Latest / Mic Drop Comedy / Josh Tyrangiel - _AI for Good_ _ The Daily Show

Transcript

Josh Tyrangiel - _AI for Good_ _ The Daily Show · Mic Drop Comedy

• 0:00
  Speaker 1: How are you? I'm good, thanks.
• 0:02
  Speaker 2: This is, listen, you have taken on a task.
• 0:05
  Speaker 1: AI for good.
• 0:06
  Speaker 2: AI is reputationally right now in the country, maybe in
• 0:11
  Speaker 2: the world, people are very concerned. It is, I saw
• 0:17
  Speaker 2: it booed at a graduation speech. The valedictorian said something
• 0:20
  Speaker 2: about AI being the new industrial revolution, and the kids
• 0:24
  Speaker 2: literally just wanted to go up on stage and drag
• 0:27
  Speaker 2: her out. Why take up the mantle of it to show,
• 0:32
  Speaker 2: to take down the temperature?
• 0:33
  Speaker 1: Yeah, I mean, I didn't do that intentionally. I was
• 0:36
  Speaker 1: reporting on AI, and a couple months into reporting, this
• 0:39
  Speaker 1: is early 2024, I was like, what the hell is
• 0:43
  Speaker 1: this good for? Because I was talking to people who
• 0:45
  Speaker 1: were telling me two things. One, it's going to cure
• 0:49
  Speaker 1: cancer and mitigate climate change, okay? It's going to end
• 0:52
  Speaker 1: human existence, right? And I was like... Are those the
• 0:58
  Speaker 1: only two options? Is there anything in between? And so
• 1:01
  Speaker 1: I was actually talking to a guy named Danny Hillis,
• 1:04
  Speaker 1: who invented cloud computing. And Danny's now in his 70s.
• 1:08
  Speaker 1: And he's kind of like the Buddha of Silicon Valley.
• 1:10
  Speaker 1: He's just this lovely guy. And I'm complaining to him.
• 1:12
  Speaker 1: I'm like, Danny, this is bullshit. What is this good for?
• 1:16
  Speaker 1: And Danny just laughs. And he's like, you need to
• 1:18
  Speaker 1: imagine the tech without the tech companies. And I was like, uh.
• 1:24
  Speaker 1: I'm a little embarrassed that I didn't realize one could
• 1:26
  Speaker 1: do that. Can one do that? One can.
• 1:29
  Speaker 2: You can separate AI and the technology from Palantir and
• 1:34
  Speaker 2: Altman and Musk.
• 1:35
  Speaker 1: You can. And so one thing that's really important to
• 1:38
  Speaker 1: know is like AI is not one thing. So there
• 1:41
  Speaker 1: are many things that are infuriating about AI.
• 1:43
  Speaker 2: Yes.
• 1:43
  Speaker 1: But one of them is the term AI. Okay. So
• 1:45
  Speaker 1: AI is actually a series of overlapping, but different capabilities
• 1:51
  Speaker 1: and techniques. Right. So, uh, there's AI that can predict
• 1:55
  Speaker 1: behavior from patterns it sees in data. Right. Those are
• 1:59
  Speaker 1: the large language models, or that's something different? There's a
• 2:02
  Speaker 1: part of it that's large language, but not entirely, right?
• 2:04
  Speaker 1: There's classifiers, and what classifiers can do is like, you know,
• 2:09
  Speaker 1: tag an image, sort your email, flag a tumor on
• 2:13
  Speaker 1: an MRI.
• 2:14
  Speaker 2: Okay.
• 2:14
  Speaker 1: Let's use that one. Yeah, exactly. Right. Then there's, then
• 2:17
  Speaker 1: there's optimizers and they work in logistics and they can
• 2:20
  Speaker 1: change the direction of something or change pricing really quickly.
• 2:23
  Speaker 1: And then there's gen AI generative AI, which is really
• 2:26
  Speaker 1: what has sparked this whole thing in the last three years.
• 2:29
  Speaker 1: And that's AI that generates words and images and can
• 2:32
  Speaker 1: create code. And so what they all have in common
• 2:36
  Speaker 1: is speed. They process information faster than any human being
• 2:40
  Speaker 1: possibly could. and with many fewer doubts. And so that's
• 2:44
  Speaker 1: sort of the key.
• 2:46
  Speaker 2: The lack of doubt.
• 2:47
  Speaker 1: Both the speed and the lack of doubt.
• 2:49
  Speaker 2: So what you do in the book, and I think
• 2:50
  Speaker 2: you do it really well, is you show AI as
• 2:55
  Speaker 2: a tool, whether it be in schools. There's a great
• 2:58
  Speaker 2: chapter in there about some teachers in Indiana who are
• 3:01
  Speaker 2: starting to use it with heart MRIs, which are apparently
• 3:05
  Speaker 2: very difficult to get but are diagnostically superior. non-verbal humans
• 3:12
  Speaker 2: that communicate and ways to help crack that code to
• 3:16
  Speaker 2: bring those people in. But it's hard to imagine it
• 3:21
  Speaker 2: will remain our assistant.
• 3:24
  Speaker 1: Yeah. And I think what you're getting at is that
• 3:29
  Speaker 1: we have all been focused on Gen AI and the
• 3:32
  Speaker 1: people who are bringing it to us.
• 3:33
  Speaker 2: Yes.
• 3:34
  Speaker 1: And we're human beings. So we focus on human beings,
• 3:36
  Speaker 1: which I think is very wise because the tech is complicated.
• 3:38
  Speaker 1: And so what is the first thing we do? We
• 3:40
  Speaker 1: assess the people who are bringing it to us.
• 3:42
  Speaker 2: Right, who are in control of it.
• 3:43
  Speaker 1: Who are in control of it. And we don't like them.
• 3:46
  Speaker 2: Thank you.
• 3:47
  Speaker 1: Yeah. I didn't want to say anything, but thank you. No, no, no,
• 3:50
  Speaker 1: it's fine. It's fine. By the way, hard to like.
• 3:53
  Speaker 1: Hard to like, okay. Hard to like. I'm going to
• 3:56
  Speaker 1: take a mildly unpopular stance and say I'm not sure
• 3:59
  Speaker 1: they're necessarily evil, but they're very alike. They're very similar.
• 4:04
  Speaker 2: Yes.
• 4:04
  Speaker 1: And so they're all kind of like racing each other,
• 4:08
  Speaker 1: obsessed with each other. intent on dominating, it doesn't help
• 4:12
  Speaker 1: that they all kind of look like they could be
• 4:14
  Speaker 1: in the Legion of Doom?
• 4:15
  Speaker 2: Yes! Right?
• 4:17
  Speaker 1: Like, you got the guy who wants to colonize Mars,
• 4:20
  Speaker 1: the guy who flies into his secret bunker in Hawaii.
• 4:22
  Speaker 1: None of that's great.
• 4:23
  Speaker 2: You got a guy who literally had to pause when
• 4:26
  Speaker 2: asked if humanity should continue.
• 4:28
  Speaker 1: Right. Should humanity continue?
• 4:31
  Speaker 2: And he was like, ooh, interesting question.
• 4:33
  Speaker 1: Yeah. So, like, not great. No! Okay? But at the
• 4:39
  Speaker 1: same time, there's more than one baby with this bathwater.
• 4:43
  Speaker 1: There's a lot of possibility in here. And we want
• 4:46
  Speaker 1: to be really careful in our response to that, which
• 4:48
  Speaker 1: I think is actually a response to capitalism and a
• 4:51
  Speaker 1: skepticism of capitalism, with what this stuff can actually do.
• 4:55
  Speaker 2: But it's also, we have been hurt before. But let
• 4:57
  Speaker 2: me take you then, we keep saying, well, it is
• 5:00
  Speaker 2: really these five guys in these five companies.
• 5:02
  Speaker 1: It's Google, OpenAI.
• 5:05
  Speaker 2: AIX and Anthropic and Meta. Why did they get it?
• 5:11
  Speaker 2: Why did all the profits of it? AI is by
• 5:15
  Speaker 2: necessity a strip mining of us. It is in fact
• 5:19
  Speaker 2: extractive in the way that oil companies, oil might be
• 5:22
  Speaker 2: a resource and you extract it from the ground and
• 5:25
  Speaker 2: they get it and okay, that's theirs. But in Alaska,
• 5:27
  Speaker 2: they get a dividend. Why does Palantir and OpenAI, and
• 5:33
  Speaker 2: why do they get proprietary knowledge that is our knowledge
• 5:38
  Speaker 2: that they stole from us that they used to make their,
• 5:41
  Speaker 2: but they're the ones who get all the money?
• 5:43
  Speaker 1: That makes no sense to me. Yeah, listen, I agree.
• 5:46
  Speaker 1: And I actually think that politically speaking, there's movement on
• 5:50
  Speaker 1: the far left and on the far right for the
• 5:53
  Speaker 1: federal government to essentially nationalize these products and basically say, look,
• 5:58
  Speaker 1: You're gonna have tremendous amounts of control. We're gonna have
• 6:00
  Speaker 1: all sorts of labor issues. I did a story for
• 6:04
  Speaker 1: The Atlantic about AI and the future of employment, right?
• 6:08
  Speaker 1: Nobody in the middle knows what I'm talking about. They're
• 6:12
  Speaker 1: not interested in AI. They're like, well, it's kinda on
• 6:15
  Speaker 1: its way, right? The two people who said, who agreed, right?
• 6:18
  Speaker 1: On the left, Bernie Sanders said, we need to tax it,
• 6:22
  Speaker 1: we need a robot tax, we need a shorter work week,
• 6:25
  Speaker 1: and we're gonna need to nationalize it, right? And so
• 6:27
  Speaker 1: then I went to Steve Bannon's townhouse. I sat with
• 6:30
  Speaker 1: Steve Bannon.
• 6:31
  Speaker 2: What'd that smell like?
• 6:38
  Speaker 1: You know, we didn't get that close. This was not
• 6:44
  Speaker 1: my first reporting trip. So I wasn't like, do we
• 6:47
  Speaker 1: start by smelling each other?
• 6:50
  Speaker 2: I just thought it would be evidence.
• 6:53
  Speaker 1: So we're talking and he says, I agree with Bernie Sanders.
• 6:59
  Speaker 1: And he said, but he doesn't go far enough. I
• 7:02
  Speaker 1: think we should get 50%, and I think we need
• 7:04
  Speaker 1: to have government control over the boards of these companies.
• 7:08
  Speaker 1: And he said, I know that's not going to be
• 7:10
  Speaker 1: popular with the right, but that's what I believe.
• 7:12
  Speaker 2: And he doesn't believe it either, is my guess. And
• 7:14
  Speaker 2: he certainly doesn't believe it if a Democrat is in office.
• 7:17
  Speaker 2: I think it's very convenient for them to do the
• 7:20
  Speaker 2: faux populism when it's really about.
• 7:24
  Speaker 2: Control. But I wouldn't even think it's that radical. It's
• 7:27
  Speaker 2: the sense of this is a resource. Look, the guys
• 7:31
  Speaker 2: from AI are the ones who agree with you that
• 7:34
  Speaker 2: this is going to be incredibly disruptive for labor. This
• 7:38
  Speaker 2: isn't something that populists have made up. Dario from Anthropic
• 7:43
  Speaker 2: has said, yeah, this is going to be a bear. Yes.
• 7:47
  Speaker 1: And what I found in my reporting on that story
• 7:49
  Speaker 1: is that the people who really didn't want to talk
• 7:52
  Speaker 1: about it were the Fortune 100 CEOs. Oh, that's interesting.
• 7:56
  Speaker 2: Now, why do you say that?
• 7:57
  Speaker 1: What do you think that's about? I think it's two responses.
• 7:59
  Speaker 1: The first is they're actually really scared. Really scared.
• 8:02
  Speaker 2: That it's going to come for them?
• 8:04
  Speaker 1: They're scared that they look at their investments in AI
• 8:07
  Speaker 1: and they know what Wall Street wants. Wall Street wants growth.
• 8:10
  Speaker 1: And if they can't show growth through using AI in
• 8:13
  Speaker 1: their products, they can grow by shrinking their workforce. And
• 8:17
  Speaker 1: so they are desperately afraid... that that's something that they're
• 8:21
  Speaker 1: gonna do and that it's gonna be really, really unpopular.
• 8:24
  Speaker 1: And what I would say is like I was impressed,
• 8:26
  Speaker 1: they were aware that they're not acting alone and that
• 8:29
  Speaker 1: if all of them do this at the same time,
• 8:30
  Speaker 1: the consequences for the country are gonna be enormous. However,
• 8:35
  Speaker 1: if they don't do it, they're gonna be the ones
• 8:37
  Speaker 1: out of a job because the street expects certain things
• 8:40
  Speaker 1: from these public companies.
• 8:41
  Speaker 2: So they're incentivized to the breakdown of our social fabric.
• 8:46
  Speaker 1: Correct.
• 8:46
  Speaker 2: Through a great amount of unemployment. The thing that I
• 8:52
  Speaker 2: respond to is there's a sense of inevitability around all this,
• 8:56
  Speaker 2: as though we've created a tool that's unbelievably powerful, but
• 9:01
  Speaker 2: we have no control over how it's deployed or how
• 9:04
  Speaker 2: it's used. Imagine if atom bombs were controlled by five
• 9:08
  Speaker 2: capitalists in Silicon Valley, and we were all like, they're
• 9:11
  Speaker 2: probably gonna start a war at some point, but what
• 9:13
  Speaker 2: are you gonna do, capitalists?
• 9:15
  Speaker 1: Like. And so that's why I actually think it's really
• 9:18
  Speaker 1: important to show people what the positive uses of AI are,
• 9:22
  Speaker 1: because we're going to have this moment, and it's not
• 9:25
  Speaker 1: going to last very long, when we can demand and
• 9:28
  Speaker 1: insist on certain kinds of AI in our lives. If
• 9:32
  Speaker 1: we try to take on all of capitalism, I got
• 9:35
  Speaker 1: bad news for you. It's had a 400-year head start,
• 9:38
  Speaker 1: and that boulder is rolling down the hill. And sometimes
• 9:40
  Speaker 1: we've directed it, and sometimes we haven't. But I wouldn't
• 9:43
  Speaker 1: take that on. I think right now, we have this
• 9:46
  Speaker 1: brief period where if we can understand what we want
• 9:48
  Speaker 1: from it, how it works, how to use it in
• 9:51
  Speaker 1: ways that it works with human beings and doesn't replace
• 9:54
  Speaker 1: human beings. How it amplifies meaningful problems, doesn't create slop.
• 9:59
  Speaker 1: We got a shot. And I would rather focus on
• 10:01
  Speaker 1: the ways in which we have a shot than, as
• 10:03
  Speaker 1: you say, like hunker down, curl up in a ball
• 10:05
  Speaker 1: and just be like.
• 10:05
  Speaker 2: Ah, it's over. But they're doing their best to prevent
• 10:09
  Speaker 2: that process from happening. And as trillions get funneled into
• 10:13
  Speaker 2: their coffers, Millions get funneled back into politicians' coffers to
• 10:17
  Speaker 2: prevent that very thing from happening. They've created a machine
• 10:21
  Speaker 2: that prevents us and is opaque. And again, I go
• 10:24
  Speaker 2: back to, but what is the fundamental driver of their product?
• 10:29
  Speaker 1: Us. For sure.
• 10:31
  Speaker 2: And so I don't understand how we have no shareholder voice. Right.
• 10:37
  Speaker 1: And I think, to me, that's why this moment matters.
• 10:40
  Speaker 2: Right.
• 10:40
  Speaker 1: We actually... and I hate to say this, we're gonna
• 10:44
  Speaker 1: have to use the political system to make change. I know.
• 10:48
  Speaker 2: Wash your mouth out with soap, young man.
• 10:51
  Speaker 1: One source of optimism. So up until like a week ago,
• 10:55
  Speaker 1: I would have told you, well, that's just terrible, right?
• 10:58
  Speaker 1: So everybody's, the booing at graduations, the fact that we're
• 11:02
  Speaker 1: talking about this in such negative terms. So for the
• 11:05
  Speaker 1: first 14 months of the Trump two, David, a guy
• 11:09
  Speaker 1: named David Sachs was in charge of AI. He also
• 11:11
  Speaker 1: was the crypto czar. He also was an investor in
• 11:14
  Speaker 1: all the AI companies. So he's basically like the, you know,
• 11:17
  Speaker 1: the Cy Sperling of AI, right? He's the president, he's
• 11:19
  Speaker 1: the investor. He stepped down about a month ago. And
• 11:23
  Speaker 1: last week, you know, we reported, other people reported, the
• 11:26
  Speaker 1: Trump administration is thinking of suddenly regulating AI models above
• 11:30
  Speaker 1: a certain power. And I think they're doing that because
• 11:33
  Speaker 1: they know if there is an AI related disaster, You know,
• 11:37
  Speaker 1: there's photos of all these guys cozying up to Donald Trump.
• 11:40
  Speaker 2: At the inauguration. They finally realized that.
• 11:42
  Speaker 1: At the inauguration, at dinners. And they're like, well, that's
• 11:45
  Speaker 1: not great for us. Yes. And so the political movement
• 11:48
  Speaker 1: is having an impact. And at the same time as
• 11:51
  Speaker 1: that's working, people are actually making their point of view heard.
• 11:55
  Speaker 1: We do need to demand that the people we elect
• 11:58
  Speaker 1: to Congress have used AI. Nope. the basic functionality of technology.
• 12:03
  Speaker 2: I'm sure they have.
• 12:04
  Speaker 1: Oh, they all have. I mean, they're all training their
• 12:08
  Speaker 1: own models.
• 12:09
  Speaker 2: What color tie makes voters most comfortable?
• 12:15
  Speaker 1: So we do have that available to us. And I actually,
• 12:17
  Speaker 1: for the first time, I'm like, oh, that lever is
• 12:20
  Speaker 1: working a little bit.
• 12:21
  Speaker 2: The problem is, again, I don't trust them to regulate
• 12:24
  Speaker 2: it either because of the corrupt nature of their dealings.
• 12:26
  Speaker 2: I mean, when your two kids suddenly have a multi-million
• 12:31
  Speaker 2: dollar drone contract with the Pentagon, and they ran golf courses,
• 12:37
  Speaker 2: and you're like, I don't think that makes any sense.
• 12:40
  Speaker 2: The corruption endemic to this administration doesn't give you, we
• 12:44
  Speaker 2: need to establish some form of a way to adjudicate
• 12:50
  Speaker 2: it that is apolitical. And more akin to a commission
• 12:56
  Speaker 2: of trusted advisors.
• 12:58
  Speaker 1: Yeah, and I think what's happening with the five guys,
• 13:01
  Speaker 1: as we call them in charge, is that I don't
• 13:03
  Speaker 1: think anybody wants to own the disaster. And I think
• 13:06
  Speaker 1: the more unpopular this gets, there is more of an
• 13:10
  Speaker 1: incentive for everybody to sit down and say, how do
• 13:12
  Speaker 1: we want to handle this? And by the way, I
• 13:14
  Speaker 1: would include the Chinese in this. From the beginning, we've
• 13:16
  Speaker 1: been told, no, no, no, we have to beat the Chinese.
• 13:19
  Speaker 1: That's why we can't regulate it. Oh, that's so interesting.
• 13:22
  Speaker 2: You think there's a self-preservation that has suddenly occurred to
• 13:26
  Speaker 2: them that we're at the forefront? Because I was of
• 13:29
  Speaker 2: the impression that they felt immune to the vagaries of
• 13:34
  Speaker 2: accountability in our government and also in those countries. Silicon
• 13:38
  Speaker 2: Valley boardrooms.
• 13:39
  Speaker 1: I would not speak for them. I would only say
• 13:42
  Speaker 1: the pressure has increased a lot. The consequences have suddenly
• 13:45
  Speaker 1: become much more clear of what failure looks like. So
• 13:49
  Speaker 1: I think there's a window here. Do I trust the
• 13:51
  Speaker 1: Trump administration to push them through the window? No.
• 14:05
  Speaker 2: Somehow, the metaphor has me confused.
• 14:10
  Speaker 1: The window is good.
• 14:11
  Speaker 2: Push him through the window. But then I can't help
• 14:13
  Speaker 2: but think of, like, isn't that how Putin gets rid
• 14:14
  Speaker 2: of everybody? Let me ask you, though, because this is
• 14:18
  Speaker 2: a purely philosophical question, because I understand the challenges that
• 14:21
  Speaker 2: are ahead for us. But I think as human beings
• 14:25
  Speaker 2: on this Earth, the idea that we've designed something that's
• 14:31
  Speaker 2: better than us throws into disarray our entire understanding of
• 14:37
  Speaker 2: almost even faith. Because what have we said about God?
• 14:42
  Speaker 2: What do we say about the human eye? Well, the
• 14:44
  Speaker 2: human eye couldn't have been created unless it was created
• 14:48
  Speaker 2: by something greater than us. But we've just created something
• 14:53
  Speaker 2: greater than us and we're less. So does that mean,
• 14:58
  Speaker 2: does that change almost, do they reckon with that?
• 15:01
  Speaker 1: I don't think they do, but... Want to get high
• 15:04
  Speaker 1: and reckon with that?
• 15:05
  Speaker 2: Do you know what I mean? But how is it
• 15:11
  Speaker 2: that a lesser being creates a better being?
• 15:15
  Speaker 1: But I guess I look at it differently, okay? So
• 15:16
  Speaker 1: remember a couple weeks ago, the robot that won the
• 15:20
  Speaker 1: half marathon in China? And everybody's saying, oh, the robot
• 15:23
  Speaker 1: beat the humans. I was like, that robot probably had 2,000
• 15:28
  Speaker 1: programmers and manufacturers who worked on it. So 2,000 people
• 15:31
  Speaker 1: and a robot won that half marathon, in my mind.
• 15:36
  Speaker 1: You just blew it.
• 15:37
  Speaker 2: Oh, hold on.
• 15:40
  Speaker 1: It's a great point.
• 15:41
  Speaker 2: I keep thinking of it as every iteration of it
• 15:46
  Speaker 2: blows my mind more than the next. And it's moving
• 15:49
  Speaker 2: so quickly. I'll tell you a quick story before we go.
• 15:53
  Speaker 2: I was at one of those, like, Illuminati get togethers
• 15:59
  Speaker 2: in the mountains of Colorado. Like one of those things
• 16:01
  Speaker 2: that you're like, that doesn't happen. But it happened. It's
• 16:06
  Speaker 2: always like a conference. And they have like, it's like
• 16:09
  Speaker 2: Bill Gates and Musk. And they're all in the room.
• 16:12
  Speaker 2: And there's a good amount of security. And the food
• 16:15
  Speaker 2: is fantastic. And so I thought, well, I'll get a chance.
• 16:21
  Speaker 2: I go up to SAML. And these guys are always like,
• 16:26
  Speaker 2: you couldn't find a bag? You're like, I was dressed nicer.
• 16:30
  Speaker 2: And that never happens. You're literally like, really? Corduroy?
• 16:35
  Speaker 1: Come on, dude.
• 16:38
  Speaker 2: And I said, the Industrial Revolution displaced. The globalization displaced.
• 16:42
  Speaker 2: But all those took place over decades. And America still
• 16:45
  Speaker 2: and the world still didn't adjust to it in the
• 16:47
  Speaker 2: same way. AI may have that same level of displacement
• 16:53
  Speaker 2: over three years. And we can't in any way be
• 16:58
  Speaker 2: prepared for that kind of disruption in that kind of timeframe.
• 17:01
  Speaker 1: And I laid.
• 17:02
  Speaker 2: It out so beautifully. Josh, I was sure that I
• 17:06
  Speaker 2: was the one that he was gonna be.
• 17:08
  Speaker 1: Like, let's get some coffee and talk. And what did
• 17:11
  Speaker 1: he say?
• 17:12
  Speaker 2: He goes, we'll be good.
• 17:16
  Speaker 1: And then he walked away.
• 17:19
  Speaker 2: And I was like, huh.
• 17:21
  Speaker 1: Okay. That sucked. Yeah. Yeah. Sorry. But thank you for
• 17:33
  Speaker 1: wrestling with these questions.
• 17:35
  Speaker 2: Man, this is going to be the book. Start to
• 17:38
  Speaker 2: read how we can use this as a tool because
• 17:40
  Speaker 2: you're going to need it.
• 17:41
  Speaker 1: AI for Good. Available now. Josh Terengough.