How do you use LLMs for your research?
I'm doing several researches such as trading, ML which are domains that I'm not really familiar for. however, I think problems I want to solve is relatively clear.
In these cases, how do you guys use LLMs for them to find kinda "optimal" solution ?
I wouldn’t use an LLM as a one-off chat. I’d turn the research into a long-running Codex or Claude Code project, keeping the facts, findings, decisions, and actual outcomes as project documents.
There usually isn’t an “optimal” answer from day one. But as the project builds a history of what you believed versus what actually happened, the LLM can learn from it, refine its judgment, and gradually get closer to the optimal solution.