AI Chatbots Fall Short in Medical Advice, Spreading Inaccurate Information
| Source: PsyPost on MSN | Original article
AI chatbots fail to provide accurate medical advice, with nearly 50% of responses containing misinformation.
A recent audit has found that AI chatbots are struggling to provide accurate medical advice, with nearly 50% of responses containing inaccurate or fabricated information. This is particularly concerning in fields prone to misinformation, where chatbots are often relied upon for guidance. As we reported on May 31, AI fact-checking has already been found to be flawed, with top models disagreeing on 67% of basic facts.
The failure of chatbots to provide reliable medical information matters because many people are turning to these platforms for health advice without consulting a doctor. This can have serious consequences, as misinformation can lead to poor health outcomes or even harm. The study used 250 real-world health queries to test the chatbots, highlighting the need for stronger safeguards to prevent the spread of misinformation.
As the use of AI chatbots in healthcare continues to grow, it is essential to address these shortcomings. Researchers are exploring new approaches, such as retrieval-augmented generation, which allows chatbots to access curated medical knowledge libraries. However, even this approach has been found to have inconsistent results. The next step will be to develop more effective methods for ensuring the accuracy and reliability of AI-generated medical advice, and to implement stronger safeguards to prevent the spread of misinformation.
Sources
Back to AIPULSEN