AI coding tools are replacing take-home exams and second opinions

Two threads today show AI being used in contexts where it probably shouldn't work, and sometimes working anyway. A Brown professor reported mass AI fraud on a take-home exam, with students apparently not even trying to hide it. The comments split: some say the professor should have expected this, others say students are cheating themselves out of learning they paid for. A separate thread shows someone using Claude Code to get a second opinion on an MRI, with a radiologist in the comments saying they 'would not use Claude to get a second opinion on anything that's an image.'

The pattern: AI is being used as a shortcut in high-stakes, expert domains (medicine, education) faster than any institution has adapted. The take-home exam is effectively dead as an assessment format. Medical AI second opinions are happening whether radiologists endorse them or not.

Commenters in the education thread argue that if graduates will use AI on the job anyway, maybe the exam format is the wrong thing to defend. That's a legitimate point, but it sidesteps the question of whether students are learning the underlying reasoning that makes AI output interpretable.