Using AI to accelerate your scientific research

I will present an interactive demo of Claude Code and OpenAI Codex. Bring any outstanding research questions that you might have. We will use these tools to first do a thorough literature review to position our research question against prior art. Then, we will experiment with which research methodology we want agents to pursue to address the research question. We will then collectively interpret the output, and, in the circumstance we think there is a novel result, ask the agent to write a manuscript end to end. We will then critique the manuscript harshly, see what survives, and then collectively reassess. We will push every code and every written deliverable to a GitHub repo. If the experiment produces a genuinely promising result, we will discuss what additional verification, human contribution, and AI disclosure would be required before it could become a manuscript.