AI Round Up: Ari Morcos from Datalogy AI and Rob Toews from Radical VC on Karpathy Reactions, OpenAI’s Dealmaking, & Bubble Reality Check

**Jacob Effron** (0:00)
We're back on Unsupervised Learning with another part of our recurring series with Ari from Datalogy and Rob from Radical, where we hit on the big questions in AI today. We talked about Andrej Karpathy's recent comments about agents and how they're overhyped in the short term, as well as whether we're in a bubble today. We talked about OpenAI, all the things they've shipped from Sora to the apps SDK, to the browser and a bunch of other things. We talked about the compute deals that have been struck and getting reactions to those, as well as potential acquisition targets for Apple. And then finally, we hit on the AI app ecosystem and what's interesting and not as interesting from an investing perspective. It's just a ton of fun to get to do this with two good friends who are so smart in the AI space. I think folks will really enjoy it. Without further ado, here's our conversation. Guys, this is awesome. I'm so glad we're doing this again.

**Rob Toews** (0:42)
We're running it back.

**Ari Morcos** (0:43)
Yeah, really excited to get to do it again. And in person this time.

**Jacob Effron** (0:45)
Yeah, this time in person always way better. That way we can get, you know, like, you know, it's super dramatic reactions to especially bad takes. You know, you can stand up, you can kind of like, you know, just storm out of the person.

**Ari Morcos** (0:58)
So what's the over under until one of us, like, storms out of the room?

**Jacob Effron** (1:00)
Hopefully if the questions are good, by mid-interview, something bad should have been said. I think to kick it off, I feel like the dominant question of the last few weeks, really the last few months has been, you know, are we in a bubble? And there always seems to be some net new data point that people used to react to that. But the latest is another podcast appearance from Andrej Karpathy, where he said some really interesting things. You know, I'll read some of the quotes that I found compelling.
You know, I think it's probably best summarized by him saying, you know, look, overall, the models are not there. I feel like the industry is making too big of a jump and is trying to pretend like this is amazing, and it's not. It's slop. You know, he also said it'll take at least a decade until AI can meaningfully automate entire jobs, like an employer intern that you'd hire to work with. And so obviously a pretty different narrative than some of the stuff we've been hearing out of the labs. Curious both of you, what was your reaction to what Andrej was saying?

**Ari Morcos** (1:50)
I think, you know, I would start by saying, I think it's still a pretty hypey, like Andrej Karpathy is still pretty hypey on AI. I think it's kind of wild that's saying like, oh, like AGI will get here in 10 years and we'll be able to fully automate all work in 10 years means there's a bubble. Like that's a crazy statement. Like if you had told me that five years ago, you know, with high confidence in 15 years, we would automate all work. Like that would be the biggest technological advancement ever. So maybe it's like more a commentary on how ridiculously overhyped things have been, but you know, my take on this generally is that like, yes, there's absolutely some overhyping going on. I think, you know, we talk to people who think that they have two years to make money before nobody makes money anymore because everything is fully automated. Like that's obviously not true. If nothing else, like it takes a million years for anything to actually disseminate through the enterprise. And like just enterprises and big companies are slow and it will take a long time to actually see the real world impacts in a lot of these cases just because of that pace.
So you know, my take is, yeah, there are some people overhyping things, but there's like still very, very real substance going on here. If we stop making models better today, like they just, this is fixed. This is all we can ever do. We would still see a massive, massive transformation throughout the economy. This is putting a splash of water on some of the really overhypy stuff, sure. But there's still real substance where we are in a way that is like very different, for example, than the.com bubble.

**Jacob Effron** (3:17)
Ten years to AGI is pretty exciting. But obviously, there's concerns in the short term that are we overshooting with the amount of capital being put into some of these data centers. I think what's not clear is right now, do you need enterprise agents and the ability to replace work to justify this latest level of reported capex spend? We could be headed on this incredible long-term path, and if some of the stuff doesn't show up in the short term, it's kind of unclear if we'll see the revenues needed to justify this next wave of capex investment.