Clinician use of language models diverges from how the models are evaluated

Key points

• We characterize each query with RCQ-Map, a clinician-validated framework grounded in taxonomies of clinical questions and of LLM evaluation, which records its task, intent, answerability, missing information and potential harm.1
• Documentation and administration (36.2%) and knowledge retrieval (28.9%) made up nearly two-thirds of use, and diagnosis 3.7%; more than a third of queries could not be answered well as posed.1
• Applying RCQ-Map to 58 public benchmarks drawn from major evaluation suites and frontier model reports, which we assemble into the Clinical AI Benchmark Atlas, showed that the median benchmark contained no documentation requests and shared 31% of the task mix of real use, less than an even spread across task categories would.1
• Benchmark scores therefore say little about how clinical AI performs on most of the work it is actually given, and evaluation should be matched to real clinical use.1

Sources (1)

• [1]Clinician use of language models diverges from how the models are evaluated
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 8, 01:29 AM

  Large language model (LLM) assistants are being deployed to clinicians across health systems, and judgments about their readiness rest largely on benchmark scores, most of them derived from examination questions or curated cases.

  We characterize each query with RCQ-Map, a clinician-validated framework grounded in taxonomies of clinical questions and of LLM evaluation, which records its task, intent, answerability, missing information and potential harm.

Extractive summary: sentences quoted from the sources.