Get the latest tech news

A.I. Has Become a Technology of Faith

Sam Altman and Arianna Huffington told me that they believe generative AI can help millions of suffering people. I’m not so sure.

Altman told me that his decision to join Huffington stemmed partly from hearing from people who use ChatGPT to self-diagnose medical problems—a notion I found potentially alarming, given the technology’s propensity to return hallucinated information. For the past two years, shortcomings in generative-AI products—hallucinations; slow, wonky interfaces; stilted prose; images that showed too many teeth or couldn’t render fingers; chatbots going rogue—have been dismissed by AI companies as kinks that will eventually be worked out. When I asked about hallucinations, Altman and Huffington suggested that the models have gotten much better and that if Thrive’s AI health coaches are focused enough on a narrow body of information (habits, not diagnoses) and trained on the latest peer-reviewed science, then they will be able to make good recommendations.

Get the Android app