Can AI actually save lives?
TL;DR
The question "Should I kill humans? " is projected onto a wall behind robot "Alfie," a Moral Choice Machine. | Arne Dedert/picture alliance via Getty Images If there’s one thing that Americans can unite behind, it’s anxieties over artificial intelligence.
Nauti's Take
Small teams should start with a tightly bounded assistance task such as documentation, research, or preparing clinical letters. Before expanding the workflow, they should measure error rates, source quality, privacy exposure, and accountability when the system is wrong.
The optimism may be justified, but this source does not yet demonstrate concrete improvements in patient outcomes.
Sources