Logical methods for AI
AI disclosure
The course material
The course material began as a manuscript composed by hand, without AI. It has
since been through extensive copy-editing for both style and content, by human
and agentic readers alike. The agents were
OpenAI’s Codex (opens in a new tab)
mainly with gpt-6-astra, and
Anthropic’s Claude (opens in a new tab)
mainly with Opus 5. Every agent edit
passed a human review for correctness and style before publication.
The website
The site’s code, its interactive apps and the tooling behind the slides rest on human-written design specifications. Agents carried out the implementation. Nothing reaches the site without a human review against the specification and for the general soundness of the code.
Responsibility for the result, and for any errors in it, remains with the author.
Further reading
The Leiden Declaration on Artificial Intelligence and Mathematics (opens in a new tab) concerns the use of AI in mathematical research rather than in teaching material, so it does not govern a course like this one. Its recommendations nevertheless set the standard this page follows: disclose the tools used, take responsibility for the correctness of published work, and affirm human authorship.
Anyone working in mathematics or the formal sciences should read it in full.