AI Chatbots Vary Widely on Fake Medical References - Medscape
A study of nine AI chatbots found hallucination rates from 0% to 34%, with Grok 3 worst and Perplexity Research most reliable.
Read the original at Medscape ↗
- Topics
- Hallucinations & FabricationChatbots & LLMsHealthcare
- Mentions
- GrokPerplexityxAI
- Tracked via
- “AI hallucination”