"As Joel Pick from the Chair of the Advocacy Committee and myself, along with 22 data editors from six journals, we've written some guidelines for standardized data and code editing. One of the sections that we have is about whether the data and the code match what's written in the article. Currently, there are some journals that use an AI program that essentially scrapes the methods section and then matches the data with the methods. It would make natural sense for there to be something similar for code. I do think AI has its place, but I'm always quite reticent to say you should use AI to do it. Because whilst AI is very good at describing code, and it can tell you, like, OK, this code is running this loop, it doesn't really tell you, A, why you're doing it. And it can't also link it to the manuscript, as in saying, like, this produces figure one. I do think humans have to be involved in terms of, like, double-checking everything."