Latest / Stuff You Should Know / Selects: A List Of Games You Would Surely Lose to a Computer

Transcript

Selects: A List Of Games You Would Surely Lose to a Computer · Stuff You Should Know

0:00
Speaker 1: Hi everyone, it's Josh and for this week's select, I've
0:03
Speaker 1: chosen our twenty eighteen episode Some Games You would Surely
0:08
Speaker 1: Lose to a Computer. It's a philosophical discussion about AI
0:13
Speaker 1: that's disguised as an episode on computer games. Honestly, we
0:16
Speaker 1: didn't plan it to be like that. It just turned
0:19
Speaker 1: out that way. We're pretty happy that it did. And
0:22
Speaker 1: in light of the recent advances with machine learning like
0:24
Speaker 1: chat GPT, a few of the things we say seemed
0:27
Speaker 1: naively quaint now. Plus it has a doll up of
0:30
Speaker 1: our tech stuff colleague Jonathan stricklanet and so that's a bonus.
0:34
Speaker 1: I hope you enjoy.
0:39
Speaker 2: Welcome to Stuff you should know, a production of iHeartRadio.
0:49
Speaker 1: Hey, and welcome to the podcast. I'm Josh Clark. There's
0:52
Speaker 1: Charles w Chuck Bryant, there's Jerry over there. I'm just
0:56
Speaker 1: gonna come out and tell everybody making fun of me
0:59
Speaker 1: for some weird reasons, vaguely weird ways. But I'm all right,
1:05
Speaker 1: So check out his story for you. Okay, I'm going
1:07
Speaker 1: to take us back to the seventeen seventies and they'll
1:12
Speaker 1: swing in town of Vienna, not Virginia, not Vanna Georgia,
1:18
Speaker 1: which you know, that's how they pronounce it, right, Yanna,
1:21
Speaker 1: Vienna sausages, right, Vienna Austria.
1:24
Speaker 3: Have you ever been there?
1:26
Speaker 1: Vienna, Austria. No, been to Brussels. That was pretty close.
1:30
Speaker 3: Vienna's lovely, I'm sure.
1:32
Speaker 1: I think it's a lot like Brussels.
1:35
Speaker 3: Very clean, lovely town. I just remember it being very clean.
1:39
Speaker 1: Yeah, very clean, gorgeous architecture, weird little angled side streets.
1:44
Speaker 1: They're very narrow, very pretty town. So we're in Vienna
1:48
Speaker 1: and there is a dude skulking about going to the
1:52
Speaker 1: royal Palace in Vienna. His name is Wolfgang von Kempelin,
1:59
Speaker 1: and he's he's an inventor, he's an engineer. He's a
2:02
Speaker 1: pretty sharp dude. And he's got with him what would
2:05
Speaker 1: come to be known as the Turk, but he called
2:08
Speaker 1: it the mechanical Turk or the automaton chess player, and
2:13
Speaker 1: that's what it was. It was a wooden figure that
2:17
Speaker 1: moved mechanically, seated at a cabinet, and on top of
2:22
Speaker 1: the cabinet was a chessboard. And when he brought it
2:25
Speaker 1: out to show to the royal court, he would it
2:30
Speaker 1: was cool kind of but nothing they hadn't seen before,
2:34
Speaker 1: because automata was kind of a hip thing by then.
2:38
Speaker 3: Yeah, people loved building these engineering these automata machines to
2:45
Speaker 3: do various things. And people are just knocked out by
2:48
Speaker 3: the fact that, you know, you hide these gears and
2:51
Speaker 3: levers behind wood or a cloth, and it looks as
2:54
Speaker 3: though there's a real well not real, but you know.
2:57
Speaker 1: What I mean that it's like a real machine.
3:00
Speaker 3: Yeah, but not. They weren't fooled anything like is that
3:02
Speaker 3: a real man? It was, but it was for their time.
3:06
Speaker 3: It was so advanced looking that it's like us seeing
3:10
Speaker 3: x Makina in the movie theater.
3:12
Speaker 1: Sure does that make sense? Yeah, no, it does make sense.
3:15
Speaker 1: But imagine seeing like X makinan being like I've seen
3:19
Speaker 1: this before. This isn't anything special, okay.
3:22
Speaker 3: Yeah, And this thing, to be clear, looked like a
3:25
Speaker 3: is it Zoltar or Zultan from Big Zoltar Zultan?
3:31
Speaker 1: I don't know. It's one of those two.
3:32
Speaker 3: One of those two. Like this, this guy's wearing a
3:34
Speaker 3: turban and it's in a glass case like a bust like,
3:39
Speaker 3: you know, like a chest up thing.
3:41
Speaker 1: Yeah, he's seated at this cabinet, so there's no need
3:44
Speaker 1: for legs or anything like that.
3:45
Speaker 3: Yeah.
3:46
Speaker 1: But the thing, this is what was amazing about the Turk.
3:50
Speaker 1: He could play chess, and he could play chess really well.
3:54
Speaker 1: So yeah, he was like an automaton and he moved
3:57
Speaker 1: all herky jerky or whatever. But he could play you
4:00
Speaker 1: in chess, which was a huge, huge advance at the time. Like,
4:07
Speaker 1: this is something that wouldn't come up again until the
4:10
Speaker 1: nineteen nineties, more than two hundred years later. This thing,
4:13
Speaker 1: this automaton could play a human being in chess and
4:17
Speaker 1: beat them.
4:18
Speaker 3: Well. Yeah, and it looked like when the game started,
4:20
Speaker 3: it would look down at the chessboard and like cock
4:23
Speaker 3: his head, like what should my first move be?
4:25
Speaker 1: Right?
4:26
Speaker 3: And if people I love this part. If people tried
4:28
Speaker 3: to cheat. Apparently Napoleon tried to cheat this thing because
4:32
Speaker 3: this guy he debuted at the VI and East Court.
4:35
Speaker 3: But then it, you know, it went on a world tour.
4:38
Speaker 1: Yeah, and he was even it was taken over by
4:41
Speaker 1: a successor to the guy who toured with it.
4:43
Speaker 3: Even further, people went nuts for this stuff.
4:44
Speaker 1: They did. They loved it because they were like, this
4:46
Speaker 1: is crazy, I can't believe what I'm seeing. Most people, though,
4:49
Speaker 1: were not taken in by it, they're like, there's some
4:51
Speaker 1: trick here. Sure, but von Kempelin and the guy who
4:54
Speaker 1: came after him, I don't remember his name, they would
4:58
Speaker 1: demonstrate you could open this happened and you could see
5:01
Speaker 1: all the workings of the mechanical turk inside.
5:05
Speaker 3: Right, So what I was saying is if this thing's
5:08
Speaker 3: since a cheater like Napoleon supposedly did it, would you
5:13
Speaker 3: know Napoleon would move a piece out of turner illegally
5:15
Speaker 3: or something. This dude, the turk Turk one eighty two
5:19
Speaker 3: would pick up the chess piece, move it back as
5:21
Speaker 3: if to say like, no, no, Napoleon, let' see what
5:23
Speaker 3: you're doing. And then if the person attempted to move
5:26
Speaker 3: it again, I don't know how many times on you
5:28
Speaker 3: two or three times, eventually it would just go ah
5:32
Speaker 3: and wipe his hand across the board and knock off
5:34
Speaker 3: all the pieces. Game over, right, Which is pretty great. Yeah,
5:37
Speaker 3: it's a nice little feature.
5:39
Speaker 1: Yeah it is. But it even showed even more that
5:43
Speaker 1: this thing was thinking for itself. Yeah, that's the key here, right. Sure,
5:47
Speaker 1: chess had been for a very long time viewed as
5:51
Speaker 1: only something that a human would be capable of, because
5:55
Speaker 1: it took a human intellect, and there was actually a
5:58
Speaker 1: guy in English engineer, I think he was a mechanical engineer.
6:03
Speaker 1: His name was Robert Willis. He said that chess was
6:08
Speaker 1: in the province of intellect alone. So the idea that
6:12
Speaker 1: there was this automaton playing chess blew people away. But
6:16
Speaker 1: again people figured out, like, okay, there's something going on here.
6:19
Speaker 1: We think that von Kempelin is controlling this thing remotely somehow,
6:24
Speaker 1: maybe using magnets or whatever. Other people hit upon the
6:27
Speaker 1: idea that there was a small person inside the cabinet
6:31
Speaker 1: who would hide when the cab when the workings were shown,
6:34
Speaker 1: when the cabinet was open to show the workings, and
6:37
Speaker 1: then when the cabinet was closed again and the mechanical
6:39
Speaker 1: turk started playing, the person that crawled back out and
6:42
Speaker 1: was actually controlling it. This seems to be the case
6:45
Speaker 1: that there was a person controlling it, but the idea
6:48
Speaker 1: that it was it was a machine that could think
6:53
Speaker 1: and beat humans in chess had like kind of unsettling implications.
6:57
Speaker 3: Yeah, this author Philip thick Ness, great name, British author
7:04
Speaker 3: for sure, Philip Thickness. Yeah, he said, and you know,
7:09
Speaker 3: people like you said, all those more complicated explanations. In
7:12
Speaker 3: this article you sent, Astuteley points out that he followed
7:16
Speaker 3: Occam's razor and basically said, he's got a little kid
7:19
Speaker 3: in there. He's got a little a little Bobby Fisher
7:21
Speaker 3: in there that's really good at chess, and that's what's
7:25
Speaker 3: going on. And other people speculated that other, you know,
7:28
Speaker 3: little people might be in there, just adults who would
7:31
Speaker 3: fit in there. But then you know, there's the explanation
7:34
Speaker 3: that he would open it up and shine a candle
7:36
Speaker 3: around and say, you know, nothing to see here everyone,
7:42
Speaker 3: So what should we reveal the real deal?
7:46
Speaker 1: Sure, I think I did already.
7:49
Speaker 3: Well I don't think you spelled it out as.
7:51
Speaker 1: Oh, we'll spell it out.
7:52
Speaker 3: There was a little person in there, Yeah, not just
7:54
Speaker 3: one little person, but they would travel around and recruit people.
7:58
Speaker 3: I guess people would get tired of being in there.
8:00
Speaker 1: Or they'd forget about them and they'd starve and have
8:02
Speaker 1: to replace them.
8:03
Speaker 3: But it really was a trick. There was a little
8:05
Speaker 3: person in there. They did the same thing as like
8:08
Speaker 3: the magic acts, you know, when they saw a person
8:10
Speaker 3: in half. It's that the lady just gets into a
8:13
Speaker 3: tiny little ball right in one section of that box.
8:16
Speaker 1: But my thing is this, like, this is not a
8:18
Speaker 1: satisfying explanation to me, Chuck.
8:20
Speaker 3: I think it's great.
8:21
Speaker 1: How did the person keep up with the board above?
8:25
Speaker 3: Well, I mean some I don't know if they ever
8:28
Speaker 3: proved exactly how it was going.
8:30
Speaker 1: That's what I'm saying.
8:31
Speaker 3: Oh, okay, whether or not I think they that the
8:35
Speaker 3: zoltar or I'm sorry, the turk was just hollowed out
8:39
Speaker 3: and you you would just put your arms through the
8:42
Speaker 3: arm hole turk, you would become the turk and the
8:47
Speaker 3: churk would fuse.
8:48
Speaker 1: That's what some people thought. I think that's what Egar
8:50
Speaker 1: Allen Poe thought too.
8:52
Speaker 3: He Kidney right.
8:54
Speaker 1: Other people thought that there was the the the little
8:57
Speaker 1: person was underneath the in the cabin operating the trick
9:01
Speaker 1: with levers and stuff like that.
9:02
Speaker 3: Well, there could have been a mirror or something, you know,
9:04
Speaker 3: I guess, like a telescopic mirror.
9:07
Speaker 1: That's what's getting me is how would they keep up
9:09
Speaker 1: with the game?
9:10
Speaker 3: Right?
9:10
Speaker 1: You could keep track of the game, but how could
9:12
Speaker 1: you see where the other person moved? You would know
9:14
Speaker 1: where you moved, but you wouldn't be able to see
9:15
Speaker 1: where the other person moved. That's what I don't get.
9:17
Speaker 3: Just mirrors smoking mirrors.
9:19
Speaker 1: Maybe so, But the point is is it was a fake.
9:22
Speaker 1: It was a fraud, but it raise some really big
9:24
Speaker 1: questions about the idea of a machine beating a person
9:28
Speaker 1: at something like chess.
9:30
Speaker 3: Yeah, and it really peaked the mind of one Charles
9:34
Speaker 3: Babbage was he was a kid or young at least
9:38
Speaker 3: at the time when he saw the Turk in person,
9:41
Speaker 3: and a few years afterward he began work on something
9:44
Speaker 3: called the Difference Engine, which was a machine that he
9:48
Speaker 3: designed to calculate mathematics automatically. So some point to this
9:54
Speaker 3: is kind of maybe the beginnings of humans trying to
9:58
Speaker 3: create AI.
10:00
Speaker 1: Well yeah, with Babbage's differential machine or difference machine.
10:04
Speaker 3: Yeah, difference engine. But at the very least, what this
10:06
Speaker 3: is is the first that I know of example of
10:10
Speaker 3: man versus machine, even though it was really man versus
10:14
Speaker 3: man because it was a man in the machine.
10:16
Speaker 1: Right. It was a fraud, Yeah, but it sparked that idea,
10:20
Speaker 1: It definitely did. And that's something that like chess in particular,
10:25
Speaker 1: has always been like this idea of like, if you
10:28
Speaker 1: can teach a machine to play chess, you have really
10:30
Speaker 1: achieved a milestone. And there's been you know, plenty of programs,
10:35
Speaker 1: most notably Deep Blue, which we'll talk about. But there's
10:40
Speaker 1: there's been this idea that like part of AI is chess,
10:45
Speaker 1: teaching it to play chess. But they, the people who
10:49
Speaker 1: develop AI, never set out to make a chess playing AI,
10:53
Speaker 1: just to make a machine that can play chess. That's
10:56
Speaker 1: not the point. Chess has always been this way to
10:59
Speaker 1: demons straight the progress of artificial intelligence.
11:02
Speaker 3: Yeah, because it's a complex game that you can't just
11:06
Speaker 3: program it, like it almost has to learn.
11:09
Speaker 1: Well, it depends on how you come at it at first, right,
11:12
Speaker 1: So initially they did try to program it. Okay, there's
11:17
Speaker 1: this from basically nineteen fifty to the about the mid
11:23
Speaker 1: like about say nineteen fifty to twenty ten, sixty years right,
11:27
Speaker 1: That is how they approached AI and chess is you
11:31
Speaker 1: figured out how to break chess down and explain it
11:35
Speaker 1: to a computer. Now, what if you could, ideally you
11:38
Speaker 1: would have this computer or this AI, this artificial intelligence
11:44
Speaker 1: be able to think about the outcome of every possible
11:49
Speaker 1: or every possible outcome of a move before making it. Right,
11:52
Speaker 1: that's just not possible. Still today we don't have computers
11:55
Speaker 1: that can do that. Right, So what you have to
11:57
Speaker 1: do is figure out how to create shortcuts for the machine,
12:00
Speaker 1: give it best practices, that kind of thing. Yeah, And
12:03
Speaker 1: that was actually laid out in nineteen fifty by a
12:06
Speaker 1: guy named Claude Shannon who's a father of information theory,
12:10
Speaker 1: and he wrote a paper with a pretty on the
12:12
Speaker 1: nose title called Programming a Computer for Playing Chess. And
12:17
Speaker 1: you have to say it like that when you say
12:18
Speaker 1: the name.
12:19
Speaker 3: Yeah, it's got a question mark at the end, right, But.
12:21
Speaker 1: He laid out two big things. One is creating a
12:26
Speaker 1: function of the different moves, and then another one is
12:31
Speaker 1: called a mini max. And if those were the two
12:35
Speaker 1: things that Shannon laid out, and they established about fifty
12:39
Speaker 1: or sixty years of development in teaching an AI to
12:42
Speaker 1: play chess.
12:43
Speaker 3: Yeah. So this evaluation function is just sort of the base,
12:48
Speaker 3: the very basis of it all kind of where it starts,
12:50
Speaker 3: which is you a kind of give a number to
12:52
Speaker 3: create a numerical evaluation based on the state of the
12:57
Speaker 3: board at that moment, right, and assign a real number
13:01
Speaker 3: evaluation to it. So the highest number that you would
13:06
Speaker 3: shoot for is obviously getting checkmate, getting a king and checkmate.
13:11
Speaker 1: Right, right, So what you've just done now is by
13:14
Speaker 1: assigning a number to a state like the pieces on
13:18
Speaker 1: a board. What you what you've done is to say,
13:22
Speaker 1: like shoot for this number. Right, the higher the number,
13:24
Speaker 1: like you're going to give this aither rule. Now, the
13:26
Speaker 1: higher the number, the more desirable that this move that
13:30
Speaker 1: could lead to that higher number. Function. Evaluation function is
13:34
Speaker 1: what you want to do.
13:35
Speaker 3: Right, like capture the night or capture the queen. Capture
13:38
Speaker 3: the queen would have a higher evaluation number.
13:40
Speaker 1: Right exactly. So that's the function. Then there's another one
13:43
Speaker 1: called the mini max. Yeah, this is pretty great where
13:46
Speaker 1: you want to minimize the maximum And this is another
13:48
Speaker 1: shortcut that they taught computers.
13:50
Speaker 3: The maximum loss that is right, Yeah.
13:52
Speaker 1: So what they what they taught computers to do is
13:55
Speaker 1: so you no computer can look through an entire game
13:58
Speaker 1: every possible outcome, but there are computers that can look
14:04
Speaker 1: pretty far down the line at every possible outcome. And
14:09
Speaker 1: what you can say is, Okay, you want to find
14:13
Speaker 1: the evaluation function that is the worst case scenario, the
14:16
Speaker 1: maximum loss, and then find the move that will minimize
14:22
Speaker 1: the possibility for that outcome.
14:24
Speaker 3: Yeah bye, And this is you're only limited by your
14:26
Speaker 3: programming power. But by looking not only at the state
14:29
Speaker 3: of the board right now, but if I make this
14:31
Speaker 3: move and I move the PAWD to this spot, what
14:35
Speaker 3: are the next like three moves possibly that could happen
14:38
Speaker 3: as a result of this move. And you're only limited,
14:41
Speaker 3: like I said, by programming power. So obviously, the more
14:45
Speaker 3: juice you have, the more moves ahead that you can
14:48
Speaker 3: look exactly.
14:49
Speaker 1: And then they just shy away from ones with the
14:52
Speaker 1: higher function number exactly or lower function number, depending on
14:55
Speaker 1: how you've programmed it. But they're making these decisions based
14:59
Speaker 1: on these rules. And then there's other things you can do,
15:02
Speaker 1: like little shortcuts to say, if a decision tree leads
15:07
Speaker 1: to the other players King being in checkmate, don't even
15:13
Speaker 1: think about that move any further, don't evaluate any longer,
15:17
Speaker 1: just abandon it because we would never want to make
15:18
Speaker 1: that move. So there's all these shortcuts you can do.
15:21
Speaker 1: And that's what they did to teach computers. That's what
15:24
Speaker 1: Deep Blue did when it beat Gary Kasparov in nineteen
15:27
Speaker 1: ninety seven. It was this huge, massive computer that knew
15:30
Speaker 1: a lot of chess, a lot about chess. It had
15:35
Speaker 1: a lot of rules, a lot of incredibly intricate programming
15:38
Speaker 1: that was extremely sharp, and it actually won. It became
15:43
Speaker 1: the first computer to beat an actual human chess grand
15:47
Speaker 1: master in like regulation match play.
15:51
Speaker 3: Yeah. I mean, and I don't think Kasparov gets enough
15:54
Speaker 3: credit for being willing to do this, because it was
15:58
Speaker 3: a big deal for him to lose. It was in
16:01
Speaker 3: this community and the AI community. It sent shock waves,
16:04
Speaker 3: and everyone that was alive remembers, even if you didn't
16:09
Speaker 3: know anything about either, one remembers Deep Blue being all
16:12
Speaker 3: over the news. It was a really big deal. And
16:14
Speaker 3: Kasparov put his name on the line and lost.
16:17
Speaker 1: Yeah, And I was wondering, Chuck, how how like you
16:20
Speaker 1: would get somebody to do that.
16:22
Speaker 3: I'm sure the Mountain of catch.
16:25
Speaker 1: I guess that would probably be part of it. But
16:26
Speaker 1: I mean, shit, I don't know.
16:29
Speaker 3: I bet, I bet that's out there. We just I
16:31
Speaker 3: just didn't look it up.
16:32
Speaker 1: So that's possible. It's also possible that they said, look, man,
16:36
Speaker 1: like this is chess we're talking about or whatever. But really,
16:38
Speaker 1: what you're doing is helping advance artificial intelligence.
16:42
Speaker 3: Right, because we're not really trying ultimately to win chess games.
16:46
Speaker 3: We're trying to cure cancer.
16:47
Speaker 1: I mean, yeah, we're going to take your title because
16:50
Speaker 1: we're going to beat you, or our machine's going to
16:52
Speaker 1: beat you. But even still, you're going to be helping
16:55
Speaker 1: with cancer. Think of the cancer Casparov. That's probably what
16:58
Speaker 1: they said.
16:59
Speaker 3: Should we take a break?
17:00
Speaker 1: Yeah?
17:00
Speaker 3: Uh wait, well should we tease our special guests first?
17:04
Speaker 1: Is he okay? I can smell him.
17:06
Speaker 3: I don't think we even said we're gonna have a
17:08
Speaker 3: special guest later in the episode, Mister Jonathan Strickland of
17:12
Speaker 3: Tech Stuff.
17:13
Speaker 1: Nice.
17:13
Speaker 3: It's just been a long time since, like years since
17:15
Speaker 3: we had Stricken.
17:16
Speaker 1: The last time we had Stricken was like two thousand
17:19
Speaker 1: and nine with the Necronomicon episode.
17:21
Speaker 3: What is what? Where's he been besides sitting in between
17:24
Speaker 3: us every day?
17:24
Speaker 1: It's been a strickling drought, is what it's been.
17:27
Speaker 3: Yeah, so Strickland's coming later, but we're going to come
17:29
Speaker 3: back after this and talk a little bit more about
17:32
Speaker 3: a man versus machine.
17:55
Speaker 1: Okay, dude, So what we just described was how AI
17:59
Speaker 1: was to play things like chess or to think like
18:03
Speaker 1: you take something, you figure out how to break it
18:05
Speaker 1: down into little rules and things that a computer can
18:10
Speaker 1: think of, right, and then follow these kind of rules
18:13
Speaker 1: to make the best decision. That's how it used to be.
18:16
Speaker 1: The way that it's done now that everybody's doing now
18:21
Speaker 1: is where you are creating a machine that teaches itself.
18:24
Speaker 3: Yeah, that's the jam.
18:25
Speaker 1: That was the breakthrough. You may have noticed back in
18:29
Speaker 1: about ty thirteen twenty fourteen, all of a sudden, things
18:34
Speaker 1: like Siri and Alexa got way better at what they
18:39
Speaker 1: are doing. They got way less confused. Really, your navigation
18:44
Speaker 1: app got a lot better. And the reason why is
18:47
Speaker 1: because this type, this new type of AI, this new
18:50
Speaker 1: type of machine learning that can teach itself and learn
18:54
Speaker 1: on its own, just hit the scene and they just
18:57
Speaker 1: started exploding. And one of the things that they were
19:00
Speaker 1: first trained on was games.
19:02
Speaker 3: Yeah, and it makes sense. And if you thought chess
19:04
Speaker 3: was complicated and difficult, when it comes to these new
19:08
Speaker 3: AIS that they're teaching to teach themselves game strategy. They said,
19:13
Speaker 3: we might as well dive in to the Chinese strategic
19:16
Speaker 3: game Go because it has been called the most complex
19:21
Speaker 3: game ever devised by humans. Yeah, and this was actually
19:24
Speaker 3: that was actually a quote from Demi Hasabi, a neuroscientist
19:30
Speaker 3: and the founder of deep Mind, which was deep Mind.
19:35
Speaker 3: They were purchased by Google or were they always part
19:38
Speaker 3: of Google.
19:39
Speaker 1: I don't know if they were a spun off branch
19:41
Speaker 1: or where they were purchased, but it's one of Google's
19:43
Speaker 1: AI outfits.
19:45
Speaker 3: Well, they're one of the teams, yeah, that are designing
19:47
Speaker 3: these new programs. And to give you an idea of
19:50
Speaker 3: how complex Go is, it deals with a board with
19:54
Speaker 3: different stones and there are ten how do you even
19:59
Speaker 3: say that?
20:00
Speaker 1: Ten to one hundred and seventieth power, So.
20:02
Speaker 3: That means one hundred and seventy zeros and take that
20:06
Speaker 3: number and that's the number of possible configurations of a
20:09
Speaker 3: go board.
20:10
Speaker 1: Right, So, like you say, chess is very complex and complicated,
20:14
Speaker 1: and it's very difficult to master Go. And I've never
20:17
Speaker 1: played Go, of you no, So it's supposedly it's easy
20:21
Speaker 1: to learn.
20:21
Speaker 3: Right, but very complicated in its simplicity.
20:24
Speaker 1: Right, right, exactly, it's extremely difficult to master. And there
20:28
Speaker 1: was a guy in the late nineties and I'm guessing
20:31
Speaker 1: that he was saying this after Deep Blue beat Caspar
20:36
Speaker 1: ofv It was an astrophysicist from Princeton. He said that
20:41
Speaker 1: it would probably be one hundred years before a computer
20:43
Speaker 1: beats a human a go. To give you an idea
20:45
Speaker 1: of just how complex GO is that deeply would just
20:50
Speaker 1: be caspar OFV. And this guy's saying it'll still be
20:52
Speaker 1: one hundred years before anyone gets beat at GO by
20:55
Speaker 1: a computer.
20:56
Speaker 3: And he was someone who knew about this stuff, who
20:58
Speaker 3: was an astrophysicist. He was just some schmoe at home
21:01
Speaker 3: and drunken as reclining.
21:03
Speaker 1: Is making asinine predictions.
21:07
Speaker 3: So and again we've said this before, but I want
21:10
Speaker 3: to reiterate the people that I think Alpha Go is
21:14
Speaker 3: the name of this program. The people that created this
21:16
Speaker 3: at Deep Mind, they wanted to stress that this is
21:21
Speaker 3: a problem solving program. We're just teaching it this game
21:24
Speaker 3: at first, just to make it learn and to see
21:27
Speaker 3: if it can get good at what it does. But
21:30
Speaker 3: they said it is built with the idea that any
21:33
Speaker 3: task that has a lot of data that is unstructured
21:36
Speaker 3: and you want to find patterns in the data and
21:38
Speaker 3: then decide what to do right, And that's kind of
21:40
Speaker 3: like what we were talking about. It It crunches down
21:44
Speaker 3: all these possible options aka data to decide what move
21:48
Speaker 3: should I make right? And you could apply that Ideally,
21:51
Speaker 3: they're going to apply this to Alzheimer's and cancer and
21:53
Speaker 3: all sorts of things.
21:54
Speaker 1: Right, it's general purpose thinking, right, Yeah, and thinking on
21:58
Speaker 1: the fly too, and face with novel stuff. So one
22:02
Speaker 1: of the reasons why it's good to use games like
22:05
Speaker 1: chess or Go or whatever, those are called perfect information
22:09
Speaker 1: games where both players or anybody watching has all the
22:13
Speaker 1: information that's available on it. There are definite rules or structure.
22:18
Speaker 1: It's a good proving ground. But as we'll see, AI
22:21
Speaker 1: makers are getting further and further away from those structure
22:24
Speaker 1: games as their AI becomes more and more sophisticated, because
22:28
Speaker 1: the structure and the limitations aren't necessarily needed anymore, because
22:33
Speaker 1: these things are starting to be able to think on
22:35
Speaker 1: their own in a very generalized and even creative way.
22:40
Speaker 3: Yeah, it's really really interesting. Yeah, the way that they're
22:44
Speaker 3: like you said earlier before the break, that we don't
22:48
Speaker 3: have computers that can run all the possibilities. So what
22:51
Speaker 3: they teach in the case of Alpha Go, this program
22:54
Speaker 3: teaches itself by playing itself in these games and Go specifically,
23:00
Speaker 3: and the more it plays itself, the more it learns,
23:02
Speaker 3: and the more ability it has during a game to
23:07
Speaker 3: choose a move by narrowing down possibilities. So instead of like,
23:11
Speaker 3: well there are twenty million different variations here. By playing itself,
23:15
Speaker 3: it's able to say, well, in this scenario, they're really
23:18
Speaker 3: only fifty different moves that I could or should make, right,
23:22
Speaker 3: or that's kind of a simplified way to say it.
23:24
Speaker 1: But right, No, but it's true. But that's exactly right.
23:27
Speaker 1: And what they're doing is basically the same thing that
23:29
Speaker 1: a human does. It's going back to its memory banks, Yeah,
23:33
Speaker 1: exact experience, huh, and saying well, I've been faced with
23:36
Speaker 1: something like this before, and this is what I used
23:38
Speaker 1: and it was successful. Forty out of fifty times. I'll
23:41
Speaker 1: do this one. This is a pretty reasonable move. Yeah,
23:45
Speaker 1: that is what humans do.
23:46
Speaker 3: Yeah. Not only I mean, boy, we screwed up the
23:48
Speaker 3: chess episode, but I get the idea that when you're
23:51
Speaker 3: a chess master, you don't just think what do the
23:53
Speaker 3: numbers say and what does the book say?
23:55
Speaker 1: Right?
23:55
Speaker 3: But man, I did this move that one time and
23:59
Speaker 3: it didn't go as the book said.
24:02
Speaker 1: Right.
24:02
Speaker 3: So that's now factored into my thinking.
24:04
Speaker 1: Right, except imagine being able to learn from scratch and
24:09
Speaker 1: get to that point in eight days or eight hours. Yeah,
24:13
Speaker 1: So that go team the alpha go the first the
24:16
Speaker 1: first iteration of alpha go, I think they started working
24:19
Speaker 1: on it in twenty fourteen and in twenty sixteen. At
24:23
Speaker 1: the end of twenty sixteen, they unleashed it secretly onto
24:28
Speaker 1: an Alpha Go website and it started just wiping the
24:32
Speaker 1: floor with everybody. Yeah, everybody's like, this thing's pretty good. Oh,
24:36
Speaker 1: it's Alpha Go. That was the end of twenty sixteen.
24:39
Speaker 3: Okay, so chess had already come and gone. Like, oh,
24:43
Speaker 3: by this point, you can download a program that's like
24:46
Speaker 3: Deep Blue, right.
24:47
Speaker 1: That was That's a great point. Yeah, like today the
24:50
Speaker 1: stuff you played chess with on your laptop is even
24:54
Speaker 1: more advanced than Deep Blue was in the nineties, and
24:57
Speaker 1: it's just on your laptop. But this is so, this
25:00
Speaker 1: is Go. This is the end of twenty sixteen. The
25:02
Speaker 1: end of twenty seventeen, Alpha Go was replaced with Alpha
25:08
Speaker 1: Go zero. It learned what Alpha Go had taken two
25:12
Speaker 1: years or three years to learn in forty days by
25:17
Speaker 1: teaching itself, and.
25:19
Speaker 3: It beat the master. Yeah, and finally in May of
25:22
Speaker 3: twenty seventeen, Alpha Go took on Key g the highest
25:28
Speaker 3: ranked Go player in the world. Don't know if he
25:31
Speaker 3: or she still is.
25:33
Speaker 1: No. Lisa A. Doll is the current or was until
25:37
Speaker 1: Alpha Alpha Go beat him?
25:39
Speaker 3: Oh man? Yeah, did they get knocked off and Alpha
25:42
Speaker 3: Go is the champion? Yeah? Like that's that's not fair.
25:45
Speaker 1: I if it's match play and the player, the human
25:49
Speaker 1: player is accepted a challenge from the computer, I don't
25:53
Speaker 1: see why it wouldn't be the world champion.
25:56
Speaker 3: Or do they just now say on websites like human champion?
26:00
Speaker 3: Maybe in italics? What's like a sneer?
26:03
Speaker 1: Right? Maybe? Yeah?
26:06
Speaker 3: Interesting?
26:06
Speaker 1: What do they call that? Wetwear? Like your brain, your
26:10
Speaker 1: neurons and all that. What instead of hardware? It's wetwear?
26:13
Speaker 3: Oh? I don't know about that.
26:14
Speaker 1: I think that's the term for it.
26:16
Speaker 3: What does that mean? Though?
26:17
Speaker 1: It means like you you have a substrate, right your intelligence?
26:21
Speaker 1: Your intellect is based on your neurons and they're firing
26:25
Speaker 1: all that stuff and it's wet and squishy and meat.
26:28
Speaker 1: Then there's hardware that you can do the same thing on,
26:31
Speaker 1: you can build intelligence on, but it's hardware, it's not wetwear.
26:35
Speaker 3: Oh.
26:35
Speaker 1: Interesting, so that's probably it. It's the wetwear champion versus
26:39
Speaker 1: the hardware champion. But wetwear is italicized with the sneer.
26:44
Speaker 3: So where things really got interesting because you were talking
26:46
Speaker 3: earlier about what is it with the chest and go?
26:51
Speaker 3: What do they called? What kind of games?
26:53
Speaker 1: Perfect information games?
26:55
Speaker 3: Right? Then you think And my first thought when you
26:58
Speaker 3: said that was well, yeah, and then there's there's games
27:01
Speaker 3: like poker like Texas Hold Them where there are a
27:04
Speaker 3: set of rules. But poker is not about the set
27:07
Speaker 3: of rules. It is about sitting down in front of
27:10
Speaker 3: whatever five or six people and lying, bluffing and getting
27:15
Speaker 3: away with it in your game face being bluff Like
27:18
Speaker 3: there's so many human emotions and contextual clues and micro
27:25
Speaker 3: expressions and all these things, like surely you could never
27:29
Speaker 3: ever teach a machine to win at Texas Holding Poker.
27:33
Speaker 1: Yeah, it'll be one hundred years at least before that happens,
27:35
Speaker 1: I predict.
27:37
Speaker 3: No, they did it, and more than one team has
27:41
Speaker 3: done it.
27:42
Speaker 1: Yeah. I read there was one from Carnegie Mellon called
27:46
Speaker 1: Liberatus AI Go melon Heads, Yeah, go the Thornton Melons.
27:53
Speaker 3: Yeah, I mean that's was The University Alberta has one
27:57
Speaker 3: called deep Stack.
27:58
Speaker 1: That was the one I read about. And it actually
28:01
Speaker 1: here's the thing, like if you read the release on it,
28:05
Speaker 1: you're like, you don't know how this thing works?
28:07
Speaker 3: Do you really?
28:08
Speaker 1: Yeah? And I'm pretty sure they don't fully get it,
28:11
Speaker 1: because that's one of the problems. I actually talk about
28:13
Speaker 1: this in the Existential Risks series.
28:15
Speaker 3: That's scary that is to be released right.
28:18
Speaker 1: That there is a type of machine learning where the
28:22
Speaker 1: machine teaches itself but we don't really understand how it's teaching.
28:26
Speaker 1: Probably the scariest one, right, or what it's learning, but
28:28
Speaker 1: that's the most prevalent one. That's what a lot of
28:31
Speaker 1: this is is like these machines. It's like here's chess,
28:36
Speaker 1: go figure it out and they go, okay, got it.
28:39
Speaker 1: How'd you do that? Wouldn't you like to know?
28:42
Speaker 3: So that's the scariest presentation you will see on AI
28:45
Speaker 3: is when someone says, well, how does all this work?
28:47
Speaker 1: And they go, but we just know it can be
28:50
Speaker 1: to human at poker. But the thing about deep Stack
28:54
Speaker 1: at the University of Alberta is that it learned somehow
28:58
Speaker 1: some sort of intuition. Yeah, because that's what's required is
29:02
Speaker 1: not just the perfect information where you have all the
29:05
Speaker 1: information on the board. It's with poker, you don't know
29:08
Speaker 1: what the other person's cards are, and you don't know
29:10
Speaker 1: if they're lying or bluffing or what they're doing so
29:14
Speaker 1: that's an imperfect information game, so that would require intuition,
29:18
Speaker 1: and apparently not one, but two different research groups taught
29:22
Speaker 1: AI to into it.
29:24
Speaker 3: Yeah, Carnegie Mellon came out in January of twenty seventeen
29:28
Speaker 3: with its Liberatus AI and they said they spent twenty
29:32
Speaker 3: days playing one hundred and twenty thousand hands of Texas
29:36
Speaker 3: hold them with four professional poker players and one and
29:42
Speaker 3: smoked them. Basically got up to They weren't playing with
29:45
Speaker 3: real money, obviously, but they they that would have been great.
29:48
Speaker 1: They were playing with skittles like me as a kid.
29:50
Speaker 3: Funded their next project, Liberatus was up by one point
29:53
Speaker 3: seven million, and one of the quotes from one of
29:56
Speaker 3: the poker players that he made to Wired magazine said,
30:00
Speaker 3: felt like I was playing against someone who was cheating,
30:02
Speaker 3: Like it could see my cards. I'm not accusing it
30:04
Speaker 3: of cheating. It was just that good.
30:06
Speaker 2: Right.
30:07
Speaker 3: So that's a really interesting thing, man, that they could
30:11
Speaker 3: teach self teach a program, or a program could teach
30:15
Speaker 3: itself intuition. Right, it's creepy. I thought this part was interesting,
30:20
Speaker 3: the Atari stuff. This gets pretty fun. Google deep mind.
30:26
Speaker 3: Let it's AI wreak havoc on Atari forty nine different
30:32
Speaker 3: Atari twenty six hundred games. See, they could figure out
30:35
Speaker 3: how to win, and apparently the most difficult one was
30:39
Speaker 3: Miss pac Man, which is a tough game. Still man,
30:43
Speaker 3: misspac Man. They nailed it. It's still one of the
30:46
Speaker 3: great games.
30:46
Speaker 1: But their their game, or their Q deep Q network
30:52
Speaker 1: algorithm beat it.
30:54
Speaker 3: Yeah.
30:55
Speaker 1: I think it got the highest score nine nine points,
30:59
Speaker 1: and no human or machine has ever achieved that high
31:02
Speaker 1: score from what I.
31:03
Speaker 3: Understand, amazing. And the way this one does it, the
31:05
Speaker 3: hybrid reward architecture that it uses, is really interesting. It
31:11
Speaker 3: says here, it generates a top agent that's like a
31:14
Speaker 3: senior manager, and then all these other one hundred and
31:17
Speaker 3: fifty individual agents. So it's almost like they've devised this
31:21
Speaker 3: artificial structural hierarchy of these little worker agents that go
31:26
Speaker 3: out and collect I guess data and then move it
31:30
Speaker 3: up the chain to this top agent.
31:34
Speaker 1: Right, and then this thing says, Okay, you know, I
31:38
Speaker 1: think that you're probably right what these agents are probably doing.
31:43
Speaker 1: And I don't know this is exactly true, but there's
31:46
Speaker 1: there are models out there like this where the agent
31:49
Speaker 1: says this is you have a ninety percent chance of
31:52
Speaker 1: success at getting this pellet. If we take this action,
31:57
Speaker 1: somebody else says, you've got an eighty two percent chance
31:59
Speaker 1: of av this ghost if we go this way. And
32:02
Speaker 1: then the top agent, the senior manager, can put all
32:06
Speaker 1: this stuff together and say, well, if I listen to
32:08
Speaker 1: this guy and this guy not only while I evade
32:10
Speaker 1: this ghost, I'll go get this pellet. And it's based
32:13
Speaker 1: on what confidence level that the lower agents have in
32:19
Speaker 1: success in recommending these moves. And then the top agent
32:23
Speaker 1: weighs these things.
32:24
Speaker 3: Wow, they should give him a little cap.
32:27
Speaker 1: But all this is happening like that. Oh yeah, you
32:29
Speaker 1: know what I'm saying. This isn't like well, hold on,
32:31
Speaker 1: hold on, everybody, what is Harvey? What do you have
32:33
Speaker 1: to say? Well, let's get some Chinese in here and
32:36
Speaker 1: hash it out, and everybody sits there in order some
32:39
Speaker 1: Chinese food, and then you wait for it to come,
32:41
Speaker 1: and then you pick up the meeting from that point on.
32:44
Speaker 1: And then finally Harvey gives his idea, but he forgot
32:47
Speaker 1: what he was talking about, so he just sits down
32:50
Speaker 1: and eats his a roll.
32:51
Speaker 3: Well, here's a pretty frightening survey. There was a survey
32:55
Speaker 3: of more than three hundred and fifty AI researchers, and
32:58
Speaker 3: they have the following things to say, And these are
33:00
Speaker 3: the pros that are doing this for a living. They
33:03
Speaker 3: predicted that within ten years AI will drive better than
33:06
Speaker 3: we do. By twenty forty nine they will be able
33:09
Speaker 3: to write a best selling novel. AI will generate this
33:13
Speaker 3: and by twenty fifty three be better at performing surgery
33:17
Speaker 3: than humans are.
33:19
Speaker 1: You know. So again, one of the things that about
33:22
Speaker 1: the field of artificial intelligence.
33:25
Speaker 3: You know a lot about now famous.
33:27
Speaker 1: It is famous for making huge predictions that did not
33:31
Speaker 1: pan out. Sure, but you've also seen it's also famous
33:35
Speaker 1: for beating predictions that you know have been levied against it.
33:40
Speaker 1: But there is something in there, Chuck, that stands out
33:43
Speaker 1: to me, and that's the idea of an AI writing
33:46
Speaker 1: a novel. Like for a very long time I thought, well, yeah, okay,
33:50
Speaker 1: you can teach a robot arm to like put a
33:53
Speaker 1: car part or something somewhere if you wanted to just
33:56
Speaker 1: follow these mechanical things. Or it can use in auition,
33:59
Speaker 1: or it can use logic in reason. But to create
34:02
Speaker 1: that's different, right. That was like the new frontier. It
34:05
Speaker 1: used to be chess and then then was go. The
34:08
Speaker 1: next frontier is creativity and they're starting to bang on
34:11
Speaker 1: that door big time. There's a game designing AI called
34:15
Speaker 1: Angelina out of the University of Foulmouth, which I always
34:19
Speaker 1: want to say Foulmouth, Yeah, but we'll just call it
34:21
Speaker 1: Foulmouth like it's supposed to. And Angelina actually comes up
34:26
Speaker 1: with ideas for new games, not like a different level
34:32
Speaker 1: or something like you should put a purple loincloth on
34:36
Speaker 1: that player, you know, that'll look kind of cool like
34:38
Speaker 1: new games, but whacked out games that humans would never
34:42
Speaker 1: think of. One example I saw is in a dungeon
34:45
Speaker 1: Battle Royale game, a player controls like ten players at once,
34:50
Speaker 1: and some you have to sacrifice to be killed to
34:52
Speaker 1: save the others, like the stuff that human wouldn't necessarily
34:55
Speaker 1: think of. This AI is coming up with.
34:58
Speaker 3: Well, I mean, when you think of creatively, especially something
35:00
Speaker 3: like writing a novel or a film, if there are
35:03
Speaker 3: only seven stories, I mean, and that's sort of the
35:07
Speaker 3: thinking that they're basically every every dramatic story is a
35:10
Speaker 3: variation of one of seven things.
35:12
Speaker 1: Yeah. So I mean you can look at like AI
35:16
Speaker 1: is scary, and in some ways it very much is
35:20
Speaker 1: and can be, but there's also like, definitely a level
35:23
Speaker 1: of excitement of the whole thing, and the idea that
35:26
Speaker 1: there are artificial minds that are coming online or that
35:30
Speaker 1: have come online now, that are out there that are
35:33
Speaker 1: they'll they'll just naturally by definition, see things differently than
35:36
Speaker 1: we do. Yeah, and the idea that they can come
35:39
Speaker 1: up with stuff that we've never even thought of there
35:40
Speaker 1: is just gonna knock our socks off, hopefully in good ways.
35:45
Speaker 1: That's a really cool thing. And so maybe there's just
35:48
Speaker 1: seven as far as humans know, but there's an unlimited amount.
35:52
Speaker 1: Is if you put computer minds to thinking about these
35:55
Speaker 1: kind of things, that's the premise of it.
35:57
Speaker 3: Right, so the robot would be like you never thought
35:59
Speaker 3: of boy meets girl meets well trilobite. But see even
36:06
Speaker 3: that's a variation of right.
36:09
Speaker 1: Just imagine something that we've never even thought of.
36:12
Speaker 3: Well, do you know how they should do this? If
36:13
Speaker 3: they do do that is uh? Is not is just
36:17
Speaker 3: release a book and not tell anyone that it was
36:20
Speaker 3: written by an AI program because if they do that
36:24
Speaker 3: then it's going to be so under scrutiny. Oh yeah,
36:26
Speaker 3: they should secretly release this book and then after it's
36:29
Speaker 3: a New York Times bestseller, say meet the Whopper, the
36:34
Speaker 3: author of this.
36:36
Speaker 1: You know his interests are roller skating, playing Tic tac toe,
36:39
Speaker 1: and global thermal nuclear war.
36:42
Speaker 3: All right, should we take a break and get strickland
36:44
Speaker 3: in here.
36:44
Speaker 1: Yeah, we're going to end the Strickland drought because it
36:46
Speaker 1: is about to rain strickland in this piece.
36:49
Speaker 3: You gross.
37:12
Speaker 1: Okay, we're back and get this. The scent of strick
37:17
Speaker 1: has permeated our place. That's a beautiful scent.
37:20
Speaker 3: It smells like a soldering gun and a circuit board
37:25
Speaker 3: and feel a lavender in a protein bar.
37:28
Speaker 2: That's fair. I was gonna say Draco noir that would
37:31
Speaker 2: have been a lie.
37:32
Speaker 1: Is that how you said? I always called a drakar dracr.
37:36
Speaker 2: That's that's fair.
37:38
Speaker 3: I always pronounced it Benetton colors. That was what I wore.
37:43
Speaker 1: Oh is that what you wore?
37:44
Speaker 3: Yeah? During my what I call the Year of Cologne,
37:47
Speaker 3: I had a couple of seven.
37:51
Speaker 1: Uh.
37:51
Speaker 2: This is scintillating. Why why am I here?
37:55
Speaker 1: So we know that you already know because we talked
37:58
Speaker 1: via email about this, but we'll tell everybody nobody else.
38:01
Speaker 1: We have brought you in here because you are the
38:03
Speaker 1: master of tech and we were talking tech today, which
38:06
Speaker 1: we've talked about without you before. But frankly, Chuck and
38:08
Speaker 1: I and Jerry huddled and we said, this is not
38:11
Speaker 1: quite as good with that strict so let's try something different, gotcha.
38:14
Speaker 2: And we're talking about games and machine versus man and
38:20
Speaker 2: that that whole evolution and how that's gone super crazy
38:24
Speaker 2: over the last few years.
38:24
Speaker 3: Games without frontiers, as Peter Gabriel would say, yeah, or
38:28
Speaker 3: without fear.
38:29
Speaker 1: And we've talked, I mean, we've talked a lot about
38:31
Speaker 1: the evolution of machine learning and how now it's starting
38:34
Speaker 1: to take off like a rocket because they can teach themselves, right.
38:38
Speaker 1: But one thing we haven't really talked about are solved games.
38:42
Speaker 1: I mean, we talked about chess. Yeah, we talked about
38:44
Speaker 1: go right, would those constitute solve games?
38:48
Speaker 3: Not really?
38:49
Speaker 2: So a solved game is the concept where if you
38:52
Speaker 2: were to assume perfect play on either sides of the game,
38:56
Speaker 2: you would always know how it was going to end.
38:58
Speaker 3: Which we always assume perfect play, right, Yeah, it's kind
39:01
Speaker 3: of our.
39:01
Speaker 1: Bags, right stuff.
39:02
Speaker 2: You should know motto so perfect play just meaning that
39:05
Speaker 2: no one ever makes a mistake, so very much the
39:07
Speaker 2: way I do my work right stuff, you should know
39:09
Speaker 2: exactly So if you were to take a game like
39:12
Speaker 2: Tic Tac Toe and you assume perfect play on both sides,
39:16
Speaker 2: it is always going to end in.
39:17
Speaker 3: A draw, which is what was in war games.
39:19
Speaker 2: Yes, right, the only way to win is not to
39:22
Speaker 2: play right. Yes, so a game with like a game
39:26
Speaker 2: like Connect four, whoever goes first is always going to win,
39:30
Speaker 2: assuming perfect play both sides.
39:33
Speaker 3: Yes, what, I don't think I've played Connect for it.
39:36
Speaker 3: That's where you drop ever a long time. That's wh
39:38
Speaker 3: where you drop the little tokens.
39:41
Speaker 2: Yeah, like it kind of like Checkers.
39:42
Speaker 1: You did an interstitial playing Connect four. Remember I was.
39:45
Speaker 3: Faking it though, and you had perfect play, so I
39:47
Speaker 3: knew it was useless.
39:48
Speaker 1: I was going to say that I'm so humiliated by
39:51
Speaker 1: all the Connect four games that I've lost starting even.
39:55
Speaker 2: Yeah, but I mean perfect play. That's something that that
39:59
Speaker 2: obviously only the best players typically achieve with significantly complex games.
40:05
Speaker 2: Obviously the simpler the game, the easier it is to
40:07
Speaker 2: play perfectly. Right tic Tac toe, if you know, once
40:11
Speaker 2: you've mastered the basics of Tic Tac Toe and the
40:14
Speaker 2: other person has, you're never really going to win unless
40:17
Speaker 2: someone has just made a silly mistake because they weren't
40:20
Speaker 2: paying attention.
40:21
Speaker 3: They put a star instead of an xer right.
40:23
Speaker 1: Which doesn't count automatically disqualified you. One thing I've found
40:27
Speaker 1: that's very enjoyable is playing with little kids who haven't
40:29
Speaker 1: figured out that tic tac toe is very easy.
40:31
Speaker 3: To yes and smash their face on the board and
40:34
Speaker 3: rub it in.
40:35
Speaker 1: Yeah.
40:35
Speaker 2: I mean the same reason why I like to join
40:38
Speaker 2: in on little league games because I can really whale
40:40
Speaker 2: that ball out of the park. Yeah, I really missed
40:42
Speaker 2: me feel like a man.
40:43
Speaker 3: That's the most tech stuffy thing you've ever said. You
40:46
Speaker 3: really whaled that ball out of the park.
40:47
Speaker 2: Well, to be fair, I did just do a textuff
40:50
Speaker 2: episode about the technology behind baseball bats, so it's so
40:52
Speaker 2: fresh on mine.
40:53
Speaker 1: Nice.
40:53
Speaker 3: Yeah one.
40:54
Speaker 2: Actually, it's a lot of fun. So there have been
40:57
Speaker 2: a lot of games that have been solved, but Checkers
40:59
Speaker 2: will one that was recently solved back in recently by
41:02
Speaker 2: the early nineties when it was played against a computer
41:07
Speaker 2: called Chinook and chi in Ok.
41:11
Speaker 1: Yeah, like the Helicopter or the wins that blow through
41:13
Speaker 1: Alberta exactly.
41:15
Speaker 2: And so there are certain games that are more easily
41:20
Speaker 2: solved than others. You do it through an algorithm, but
41:23
Speaker 2: other games like chess, are more complicated because you can
41:26
Speaker 2: In chess you have multiple moves that you can do
41:29
Speaker 2: where you can you can move a piece back the
41:31
Speaker 2: way you win, right, it's not you're not committed to
41:34
Speaker 2: going a specific direction with certain pieces, like with a
41:37
Speaker 2: night you know you can you could go right back
41:39
Speaker 2: to where you started on the next move if you
41:41
Speaker 2: wanted to, and that creates more complexity. So the more
41:46
Speaker 2: complex the game, the more difficult it is to solve.
41:48
Speaker 2: And some games are not solvable simply because you'll never
41:52
Speaker 2: know what the full state of the game is from
41:56
Speaker 2: any given moment. Did you have a chance to talk
42:00
Speaker 2: about the difference between perfect knowledge and imperfect knowledge in
42:03
Speaker 2: a game.
42:04
Speaker 3: Yeah, yeah, we talked about that some. Yeah.
42:06
Speaker 2: So computers, obviously they do really well if they understand
42:11
Speaker 2: the exact state of the game all the way through,
42:13
Speaker 2: if they have perfect knowledge.
42:15
Speaker 1: All of the informations there on the board.
42:17
Speaker 2: Right and all players can see all information at all times.
42:20
Speaker 2: But games like poker, which you guys talked about, obviously
42:23
Speaker 2: you have imperfect information. You only know part of the
42:27
Speaker 2: state of the game. That's why those games have been
42:30
Speaker 2: more difficult, more challenging for computers to get better than
42:34
Speaker 2: humans until relatively recently, and there have been two major
42:38
Speaker 2: ways of doing that. You either throw more processing power
42:40
Speaker 2: at it, like you get a supercomputer, or you create
42:43
Speaker 2: neural networks, artificial neural networks, and you start teaching computers
42:47
Speaker 2: to quote unquote learn the way people do.
42:51
Speaker 1: So we talked about that. Yeah, and one of the
42:53
Speaker 1: things that we talked about was how there's this idea
42:56
Speaker 1: that the programmers, especially say the people who are making
43:01
Speaker 1: programs that are playing poker and or getting good at poker,
43:03
Speaker 1: aren't exactly sure how the machines are learning to play
43:07
Speaker 1: poker or what they're learning. They're just getting better at poker. Yeah,
43:11
Speaker 1: do they know how they're learning poker? They just know
43:14
Speaker 1: that they're learning poker and that they're good at it. Now, like,
43:17
Speaker 1: where's the intuition? How is that being learned?
43:20
Speaker 2: An excellent question. The way it typically has learned, especially
43:23
Speaker 2: with artificial neural networks, is that you set up the
43:26
Speaker 2: computer to play millions of hands of poker that are
43:32
Speaker 2: randomly assigned, so it's truly as random as computers can get.
43:37
Speaker 2: That's a whole philosophical discussion that I don't think we're
43:39
Speaker 2: ready to go into right now. But you have games
43:44
Speaker 2: come up where the computer is playing itself millions upon
43:48
Speaker 2: millions of times and learning every single time how the
43:52
Speaker 2: statistics play out, how different betting strategies play out. It's
43:57
Speaker 2: sort of partitioning its own mind to play against itself.
44:01
Speaker 2: And through that process, it's as if you, as a
44:04
Speaker 2: human player, were playing thousands of games with your friends
44:08
Speaker 2: and you start to figure out, Oh, when I have
44:10
Speaker 2: these particular cards and they're in my hand, and let's
44:13
Speaker 2: say we're playing Texas, hold them and the community cards
44:16
Speaker 2: are are these? Then I know that generally speaking, maybe
44:20
Speaker 2: three times out of ten I end up winning. Maybe
44:22
Speaker 2: I shouldn't bet. Well, the computer is doing that, but
44:25
Speaker 2: on a scale that far dwarfs what any human can do,
44:29
Speaker 2: and in a fraction in the amount of time, and
44:31
Speaker 2: so it's sort of well, it's intuition in the sense
44:35
Speaker 2: of it's just done it so much.
44:38
Speaker 3: Right, But does that mean it's completely ignoring micro expressions
44:44
Speaker 3: and facial cues, so that doesn't even come into play.
44:47
Speaker 1: You should say, Strickland just nodded yet, Yeah I was.
44:49
Speaker 2: I was waiting for John Well. I still nod when
44:53
Speaker 2: I do a solo show. And I do a lot
44:54
Speaker 2: of expressive dance.
44:56
Speaker 3: What do you think, Jonathan, I don't know, Jonathan.
44:59
Speaker 2: It gets lonely in here, guys. No, but yes, what
45:02
Speaker 2: you're saying all the tells, right, Yeah, tells that you
45:05
Speaker 2: would use as a human player. The computer does not
45:08
Speaker 2: pick up.
45:08
Speaker 3: A speypically taking data.
45:10
Speaker 2: Yes, typically, what it would do is it would study
45:12
Speaker 2: the outcomes of the games from a purely statistical expression,
45:17
Speaker 2: so that most of these poker games tend to be
45:19
Speaker 2: computer based poker games. So it's not that it's playing
45:22
Speaker 2: like it's not like there's a computer that says, pushed
45:25
Speaker 2: ten more chips into the table. You know, I tick
45:29
Speaker 2: right exactly a little it's a little winky face emoticon,
45:33
Speaker 2: like I don't have good cards. No, it's it's all
45:36
Speaker 2: usually over, sort of like internet poker, which a lot
45:40
Speaker 2: of the people who play professional poker cut their teeth on,
45:43
Speaker 2: especially you know, in the the more recent generations of
45:46
Speaker 2: professional poker players.
45:48
Speaker 3: Kids today, Yeah, they don't know what it's like being
45:50
Speaker 3: a smoky saloon like Moneymaker.
45:53
Speaker 2: When Moneymaker rose to the top a few years ago,
45:55
Speaker 2: more than like a decade ago, now, he had come
45:58
Speaker 2: from the world of internet poker, and so he was
46:01
Speaker 2: using those same sort of skills in a real world setting.
46:05
Speaker 2: But obviously there are subtle things that we humans do
46:09
Speaker 2: in our expressions their computers do not pick up on it.
46:11
Speaker 2: In fact, that leads us sort of into the realm
46:14
Speaker 2: of games where computers don't do as well as humans.
46:19
Speaker 3: Yeah, is that list you sent a joke or is
46:21
Speaker 3: it real? No, that's real.
46:23
Speaker 2: It does seem like it's weird, like one of the
46:25
Speaker 2: games on there is pictionary, for example, rag or tag. Yeah,
46:28
Speaker 2: but these are some of these are They sound silly,
46:32
Speaker 2: But when you start to think about them in terms
46:33
Speaker 2: of computation and robotics, you start to realize how incredibly
46:37
Speaker 2: complex it is from a technical perspective, but incredibly easy
46:41
Speaker 2: it is for your average human being. Okay, so with
46:44
Speaker 2: humans a game of tag, once you know the basics,
46:47
Speaker 2: it's it's all an instinct.
46:49
Speaker 1: You know what to do.
46:50
Speaker 2: You run after the person you tried to catch up
46:52
Speaker 2: with them and tag them. But you also know push
46:54
Speaker 2: them ind as you can. Well, if you're Josh, you
46:57
Speaker 2: push them as hard as you can. But most of
46:59
Speaker 2: us we tag and we're not trying to cause harm. Robots, however,
47:04
Speaker 2: robots not so good on you said, I'm just saying
47:09
Speaker 2: Isaac Asimov Isaac Asimov's rules of robotics. Aside, robots are
47:13
Speaker 2: not very good at judging how hard they have to
47:16
Speaker 2: hit something in order to make contact, right, They're not
47:19
Speaker 2: as good at even your bipedal robots that walk around
47:23
Speaker 2: like people, even the ones that can run and do
47:25
Speaker 2: flips and stuff.
47:26
Speaker 3: Have you seen that one the other day that the
47:28
Speaker 3: footage of that thing running and jumping, it's really impressive
47:31
Speaker 3: and super creepy.
47:32
Speaker 2: Yeah, but even so, that's that's a clip of the
47:36
Speaker 2: best of If you ever if you ever see the
47:38
Speaker 2: clips where they show all the times the robots fallen over.
47:41
Speaker 3: Yeah, we're pouring hot coffee in someone's head.
47:44
Speaker 1: Yes, but they always play those clip shows, toy.
47:48
Speaker 2: Yes, this is true. So DARPA had its big robotics
47:52
Speaker 2: challenge a few years ago where they had bipedal robots
47:56
Speaker 2: tried to go through a scenario that was simulating Fukushima
48:01
Speaker 2: nuclear disaster. So the interesting thing was the robot had
48:05
Speaker 2: to complete a series of tasks that would have been
48:08
Speaker 2: mundane to humans, things like open up a door and
48:11
Speaker 2: walk through it and pick up a power tool and
48:14
Speaker 2: use it against a wall, and you can watch the
48:17
Speaker 2: footage of some of these robots doing things like being
48:21
Speaker 2: unable to open the door because they can't tell if
48:25
Speaker 2: they need to pull or push or they open the door,
48:28
Speaker 2: but then immediately fall over the threshold of the door.
48:31
Speaker 2: And when you see that, you realize, as advanced as
48:34
Speaker 2: robotics is, as advanced as machine learning has become, and
48:38
Speaker 2: as incredible as our technology has progressed, there are still
48:42
Speaker 2: things that are fundamentally simple to your average human right
48:46
Speaker 2: that are incredibly complicated from a technical standpoint, Like a.
48:49
Speaker 3: Six year old can play jinga better than a robot.
48:52
Speaker 1: Right right, right, Okay, But the thing is we're talking
48:54
Speaker 1: robots here, and as we go more and more and
48:57
Speaker 1: more online and our world becomes more and more web
49:00
Speaker 1: based rather than reality based, doesn't the the fact that
49:06
Speaker 1: a robot can't walk through a door matter less and less,
49:09
Speaker 1: and the idea that that machines are learning intellect and
49:14
Speaker 1: the robotivity, and you just blew my mind that that's
49:18
Speaker 1: becoming more and more vital and important and something we
49:21
Speaker 1: should be paying attention.
49:22
Speaker 2: It absolutely is something we should pay attention to. I mean,
49:24
Speaker 2: we have robotic stock traders, the trading on thousands of
49:30
Speaker 2: trades per second, right fast, so fast that we have
49:33
Speaker 2: had stock market booms and crashes that last less than
49:37
Speaker 2: a second long due to that.
49:39
Speaker 3: So that the robot army that will ultimately defeat us
49:41
Speaker 3: is not something from the terminator.
49:43
Speaker 1: It's invisible, right, it's online, it will be online, it's.
49:48
Speaker 2: It's it's what's determining our retirement.
49:50
Speaker 1: Right, Yeah, the global economy or our municipal water supply
49:56
Speaker 1: or whatever.
49:57
Speaker 3: Yeah.
49:57
Speaker 2: No, There's the fascinating thing to me about this is
50:01
Speaker 2: not just that we're training machine intelligence to learn and
50:06
Speaker 2: to perform at a level better than humans, but that
50:09
Speaker 2: we're putting a lot of trust in those devices and
50:13
Speaker 2: things that have real incredible impact on our lives, significant
50:19
Speaker 2: enough impact where if things were to go south, it
50:21
Speaker 2: would be really bad for us, and not in that
50:24
Speaker 2: Terminator respect. Terminator is a terrifying dystopian science fiction story.
50:31
Speaker 2: But then when you realize what could really happen behind
50:34
Speaker 2: the scenes, you think, oh, robots don't have to do
50:37
Speaker 2: any physical harm to us to really mess things up.
50:40
Speaker 2: So there are certainly some cases for us to be
50:44
Speaker 2: very vigilant in the way we deploy this artificial ants
50:48
Speaker 2: right from the outset exactly, and too depends not necessarily.
50:55
Speaker 2: I think I think it's I don't think it's too late,
50:59
Speaker 2: but I think it's getting to that point of no
51:02
Speaker 2: return very very quickly.
51:04
Speaker 3: By December this year. Yeah.
51:06
Speaker 2: Well, if you're if you're someone like if you're someone
51:09
Speaker 2: like Elon Musk, you'd say, if we don't do something
51:11
Speaker 2: now where we're totally going to plummet off the edge
51:15
Speaker 2: of the cliff.
51:15
Speaker 1: But now is a window that is rapidly closing.
51:18
Speaker 2: Yes, yeah, yeah, the now is the now is a
51:21
Speaker 2: time where we've got a deadline. We don't know exactly
51:23
Speaker 2: when that deadline is going to be up, but we
51:27
Speaker 2: know that it's not getting further out it. We're just
51:30
Speaker 2: getting closer to that deadline. So, and a lot of
51:33
Speaker 2: this is covered in deep conversations and the artificial intelligence
51:38
Speaker 2: and machine learning fields that has been going on for ages,
51:42
Speaker 2: to the point where you even have bodies like the
51:45
Speaker 2: European Union that have debated on concepts like granting personhood
51:50
Speaker 2: to artificial intelligence. So this is a really fascinating and
51:55
Speaker 2: deep subject that and the games thing is a great
51:59
Speaker 2: entry point into have that conversation. Uh, you know. I'm
52:03
Speaker 2: lucky if I can win a game of chess against
52:04
Speaker 2: another human being.
52:05
Speaker 1: Oh yeah, right, so I can't even describe chess.
52:09
Speaker 3: Did I did? My big thing is I do that
52:11
Speaker 3: night thing. I call it the night shuffle. I just
52:13
Speaker 3: move them back and forth. I just cast.
52:16
Speaker 2: If I can castle, then I'm so happy.
52:20
Speaker 3: And that's the third tech stuffiest thing. They come in threes.
52:24
Speaker 1: Well, Strick, thank you for stopping by.
52:26
Speaker 3: Should stick around for listener mail.
52:28
Speaker 1: I think you should too. I love to and throw
52:31
Speaker 1: out any funny comments that you have.
52:33
Speaker 2: I'll throw out comments and then Jerry can decide which
52:36
Speaker 2: ones are funny.
52:36
Speaker 1: Okay, all right, fair enough, all right, So if you
52:39
Speaker 1: want to know more about AI, go listen to tech Stuff.
52:42
Speaker 1: Strict does this every week what days.
52:45
Speaker 2: Monday, Tuesday, Wednesday, Thursday and Friday.
52:48
Speaker 1: Wow, that's amazing, buddy. And wherever you find your podcast yep, okay,
52:54
Speaker 1: And you've been doing it for years, so if you
52:56
Speaker 1: love this, there's a whole big backlog, nine hundred plus episode.
52:59
Speaker 3: You're celebrating your ten year as well, right yep, by.
53:02
Speaker 2: Sure am, I'll be We'll be turning ten and tex
53:05
Speaker 2: stuff on June eleventh.
53:07
Speaker 1: Congratulations, well, since I said happy anniversary, it means it's
53:11
Speaker 1: time for listener mail.
53:14
Speaker 3: Guys, I'm gonna call this Matt Groening and cultural relativism
53:19
Speaker 3: about that?
53:20
Speaker 1: Nice?
53:20
Speaker 3: Hey, guys, love your podcast so much. The massive archive
53:23
Speaker 3: makes for endless learning and entertainment. My favorite part is
53:26
Speaker 3: you were such rad guys, including Strickland, and I could
53:29
Speaker 3: totally imagine how did they know? I can totally imagine
53:33
Speaker 3: myself getting a beer with you two, but without Strickland,
53:36
Speaker 3: your Simpsons episodes were absolutely perfect. I still live in
53:40
Speaker 3: Portland and drove on Flanders and Lovejoy Streets a lot.
53:42
Speaker 1: Wait, is this Matt Groening? Okay?
53:45
Speaker 3: Matt Granning Drew Bart in the sidewalk cement behind Lincoln
53:49
Speaker 3: High School in downtown Portland. You can google that. I
53:52
Speaker 3: would like to offer one interesting observation though, I've noticed
53:55
Speaker 3: that on several episodes you guys have said that you
53:57
Speaker 3: are cultural relativists. Is that pronounce right? Yeah? Yeah? But
54:01
Speaker 3: then in nearly every episode I hear you pass moral
54:04
Speaker 3: judgments on all the messed up stuff that people do,
54:07
Speaker 3: whether it's racism, preak shows, or crematoriums bearing bodies on
54:11
Speaker 3: the sly. You guys are never shy to condemn something
54:14
Speaker 3: that deserves to be condemned. Reminds me of something I
54:16
Speaker 3: read from Yale's sociologist Philip Gorsky, who points out that
54:21
Speaker 3: our own relativism is rarely as radical as our theory requires.
54:26
Speaker 3: We can't be complete relativists in our daily lives. He
54:30
Speaker 3: then gives the example of how academic social scientists, where
54:33
Speaker 3: diehard relativists, get furious and moralistic at the data fudging
54:38
Speaker 3: of other researchers. Anyway, love the show, guys, love tech
54:42
Speaker 3: stuff especially, and will forever be indebted to you for
54:46
Speaker 3: your hilarity and knowledgeability. Cheers Jesse Lusco.
54:50
Speaker 1: Ps go tech stuff.
54:52
Speaker 2: That's sweet.
54:53
Speaker 1: Love that. Yeah, thanks a lot, Jesse. There was an
54:56
Speaker 1: actual episode, and I don't remember which one it was,
54:59
Speaker 1: where we a and in our cultural relativism, do you remember,
55:02
Speaker 1: because we used to just be like, no judgment, no judgment, right,
55:06
Speaker 1: we just can't judge, you know, And then finally we
55:08
Speaker 1: were like, you know what, No, that's not true. We
55:10
Speaker 1: changed our philosophy to include the idea that there are
55:14
Speaker 1: moral absolutes that are universal, although sometimes we are just
55:18
Speaker 1: judge even beyond that. Look at us, Yeah, Well, if
55:23
Speaker 1: you want to get in touch with us, you can
55:24
Speaker 1: send us an email to Stuff podcast at HowStuffWorks dot com.
55:28
Speaker 1: You can send John an email to.
55:30
Speaker 2: Tech stuff at HowStuffWorks dot com.
55:32
Speaker 1: Nice, and then hang out with us at our home
55:34
Speaker 1: on the web. Stuff youshould know dot com and just go.
55:38
Speaker 2: To tech Stuff. Just search it in Google. I come
55:40
Speaker 2: up all the time.
55:41
Speaker 1: Fair enough.
55:46
Speaker 2: Stuff you Should Know is a production of iHeartRadio. For
55:49
Speaker 2: more podcasts my heart Radio, visit the iHeartRadio app, Apple Podcasts,
55:53
Speaker 2: or wherever you listen to your favorite shows.