The Ledger, every show

“The prediction that I made Is basically that there will be a death tied to a large language model in the next year.”

“So OpenAI tried to build GPT-V and they had a thing called Project Orion and it actually failed. And eventually got released as GPT four and a half. So what they thought was going to be GPT five just didn't meet expectations.”

“Models like O-I are not systematically better than GPT-IV. There's, they're better in certain use cases. Especially ones where you can create data in advance.”

“Now, the reason I wouldn't call them reasoning models, though you're right that many people do, is what I think they're doing is basically copying patterns of human reasoning. They're getting data about how humans reason certain things, but the depth of reason…”

“So the scaling quote laws were empirical guesses about how these models work and they were true for a little while, which was amazing. And they're not true anymore, which is also amazing in a way.”

“I don't think LLMs will disappear. I think they're useful, but this, yeah, the valuations don't make sense. I mean, I don't see open AI being worth three hundred billion dollars. And you have to remember that venture capitalists have to like 10 X to be happy o…”

“Nobody has a technical mode. OpenAI has a user mode.”

“The business model of gen AI will be surveillance and hyper-targeted ads, just like it has been for social media.”

“And then the other really disturbing thing is that apparently they tested in India and got, you know, customer service requests saying it's not ready for prime time. And it was, you know, still put out.”

“I think that the Turing test was an error in, in AI history. I mean, Turing's obviously a brilliant man, and he, you know, made enormous contributions to computer science, but I think that the Turing test has been an exercise in fooling people. We've now solve…”

“When I'm going through this in some depth, because when we get to Lambda, Lambda doesn't have a model of the world.”

“My turn-by-turn system has more elements of what I would actually ascribe sentience than Lambda, which is really just autocomplete on steroids. That's all it is, right? You type in your phone, I will meet you at, and it guesses that, you know, you might say th…”

“Nobody knows, you know, how to make their chatbots constrained and not toxic and not spew misinformation.”

“Despite all the hype about, you know, we're so close to solving AI or whatever. We're not, we're facing the same problems.”

“I don't think it's true at all, so there are some things that people can do in a second that AI can automate, and there are other things that it can't”

“So you should not be trusting a Tesla. Elon might get on 60 minutes and take his hands off the wheel, but you should not be doing the same because the technology is not really there yet.”

“Most of why I don't think we should trust AI right now is that the current techniques that people are using are simply too brittle to earn the trust that we're giving them.”

“These systems are just giant correlation machines. They have no understanding of causation.”

“Perception is what deep learning does, and actually it only does a small part of perception. There's lots of parts of perception where we use our knowledge about the world that it doesn't capture. And then there are all these other things like common sense and…”

“Deep learning is a better ladder for sure, but a better ladder doesn't necessarily get you to the moon.”

“But the technology for Siri is not really that different from the technology for ELISA, and Siri is still very far from artificial general intelligence.”

“Deep learning doesn't seem to be getting us there. The old systems, in some ways, we're better at that. We do need a marriage between the two.”

“So he built Grok three and by his own testimony, it was 10 times the size of Grok two. It's a little better, but it's not night and day, right? Grok two was night and day better than the original Grok. GPT four was night and day better than GPT three. GPT thre…”

“That's their sweet spot is regurgitation. And so Yeah, they can build the stuff that's out there, but if you want to code things in the real world, you usually want to code something that's new, and these systems have a lot of problems with that. And another r…”

“If you look at a completely pure case of a transformer model trained on a bunch of data, It doesn't have any mechanisms for truth. Now, except the sort of accidental contingency, and there are inherent reasons why these systems hallucinate, and maybe I can, in…”

“The way I think about it is that these things don't understand the difference between individuals and kinds. So I actually wrote about this twenty-some years ago in my book, The Algebraic Mind, and I gave an example there, which is I said, suppose it was a dif…”

“Part of the reason why ChatGPT succeeded where Galactica didn't is Galactica didn't really have those guardrails at all, and so it was just, you know, very easy to get it to say terrible things, and it's harder to get ChatGPT to say terrible things because tha…”

“We don't even know what the right controls should be. They didn't do a good job and we don't yet have a good science or engineering practice on what it should be.”

“They're too compelling in that illusion for the average person with no training in how they work to understand it.”

“I, I've been thinking about this whole thing as a dress rehearsal. And before we didn't know how to make a dress rehearsal. This is a dress rehearsal for AGI. And you know, the lesson of the dress rehearsal is like, we are not ready for prime time. Let us not …”

“And some point I'm going to say, 30 years from now. We'll be able to make machines that do podcast interviews and I won't really know. I don't really know if you're a person fake or, you know, machine faking me out or whatever at some point.”

“None of these chatbot systems are well-debugged. Nobody knows how to debug them, in fact. And so both the problem with GPT-III and with driverless cars is we don't actually have a methodology even for debugging it.”

“It turns out that Teslas have a problem with stopped vehicles, mostly emergency vehicles, on the side of the road. So they have run into, in the last year, three fire trucks, a tow truck a police car, and so forth.”

“If a typical person can do a mental task with less than one second of thought, and we can gather an enormous amount of data, of directly relevant data, we have a fighting chance. So long as the test data aren't too terribly different from the training data, an…”

“Deep learning doesn't even really fully understand the relations between parts and wholes. It certainly doesn't understand what a silhouette is, and it gets it wrong.”

“Deep learning can't do what we call compositionality. It can't put ideas together.”

“I think that deep learning doesn't just fail on reading. It completely, entirely misses the point.”

“There are a lot of different components to cognition, and so we have to have a hybrid model. There's just no way that one system is going to do all of these different things”

“I suspect that what will actually happen is that people will start to incorporate hybrid models and not call them that.”

“AI is still basically a collection of idiot savants.”

“And I have a three-part diagnosis, which is that it's fallen in love with statistics, it's fallen in love with big data, and it's forgotten its roots.”

“But it's another thing if you import those same techniques into robots. If you have your robots doing reinforcement learning, which is basically trial and error learning, they start knocking over the furniture a 100,000 times, you're probably gonna send it bac…”

“I don't think there's going to be like one special day with this big explosion. They're going to get better and better. It's going to take a while.”

“I'll give you just one more example is O three apparently hallucinates more than the models that came before it.”

“Well, Microsoft did a study, in fact, suggesting that critical thinking was getting worse as a function of them.”

“And also alpha fold is actually a neurosymbolic model, and it's probably the best thing that AI ever did.”

“I think that's actually the deepest problem here is we have no way to debug these systems, really. We have the band-aids of, like, reinforcement learning and things like that that are so indirect.”

“You have companies like Adept that are spending their whole time trying to quote, add all the world, connect all of the world's software to large language models. So if not now, soon enough. I think that that's right. Like, you know, two years if it's not toda…”