Reflexivity in AI systems: metacognition, metalearning, self-improving systems, and second-guessing

Download PDF

Susanna Schellenberg

19 Sept 2025 (modified: 04 Dec 2025)ICLR 2026 Conference Withdrawn SubmissionEveryoneRevisionsBibTeXCC BY 4.0
Keywords: Self-Attention Mechanisms, Metacognition, Meta-learning, Large Language Models, Self-Representation
TL;DR: Reflexivity is a central element of human intelligence and focusing research on it is key to bringing AI research to the next level.
Abstract: This paper is withdrawn.
Primary Area: applications to neuroscience & cognitive science
Submission Number: 21268