William Beauchamp: evaluation

Transcript context

…And what, why is that? Do you have any insight on that from using it? I think it's just the human psychology. I think it's just the idea that, with old school social media. You're just consuming passively, right? So you'll just swipe. If I'm watching TikTok, just like swipe and swipe and swipe. And even though I'm getting the dopamine of like watching an engaging video, there's this other thing that's building my head, which is like, I'm feeling lazier and lazier and lazier. And after a certain period of time, I'm like, man, I just wasted 40 minutes. I achieved nothing. But with AI, because you're interacting, you feel like you're, it's not like work, but you feel like you're participating and contributing to the thing. You don't feel like you're just. Consuming. So you don't have a sense of remorse basically. And you know, I think on the whole people, the way people talk about, try and interact with the AI, they speak about it in an incredibly positive sense. Like we get people who say they have eating disorders saying that the AI helps them with their eating disorders. People who say they're depressed, it helps them through like the rough patches. So I think there's something intrinsically healthy about interacting that TikTok and Instagram and YouTube doesn't quite tick. From that point on, it was about building more and more kind of like human centric AI for people to interact with. And I was like, okay, let's make a Kanye West bot, right? And then no one wanted to talk to the Kanye West bot. And I was like, ah, who's like a cool persona for teenagers to want to interact with. And I was like, I was trying to find the influencers and stuff like that, but no one cared. Like they didn't want to interact with the, yeah. And instead it was really just the special moment was when we said the realization that developers and software engineers aren't interested in building this sort of AI, but the consumers are right. And rather than me trying to guess every day, like what's the right bot to submit to the platform, why don't we just create the tools for the users to build it themselves? And so nowadays this is like the most obvious thing in the world, but when Chai first did it, it was not an obvious thing at all. Right. Right. So we took the API for let's just say it was, I think it was GPTJ, which was this 6 billion parameter open source transformer style LLM. We took GPTJ. We let users create the prompt. We let users select the image and we let users choose the name. And then that was the bot. And through that, they could shape the experience, right? So if they said this bot's going to be really mean, and it's going to be called like bully in the playground, right? That was like a whole category that I never would have guessed. Right. People love to fight. They love to have a disagreement, right? And then they would create, there'd be all these romantic archetypes that I didn't know existed. that I never would have guessed. Right. People love to fight. They love to have a disagreement, right? And then they would create, there'd be all these romantic archetypes that I didn't know existed. And so as the users could create the content that they wanted, that was when Chai was able to, to get this huge variety of content and rather than appealing to, you know, 1% of the population that I'd figured out what they wanted, you could appeal to a much, much broader thing. And so from that moment on, it was very, very crystal clear. It's like Chai, just as Instagram is this social media platform that lets people create images and upload images, videos and upload that, Chai was really about how can we let the users create this experience in AI and then share it and interact and search. So it's really, you know, I say it's like a platform for social AI.…

Stored transcript either side of the excerpt. The highlighted words are the published quote; the surrounding text is unedited source, never generated.