Hi Scott,

Thought provoking post! I have a couple of thoughts:

1. When you say that ChatGPT is only good for rote copy editing, what else have you tried? Personally I think that it's coding is remarkably good. Not perfect of course, but saves hours of time and suggests ways of doing things that I didn't already know. It's also useful for outlining lectures for e.g. undergrads, or even MSc/PhD students. If you ask it to outline "10 lectures on introduction to using bioinformatics for public health microbiology" or "the role of the microbiome in colonisation resistance" I'd argue that it does a better job than most MSc or PhD students could do in a day or two of work. Again, you have to tweak it of course, but much better than starting from nothing.

2. Secondly, I'm not sure about the characterisation of what it is doing as plagiarism. Perhaps in a scientific way, where ideas have to be credited to their originator, then presenting the output of ChatGPT as your own work without any further reference would be bad practice, bordering on plagiarism. But, I don't think most people would do this. It's like wikipedia, can be very helpful, but you still need to dig into the primary sources for scientific purposes. Beyond science, I don't really think it's plagiarism anymore than a writer who writes in a similar style to another is plagiarising that person. If I ask you to write a song about malaria in the style of Dolly Parton (as I did with ChatGPT), and you do a good job (as ChatGPT did), then you're not plagiarising Dolly Parton are you?

Personally I can't imagine it being useful for writing any part of a scientific paper other than maybe the first couple of paragraphs of the introduction, and actually, I haven't found the copy editing that helpful either. I could imagine it being helpful for motivation statements for things, or pathway to impact statements and things like that though.

Expand full comment