A short break from plasma to chat about something that comes up in almost every research conversation I have lately: should scientists use AI? Here's my honest take after two decades in research — the concerns I think are overblown, the one I take seriously, the benefits that move my work 10×, and the app I asked Claude to build yesterday that's already running on my phone.

⏱️ Chapters

00:00 Today: AI in science, not plasma

00:24 Concern 1: Is AI deskilling researchers?

00:48 Why I think it's the opposite — AI breaks the "singular mindset" of searching

01:23 Use AI as a tool, not the answer

01:33 Concern 2: Hallucinations and fake citations

01:38 You shouldn't be citing papers you haven't read

02:00 What I've seen: real citations to abstracts with no content

02:08 Double- and triple-check ever ...