Lancet Digital Health review: 201 studies on medical LLM bias, 180 only tested closed models
MaziyarPanahi · x · 2026-11-21
A Lancet Digital Health systematic review on bias in medical LLMs flew under the radar, covering 201 studies. The data: 180 studies tested closed/proprietary models vs only 69 for open-weights (just 24 medical); only 44 used real-world data. Clinical decision support dominates (104 studies, 41 in informatics & policy), patient communication 72, while clinical note generation has just 7. Author's hot take: open weights aren't just cheaper — they're the only way to actually audit bias in medical AI.
Related event: Lancet Review: 90% of Medical LLM Bias Studies Test Only Closed Models(3 posts)→
More from Safety
- Christiano: Two Months of Math Progress in a Decade, and ML Could Be Next — OwainEvans_UK · 2026-11-21
- AI safety researcher: doubting China's moratorium compliance is like doubting your own marriage — DavidSKrueger · 2026-11-21
- Hinton may be softening on an AI moratorium after SF risk-scenarios panel — DavidSKrueger · 2026-11-21
- 205 Protesters Rally Outside Anthropic Against Recursive Self-Improvement — DavidSKrueger · 2026-11-21
- Study: human auditors can't detect sabotage in AI safety research — ronbodkin · 2026-11-21
- ~$50K CAD unauthorized Gemini API usage on a partner-billed account sparks routing nightmare — Big-Coffee-8899 · 2026-11-21