Men and Machines

Speaker

[music]

Andrew Keen

Hello, everybody. It's Saturday, the 10th of August, 2026. Saturdays are our day with Keith Teare, if That Was The Week. To summarize tech news of the week, it hasn't been a huge week. I was really struck with an interesting piece in the New York Times, an op-ed by Stephen Witt on Jensen Huang, the CEO of NVIDIA. The title of the piece was, he's the most powerful engineer on earth. Here's why he thinks AI is going great. Wit was on the show, my show last year, his book, The Thinking Machine, which is, I think, the authoritative book on Huang and NVIDIA, won the FT Business Book of the Year last week. Wit is, in some ways, ambivalent and critical of Huang because he thinks that... He doesn't quite understand everything about AI, which is a debatable issue. But what really struck me about the WIT piece was how engineers have become, for better or worse, all powerful in the world today. There was another important influential book brought out last year, By Dan Wang, a Canadian-Chinese journalist called Breakneck, China's Quest to Engineer the Future, in which he broke America and China down as America is a society ruled by lawyers and China is a society ruled by engineers. And I wonder whether Witt's prominence, his wealth, his power, not Witt, Huang, Witt notes that Trump has been very influenced by by Huang, whether that reflects the increasing power of engineers in our society. Keith, you often talk about, I don't know if you call it the engineering ethic, but the ways in which engineers look at the world. Huang is clearly a remarkably brilliant engineer. I mean, even Witt, who's somewhat critical of him, acknowledges that he's brilliant. Is there, in your mind, is there an engineering ethic, a way of looking at the world, which perhaps Jensen Huang epitomizes?

Keith Teare

I think there is a way engineers look at the world, although I wouldn't call it an ethic because that implies value judgments, which I think engineers are very widely spread across the spectrum of value judgments. So I don't think it's an ethic, but I think it's... Accepting the responsibility to make something work is an engineering ethos, if you will, probably more than an ethic. That's their job. And that goes way back. I don't even think that's new. Whether it was Stevenson's steam engine, the Golden Gate Bridge, or AI, I think engineers in general are trained and have the task of making things work. Usually for good, but that's an open question. There are some engineers who are quite happy to make things work for bad ends as well.

Andrew Keen

Your editorial this week focuses on, you don't call it the ethic of engineering, the principle of engineering. You focus on a really interesting, I thought, interview on the Ezra Klein Show between Klein and Arvind Naravayan, who's a Princeton computer scientist and expert on AI. Does Naravion, in your view, represent, if not the ethic of engineering, the logic of the engineer when it comes to making sense of AI?

Keith Teare

Yeah, absolutely. I thought it was a great interview. And I think Ezra Klein, actually, if you look at his interviews over the last three weeks, is double clicking on the whole discussion around AI and potential problems with AI.

Keith Teare

And he isn't happy just to talk to people that reinforce his own instincts. He's talking across the spectrum. And this was the most recent one. Of course, he talked to Jensen Huang prior to that.

Andrew Keen

And we talked about that. And that was another, I thought, really interesting interview in which Huang was very clear, very articulate about his sense that it was the frontier AI companies who had the responsibility. I'm not sure if it was a moral or a business, a combination of moral and business responsibility,

Keith Teare

not to bring their products to market if they were dangerous. Yeah, I think the best thing about this interview is that Arvind is definitive. I mean, I thought Jensen Huang was as well, but this is, if anything, even more definitive. In drawing a line between computer code and electronics hardware and organic consciousness, and just saying, you know, you can't really cross that line. There's no conversation to be had. This is something built by engineers, can be controlled by engineers, and it is engineers who have the job to do it. Now that doesn't mean there aren't political or social dimensions to it, that engineers have no expertise in discussing. Of course there are, but at root, he makes very clear that he believes at least that this is still a very early stage Technology that is largely an engineering set of tasks.

Andrew Keen

Yeah, and I don't disagree with you, although I think you slightly over, you exaggerate the disagreement between Klein and Naravi, and I think I mean, if anything, they mostly agreed. I mean, Naravayanan came into Klein's camp when he made it very clear that he thought AI needed to be regulated. And you may disagree, but I don't think Klein believes that AI can have consciousness. I think he's very clear and he said it not just in that interview, but in many other interviews that he doesn't believe it can have consciousness. So they're basically on the same page. I mean, do you think that's wrong?

Keith Teare

It's pretty hard to tell because Klein expresses his disagreement through facial gestures and, you know, mmm kind of sounds.

Andrew Keen

I only listen, so I can't see his facial gestures. But from the audio, he sounds... I mean, as a journalist, he's asking questions and he's trying to be relatively objective. There is a camp of people who believe that AI could have consciousness, so he's willing to at least talk to them. He's had a lot of those on the show too.

Keith Teare

Yeah, so let's give him the benefit of the doubt and say that you're right. Because I definitely don't have a strong case that you're wrong. And I admire him for the breadth and pace at which he's doing these interviews and shining a light on areas that are murky at the best. So it's all to the good, really. And I thought this was the most recent Example of a good interview. I do think that,

Keith Teare

let's get rid of the idea of consciousness, but there's certainly the belief, and this is due to the fact that AI is probabilistic, not deterministic. That is to say, Satya Nadella says this morning, you can't go and find the code that made a bad thing happen, because bad things are not coded. When an AI doesn't follow your directions or follows them badly, it's doing it because it isn't yet good enough to not do that.

Keith Teare

Its ability to understand context is still somewhat restricted. And it's working off of skills, which are text files that describe its goals, which are not fully comprehensive. And so there's no doubt that AI is currently not controllable, as in it won't precisely do what you want it to do unless you work hard to...

Andrew Keen

Sounds like children, right?

Keith Teare

Well, it is a little bit like children in that their frontal cortex hasn't developed so they can do random things. It is a little bit like that. And now that's why it's an engineering problem, because it's a function of the immaturity of AI, not a function of its core nature.

Andrew Keen

And let me come back then to this idea. You rejected the idea there's such a thing as an engineering ethic. I would actually disagree with that in the sense that we live in a society where, for better or worse, engineers are enormously powerful, influential, and wealthy. I mean, Huang is exhibit A, B, and C in that. And he has the ear, according to wit, and it seems from the recent conference it's true, he has the ear of the current American president. For better or worse, it is engineers making moral decisions about their products. Whether or not you agree with those moral decisions is another matter. But there seems, as you've noted many times, maybe there's an intramoral issue. Debate amongst engineers between people like Amadai and Huang about what AI can and can't do.

Andrew Keen

Isn't that fair that they're the ones who are arguing about this? I mean, some philosophers are getting involved, but it's basically a debate amongst technologists about what AI can and can't do.

Keith Teare

You know, I actually don't think that's exclusively true, but I want to take on board what you said, because I think what you said has truth in it, especially the fact that if you do want to use the word ethic, you just described two very different engineers who have very different views. So as long as ethic is not intended to be a singular ethic, but is a plural, yes, engineers do have And I would say younger engineers have more than older engineers.

Keith Teare

They're part of a generation that's grown up extremely cosseted from harsh realities of the world. And some people use the word woke to describe the impact of that, touchy feely. And a lot of engineers, especially hired by an anthropic, this week made the point that, in their terms and conditions, that humans will be banned for being cruel to Claude.

Andrew Keen

And they're seeing... Anthropic bans abusive or cruel behavior towards Claude. You know, you're obviously not a fan of that. And I'm not... So I can't claim to really understand a lot of the debates about whether or not AI has consciousness. But the one argument to be made in support of that is I take your point that maybe AI isn't conscious, it's not a human being. But a lot of people think it is, for better or worse. And we live in a culture where people are abusive or cruel, especially in social media, towards one another, towards our politicians. So it's not necessarily a bad thing. I mean, all they're doing is requiring politeness in their space. It's rather like a restaurant that requires you to wear shoes or a shirt.

Keith Teare

But Andrew, imagine, just let's replace Claude with Facebook. What if Facebook said it's going to ban people that are abusive or cruel to Facebook?

Andrew Keen

Well, some people might think that was a good thing.

Keith Teare

Banning people that are abusive or cruel to Microsoft Word. This is just software. This is just code.

Andrew Keen

Yeah, but I understand that and you make that point every week and I don't disagree, but it's the code that determines how we live. I mean, this week you also talk about how AI agents now have established a protocol. Established by Pret Taylor. So the muses of the world now becoming more and more real. We talked about that last week. So this is our, for better or worse, it's our common language. So the fact that, I mean, I agree with you in a way on Anthropic, that it's slightly ludicrous. But on the other hand, it's encouraging people to behave themselves, to be polite. And in a sense, if AI is us, if it is a mirror, Then we're just being polite to ourselves.

Keith Teare

Yeah, well, look, you and I were both raised in the UK. And one of the things about the UK, I don't know about your mum, but my mum used to hit me if I was rude.

Andrew Keen

You're from the working class. It's much more violent. I'm from the upper classes.

Keith Teare

Well, I'm still sure you weren't allowed to be rude and get away with it.

Andrew Keen

Well, we used to beat our parents up. In the working class, the parents beat the kids up.

Keith Teare

But of course, we all want to be civil. We all want to be polite. And it's not unreasonable to ask for politeness. But I don't think people even require politeness towards animals. Like if I take my dog in the park, if the dog is even a little bit unruly, other people will get aggressive and angry with it. Angry with the dog or with you? Both. And they'll blame the dog, not me, most of the time. And you have a really bad dog. What I'm saying is, we humans are perfectly capable of being nice and polite to each other, whilst also not extending human privileges to inanimate objects. And, you know, it is Bizarre to me that this discussion is even happening. I think that's

Andrew Keen

very much of, and again, kind of using this term, it's an engineering ethic in the sense that you believe that machines can't think, therefore any idea that humans should or shouldn't be allowed to insult machines is stupid, which I don't necessarily disagree with, but it seems a little... Categorical.

Keith Teare

I will say, Andrew, I don't do it. I actually say thank you and please.

Andrew Keen

Yeah, so do I. Well, that's our English, our common English background.

Keith Teare

Yeah, so I don't do it. And I haven't really got a dog in the race or a horse in the race. What about the dog? Exactly.

Andrew Keen

We keep on mentioning the dog. It's going to start barking in a minute and ruin our conversation.

Keith Teare

But going back to your core thing about ethics, these are... Completely anti-human ethics. To punish a human for how it works with software is at core, not pro-software, it's anti-human.

Andrew Keen

I don't agree. I think it's a soft humanism. Maybe it's a misplaced humanism. Maybe it's a slightly infantile humanism, but it's not anti-humanism. It's just requiring Well,

Keith Teare

it's not distinguishing.

Andrew Keen

Requiring politeness. And also, it's like a restaurant. People say, well, I go to a restaurant. I have every right to wear or not wear whatever I want. Anthropic has every right. It owns its product. It owns its space. It has every right to throw people out if they're rude to its product.

Keith Teare

You've seen these movies, I'm sure everyone will relate to this, where somebody throws their cell phone out of the car. Nobody says, hey, you're being cruel to your cell phone. The cell phone didn't work properly. Throw it out of the car.

Andrew Keen

I haven't seen a film.

Keith Teare

Or smash it on a desk. I've seen lots of movies that have scenes like that. I was actually once driving with my CFO at Real Times, Jim Strawbridge. He was a lawyer from Wilson Sonsini by trade. Great guy. Very clever. Thoughtful. Not very technical. His cell phone didn't work. And he literally was banging it and screaming at it. It's quite normal for humans to get angry with software, especially when it doesn't do what they want it to do. And that's going to happen. And I don't think you can, you can't characterize that the same as being mean to each other as humans. It's different.

Andrew Keen

But I think the thing, and Mustafa Suleiman at Microsoft has been very explicit on this. He's made it clear. I think it's a really interesting and important debate that the Companies like Anthropik and OpenAI have made a mistake in that they've humanized the bots. They've made them look or made them seem human, which is not true. And therefore, it's encouraged us to think of these bots, these AIs as human. We're seeing it in a way with the new AI bots, although I think Facebook has made an effort to Make them not so much human as cartoon-like. But the reality is that for kids in particular, when they go on to these products, when you talk to one of these butts, they do seem human. Maybe they're not, and we won't understand that.