Making sense of the debate over AI psychosis

Box founder Aaron Levie got us talking this week with a social media post suggesting that tech CEOs are “uniquely prone to AI psychosis.”

On the latest episode of TechCrunch’s Equity podcast, Kirsten Korosec, Sean O’Kane, and I did our best to unpack Levie’s comment. For one thing, we noted that he isn’t disavowing AI tools, merely insisting that CEOs need to actually use those tools to understand them.

That’s a relatively gentle note of skepticism compared to other signs of a broader backlash, whether you look at graduating college students booing any mention of AI, the bad vibes around tech industry layoffs, or the apparent surge of installs at search engine DuckDuckGo after Google’s announcement that it’s bringing more AI to the search experience.

Kirsten suggested that Google faces a dilemma where it’s “chasing that thing it feels like it has to do to keep up, but it’s messing with the thing that people attach to the brand the most, and it’s not improving it.” More broadly, she wondered “if this anti-AI moment is an opportunity for startups or other areas of business.”

Keep reading for a preview of our conversation, edited for length and clarity.

Anthony Ha: AI is incredibly polarizing. And that’s part of what’s challenging to talk about; you can feel a little crazy because [simultaneously,] everybody’s using it and everybody loves it, but also no one’s using it and everybody hates it at the same time. There are large contingents for whom both of those things are true. 

On the user side, one thing that was very striking, we [already] talked about Google’s announcements about search and how AI is becoming a bigger part of search — although it’s been interesting to see how Google has tried to walk that back a little bit, or at least add some nuance in terms of, if you want that 10 blue links experience, there are still ways you can get it. It’s not going away entirely.

But I think a lot of people are not excited about the direction Google is going in. And so you see, for example, that DuckDuckGo said installs are up 30%, which is a huge leap. Now, of course, DuckDuckGo is a much, much smaller product than Google. I don’t think Google is in any immediate trouble, but I think that’s a sign that there is a very significant audience that does not like the current AI direction.

Sean O’Kane: I will say one thing that I keep looking for when I look at all of these leading AI labs or tech companies that are really pushing AI features and products — to me, there seems to just be this collapsing towards Anthropic’s approach, this idea of really trying to understand what it is you want to offer people and sticking to that.

And Google is one of the ones that I would say is actually still pushing the other direction. They’re trying to do a lot of different things, but they don’t do themselves any favors by being so vague about it.

What I mean by that is, when Google goes onstage at IO and talks about the way that it thinks it’s going to change search, so much of what they’re talking about, they’re talking about shopping or stuff that ends in a commercial transaction. And I think so much of what we think of Google as collectively, especially people who have been using it for two or three decades, is as an information retrieval system. 

Google can struggle with that a lot, where they get reactive fears of how they may be damaging the information retrieval side of things, and their response is, “Yeah, but that’ll still be there. Let’s focus on how it’s going to help you book a flight or something like that.” 

And then they also go off and sort of shoot themselves in the foot by releasing — it must be very challenging to stress-test these systems, but they go out and they release this stuff and they’re running into the same problems they’ve run into for years.

Kirsten Korosec: We had a great article that just published about how Google doesn’t know how to spell its own name. If you ask it, “How many P’s are in Google?” it says two. 

It’s this tension between: Google is chasing that thing it feels like it has to do to keep up, but it’s messing with the thing that people attach to the brand the most, and it’s not improving it.

What I’m wondering is, we’ve already seen some early evidence of people’s fingers doing the voting or walking for them, by literally going to another service. But I wonder if there are opportunities for other startups out there or culturally speaking, if this anti-AI moment is an opportunity for startups or other areas of business that we haven’t really thought about.

Anthony: Absolutely. Again, it’s probably a challenge because there is such a range of opinions. And if you build something that’s tailored for a group that’s skeptical [of] AI, then you’re probably going to alienate other users who are much more evangelistic or gung-ho about it. But I think that’s just the moment we’re living in.

And you can see in how DuckDuckGo is promoting itself, that they’re very much emphasizing this idea of being anti-AI, which I find very striking because I’ve mentioned before, [I’ve been] moving away from Google myself, trying out other search engines. And I would say that a year ago, when I started that exploration, even these alternative search engines were still trying to experiment with AI features, emphasizing AI to some degree because they also thought they had to do it.

And now I think they’re seeing that there is actually a lane to be like, “No, we just were not interested in that stuff at all. Or inasmuch as we’re doing it, we’re very much putting it in a separate sandbox that’s not going to affect your core search experience.”

Kirsten: I think we unfairly sometimes categorize all the tech CEOs as force-feeding people AI. And there’s at least one tech CEO who has come out and said, “I think that there’s a little bit of psychosis among other tech CEOs around AI.” 

I’m talking about Box founder Aaron Levie, who has come to Disrupt many times and is a friend of TechCrunch for sure. He made these comments about how CEOs are uniquely prone to AI psychosis because they’re sufficiently, and I’m reading this, “distant from the last mile of work that still has to happen to generate most value with AI.” 

I thought that was really interesting. And I’m wondering if there are other CEOs out there who agree with it. I also wonder, as part of that shift of thinking about what has to happen to generate the most value, if they’re also thinking about how their workforce is changing, which is our other topic today — [not] just about the AI divide, it’s also how AI is changing work. And we’ve seen, certainly, some of the bad news side of that, and that is a lot of layoffs.

But I think also, we’re seeing big changes in how people work. I’m wondering in the areas that you two cover, if you’re seeing evidence of that, because I don’t think it’s just in the quote unquote “AI startup sector” or the big tech companies.

Sean: As far as the companies that I cover, a lot of them tend to be working on, if not physical transportation, then stuff adjacent to it. And it’s seemed much slower there than it is, unsurprisingly, on the software side of things. 

We’re starting to see some of that changing. We’ve talked on the show a little bit about Mind Robotics, which is the spinout from Rivian CEO RJ Scaringe. And, you know, there’s certainly more AI being applied to physical infrastructure and manufacturing and robotics and self-driving.

I think the software side is where it’s really changing things, where you have people whose job is just directly tied to producing code.

Anthony: Part of the question, I think, [involves] both AI adoption in companies and then AI-driven layoffs — to what extent are they top down or bottom up? 

Because I think a lot of other transformations in the workforce in the last couple of decades have at least been, to some extent, bottom up: These are tools that people actually like to use, they bring them in, and then at a certain point, executives and IT managers accept that.

There is some sense that a lot of the [belief that there are going to be these] AI productivity gains seem to be embraced by the executives — or, if you’re at a startup, probably by the VCs who are funding you — who love this dream that you can have just a tiny team and be as effective as a company with a much larger team.

And I don’t think that that is necessarily impossible, but I think that Aaron’s point is essentially that if you’re not really touching any of the end work, how would you know? He’s also not somebody who’s saying we should just throw out all the AI tools, but he’s saying that you actually have to use these tools and understand what they’re doing. You can’t just look at a slide and be like, “Yes, incredible efficiency, let’s go.”

Kirsten: Well, I think there’s a lot of real evidence out there that these companies are using these tools, and it is directly affecting workers in the form of layoffs, and also the way that they work. The two truths are accurate here.