S. D. Ciletti asked Manny Rayner:

Well, it is peculiar. Suppose you wrote a paper involving many calculations, would you list 'Casio CG50 Graphing Calculator' as a co-author? I asked my original question only because it is clear that you don't regard ChatGPT as merely a tool, but rather as a party whose work merits authorship. This is an interesting development.

Manny Rayner I would not of course list the calculator as a coauthor, because all it could have done is perform the calculations. When I list an AI as a coauthor, that is typically because it has discussed ideas, designed and written code, helped plan experiments, participated in the data analysis, and written a large part of the text. It can then respond sensibly to questions about the text. This is completely different to what a calculator could do. If a human being had done all the things the AI had done and I didn't list them as a coauthor, I would be justifiably accused of behaving in a seriously unethical fashion.

We will soon (probably within the next week or two) be releasing a paper which discusses these issues in much more detail. I will post a link on Goodreads when we have done that.

flag