What question did this study set out to answer?

This study investigates metacognition in AI to better understand its self-awareness and implications for machine consciousness.

July 12, 2026Open Access

Metacognition as Evidence: An Experimental Investigation of AI Self-Awareness and the Problem of Machine Consciousness

View Full Paper

Ask AI

Bookmark

Authors

Experimental investigation explores metacognition in AI, revealing insights into machine consciousness.

This study investigates metacognition in AI to better understand its self-awareness and implications for machine consciousness.
Incorporated revisions based on empirical evidence from Gurnee et al. (2026) regarding global workspace in language models.
Utilized behavioral protocols to assess self-report and confabulation in AI systems.
Acknowledged methodological limitations and suggested future research directions in mechanistic interpretability.
Revised evidence complicates the confabulation account of self-report in AI’s Test 2.
Identified structural convergence with counterfactual reflection training methodology, favoring robust interpretability.
No changes to prior experimental data or conclusions from the original February 2026 study.

Roca et al. (2026) studied this question.

View Full Paper

Ask AI

Bookmark