Jack Morris on Finding the Next Big AI Breakthrough

Joe Weisenthal (0:02)

Bloomberg Audio Studios, podcasts, radio, news. Hello, and welcome to another episode of the Odd Lots Podcast. I'm Joe Weisenthal.

Tracy Alloway (0:22)

And I'm Tracy Alloway.

Joe Weisenthal (0:23)

Tracy, have you played around with GPT-5 much?

Tracy Alloway (0:26)

Not really, I've been perplexity-pilled.

Joe Weisenthal (0:29)

Oh, that's your main?

Tracy Alloway (0:31)

Yeah, that's my main one at the moment. But is it good? I hear mixed things.

Joe Weisenthal (0:35)

So I use it because I use trans GPT every day. It does not strike me as like, obviously better for my uses than like the O3 models, which I've been very impressed by, because I want to establish I'm no hater or anything like that. But it does not strike me as like, oh, this is like an amazing step function or whatever.

Tracy Alloway (0:54)

It feels like the sort of breakthroughs, awe-inspiring breakthroughs are kind of behind us, and a lot of the progress on the models feels very incremental at this point, even though people are spending a lot of time and resources on doing it.

Joe Weisenthal (1:07)

The one thing GPT-5 does prompt me and say, oh, that's a great question, would you like to follow up more on that?

Tracy Alloway (1:12)

Does it say, oh, Joe, you're so smart? That's such a smart question.

Joe Weisenthal (1:16)

You know what it did say? I asked a follow up and it started an answer with, love it. Then it said, love it. Do you want me to look into that?

Tracy Alloway (1:23)

Yes, they are very flattering, aren't they? Actually, that's one thing I like about perplexity, is it doesn't really flatter you, it just spits out an answer.

Joe Weisenthal (1:31)

Anyway, there's so many questions I have about AI, and we talk about the business of Fairmount, and of course, NVIDIA and all that stuff. We actually don't really talk that much about the pure research side as much, but it's pretty important, I think, because I think a lot of people would agree that if the skills are like slowing down, or if there were a wall or something like that, that might change some of these business model calculations, et cetera. So I think it's good. We need to get an update on just sort of the state of the art, the science of AI.

Tracy Alloway (1:57)

Yeah, also, it would be nice just to understand what's possible in terms of the AI models and what people are actually researching, what they're working towards. Is it mostly about price? Is it mostly about the output? Is it mostly about energy use? All those things.

Joe Weisenthal (2:12)

Well, I'm really excited to say we have the perfect guest, someone who is an AI researcher. We're going to be speaking with Jack Morris. He's currently about to finish his Ph.D at Cornell in AI. He's been affiliated with Meta professionally. So presumably, he already has a $100 million pay package in the bank. No, he's shaking his head. He's not. That's a joke. But Jack, thank you so much for coming on Odd Lots.

Jack Morris (2:36)

Yeah, thanks for having me. This is going to be fun.

Joe Weisenthal (2:38)

What do you explain to me like what you're up to? Because I don't really understand how it works where people are. They're at a university and they're also at a company. And this isn't how it works in much of the world, right? People get their degree and then they get a job. I get the impression that in the AI world, it's a little fuzzier in terms of one's affiliations between industry and education and stuff like that.

Jack Morris (2:58)

Yeah, that's definitely true. I think it might be on the way out, but I can tell you about my situation. So there's kind of a public research world and like a private research world.

Joe Weisenthal (3:08)

Okay.

Jack Morris (3:08)

And all the academic institutions do public research. And the AI labs like OpenAI, Anthropic, Google DeepMind, they essentially do private research where they have these people in house that are running experiments and learning more about their systems, but they don't publish anything or share any of their knowledge. And so a cool thing about getting your Ph.D. right now is you can do research, write about it and then publicize it, like put it online. I tweet about it. I kind of like can talk to you about it. And there's a few places left that will still kind of like...

Joe Weisenthal (3:40)

The moment you go in the house, we're never going to hear from you again. It's like severance or something like that.