AI-powered
podcast player
Listen to all your favourite podcasts with AI-powered features
Steady Progress in AI and the Mirage of Emergent Capabilities
Progress in "AI" over the past 80 years has been steady, not marked by sudden leaps or "winters" and "summers." This progress has been seen in economics and problem-solving. Each time a problem is tackled, there's a sense that Artificial General Intelligence (AGI) is near, but it turns out to be just another modality. AGI has been a moving target for 70 years. AI has become a useful tool, and its continued development is important. However, it doesn't represent a step change or alter the nature of computers and software in unprecedented ways.
AI Progress: "the past 80 years of quote unquote AI, it's been studies progress independent of there being winters and summers."
Steady and Incremental: "It's been very, very studies progress. It's been progress on economics, progress on problem solves."
AGI a Moving Target: "every time we like tackle a problem, like everybody's, oh, goodness, this is it. We're almost at AGI. And then it just turns out like that's one modality."
No Step Change: "I don't think that there's any step change or it changes the nature of computers or software in ways that we haven't seen before."
Threshold Effects in Utility: "I'm a big believer in threshold effects. I mean, the topics of emergency are super hotly debated and is emergence of mirage or non mirage and maybe the underlying curves are smooth. But in terms of like practical utility, even if the capability of the model hasn't taken a sudden shift, the utility and what it can actually practically do in the world sometimes, seems to take a notable"
While AI capabilities may develop smoothly, their practical applications can exhibit threshold effects, leading to perceived sudden leaps in utility.