Health Exec

Real-world use of LLM artificial intelligence for consumer-level health questions exhibits error rate that ‘far exceeds the acceptable margin in most healthcare settings.’

When used in the ‘wild’ conditions of the real world, large language AI stumbles Asking medical questions of AI with language spoken “in the wild”—meaning with LLM prompts from everyday consumers—brings back answers with decidedly mediocre accuracy, a new study shows.