Cesar A. Hidalgo at his own website:
Much of the public conversation about AI focuses on chat interfaces like ChatGPT. But a quieter revolution is happening in command line AI systems such as Claude Code or Codex. Unlike chatbots, these systems can act autonomously. Given a goal, they can download data, write code, install software, run analyses, and generate figures, working like a research assistant in a can.
When we pair these “research assistants in a can” with AI reviewer systems, we can explore what happens when feedback cycles that traditionally take months collapse into hours. This is leading to a new type of research workflow, one producing what I call AI generated papers, or AIGPs.
To explore this idea, I did exactly this. placed a CLI AI tasked with writing a paper in a loop with an AI reviewer.
My conclusion is that an AI tsunami is approaching the sciences, and while some researchers are running towards the wave with their surfboards, many are still sleeping on the beach. This essay is an attempt to explore this emerging technology using a hands on example, using this as an opportunity to reflect on what it means to be a scholar while sounding the tsunami alarm system.
More here.
Enjoying the content on 3QD? Help keep us going by donating now.