Coding with Gemini: Accelerating Development in the Lab
Introduction We had several attempts at using AI—either for physics problems, for trading signals, or for general-purpose fuzzy automations. They worked well. We’ll skip those here. We put significant emphasis on understanding the inner workings of the systems we build—whatever “understanding” might mean algorithmically speaking (a concept that, between Charles Peirce and logicians, remains up in the air). We also admire the recent advances in LLMs, and we plan to deploy them whenever and wherever they are truly relevant. ...


