Reposted by Becky Kazansky
Regarding the OpenAI dump of 722 supposed proofs, I was just visiting a group of mathematicians in Barcelona, and my host had already found that some of the proofs were absolute junk. Others may take longer to evaluate, but as was pointed out at the panel discussion on "AI and Mathematics" I
![on this issue in the context of GenAI-assisted knowledge work.
High task confidence is associated with users’ ability to delegate
tasks effectively, fostering better stewardship while maintaining
accountability. Conversely, lower self-confidence may lead users to
rely more on AI, potentially diminishing their critical engagement
and independent problem-solving skills. This reliance on AI can be
seen as a form of cognitive offloading [8], where users depend on
AI to perform tasks they feel less confident in handling themselves.
Confidence in AI is associated with reduced critical thinking
effort, while self-confidence is associated with increased critical
thinking effort. This duality indicates that design strategies should
focus on balancing these aspects. The aims are both to improve
the quality of AI-assisted tasks and also to empower users to develop their skills and maintain a balanced “relationship” with AI.
To address task confidence recalibration, AI tools could incorporate feedback mechanisms that help users gauge the reliability of
AI outputs, when to trust the AI and when to apply their critical
thinking skills. This aligns with the goals of explainable AI [33].](https://cdn.bsky.app/img/feed_thumbnail/plain/did:plc:66lbtw2porscqpmair6mir37/bafkreicespkczdpxpe3zqcbrxiac3rs2nluxsnqykkmeidacnweas4fcwu)
