How can I teach ethics of AI use?

Hej, great question!

Since it sounds like your course is not about ethics or AI directly, one way to address the topic is to not make it an explicit unit (which might seem disconnected from the rest of your course, and therefore possibly not very relevant to your students), but to bring up the topic with smaller comments or considerations over and over again, whereever it seems appropriate, and from whichever angle you are most concerned about. That could mean to mention data center water use in regions that are fighting draughts already without the added stress, or mental health effects that training LLMs has on people doing the training, or how intellectual property is being used without consent. These could be just one or two sentences each instance, combined with the willingness to discuss each a bit more if students have questions. And once you start looking for connections between your subject and AI ethics, there are probably many instances where it is actually not far fetched to address the topic explicitly!

If you do want to make AI ethics a bigger topic because it deserves that attention, there are many resources available that you could lean on for inspiration, for example

• https://teachingaiethics.com/resources/ is a database that you can filter by 25 different subjects and eight ethics topic (bias, copyright, data, emotions, environment, human labour, privacy, truth) and then get case studies and articles to build on
• The podcast episode https://creators.spotify.com/pod/profile/podnetwork-podcast/episodes/What-If-AI-Ethics-Were-a-Team-Sport--With-Katie-Chiou-and-Lindsay-Doukopoulos-e3oak5t discusses the idea of an "Ethics Bowl", a debate event where teams compete on having throught through ethics topics in the most comprehensive way. They also present more collaborative round-table approaches if competition isn't what you want to do...

Why not have students suggest 'dilemmas' that they themselves encounter with respect to AI use? Maybe have them spend 5-10 minutes of a lecture to discuss in pairs, and let them submit such dilemmas (anonymously) using some Padlet or Mentimeter system. Discuss a few of these dilemmas with the class (without aiming at 'solving' them), and then let them work on their 'favourite' dilemma in small groups for 10-15 minutes. After this, have a group discussion, before asking them to submit their reflections using the same system (here Padlet really works well, as one can post 'comments' to the posted dilemmas). Share these with the students and write some type of "teacher's reflection". Be humble and explicit about neither you, nor anyone else, having good answers to these dilemmas, and it the real learning goal is to reflect on these issues. If you do this in a couple of courses, and maybe discuss with colleagues doing the same thing (and even reading what researchers have to say about this -- and people in other fields), then you will maybe develop some type of competency on the matter -- and maybe you'll end up giving expert advice to others on these matters :-)

I think it's good to be transparent with the student in letting them know that there is no easy answer to 'what is ethical use of AI'. I recently had a focussed seminar on ethics in engineering for first year students and started off by having students actively discuss in smaller groups ethical dilemmas that they have experienced, heard of, or could imagine to be relevant for them as students. Generative AI use came up in some of the examples mentioned and it was clear from further probing that many of the students were well aware of the ethical conundrums that AI can impose on us. As the rest of the seminar was focused more specifically on preventative ethics in an engineering context, we were only able to scratch the surface of the ethical issue associated with AI (I think next year I will try to direct some more focus on this topic). In any case, I realized that having these types conversations, and normalizing them both inside the classroom (with our students) as well as outside (with colleagues) is an important prerequisite to actively working towards a shared best practice for what is ethical use. In this paradigm, we as teachers maybe also need to let go of the necessity of always being the experts on the topics we bring up with students. I would prefer that students actively come to me with their questions and concerns about what is ethical AI use, us having an open discussion about it - weighing various pros and cons - without providing a definitive answer. Hopefully over time we may come up with clearer answers to what is ethical use of AI. In the meantime, I think it's important that we all talk about it!