I think this genuinely can be a good thing. An automated agent that a) tries to run the code to see if it does what it is supposed to do and b) checks it for conceptual or implementation errors (code can run but be very wrong nevertheless).
Andrey Chetverikov
Many journals already require authors to upload data and code. A near-future where an LLM systematically checks the analysis and the code of every published study is realistic. Believe it or not, I think this is the future of scientific publishing.
