Like all educators these last few years, we are figuring out how to handle students offloading their learning to generative AI and our corresponding challenges in having to read the AI's writing instead of our students. I wrote this as a statement for my students as guidelines for writing their graded MSc literature study and thesis. My lab gave nice feedback on my first drafts and this is the current version present in our lab guide:

Engineering Thinking

We ask you to write reports in school so that you learn how to write. You may not see it yet, but writing is also a key to improving your engineering thinking. The act of communicating each and every sentence, helps you think out the complexity of the engineering you are doing. If you let an LLM write for you, you remove a large amount of the thinking you need to be doing to become an engineer.

We cannot assess whether you can write if an LLM did any of the writing for you. We may be able to asses whether you can use an LLM, but that is not the same as assessing whether you can write.

LLMs can aid you during writing, in the same way discussing a paragraph with a person may aid you, but the words on the paper of your reports should be those that came from your brain and your fingers. When you are outside of school, you can use any tool you desire to produce the thing you are making. But in school, we are trying to get you to learn specific things and thus we put boundaries on what you can use in different tasks.

For example, you might be learning about integration of a mathematical function. Almost anyone can type a mathematical function into Wolfram Alpha and ask it to integrate the function but that does not teach you what integration is or means or what logic may produce Wolfram's result. You also have no idea if Wolfram gave you a correct answer or not. We teach you integration in almost the same way Newton and Leibniz taught themselves 400 years ago for a reason.

Generative AI is a tool that you can use to help you in your work. But it should not be used to directly produce content we use to assess specific learning objectives. The internship report, literature study report, and thesis report are tasks in your MSc designed to teach you to write through manual practice and feedback.

Writing in English

Using an LLM to help you write in English (whether it is a second language or your mother tongue) is useful. It can check spelling, grammar, translations, and other legibility aspects. We recommend using these tools to give you suggestions for such fixes and then you evaluate and incorporate them under your interpretation. Automatic translation saves times, but it does little to teach you a language, only using the language will. Once again, if the LLM does the writing for you, you lose the opportunity to learn how to write.

Take Grammarly, for example: it is a tool for spellcheck and grammar. It produces suggestions and explanations of in-line grammar fixes to help you to write correct and engaging texts while at the same time teaches you to write better. But it doesn't generate the whole text for you.

Lastly

If you cannot be bothered to write the text yourself, should your supervisors be bothered to read it for you? Your supervisors can just as easily talk to an LLM themselves if they want to read something it produced. Our role is to assess whether you can do technical writing and communicate your ideas and work yourself.

Permitted	Non Permitted
type a paragraph, provide it to a LLM and ask for adjusments and feedback, manually rewrite the paragraph based on the LLM's feedback	type a paragraph, provide it to a LLM and ask it to improve the paragraph, paste the paragraph from the LLM into your report
use the LLM to translate a word, sentence, or sentence fragment as you are writing because you are struggling to get the write words and phrasing in English	write your report in your mother tongue, ask LLM to translate it to English, turn in the translation for assessment
ask the LLM to check the grammar in your report, manually fix grammar that is suggested by the LLM	pass your report to an LLM and tell it to fix all grammar and spelling, turn in the updated report

Statement On AI Assistance

If you make use of writing tools that can essentially write text for you you must make a statement in your document explaining how you used such a tool. You should also be cautious when using such tools because the copyright of the text that it produces is ambiguous. The text is a derivative of other people's (copyrighted) writing and you may not want to laden your writing with possible copyright infringements. Nonetheless, make a statement on the use of the tool, for example:

During the preparation of this work the author(s) used ChatGPT 12x in order to improve the grammar in the writing. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.

More information on use of AI in academic works..

The writing in the MSc theses (particularly of my English-as-a-second-language students) has noticeably improved in the recent year or so. The AI disclosures are typically so vague it isn't easy to gather the extent of the AI use. Pangram, if trusted, supports my suspicion of heavy AI use. I've also had a student provide our guide to the generative AI tool so that it does anything to assist the writing without technically violating anything in the guide.

Prior to generative AI, a student could have someone else write their papers for them, but I think this was quite rare at the schools I've taught at. Mostly because the effort, cost, likelihood of being caught, and consequence were much higher when having a real person write for you. Given the significant lowering of risk or cost, I fear that many (or most) students will have no problem having something else write for them. This is a sad realization, as I have long believed the integrity of students to be high and effectively a constant. If integrity is a function of convenience, then my guide above may not have much effect.

I share this to see if any feedback emerges more broadly. I guess many others are writing such things for their students...or maybe not?