Interviewed by TribLive on the risks of AI

I was interviewed by Kellen Stepler of TribLive for a story on whether AI could threaten humanity, alongside researchers from Carnegie Mellon and Pennsylvania Western University.

My view is that AI safety and alignment need to be discussed openly and, ideally, regulated. Finding a solution won’t be easy, though: the issue is global, and the right safeguards will vary by job and discipline, so no one-size-fits-all solution may exist.

I’m less worried about AI going rogue on its own than about how people might misuse it (e.g., to stage cyberattacks). And I’m more worried still about the nearer-term risks of job displacement and economic impact. Anthropic’s own economic scenarios point to significant knowledge-worker job losses under an extreme case by 2030. That said, AI in the workplace isn’t inherently bad: used well, it can make work more efficient, even as many tasks will still require human verification.