Yesterday we had the first lecture of the mandatory ethics course for doctoral students. Very excited and diverse group of 200 researchers, maybe not all super enthusiastic about the topic but they do recognise its importance. I think it’s great that the Council of Rectors of Finnish Universities recommended to make it mandatory to take research ethics and research integrity training for all doctoral students in Finland. And with this change, students can maybe teach something to their supervisors who might have taken their ethics trainings too many years ago (if they even took ethics trainings…).
The final task for the course is an essay. Write a reflection on the ethicality of your research, research integrity, or any other considerations on the topics that have inspired you during the course.
Yes … you read correctly. An essay. Now you’re like: “an essay? In the age of LLMs? Nobody’s gonna write an essay! Everyone will cheat with AI!! We can’t give essays to students anymore. Who will read 200 essays?? How can you control they don’t cheat?”
Well you know what? I love essays! And me and the other two co-teachers (the super amazing Essi Viitanen and Annukka Jyrämä) love to read all 200 essays. Sometimes we even share the good ones with each other and read them multiple times. Why should we get rid of essays as a form of assignment only because generative AI became a thing?
I asked a question to the students: raise your hand if you like to write essays… one person raised their hand.. it’s a pity! Because writing is the best method to give form and structure to your thoughts, even me in this unstructured post I can figure out what I wanted and will want to say to students and researchers when they ask me about AI.
Ok, I get it. Sometimes we get writers’ block. Maybe AI can help you suggesting you a structure or list a series of bullet points before you expand on them, but don’t let AI touch your words and write for you unless you’re writing some technical documentation of some software or protocol you’ve developed. We don’t forbid the use of AI in the course, but we ask our students to disclose the use of AI and follow the ALLEA European Code of Conduct for Research Integrity that they had to read as part of course materials (in the list of unacceptable practices there is “Hiding the use of AI or automated tools in the creation of content or drafting of publications”).
Yes, writing is challenging, communication is challenging, and giving form to ones’ thoughts is not easy, but who would argue that we should stop practicing the art of writing essays just because some generative AI can write plain soulless text that sounds like what you think is expected to pass the course but in practice even the manual of my dishwasher feels more interesting to read than a series of leverage harness delve into pivotal realms and so many em-dashes that I barely even remember the shortcut to type one and now they are everywhere, in which part of the training data did all these emdashes appear???
At the beginning of the lecture we teachers introduced ourselves. I showed this picture of me:
Privacy enhancing technologies are my cup of tea so I immediately wanted them to show the pseudonymised (poor pseudonomysation really) version of my face and the AI reconstructed one. Beware:! your anonymisation tool is flawed because AI can reconstruct what you thought was anonymous. Laugh!: my AI re-identified face had blue eyes and pretty eyebrows. Is that even me? 😂
And then I told them: this is what you get when you pass your essay drafts through an LLM gen AI tool to polish your text. You get fake blue eyes and pretty eyebrows and the true you does not emerge anymore from the essay. I told them: I don’t want to read perfect essays, I wanna see their real self in those essays, with typos and wrong English grammar because their native language is what it is and I want to hear those broken English sentences of real students with very different stories behind them coming from 100 different countries to tell me about their concerns and fears and hopes on the ethicality of their research and the impact they might have on society. I don’t need to read washed out standardised westernised essays and nobody really needs to read them really. Rather just give me your prompt and your broken draft and don’t waste time generating the polished fake output. I’d rather read 200 stream of consciousness internal monologues of future PhDs reflecting on the ethicality of their work and anything else around them, rather than another delve into harness and leverage emdash emdash and emdash.
Hopefully they got the message and they will have fun writing the essay without worrying about daily limits of whatever AI tool they were planning to use. Let’s see how many delve, leverage, and harness we will count by the end of the course…