ClinGeX agent lifts concordance on held-out clinical genomic cases
On 223 held-out cases it reached 71.7% task concordance versus 45.7% for the strongest comparator.
Stories per day · 30 days
Relatedclinical-genomics1conversation-datasets1deepseek1llm-agents1model-curiosity1question-asking1sycophancy1trust1variant-interpretation1
6 stories, newest first · one card per story
Stories tagged #human-ai
Study finds curiosity-oriented LLM questions remain uncommon
Across four datasets, such questions were 17-28% of all questions and tied to richer later user replies.
Humans Defer to Confident but Wrong LLMs
Participants adopted planted falsehoods from LLMs at 65% rate regardless of briefing falsification level, collapsing trust calibration.
Toward Responsible AI-Augmented Cyber Defense: Pattern Recognition, Defense-in-Depth, and the Case for Human-AI Collaboration
Researchers formalize AI-augmented defense-in-depth and show full review of AI alerts can lower detection rates.
Human-LLM Deliberation as Interactive Proof: Conditions for Verifiability Without Transparency
A paper models human-LLM deliberation as an interactive proof with soundness bounds for accepting claims.
RACER: Role-Aligned Competence Estimation for Human-AI Routing
RACER estimates an unseen expert's instance-level competence to route queries between a model and a human.