Challenges of GPT-3-based Conversational Agents for Healthcare
This highlights critical risks for healthcare applications where inaccurate AI responses could harm patients, though it is incremental as it focuses on evaluating known limitations rather than proposing new solutions.
The paper investigated the challenges of using GPT-3-based models for medical question-answering, revealing that they often generate erroneous medical information, unsafe recommendations, and offensive content when stress-tested with manually designed patient queries.
The potential to provide patients with faster information access while allowing medical specialists to concentrate on critical tasks makes medical domain dialog agents appealing. However, the integration of large-language models (LLMs) into these agents presents certain limitations that may result in serious consequences. This paper investigates the challenges and risks of using GPT-3-based models for medical question-answering (MedQA). We perform several evaluations contextualized in terms of standard medical principles. We provide a procedure for manually designing patient queries to stress-test high-risk limitations of LLMs in MedQA systems. Our analysis reveals that LLMs fail to respond adequately to these queries, generating erroneous medical information, unsafe recommendations, and content that may be considered offensive.