Q: Can I submit a manuscript with a high artificial intelligence score?

What an interesting question. Artificial intelligence (AI) detection tools, especially free online versions, are not 100% accurate. They often produce false positives because AI tools are trained to provide text that is human-like. Unlike for plagiarism detection, it is very difficult to define parameters that are specific to text generated by AI. There are shortcomings in AI detection methods, and advanced models are increasingly making them obsolete.

To answer your question, let’s consider two scenarios:

(i) You wrote the manuscript yourself without using AI text generation tools

In this case, the high AI score is a false positive. Your writing is your own original work. You can mention to the editor that you happened to run your text through an AI detector and were surprised to find a high AI content although you did not use AI to write any part of your text. Be prepared to explain your writing process if questioned: keep records of your drafts and research notes as evidence of your work. You can add that other authors are facing a similar situation because AI detection tools are not accurate. An interesting example is how an AI detection tool flagged the text of the US Declaration of Independence, saying that 97% of this document was AI generated!

(ii) You used AI tools to assist in any part of the writing process

In this case, be transparent about your use and specify which AI tool you used and for what purpose (e.g. for editing, paraphrasing, or generating initial drafts). You could indicate the portions where this was done. Honest disclosure is mandatory. Most journals now have policies requiring disclosure of AI use.

Depending on whether or not you used an AI tool, make the necessary declarations (as described above) and go ahead with the submission. Good luck!