New Book Reveals AI Leaders Kept Underestimating Their Own Creation
Even the experts didn't see ChatGPT coming—what does that mean for the rest of us?
The AGI Chronicles, by Kevin Roose, tells the story of AI through the eyes of the people creating it. A striking pattern in the book is that AI experts have more often underestimated AI in general or transformers in particular than overestimated them.
Roose describes CEOs racing to reach key thresholds, with at least Musk and Altman influenced by an expectation that one person will end up ruling the world by developing the most powerful AI; he summarizes their feelings as “The machines of loving grace, it seemed, would be built by humans of bitter spite.” But he describes average AI-company employees differently: at parties where researchers from rival AI labs got drunk and leaked trade secrets, OpenAI employees kept expecting to learn that their rivals were working on something similar. The article says the author’s observations agree with Roose’s characterization that these employees are not very adversarial toward employees of competing companies.
- AI experts have repeatedly underestimated how fast AI would improve—even the CEO of OpenAI thought ChatGPT would only get a million users.
- Some tech leaders believe the first to build superintelligent AI could rule the world, which drives fierce competition.
- Despite the hype, many AI employees collaborate and share information, suggesting the race is not as cutthroat as it seems.
Why It Matters
AI leaders' past mistakes show they can't fully predict AI's impact—so we should stay informed and cautious.