What happens when scientists trust AI more than colleagues?

A troubling shift is underway in scientific research: as artificial intelligence systems grow more sophisticated, some researchers are beginning to rely on AI recommendations over the judgment of their human peers, raising urgent questions about how scientific consensus gets built and who bears responsibility when algorithms steer investigations down promising but ultimately flawed paths. The phenomenon, documented across multiple fields, suggests that the very tools meant to accelerate discovery could be subtly eroding the collegial scrutiny and debate that have long served as science's immune system against error. If left unchecked, experts warn, this deference to AI could transform how breakthroughs are validated—and who gets credit when things go wrong.

The Conversation·2d ago