The AI Pioneers Who Started It All | 3 Wows and 1 Promise

Summary

This special episode of “3 Wows and 1 Promise” takes viewers back to the origins of artificial intelligence through archival footage and interviews with the field’s founding pioneers: Norbert Wiener, Marvin Minsky, and John McCarthy. The video provides invaluable historical context for understanding today’s AI debates through the words of those who created the field before the term “artificial intelligence” was even coined.

The first wow features Norbert Wiener in 1954 discussing “thinking machines” - remarkably stating that machines are being made smarter with better memories and the possibility of learning, and that these advances are “happening very fast.” The second wow presents Marvin Minsky’s revolutionary perspective that consciousness is not a single thing but “a messy label for dozens of mental processes” - a view that challenges today’s AGI debates to deconstruct rather than mystify intelligence. The third wow showcases John McCarthy, who coined the term “artificial intelligence” at Dartmouth in 1956, revealing the surprising fact that he was interested in AI before computers, and that almost no one in the early AI community had read Alan Turing’s foundational 1950 paper “Computing Machinery and Intelligence.”

The “promise” segment introduces Felix, an early MIT robot designed to demonstrate machine vision - a “cute promise that actually never came” that nonetheless showcases the imaginative spirit of early AI research.

Highlights

”Machines That Think Like Animals”

yt-dlp --download-sections "*1:00-1:36" "https://www.youtube.com/watch?v=shKHSAf_KAY" --force-keyframes-at-cuts --merge-output-format mp4 -o "shKHSAf_KAY-1m00s.mp4"

“Would you say that machines are becoming smarter and smarter? Yes. We’re making them smarter and smarter. We’re building into them better memories. We’re building into them the possibility of learning and those things are happening very fast… The word ‘think’ is one of the words like ‘life’ and ‘soul’ which are bad words. They mean just what we want them to mean. But I would say that we have machines which if they were to be for the first time discovered by an animal psychologist would be said to think in the way that animals do.” — Norbert Wiener (1954), 1:00

”Consciousness Is About 50 Things”

yt-dlp --download-sections "*2:28-3:48" "https://www.youtube.com/watch?v=shKHSAf_KAY" --force-keyframes-at-cuts --merge-output-format mp4 -o "shKHSAf_KAY-2m28s.mp4"

“There is no such thing as consciousness, but there’s dozens of processes that involve memories of what you’ve been recently doing and new ways to represent things and describing things in language rather than images… The existence of the word consciousness is the main reason why psychology… they got stuck thinking that these words like consciousness and ethics are things rather than big complicated structures that might have a better explanation.” — Marvin Minsky, 2:28

”No One Read Turing’s Paper”

yt-dlp --download-sections "*7:50-8:30" "https://www.youtube.com/watch?v=shKHSAf_KAY" --force-keyframes-at-cuts --merge-output-format mp4 -o "shKHSAf_KAY-7m50s.mp4"

“The one person who already had expressed the view that computers were the key to artificial intelligence was Turing who had written this article ‘Computing Machinery and Intelligence’ that was published in 1950. But I can’t find anyone who was interested in artificial intelligence who had read that article - with the exception of Selfridge who claimed he’d read it but didn’t believe it.” — John McCarthy, 7:50

”Interactive Computing Was Obvious”

yt-dlp --download-sections "*8:32-9:30" "https://www.youtube.com/watch?v=shKHSAf_KAY" --force-keyframes-at-cuts --merge-output-format mp4 -o "shKHSAf_KAY-8m32s.mp4"

“I became convinced that computers were the key to artificial intelligence in 1955. And at that point the idea of time sharing was immediately apparent to me because I was thinking about artificial intelligence already - the question was ‘well how are you going to interact with it?’ The idea that one should be sitting comfortably in one’s office and interacting with the computer… I began to look around and talk to people and was very surprised that that wasn’t what everybody thought.” — John McCarthy, 8:32

Key Points

• Norbert Wiener on Learning Machines (1:00) - In 1954, Wiener declared machines are becoming smarter with better memories and learning capabilities, happening “very fast”
• “Think” Is a Bad Word (1:20) - Wiener argued that words like “think,” “life,” and “soul” are “bad words” that mean whatever we want them to mean
• Consciousness as Multiple Processes (2:28) - Minsky argued there’s no single thing called consciousness - instead there are about 50 different processes
• AGI/ASI Debate Trap (1:55) - Today’s AGI and ASI debates fall into the same trap of projecting grand unity onto complex architectures
• Five or Six Levels of Thinking (3:30) - Minsky described multiple levels where the higher one involves making simplified models of your whole self
• McCarthy’s 1949 Origin (4:10) - McCarthy became interested in AI at the 1949 Hixon Symposium at Caltech on cerebral mechanisms
• AI Before Computers (5:20) - McCarthy was interested in AI before becoming interested in computers - initially thinking about special purpose computers
• Bell Labs Summer 1952 (7:22) - McCarthy and Minsky were both hired by Shannon at Bell Labs to work on AI, but still didn’t focus on computers
• Turing’s Unread Paper (7:50) - Turing’s 1950 paper was the only work arguing computers were key to AI, but almost nobody in the US AI community had read it
• Time-Sharing Invention (8:35) - McCarthy invented time-sharing because interactive use of computers seemed obvious for AI work
• Felix the Robot (9:30) - Early MIT robot demonstrating machine vision, representing “an advance in evolution”

Mentions

People

• Norbert Wiener (0:28) - Mathematician and philosopher, founder of cybernetics, speaking in 1954
• Marvin Minsky (0:25) - AI pioneer who saw consciousness as multiple processes, not a single thing
• John McCarthy (0:24) - Coined the term “artificial intelligence” at the 1956 Dartmouth conference
• Alan Turing (4:35) - Only person who from the beginning was sure computers would enable AI
• John von Neumann (4:20) - Present at the 1949 Hixon Symposium
• Warren McCulloch (5:40) - Present at the Hixon Symposium
• Claude Shannon (7:28) - Hired McCarthy and Minsky at Bell Labs in summer 1952
• Oliver Selfridge (5:00) - Only American AI researcher who claimed to have read Turing’s paper, but didn’t believe it
• Aristotle (3:05) - Referenced by Minsky as writing about ethics as well as modern ethicists
• William James (3:00) - Referenced by Minsky as writing popular psychology as well as modern authors

Institutions

• MIT (9:45) - Where Felix the robot was developed
• Caltech (4:15) - Site of the 1949 Hixon Symposium
• Bell Labs (7:22) - Where McCarthy and Minsky worked in summer 1952
• Princeton (6:05) - Where McCarthy was a graduate student in mathematics
• Dartmouth (3:55) - Site of the 1956 conference where AI was named

Publications

• “Computing Machinery and Intelligence” (4:40) - Turing’s 1950 paper that laid out all the arguments for computer-based AI

Surprising Quotes

“The word ‘think’ is one of the words like ‘life’ and so on and ‘soul’ which are bad words. They mean just what we want them to mean.” — Norbert Wiener, 1:20

“The existence of the word ‘consciousness’ is the main reason why psychology… they got stuck thinking that these words like consciousness and ethics are things rather than big complicated structures that might have a better explanation.” — Marvin Minsky, 2:45

“I became interested in artificial intelligence before I became interested in computers. This was in 1949.” — John McCarthy, 5:20

“Bell Labs at that time had IBM 603 or 4 which were calculating punches and again no one thought that computers had much to do with it.” — John McCarthy, 7:40

“I was thinking that if we dress up our robots in the supermarkets the way they dressed up Felix for the movie, that would be so cool, right? Just imagine these robots in the aisles looking like a bee, like a butterfly, like a bouquet of flowers. They would not even perceive as robots.” — Ksenia Se, 10:15

Transcript

0:00 Hello everybody. Today I want to offer you a very special episode of Three Wows and One Promise. Going back to the history of artificial intelligence and meeting people who created the whole field even when the word artificial intelligence was not yet coined. Please join me in meeting John McCarthy, Norbert Wiener, and Marvin Minsky.

0:28 The first wow - Norbert Wiener about thinking machines. This video is from 1954 and Norbert Wiener, very serious, kind of grumpy, talks about what does it mean “thinking machines.” And he says that they add memory, they’re building the possibility of learning, and those things are happening very fast. He says in 1954. It’s an interesting piece because he also discusses the word “think.” Watch it.

1:00 “Would you say that machines are becoming smarter and smarter?” “Yes. We’re making them smarter and smarter. We’re building into them better memories. We’re building into them the possibility of learning and those things are happening very fast.” “By that you mean that we have machines that actually think?” “The word ‘think’ is one of the words like ‘life’ and so on and ‘soul’ which are bad words. They mean just what we want them to mean. But I would say that we have machines which if they were to be for the first time discovered by an animal psychologist would be said to think in the way that animals do.” “Yes.”

1:38 Marvin Minsky saw consciousness not as a thing but as a messy label for dozens of mental processes that track, describe, and model our own thinking. He warned that treating it as a single mysterious entity stalls progress.

1:55 Can you feel me now? Can you feel my emotions, like my rage basically coming out of this screen onto you? Because that’s exactly what today’s artificial general intelligence and artificial super intelligence debates fall into - a similar trap. It’s just projecting grand unity onto what may be just the complex architectures of memory, reasoning, and modeling.

2:20 And Minsky’s perspective reminds us to deconstruct, not to mystify intelligence, be it human or machine.

2:28 “There is no such thing as consciousness, but there’s dozens of processes that involve memories of what you’ve been recently doing and new ways to represent things and describing things in language rather than images and so forth. And in fact the existence of the word ‘consciousness’ is the main reason why psychology - you could ask why is Aristotle writing and William James writing just as good as popular writing today in psychology.”

3:00 “If you read William James you say ‘oh he’s better than this guy who’s telling you how to think.’ And if you look at Aristotle, you’ll see his discussion of ethics is just as good as this president’s ethics advisor and so forth. It’s because they got stuck thinking that these words like ‘consciousness’ and ‘ethics’ are things rather than big complicated structures that might have a better explanation.”

3:30 “So anyway, we have five or six levels of thinking. And the higher level is one where you make models of your whole self, but they’re simplified, and you say ‘What would happen if I did this?’ without actually doing it - stuff like that. Those are the things we call consciousness. And I think there’s about 50 of them.”

3:50 The third wow is about John McCarthy who coined the term “artificial intelligence” at the Dartmouth conference in 1956 that he co-organized. The surprise for me was that he thought about artificial intelligence way before that. Actually he started to think about it in 1949 after he visited the conference in Caltech where all big shots of those times like John von Neumann and McCulloch were discussing cerebral mechanisms and behavior.

4:25 The craziest part about it is that people did not believe that the computers would become the key to solve artificial intelligence. The only person who from the very beginning was absolutely sure that the computers are the part that will make artificial intelligence possible was Alan Turing in his article “Computing Machinery and Intelligence” in 1950.

4:52 He laid out all the arguments and discussed it in every detail and asked so many right questions. But according to John McCarthy, no one who was interested in artificial intelligence in the late 40s and the early 50s in the states - no one read that article by Alan Turing. The only person who read was Selfridge, but Selfridge did not believe that the things that were described in “Computing Machinery and Intelligence” were possible.

5:18 That’s an absolutely invaluable piece of history. I offer you little bits of it throughout this video, but I do encourage you to watch every video that I recommend in full.

5:30 “I became interested in artificial intelligence before I became interested in computers. This was in 1949 and there was a symposium at Caltech on cerebral mechanisms and behavior, the so-called Hixon Symposium. And there were various big shots there - von Neumann and McCulloch and psychologists and physiologists.”

5:58 “And at that time I was thinking about or began to think about artificial intelligence from the point of view of special purpose computers. And however this was by no means my main activity since I was a graduate student in mathematics and went to Princeton as a graduate student in mathematics. And felt that these ideas were too indefinite although I continued to think about them for serious thesis work.”

6:30 “At that time I showed some ideas to von Neumann who said ‘write them up, write them up.’ But they weren’t good enough and I didn’t write it up. And I had taken, listened to some lectures on computers at Caltech that time which were actually - I remember the exercise was to program divide for the SWAC which was a computer which was going to appear.”

7:00 “I don’t think I understood it very well, or if I did I forgot, because - I don’t think - well, there wasn’t any SWAC to try it out on anyway. There was eventually, wasn’t there? What? There was eventually. Sure there was eventually but there wasn’t at the time these lectures were given. At least we certainly weren’t offered that opportunity.”

7:22 “My first actual contact with computers - well, let’s see. Then I thought some more. The first time I spent full time on artificial intelligence was in the summer of ‘52 at Bell Labs where Minsky and I were hired by Shannon. And again I was interested in artificial intelligence and machines for doing it but paid no attention to computers.”

7:50 “Bell Labs at that time had IBM 603 or 4 which were calculating punches and again no one thought that computers had much to do with it. Now the one person who already had expressed the view that computers were the key to artificial intelligence was Turing who had written this article ‘Computing Machinery and Intelligence’ that was published in 1950. But I can’t find anyone who was interested in artificial intelligence who had read that article - with the exception of Selfridge who claimed he’d read it but didn’t believe it. At least his approach towards artificial intelligence was computers.”

8:32 “Now I became convinced first that computers were the key to artificial intelligence in ‘55 and at that point the idea of time sharing was immediately apparent to me because, being as I was thinking about artificial intelligence already, the question was ‘well how are you going to interact with it?’ was the key question.”

8:58 “And therefore the idea that one should be sitting comfortably in one’s office and interacting with the computer - rather than thinking of it as people did at that time, of computation, is that you have this computation that you wanted to do and you got a program to run this computation and you ran the computer. So the interactive use was the immediate notion to me. And so then I began to look around and talk to people and was very surprised that that wasn’t what everybody thought.”

9:30 And now to one big promise - one very cute promise that actually never came. Meet Felix.

9:40 “That’s Felix.” “Felix? What is he? A pet or a mascot here at MIT?” “Well, you could call him a pet at the institute. He doesn’t greet everybody that way. In fact, he’s not greeting you at all. He’s just… Felix is a device that shows how a machine can take over one of the human senses - vision. He is a machine that represents an advance in evolution.”

10:12 I was thinking that if we dress up our robots in the supermarkets the way they dressed up Felix for the movie, that would be so cool, right? Just imagine these robots in the aisles looking like a bee, like a butterfly, like a bouquet of flowers. They would not even perceive as robots.

10:35 Anyway, thank you for watching. Let me know what you think. I hope I brought you some fun knowledge. Dive deeper into the videos. You can find the links for all of them in the description. Hit that subscribe button.