LO
CapabilityCognitive automation2025-01-09
GPT-4o answered about 92% of 1,050 registered dietitian exam questions correctly, but results varied with prompts and topics, a study found
Dietitian / nutritionistoccupation page →
Event date / reported
2025-01-09
Evidence stage
CapabilityA demo, benchmark or paper shows the task can be done. Updates what the technology can do — not what employers will do.
Tasks this bears on
Medical nutrition therapy
Assessing a patient's nutritional status and prescribing a diet for their condition.
Still human-led≈ Platform inference
Where this applies
A study that tested GPT-4o, Claude 3.5 Sonnet and Gemini 1.5 Pro on 1,050 US registered dietitian exam questions from an exam-preparation bank. GPT-4o scored 91.92% with zero-shot prompting and highest overall; the authors note results varied considerably with different prompts and question domains. Exam questions test knowledge, not the care of a patient.
What this means
Models already hold the knowledge a dietitian is examined on.
What it does not yet show
Exam-bank questions are not patients; the study does not test care, calculation for a person, or counselling.
What you can check
Open the Scientific Reports article on LLMs and the registered dietitian exam and find "1050 RD exam questions".
Does it change the assessment?
No — and this stage does not move it either. A "Capability" record is real evidence, but it does not upgrade a task judgement on its own. The 1 linked judgement above stand where they were.
Source
Azimi et al. — Evaluation of LLMs accuracy and consistency in the registered dietitian exam through prompt engineering and knowledge retrieval, Scientific Reports 15, 1506 (published 09 January 2025) · verified 2026-09-30 · Claude (VOLO agent) · interpreted 2026-09-30 · Claude (VOLO agent)
Primary source — published by the party that did this, or the authority of record. No co-signature needed.