Aidan Gomez on How AI Language Models Will Shape The Future

It went through a time where our timelines moved forward enormously. And then, you know, we thought we'd have

models

who could write convincing English and, after a few decades, it suddenly appears a year later. And that leads you to this state of shock and uncertainty. And you're pretty caught off guard and I think he's talked about this publicly. The olfactory feeling of surprise, progress and pace of change. I remember having conversations with him, where we were both like those people who talked about AGI, you know, what nonsense, haha, this was back when

models

could barely spell.

But then you are surprised and surprised and your uncertainty explodes. And sometimes that can have the effect that it's okay, anything is possible. Oh my god, like I'm so far from that. But now I'm increasing my uncertainty about what a super intelligent god could be. Okay, maybe   that's even. So I think a lot of people. We are all taking that into account and recalibrating. And, you know, adjusting our own schedules and understandings about progress and pacing. AIDAN Geoff is extraordinarily thoughtful. And he's been thinking about this, at least since the beginning of Cohere. So, at least for the last three and a half years, he's been thinking about this very, very deeply.

That's why I think people should take it very seriously. I think there

will

be a lot of sensationalism. And a lot of extrapolation of what he says. But if you really listen to what he's saying, it's quite measured, he says, "I'm very sure what can happen." And that means we should take these things seriously. Because we simply don't have certain limits. We have no certainty about the

future

. That's why we should take all the different possibilities very seriously. I'm not saying they're likely to happen. But I'm just saying they can't be ruled out yet. So let's take them very seriously.

I think there is a lot of journalistic writing, headlines, clickbait and nonsense. But if you really listen to Geoff, I think his opinion is quite measured and reasonable. CRAIG 57:47 And in fact, I would love for you to talk at length again about these things, but about the idea of ​​sentience or the illusion of sentience. I mean, you know, more than anyone, having built these models, and both what they are capable of and what is behind their expressions. You think that? I mean, it's a philosophical question about what sentience or consciousness is, whether it's, you know, whether our consciousness is simply an emergent property of the neural activities of our brain and, and, and it's largely an illusion.

I mean, what would you say to all that? AIDAN 58:54 Yes, I would say, I do not position myself as a divinity above humanity. I believe that consciousness is in the brain. And it's like a physical process. And maybe it's like, maybe consciousness is what computing feels like, what processing feels like? And if that's the case, it's really hard to argue that that same phenomenon couldn't be present in silicon. I think you really would have to, I think there has to be a leap towards the right to say that the circuits in our brains, because they're human or because they're biological, have some kind of fundamental distinction.

I think you really have to take a leap of faith in that sense. And if I just say be pragmatic and reductionist. Again, we need two hours to discuss this, I think more fully, but I think it would be really, as a scientist, I think it would be very difficult for me to say: There is no way these machines can become sentient. . I just can't build an argument on this and that. CRAIG 1:00:27 Yes. Yes. Well, let's leave it there. But can I get a promise that you'll come back? You know, in a few months we

will

be able to delve deeper into that topic.

Yes, I would love for AIDAN 1:00:43. Yes. CRAIG 1:00:46 Okay. Yes. Aidan. This has been really fascinating. I am delighted. And I'm sure you heard it at the MIT Tech Review conference. Someone asked Geoff, he was virtually from the UK, but someone asked him if he would get rid of Cohere. And he said, No, no, he'll stay involved. So yes. Yeah, that's a fun question. Yes. Yes. Good, excellent. Well, I really appreciate your time and we'll talk again. CRAIG That's all for this episode. I want to thank Aiden for his time. I also want to remind him to check out NetSuite Oracle business management software for enterprise resource planning, financial management, customer relationship management, and e-commerce, among other things.

Visit netsuite.com/eyeonai to take advantage of this offer. CRAIG And remember, the singularity may not be close. But AI is about to change your world. So pay attention