News for hackers. Stuff that matters.

A comment on Mathematical exploration and discovery at scale. Parent comment

Re:Mathematical exploration and discovery at scale by adastra22 on Thursday November 06, 2025 @10:17PM (#45843219)
Yes, and that’s what a coding agent is too. Technically an agent is not (just) an LLM, but the two have become synonymous is discussions.
[ Parent | on HN ]
Re:Mathematical exploration and discovery at scale by woadwarrior01 on Friday November 07, 2025 @04:50AM (#45844935)
> that’s what a coding agent is too.

Not quite.

[ Parent | on HN ]
Re:Mathematical exploration and discovery at scale by adastra22 on Saturday November 08, 2025 @12:38AM (#45854445)
An AI agent mutates state (the repository) in order to get some metric (e.g. test suites) closer to passing. Same thing, really.
[ Parent | on HN ]
Re:Mathematical exploration and discovery at scale by woadwarrior01 on Saturday November 08, 2025 @04:06PM (#45859982)
An AI agent isn't an evolutionary optimization loop.
[ Parent | on HN ]
Re:Mathematical exploration and discovery at scale by adastra22 on Sunday November 09, 2025 @02:04AM (#45863531)
It is when you’re using it right. The most effective way to use an AI agent is to give it a metric for success (ideally tests, but even a description of success), a state upon which it can effect changes (the repo) and it moves the state towards success by executing tool calls and evaluating the success criteria at each step.
[ Parent | on HN ]