Research Fundamentals with AI Agents
9/9 parts published
Notes on working with AI agents across personal development, research, and career decisions.
9/9 parts published
Keep verified facts, working notes, open questions, and current decisions separate so you can revisit evidence, avoid repeating work, and use AI agents without treating provisional ideas as evidence.
How I use AI agents as the default for research while still using free chat-based generative AI, based on context, evidence, records, and usage allowances.
Use generative AI discussions to check questions, evidence, and alternatives, then turn the results into working notes, provisional decisions, and the next research question.
When brainstorming with generative AI, do not mistake a plausible first option for a reason to choose it. Check necessary properties, assumptions, alternatives, and evidence.
Use records, links, and history to build a research foundation that lets you and AI agents trace evidence, decisions, and changes.
Use titles and abstracts to decide which papers deserve attention. Use AI summaries only for those papers, then verify claims in the original paper.
Connect a research question, comparison, metric, and conditions before starting an experiment, so that each possible result can support a clear conclusion.
Learn the practical differences between peer-reviewed papers, preprints, journal articles, and conference papers. Use publication type as a first check, then examine the paper itself.
Use review papers, citation trails, and English search terms to build a path through the literature. Use AI to find candidates, then verify them in the original sources.
Before looking for a method, separate the social problem, technical problem, research question, and necessary properties. A practical order for keeping research on track.