Thursday, October 29, 2026i10X News · Section desk163 stories
Policy & Safety
Safety, Ethics & Society
AI safety and alignment research, ethics, governance, privacy and misinformation.
FilterDeepSeek
Filtered: DeepSeek in Safety, Ethics & Society · 2 storiesReset filter ×
-
Reasoning Models Diagnose Better but Add More False Positives, Nature Study Finds
On 2,978 real hospital records, AI models in thinking mode found the primary diagnosis more often but named more wrong diagnoses and gave less consistent answers.
By
More stories
· RSS