Use of AI Tools in Your Scientific Work

Artificial intelligence tools such as large language models have become widely available and can serve useful functions similar to spell checkers, grammar checkers, and even evaluate writing style. Using these tools to polish prose, catch typographical errors, or improve the clarity of sentences you have already written is generally acceptable. I sometimes use Claude to check this lab manual for spelling mistakes, typos, and grammatical errors.

However, in our lab it is not appropriate to use AI tools to perform the intellectual work that lies at the heart of our work as scientists. This includes synthesizing ideas from the literature, generating research questions or hypotheses, critically evaluating papers, drafting the content of a research paper, or grant proposal, or formulating arguments and interpretations. These kinds of tasks are precisely what you are here to learn to do, as scientists in training.

The purpose of graduate and postdoctoral training is to develop your capacity for independent scientific thought, and to train you in the skills necessary for scientific work. That capacity grows through intellectual struggle, through the slow and sometimes frustrating process of reading deeply, thinking carefully, writing, and rewriting, and in doing so refining your ideas. These skills include reading papers, synthesizing information, writing prose, and writing code. Your supervisor and your other mentors play a crucial role in guiding your intellectual development, through discussion and critique.

If you outsource your mental work to an AI tool, you deprive yourself of the experiences that build expertise and intellectual maturity. Science advances because human minds engage seriously with hard problems. Our aim in this lab is to cultivate that engagement, not to shortcut it.

It is also important to consider that privacy and intellectual property safeguards differ significantly between commercial AI tools, and also between free, paid, and enterprise versions. When you input your research ideas, data, or draft text into AI tools you may be disclosing sensitive information to third parties. It may not be possible to determine how that information will be stored, used, or protected.

Using AI tools to edit drafts of prose (abstracts, reports papers, theses, etc) and/or to edit code, is a reasonable thing to do. I think of this as not very different than going to talk to a fellow human and get their comments/suggestions. I may or may not take them up on those suggestions, and I may use those comments to come up with my own changes. This is OK. It is appropriate and perhaps even necessary to attribute your use of AI in this “editor/suggestion” way of working with it.

What is not OK is to use AI to generate prose, or code, and then pass it off as your own, unattributed. It would be the same as asking a fellow human to write a paragraph of your thesis, you read it, and think “OK seems good” and then pass it off as your own work, unattributed. This constitutes academic fraud and is an academic offense here at Western.