Zoë Schiffer: Yeah, I mean it really, really impacted people. I think on the most extreme ends, you see people who have what looks like perhaps like a mental health crisis, they're so attached to the model, but then you just have complete power users who are like, “This is part of my minute by minute life. What have you done? You didn't warn me.”
Jake Lahut: And this is where the introspective aspect of these tools, the kind of desire for self-understanding, the people who are not advisably from any medical perspective, but they are trying to use these bots for something akin to therapy. And what it made me think of when I saw this rolling out was, is this maybe the beginning of something bigger where there's kind of a departure between the “regular consumer” experience and demand for AI versus the business application? We may not all have the same definition of intelligence when it comes to these models, and that some of us really just want a buddy, a companion, a way to know ourselves better. And then other people are like, “No, I just need a little team of bots here to manage, get my stuff done, I'm going to babysit and I'm going to tell them what to do and live my life.” And yeah, I don't know where that goes. It does seem like it's revealing something maybe genuinely new about the human condition in a way that I would not have expected.
Zoë Schiffer: Yeah, I mean, I think it's been a learning for OpenAI. They've been kind of baffled. I've seen these conversations internally where they're like, “I guess people don't care as much about intelligence as we thought.” The narrative around intelligence is critically important for fundraising, if nothing else, they really need to raise gobs of money and being like, “We're about to achieve artificial and general intelligence, AI will be able to do all of these things,” is really important for that. But for everyday users, it really makes me think of this story that is kind of famous inside OpenAI about the night before the ChatGPT release in November of ’22, Ilya, he was testing out what was going to be ChatGPT and asked it 10 pretty hard questions. And he felt like five of them, he got pretty good responses, and five were unacceptably bad. And they had this moment where they were like, “Do we release this? I don't know if it's good enough.” And then they decided to move forward. And what we saw was the general public was like, “This is amazing.” Because they'd solved a product issue. It wasn't necessarily about the model, which had been out for a long time. It was like the interface to interact with the model was really the unlock. And I think OpenAI, that really is more and more the company's edge, even though it really sees itself as a research lab. It's a product lab in a lot of ways, and it'll be interesting to see how that changes the company moving forward.
Jake Lahut: Absolutely.
Zoë Schiffer: That's our show for today. We'll link to all the stories we spoke about in the show notes. If you're in San Francisco, don't forget to get your tickets for the September 9th event with KQED. Make sure to check out last Thursday's episode of Uncanny Valley, which is about what Palantir actually does and why it's so controversial. Adriana Tapia produced this episode, Amar Lal at Macrosound mixed this episode. Pran Bandi is our New York studio engineer. Kate Osborn is our executive producer. Condé Nast head of global audio is Chris Bannon, and Katie Drummond is WIRED's global editorial director.