PulseAugur
EN
LIVE 11:38:26

LLMs Claude Fable 5.1 and GPT-6 Astra lag behind older CNN in skin lesion diagnosis

By PulseAugur Editorial · [1 sources] ·

A dermatologist compared the diagnostic accuracy of two large language models, GPT-6 Astra and Claude Fable 5.1, against a 2015 Google image network on 200 biopsy-proven skin lesions. The older Google network achieved 62% accuracy, while both LLMs scored 42%. This study, with publicly available data and code, highlights the limitations of current LLMs in medical diagnosis, as the CNN would have missed 18 of 30 melanomas at its set threshold. AI

IMPACT Current LLMs show limited diagnostic accuracy in dermatology compared to older specialized models, indicating a need for further development in medical AI applications.

RANK_REASON Comparison of LLM performance on a specific medical task against existing models. [lever_c_demoted from research: ic=1 ai=1.0]

Read on Mastodon — sigmoid.social →

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →

LLMs Claude Fable 5.1 and GPT-6 Astra lag behind older CNN in skin lesion diagnosis

How we ranked this

Signal score
14 / 100
Composite score across the factors below. Higher = stronger signal that this story matters right now.
Newsworthiness bucket
Tool
Comparison of LLM performance on a specific medical task against existing models. [lever_c_demoted from research: ic=1 ai=1.0]
Source corroboration
Single-source cluster
Only one publisher covered this so far. Single-source stories can still rank when the publisher is high-authority, but they lack cross-source corroboration.
Topics
model release, product, other
Editorial topic classification. Feeds into how the story surfaces on /topic/<slug> hub pages and into the per-entity coverage mix.
AI-industry relevance
High
Clearly on-topic for AI-industry coverage.
Story freshness
Breaking (< 6h)
Fresh story with cross-source coverage still developing. Ranking may shift as more sources report.

Full methodology in our editorial standards.

COVERAGE [1]

  1. Mastodon — sigmoid.social TIER_1 English(EN) · [email protected] ·

    Dermatologist Magnus Lynch gave 200 biopsy-proven skin lesions to GPT-6 Astra, Claude Fable 5.1 and a 2015 Google image network fine-tuned for about twenty minu

    Dermatologist Magnus Lynch gave 200 biopsy-proven skin lesions to GPT-6 Astra, Claude Fable 5.1 and a 2015 Google image network fine-tuned for about twenty minutes. The old network named the right diagnosis for 62%. Both chatbots got 42%. Not a clean win. It's a small sample, and…

    LINKS lynchlab.net/llm-skin-cancer

RELATED ENTITIES

RELATED TOPICS