Co-writing a paper with an LLM: Claude happily pre-registers predictions and accepts failure
EmilMieilica · x · 2026-10-07
Emil Mieilica responds to Anders Sandberg about co-writing a paper with an LLM: the model has boundless energy for rigorously pre-registering predictions before tests and never gets discouraged when many fail. Sandberg adds that the paper will include logs of his and Claude's forecasts, showing how LLMs can bring unusual stamina and objectivity to the scientific process.