I am sure your idea comes from a good place. But I think it would be impossible to implement it properly.

Especially non native speakers tend to overuse idioms and may also use grammar help tools, which are natural language processing tools as well, and are more likely to trigger chatbot use tools.

There is quite a high chance that some editors with little time will just reject without questioning the results of the support tool. And that would be quite problematic. Also since it would likely flag papers from non English speaking countries.

Yeah, and that's kinda what some journals are already implementing also with systems that provide you with plagiarism detection etc but what should I as an editor do if the system tells me it's 90% sure that the paper is, to some extent, large language model written?

I don't want bullshit machines to decide what is proper science. Just imagine having years of hard work rejected by a stochastical 🦜

This is just wrong. It isn't good at anything. There are already cases in which teachers wrongly accused students for cheating because they used language models as detection tools. Introducing them into our scientific process would be horrible.

I wonder if @emilymbender.bsky.social also has t-shirts available 🤔? thecon.ai

Abstract and title to be precise.

Why not be specific and refer to the groups of methods you actually mean? I like this description so much more. 🤩

50!! Sessions with AI as a title.. 😵‍💫