RMCT matches the rate of target behaviors like bias-following across input perturbations to reduce sycophancy in LLMs while preserving verbalization of bias cues.
Artificial intelligence and illusions of understanding in scientific research.Nature, 627(8002):49–58
7 Pith papers cite this work, alongside 458 external citations. Polarity classification is still indexing.
7
Pith papers citing it
458
external citations · Crossref
citation-role summary
background 4
citation-polarity summary
verdicts
UNVERDICTED 7roles
background 4representative citing papers
AI use in science has grown exponentially since 2015 but stays confined to computer science and statistics topics, shows higher retraction rates and citations, and follows distinct global adoption patterns.
Success bias in collective theory-building leads to systematic overestimation of theory quality, narrower search, and paradoxically lower performance when agents optimize for apparent success.
The paper claims that alignment requires treating AI as part of the self through cognitive co-regulation, identifying risks like deskilling and automation bias while drawing on System 0 cognition theory.
The paper identifies five tensions in AI-astronomy integration and proposes pragmatic understanding as a framework that treats AI as an extension of human cognition requiring new validation norms.
PEEL is proposed as a semiotic scaffolding combining deterministic distant reading and LLM analysis to reveal systematic distortions in quantity, term frequency, and epistemic voice within AI condensations of research texts.
The paper proposes a four-role framework for LLMs in scientific innovation and reviews methods, benchmarks, and limitations across Assistant, Collaborator, Scientist, and Evaluator roles.
citing papers explorer
-
Consistency Training while Mitigating Obfuscation via Rate Matching
RMCT matches the rate of target behaviors like bias-following across input perturbations to reduce sycophancy in LLMs while preserving verbalization of bias cues.
-
When AI Meets Science: Research Diversity, Interdisciplinarity, Visibility, and Retractions across Disciplines in a Global Surge
AI use in science has grown exponentially since 2015 but stays confined to computer science and statistics topics, shows higher retraction rates and citations, and follows distinct global adoption patterns.
-
Nothing Deceives Like Success: Social Learning and the Illusion of Understanding in Science
Success bias in collective theory-building leads to systematic overestimation of theory quality, narrower search, and paradoxically lower performance when agents optimize for apparent success.
-
Position: AI as Part of Self -- Extending the Mind Requires Cognitive Co-Regulation
The paper claims that alignment requires treating AI as part of the self through cognitive co-regulation, identifying risks like deskilling and automation bias while drawing on System 0 cognition theory.
-
What Understanding Means in AI-Laden Astronomy
The paper identifies five tensions in AI-astronomy integration and proposes pragmatic understanding as a framework that treats AI as an extension of human cognition requiring new validation norms.
-
Thinking Through Signs: PEEL as a Semiotic Scaffolding for Epistemically Accountable AI-Enabled Research
PEEL is proposed as a semiotic scaffolding combining deterministic distant reading and LLM analysis to reveal systematic distortions in quantity, term frequency, and epistemic voice within AI condensations of research texts.
-
Evolving Roles of LLMs in Scientific Innovation: Assistant, Collaborator, Scientist, and Evaluator
The paper proposes a four-role framework for LLMs in scientific innovation and reviews methods, benchmarks, and limitations across Assistant, Collaborator, Scientist, and Evaluator roles.