. You're talking about the famous 00:19:37.970 --> 00:19:42.309 1961 speech he gave at MIT's Centennial Celebration. 00:19:42.450 --> 00:19:45.519 That's the one. He didn't just suggest timesharing 00:19:45.519 --> 00:19:47.740 was a good idea. He suggested it should be a 00:19:47.740 --> 00:19:50.019 business model. A completely new one. He suggested 00:19:50.019 --> 00:19:53.279 the concept of utility computing. This was maybe 00:19:53.279 --> 00:19:56.480 the first public proposal of its kind. He argued 00:19:56.480 --> 00:19:58.980 that computing power and specific applications 00:19:58.980 --> 00:20:02.460 should be sold through a utility model. just 00:20:02.460 --> 00:20:05.339 like electricity or water or gas. So you'd plug 00:20:05.339 --> 00:20:07.839 in, use what you need, and get a bill based on 00:20:07.839 --> 00:20:09.960 your usage. Exactly. That is just staggering 00:20:09.960 --> 00:20:13.140 foresight. To suggest that in 1961, when computers 00:20:13.140 --> 00:20:15.940 still used magnetic tape and punch cards, that 00:20:15.940 --> 00:20:18.599 computing would become a ubiquitous metered service 00:20:18.599 --> 00:20:21.619 that's visualizing an entire future economy. 00:20:21.740 --> 00:20:24.180 It was wildly ambitious. And the initial idea 00:20:24.180 --> 00:20:26.599 really took off with great enthusiasm in the 00:20:26.599 --> 00:20:29.279 late 60s and early 70s. You had companies like 00:20:29.279 --> 00:20:32.049 GE offering time -sharing services. But then, 00:20:32.069 --> 00:20:35.190 the rise of the cheap PC in the 80s made the 00:20:35.190 --> 00:20:37.509 idea of remote shared computing seem obsolete. 00:20:38.089 --> 00:20:40.009 Suddenly, people could own the resource locally. 00:20:40.210 --> 00:20:43.470 But the pendulum swung back. Traumatically. The 00:20:43.470 --> 00:20:45.769 internet explosion brought the concept roaring 00:20:45.769 --> 00:20:48.589 back, first as application service providers, 00:20:48.690 --> 00:20:52.269 or ASPs, then as grid computing, and finally, 00:20:52.369 --> 00:20:55.529 after 2000, as the massive infrastructure we 00:20:55.529 --> 00:20:58.069 now just call the cloud. And if any listener 00:20:58.069 --> 00:21:01.069 needed that one single definitive takeaway connecting 00:21:01.069 --> 00:21:04.609 McCarthy to the modern day, this is it. It comes 00:21:04.609 --> 00:21:06.670 directly from his colleague, Lester Earnest. 00:21:07.130 --> 00:21:09.430 He observed that the timesharing systems they 00:21:09.430 --> 00:21:11.250 worked on in the 60s eventually came to be called 00:21:11.250 --> 00:21:14.970 servers. And Earnest noted, and now we call it 00:21:14.970 --> 00:21:18.130 cloud computing. That is still just time sharing. 00:21:18.309 --> 00:21:20.849 John started it. Wow. It confirms that every 00:21:20.849 --> 00:21:23.269 time you use Amazon Web Services or Google Cloud 00:21:23.269 --> 00:21:25.970 or Microsoft Azure, you are using the economic 00:21:25.970 --> 00:21:28.609 and technical framework that John McCarthy conceptualized 00:21:28.609 --> 00:21:30.869 six decades ago. And beyond just the technology, 00:21:31.150 --> 00:21:33.589 the man also understood that big ideas need big 00:21:33.589 --> 00:21:36.450 institutions to nurture them. His role in institution 00:21:36.450 --> 00:21:38.650 building was just as crucial as his inventions. 00:21:38.950 --> 00:21:42.170 He was a driving force. At MIT, his efforts helped 00:21:42.170 --> 00:21:44.609 motivate the creation of Project MC, which became 00:21:44.609 --> 00:21:47.369 the cornerstone of MIT's computer science and 00:21:47.369 --> 00:21:49.650 AI research. And when he moved to Stanford, he 00:21:49.650 --> 00:21:52.730 did it again. He replicated that success. He 00:21:52.730 --> 00:21:54.690 helped establish the Stanford AI Laboratory, 00:21:55.029 --> 00:21:57.910 or SAIL. And these weren't just parallel labs, 00:21:58.069 --> 00:22:00.410 right? They were intentionally friendly rivals. 00:22:00.710 --> 00:22:03.450 Exactly. SAIL was, for many years, the primary 00:22:03.450 --> 00:22:07.349 friendly rival to MIT's Project MC. And this 00:22:07.349 --> 00:22:09.509 competition was essential. It spurred innovation, 00:22:09.890 --> 00:22:12.509 it attracted the brightest students, and it created 00:22:12.509 --> 00:22:14.869 the academic infrastructure necessary to push 00:22:14.869 --> 00:22:17.269 the boundaries of AI and computing through the 00:22:17.269 --> 00:22:20.410 60s and 70s. So the whole field benefited. The 00:22:20.410 --> 00:22:22.910 entire field benefited. And just to show that 00:22:22.910 --> 00:22:24.950 the early drive wasn't purely academic, you can 00:22:24.950 --> 00:22:27.130 look at the famous Katak -McCarthy match. A chess 00:22:27.130 --> 00:22:31.369 match against the Soviets. In 1966, yes. McCarthy's 00:22:31.369 --> 00:22:33.809 team, representing the U .S., used a program 00:22:33.809 --> 00:22:36.369 they wrote to play a series of chess games against 00:22:36.369 --> 00:22:38.650 their Soviet counterparts. They ended up losing 00:22:38.650 --> 00:22:40.490 two and drawing two, so it wasn't a victory. 00:22:40.650 --> 00:22:43.089 But it demonstrated that early practical drive 00:22:43.089 --> 00:22:45.650 for application. It was an intellectual Cold 00:22:45.650 --> 00:22:48.349 War battle waged on a chessboard, proving that 00:22:48.349 --> 00:22:50.890 symbolic programming could tackle complex, real 00:22:50.890 --> 00:22:53.690 -world strategic problems. Okay, so we've covered 00:22:53.690 --> 00:22:56.329 his contributions to naming, to programming, 00:22:56.490 --> 00:22:58.930 and to infrastructure. Now we need to explore 00:22:58.930 --> 00:23:01.710 what he really, truly cared about. The philosophical 00:23:01.710 --> 00:23:05.009 implications of AI and the fierce intellectual 00:23:05.009 --> 00:23:08.920 debates his... His logic -based optimism inspired. 00:23:09.279 --> 00:23:12.000 Right. At his core, McCarthy was a profoundly 00:23:12.000 --> 00:23:15.279 committed artificial intelligence optimist. He 00:23:15.279 --> 00:23:18.000 never wavered in his conviction that human intelligence, 00:23:18.339 --> 00:23:21.000 all of it, could eventually be formalized precisely 00:23:21.000 --> 00:23:23.500 enough to be programmed into a machine. So to 00:23:23.500 --> 00:23:26.700 him, the human mind was complex, but... fundamentally 00:23:26.700 --> 00:23:29.299 computable. Yes. And this optimism led him to 00:23:29.299 --> 00:23:31.299 develop formal logic systems that were designed 00:23:31.299 --> 00:23:34.880 explicitly to capture the messiness of human 00:23:34.880 --> 00:23:36.980 thought. His most important work in this area 00:23:36.980 --> 00:23:39.099 was his development of circumscription. Right. 00:23:39.200 --> 00:23:42.920 Between 1978 and 1986. That's right. Circumscription 00:23:42.920 --> 00:23:45.700 is a method of non -monotonic reasoning. And 00:23:45.700 --> 00:23:47.759 that's a crucial distinction. Traditional logic 00:23:47.759 --> 00:23:50.259 is monotonic. If you add a new piece of information, 00:23:50.559 --> 00:23:53.059 you never retract an old conclusion. It only 00:23:53.059 --> 00:23:55.220 grows. But that's not how human reasoning works 00:23:55.220 --> 00:23:57.380 at all. We constantly change our minds when new 00:23:57.380 --> 00:24:00.220 facts contradict our old assumptions. Precisely. 00:24:00.460 --> 00:24:03.640 If I tell you birds can fly, you conclude that 00:24:03.640 --> 00:24:06.200 Tweety the canary can fly. But if I then tell 00:24:06.200 --> 00:24:08.940 you Tweety is a statue, you retract the initial 00:24:08.940 --> 00:24:11.420 conclusion. That is non -monotonic reasoning. 00:24:11.680 --> 00:24:14.640 It's the ability to change your mind when you're 00:24:14.640 --> 00:24:17.220 presented with contradictory but more specific 00:24:17.220 --> 00:24:20.799 information. Circumscription was McCarthy's attempt 00:24:20.799 --> 00:24:23.839 to formally model that crucial everyday human 00:24:23.839 --> 00:24:25.799 ability. He also worked on something called situation 00:24:25.799 --> 00:24:28.380 calculus. What was that focused on? Situation 00:24:28.380 --> 00:24:30.839 calculus was another formal logic, and it was 00:24:30.839 --> 00:24:33.140 designed to represent and reason about dynamic 00:24:33.140 --> 00:24:36.230 situations and changes in the world. It's a language 00:24:36.230 --> 00:24:38.450 for describing actions, their preconditions and 00:24:38.450 --> 00:24:40.690 their resulting effects. Both of these were part 00:24:40.690 --> 00:24:42.950 of his lifelong campaign to formalize common 00:24:42.950 --> 00:24:45.849 sense. That vast, often implicit body of knowledge 00:24:45.849 --> 00:24:48.589 that lets us navigate the world. And this confidence 00:24:48.589 --> 00:24:50.990 in logic led to one of his most famously provocative 00:24:50.990 --> 00:24:54.930 statements, a thermostat test. In his 1979 article 00:24:54.930 --> 00:24:57.569 ascribing mental qualities to machines, McCarthy 00:24:57.569 --> 00:24:59.809 deliberately pushed the boundaries of what intelligence 00:24:59.809 --> 00:25:02.650 meant. He stated that, and I quote, machines 00:25:02.650 --> 00:25:04.589 as simple as thermostats can be said to have. 00:25:04.619 --> 00:25:08.259 A thermostat has beliefs. His argument was that 00:25:08.259 --> 00:25:10.680 if a device performs a problem solving function, 00:25:10.900 --> 00:25:13.500 the thermostat believes the room is too cold 00:25:13.500 --> 00:25:16.559 and therefore turns on the furnace. We can functionally 00:25:16.559 --> 00:25:18.980 ascribe beliefs to it, even if those beliefs 00:25:18.980 --> 00:25:21.200 are very simple. He was equating a functional 00:25:21.200 --> 00:25:24.140 behavior with a mental state. But that philosophical 00:25:24.140 --> 00:25:27.460 position immediately triggered the biggest counterpunch 00:25:27.460 --> 00:25:30.079 in the history of AI philosophy. It triggered 00:25:30.079 --> 00:25:33.599 John Searle's famous 1980 response. the Chinese 00:25:33.599 --> 00:25:36.779 room argument. Searle argued that merely manipulating 00:25:36.779 --> 00:25:39.539 symbols according to rules, which is what the 00:25:39.539 --> 00:25:42.579 thermostat or even a sophisticated AI does, is 00:25:42.579 --> 00:25:45.279 not the same as true understanding or intentionality. 00:25:45.519 --> 00:25:48.740 The machine lacks consciousness. Searle's thought 00:25:48.740 --> 00:25:50.799 experiment showed a person in a room manipulating 00:25:50.799 --> 00:25:53.460 Chinese characters based on a rulebook successfully 00:25:53.460 --> 00:25:55.599 communicating without understanding a single 00:25:55.599 --> 00:25:58.019 character of Chinese. So the core tension was 00:25:58.019 --> 00:26:00.700 McCarthy was a functionalist arguing that if 00:26:00.700 --> 00:26:02.619 it behaves intelligently, it is intelligent. 00:26:02.920 --> 00:26:05.000 And Searle was an intentionalist arguing that 00:26:05.000 --> 00:26:07.099 behavior without genuine consciousness is just 00:26:07.099 --> 00:26:10.000 simulation. That tension functionality versus 00:26:10.000 --> 00:26:12.839 intentionality, it generated a massive amount 00:26:12.839 --> 00:26:15.700 of literature, and it remains the defining philosophical 00:26:15.700 --> 00:26:19.500 boundary dispute in the field. McCarthy was challenging 00:26:19.500 --> 00:26:22.700 us to expand our definition of belief. Searle 00:26:22.700 --> 00:26:25.079 was defending the uniqueness of human experience 00:26:25.079 --> 00:26:27.700 and consciousness. And the conflict wasn't just 00:26:27.700 --> 00:26:30.670 with Searle. McCarthy had a long, defining professional 00:26:30.670 --> 00:26:33.269 debate with Hubert Dreyfus, the philosophy professor 00:26:33.269 --> 00:26:36.329 at Berkeley. Oh, yes. Dreyfus was perhaps the 00:26:36.329 --> 00:26:39.589 most famous and persistent skeptic of early logic 00:26:39.589 --> 00:26:42.549 -based AI. He fundamentally disagreed with the 00:26:42.549 --> 00:26:44.950 idea that human reasoning could be reduced to 00:26:44.950 --> 00:26:48.779 formal logical systems. Dreyfus, who was influenced 00:26:48.779 --> 00:26:52.000 by existentialist phenomenology, saw human understanding, 00:26:52.220 --> 00:26:55.079 especially expertise in intuition, as being deeply 00:26:55.079 --> 00:26:57.680 embedded in the body in experience and in contextual 00:26:57.680 --> 00:27:00.319 understanding that just transcends explicit rules. 00:27:00.809 --> 00:27:02.730 It sounds like Dreyfus was arguing that you can't 00:27:02.730 --> 00:27:05.569 program wisdom. Essentially, yes. He viewed the 00:27:05.569 --> 00:27:07.569 human experience as encompassing these deeper 00:27:07.569 --> 00:27:10.609 existential questions about the good life, about 00:27:10.609 --> 00:27:14.029 meaning, about choice. And he argued that these 00:27:14.029 --> 00:27:16.410 dimensions of reality and intuition could never 00:27:16.410 --> 00:27:19.029 be reduced to a symbolic calculus. For McCarthy, 00:27:19.329 --> 00:27:21.069 the challenge was just a matter of technical 00:27:21.069 --> 00:27:23.970 complexity. The formalization was possible, just 00:27:23.970 --> 00:27:26.910 difficult. For Dreyfus, the challenge was fundamental 00:27:26.910 --> 00:27:30.349 and philosophical. They represented the two irreconcilable 00:27:30.349 --> 00:27:33.450 poles of the AI debate for decades. And despite 00:27:33.450 --> 00:27:36.569 all this intense skepticism, McCarthy never backed 00:27:36.569 --> 00:27:39.410 down from his optimism. He always held firm to 00:27:39.410 --> 00:27:41.750 the ultimate computability of the mind. He was 00:27:41.750 --> 00:27:44.309 unshakable. He addressed the skepticism about 00:27:44.309 --> 00:27:47.130 spiritual or transcendent intuition in a 1989 00:27:47.130 --> 00:27:49.839 interview. He said that the view that such things 00:27:49.839 --> 00:27:52.200 are not physically accessible has been in steady 00:27:52.200 --> 00:27:54.420 decline for centuries, ever since the scientific 00:27:54.420 --> 00:27:56.259 revolution started chipping away at supernatural 00:27:56.259 --> 00:27:59.329 explanations. McCarthy fully expected that the 00:27:59.329 --> 00:28:01.349 difficult challenge of realizing human consciousness 00:28:01.349 --> 00:28:03.430 in machines would eventually be overcome through 00:28:03.430 --> 00:28:05.970 logical formalization and engineering ingenuity. 00:28:06.109 --> 00:28:09.190 He saw no magical barrier, only a temporary technical 00:28:09.190 --> 00:28:12.170 one. McCarthy's sheer intellectual range extended 00:28:12.170 --> 00:28:15.109 far beyond logic gates and programming languages. 00:28:15.430 --> 00:28:17.569 Before we conclude, we have to look at some of 00:28:17.569 --> 00:28:20.170 his more unique later work, including an idea 00:28:20.170 --> 00:28:22.970 for literal space travel infrastructure. The 00:28:22.970 --> 00:28:26.160 Space Fountain. This concept, which he proposed 00:28:26.160 --> 00:28:29.180 around 1982, it just showcases that McCarthy 00:28:29.180 --> 00:28:32.440 never lost that childhood interest in grand engineering 00:28:32.440 --> 00:28:35.400 problems. I always found this idea so compellingly 00:28:35.400 --> 00:28:38.140 strange. How was this massive tower supposed 00:28:38.140 --> 00:28:40.900 to function without traditional structural support? 00:28:41.279 --> 00:28:43.880 Well, it was conceived as a non -rocket space 00:28:43.880 --> 00:28:46.819 launch system, a tower extending hundreds of 00:28:46.819 --> 00:28:49.859 kilometers into space. But unlike a traditional 00:28:49.859 --> 00:28:52.880 space elevator, it wouldn't rely on tensile strength. 00:28:53.319 --> 00:28:55.200 Instead, it would be held vertical by a stream 00:28:55.200 --> 00:28:58.200 of small, high -velocity pellets propelled continuously 00:28:58.200 --> 00:29:00.740 upward from Earth. So the momentum of the pellets 00:29:00.740 --> 00:29:02.980 keeps it up. The outward momentum of that stream, 00:29:03.160 --> 00:29:05.099 constantly resupplied, would keep the structure 00:29:05.099 --> 00:29:07.359 rigid and vertical. And the payloads, how do 00:29:07.359 --> 00:29:09.579 they get up? Payloads would ride a separate parallel 00:29:09.579 --> 00:29:11.500 conveyor, belt up the inside of the structure, 00:29:11.640 --> 00:29:13.960 using the constant momentum of the pellet stream 00:29:13.960 --> 00:29:17.079 as its support and stabilization. Then the pellets 00:29:17.079 --> 00:29:18.640 would be returned to Earth to repeat the whole 00:29:18.640 --> 00:29:21.380 process. That's wild. It's a remarkable combination 00:29:21.380 --> 00:29:25.579 of... Physics, engineering, and just pure imagination. 00:29:26.160 --> 00:29:28.579 And he also embraced the early dawn of digital 00:29:28.579 --> 00:29:31.319 culture in a way that really bridged the academic 00:29:31.319 --> 00:29:34.099 and the everyday. He was active on Usenet. Very 00:29:34.099 --> 00:29:36.700 active. He was a participant in Usenet forums, 00:29:36.839 --> 00:29:39.420 which was the original social networking ecosystem 00:29:39.420 --> 00:29:42.660 of the academic world. He frequently commented 00:29:42.660 --> 00:29:45.480 on world affairs and engaged in spirited intellectual 00:29:45.480 --> 00:29:48.619 exchange. He was also a devoted book reader. 00:29:49.019 --> 00:29:53.140 actively attending SF Bay Area dinners for rabreaders 00:29:53.140 --> 00:29:57.099 -rec .arts .books, the famous rabfests. And that 00:29:57.099 --> 00:29:59.019 presence in early digital culture gave us one 00:29:59.019 --> 00:30:01.279 of his most famous personal tenets, which served 00:30:01.279 --> 00:30:04.000 as his Usenet signature. It was his mantra, his 00:30:04.000 --> 00:30:06.180 belief in the importance of grounding opinions 00:30:06.180 --> 00:30:09.279 in quantifiable fact. He who refuses to do arithmetic 00:30:09.279 --> 00:30:11.519 is doomed to talk nonsense. He even had it on 00:30:11.519 --> 00:30:14.549 his license plate cover. He did. And that quote 00:30:14.549 --> 00:30:17.170 just perfectly encapsulates his lifelong belief 00:30:17.170 --> 00:30:19.430 in mathematics, education, critical thinking 00:30:19.430 --> 00:30:22.430 and logical rigor as essential prerequisites 00:30:22.430 --> 00:30:25.329 for sound political and social discourse. He 00:30:25.329 --> 00:30:27.349 was also a staunch defender of controversial 00:30:27.349 --> 00:30:29.470 speech, wasn't he? Which reflects that early 00:30:29.470 --> 00:30:31.650 commitment to critical thinking, even when it 00:30:31.650 --> 00:30:35.250 caused friction at Stanford. Yes, in 1989, he 00:30:35.250 --> 00:30:37.410 actively defended controversial free speech, 00:30:37.589 --> 00:30:40.369 even involving European ethnic jokes at Stanford. 00:30:40.569 --> 00:30:42.950 His position wasn't an endorsement of the content 00:30:42.950 --> 00:30:45.670 of the jokes, but a principled stand. He believed 00:30:45.670 --> 00:30:47.470 that in the pursuit of intellectual and political 00:30:47.470 --> 00:30:50.589 honesty, one must tolerate speech that is offensive 00:30:50.589 --> 00:30:52.990 as long as it's part of a necessary rigorous 00:30:52.990 --> 00:30:56.329 intellectual exchange. He was consistent. Freedom 00:30:56.329 --> 00:30:59.160 of thought requires freedom of expression. His 00:30:59.160 --> 00:31:01.140 final years also saw him turn his philosophical 00:31:01.140 --> 00:31:03.200 questions into fiction, which is a wonderful 00:31:03.200 --> 00:31:06.380 full circle for a scientist. His 2001 short story, 00:31:06.500 --> 00:31:08.920 The Robot and the Baby, used a fictional scenario 00:31:08.920 --> 00:31:11.740 to explore those central philosophical challenges 00:31:11.740 --> 00:31:14.500 of AI specifically, whether robots should truly 00:31:14.500 --> 00:31:17.500 have emotions or merely simulate them. It was 00:31:17.500 --> 00:31:19.559 a farcical but really insightful look at the 00:31:19.559 --> 00:31:22.259 very questions that had defined his rivalry with 00:31:22.259 --> 00:31:25.299 Searle and Dreyfus. And it anticipated many of 00:31:25.299 --> 00:31:27.859 the cultural concerns we have today about AI 00:31:27.859 --> 00:31:30.140 transparency and social interaction. And on the 00:31:30.140 --> 00:31:32.220 professional side, his legacy of mentorship is 00:31:32.220 --> 00:31:34.539 also crucial. He didn't just invent, he trained 00:31:34.539 --> 00:31:37.680 the next generation. He advised 30 PhD graduates, 00:31:37.980 --> 00:31:40.240 including computing luminaries who would go on 00:31:40.240 --> 00:31:42.480 to shape the field themselves. People like Barbara 00:31:42.480 --> 00:31:45.059 Liskov, who won the Turing Award in 2008, and 00:31:45.059 --> 00:31:47.279 Raj Reddy, another Turing Award. winner. That 00:31:47.279 --> 00:31:49.559 is an enormous influence. It's like extending 00:31:49.559 --> 00:31:51.839 his intellectual DNA across the entire industry. 00:31:52.059 --> 00:31:54.059 And while his complex personal life included 00:31:54.059 --> 00:31:57.079 tragedy his second wife Vera Watson who was a 00:31:57.079 --> 00:31:59.700 programmer and a dedicated mountaineer tragically 00:31:59.700 --> 00:32:02.140 died attempting to scale Annapurna Central in 00:32:02.140 --> 00:32:04.859 1978. His professional contribution remained 00:32:04.859 --> 00:32:08.079 a constant powerful force. And his recognition 00:32:08.079 --> 00:32:10.500 was sustained right up until his death in 2011. 00:32:11.160 --> 00:32:13.740 He was inducted as a fellow of the Computer History 00:32:13.740 --> 00:32:17.180 Museum in 1999, specifically for co -founding 00:32:17.180 --> 00:32:20.160 AI and timesharing, and he was posthumously named 00:32:20.160 --> 00:32:23.440 a 2012 Stanford Engineering Hero. His body of 00:32:23.440 --> 00:32:25.700 work remains arguably the most comprehensive 00:32:25.700 --> 00:32:28.400 and multifaceted in the history of computer science. 00:32:28.720 --> 00:32:31.160 You know, looking back over this deep dive, we 00:32:31.160 --> 00:32:33.480 realize that John McCarthy wasn't just a participant 00:32:33.480 --> 00:32:36.140 in the computer revolution. He was an explosive 00:32:36.140 --> 00:32:38.759 polymath who defined its essential ingredients. 00:32:39.599 --> 00:32:41.880 He gave the field its name, provided the logic 00:32:41.880 --> 00:32:44.380 framework that defines symbolic computation through 00:32:44.380 --> 00:32:46.900 Lisp and functional programming, and then, decades 00:32:46.900 --> 00:32:49.759 early, drew the economic map for utility computing, 00:32:49.940 --> 00:32:51.819 which is now the cloud. His whole career was 00:32:51.819 --> 00:32:53.960 just this relentless application of logic and 00:32:53.960 --> 00:32:56.460 a profound optimism about the power of formal 00:32:56.460 --> 00:32:58.559 systems. And that optimism is the key takeaway 00:32:58.559 --> 00:33:01.380 for you. McCarthy fundamentally believed that 00:33:01.380 --> 00:33:03.980 knowledge, even the most elusive forms of human 00:33:03.980 --> 00:33:06.500 intelligence, could eventually be formalized 00:33:06.500 --> 00:33:09.509 and programmed. He solved massive formalization 00:33:09.509 --> 00:33:12.009 challenges in symbolic logic and in infrastructure. 00:33:12.269 --> 00:33:15.049 But the challenge he set for himself, to realize 00:33:15.049 --> 00:33:17.930 human consciousness in machines, that is still 00:33:17.930 --> 00:33:20.549 the defining struggle for AI today. We live in 00:33:20.549 --> 00:33:22.750 the technological world he predicted, but we 00:33:22.750 --> 00:33:25.289 are still striving to build the reasoning machine 00:33:25.289 --> 00:33:27.690 he envisioned. So we connect this back to the 00:33:27.690 --> 00:33:30.430 frontier. If the infrastructural challenge was 00:33:30.430 --> 00:33:32.970 solved by time sharing becoming the cloud and 00:33:32.970 --> 00:33:35.109 the symbolic challenge was solved by LISP and 00:33:35.109 --> 00:33:38.289 functional programming, what is the next? unformalized 00:33:38.289 --> 00:33:40.890 common sense challenge that today's AI architects 00:33:40.890 --> 00:33:42.829 need to conquer. I mean, McCarthy was working 00:33:42.829 --> 00:33:45.509 on things like elaboration tolerance, the ability 00:33:45.509 --> 00:33:47.950 to deal gracefully with vast quantities of facts 00:33:47.950 --> 00:33:50.170 without explicit guidance, and the formalization 00:33:50.170 --> 00:33:53.150 of context. Which leads to our final provocative 00:33:53.150 --> 00:33:56.509 thought for you to chew on. If John McCarthy 00:33:56.509 --> 00:33:58.630 proved that symbolic thought could be structured 00:33:58.630 --> 00:34:01.269 logically, what fundamental human characteristic 00:34:01.269 --> 00:34:04.289 beyond mere consciousness or intuition remains 00:34:04.289 --> 00:34:06.630 the ultimate programming challenge for today's 00:34:06.630 --> 00:34:09.269 AI? What's the next thing that must be formalized 00:34:09.269 --> 00:34:11.469 before a machine can truly possess common sense 00:34:11.469 --> 00:34:13.530 and wisdom rather than just impressive processing 00:34:13.530 --> 00:34:16.429 power? That, in a very real sense, is the map 00:34:16.429 --> 00:34:17.869 that McCarthy drew for us to follow.