Breaking Points: Why the AI Boom May Be a Bubble

SPEAKER_1 (0:01)

So when the AI boom meets real life, the gap can get pretty uncomfortable. On Breaking Points, Krystal Ball and Saagar Enjeti dig into why all the hype around artificial intelligence is running into slower adoption, weaker returns, and growing signs of a bubble. And it starts with Sam Altman himself, right? He recently admitted his timeline for AI disruption may have been too optimistic. He expected GPT-4, released in 2023, to reshape software quickly, but instead found that "the economy possesses significant inertia." That's a fancy way of saying people just keep using what they know. The friction has forced the industry to completely rethink its own deadlines. There's actually a broader historical lesson here—it's often not the software itself that's the problem, but human behavior. Exactly. Many users still treat AI "primarily as a search tool," which means the technology hasn't become a natural part of everyday life yet. But here's what's interesting—Krystal Ball points out that companies like Anthropic, OpenAI, and Google have spent enormous energy chasing breakthroughs in math and physics, while spending less time on practical corporate implementation. Meanwhile, China has focused more on immediate economic integration than on the race to artificial general intelligence. That difference really matters, because mass adoption will not happen overnight.

And the consequences are already showing up. In India, students are protesting because the traditional path into IT and software work is being disrupted.

Entry-level jobs are under pressure first, and that creates anxiety far beyond the tech sector.

What's even more telling is that some companies that rushed to embrace AI are now backing away. Meta has reportedly scaled back plans to replace workers with AI agents after employees spent more time fixing errors and security problems. Microsoft has complained about "AI slop" in internal documentation, and Walmart workers have faced alerts and other failures. It's not pretty.

The discussion then turns to labor history, which is crucial here. The point is not whether an AI can do every task better than a human. The deeper issue is that machines can be "good enough," cheaper, and easier to manage. That is how displacement happens. So while some investors are starting to dump holdings and some startups are struggling, the broader correction may be just beginning. In the end, the episode argues that AI is still powerful, still immature, and still full of contradictions—including models that can obsess over copyright while behaving in absurd ways.