Find the relevant discussion for your question

People tend to talk (and think) about how AI responds to questions they asked it, how it helps them do stuff, etc, as if AI/it was a person -- which it is very clearly not. What are ways to talk about AI that make it clear that there isn't some little helpful and knowledgeable person sitting inside people's computers and phones, but that outputs are just based on probabilities calculated by some algorithm that is trained on some data set in certain ways? I feel like the kind of language we use to talk about AI has an influence on how we think about it and I don't want to do that unreflectedly.

Yesterday I got into a heated discussion with colleagues over whether we should make it mandatory for students to disclose their AI use in assignments. Arguments from one side were that we need to know what they are doing, and also that having to disclose AI use will make them reflect on how they are using it. Arguments from the other side were that mandated disclosure does not mean that all students really tell us what they are doing, but rather that some will feel a lot of stress to disclose to the tiniest detail, while others will just write what they think a teacher will want to hear. So if we cannot trust the information they provide anyway, why put the stress on them. While I really want to know how my students are using AI so I can discuss it with them and make sure they use it in ways that will (likely) not harm them, is mandatory disclosure really the way to go, or what other alternatives do I have?

Even though this is really not what my course is about, I want to do something about the ethics of AI. Where do I start?

My colleague -- who I respect very much -- is "forcing" students to use AI products. He says that he is doing it because he wants them to learn what AI can and cannot do, which he considers a critically important skill for their future. Some of the students do not want to use AI on principle because of environmental and social concerns, which I believe are valid reasons for refusal (and which, I think, my colleague actually sympathises with, too). Nevertheless, he is making some AI tasks mandatory. Is that a good idea? And can he even do that?

Hej! We as teachers are encouraged to explore uses of GenAI in teaching, and obvious uses I see are to check texts that I have written for grammar, spelling, readability, or to get feedback on materials I have created. But are there any really useful tools that can do tasks like setting up a Canvas page from a course plan, or look at a schedule of assignments in a course and block appropriate amounts of time on my calendar around them so that I have time to provide careful feedback, or look through an existing Canvas and plan the talking points for weekly meetings with my Teaching Assistants to prepare them for the upcoming week, both content-wise and pedagogically? Or what are other good uses, what are the appropriate tools, and how do I use them? Any experiences or recommendations?

That’s it. Any advice?

The other day, I asked CoPilot something related to my career progression, and the output addressed me by name and knew what type of contract I am currently employed on, which I found quite scary. I know that we aren't supposed to upload student names and other sensitive data to GenAI, so the university seems to have some data privacy concerns there, but at the same time, if the university-provided (and, I am assuming, approved) tool can access my files and emails and browser, I would assume it can access all that data anyway. What happens to this data, what happens to the questions I ask, the input I provide? Can anyone see all of it directly, does it somehow become part of the training data, ...? I am concerned!

A student approached me and asked whether they could connect Claude to Canvas, as they are used to studying with Claude and this would allow them to not change their habits. I don't know what that would look like in detail, but to me it appears that it is going too far?

Hi! I keep hearing that we are supposed to write guidelines for AI use for students, but I have no idea where to start. All colleagues I talk to are super vague and nobody seems to be willing to share what they do with their courses. I don’t even know if they actually have anything written or if they mostly tell their students general things either like “just don’t do it!” or “do whatever, I don’t care”. Is there anyone out there willing to share what they do?

What does it mean when a thesis supervisor says "You can use GenAI to polish the text, but not for anything else". What are the limits of polishing?

I am highly sceptical of generative AI and don't use it myself, but I know there's a general impression among students that it 'knows everything'. The discussions I have seen have focused on text (logical enough for 'large language models'), and images (many of which are disastrous), but I cannot see how LLMs can reliably do (complex) mathematics where there are many steps to a derivation, some of which are nuanced or require particular systematic techniques, and the final result should be as general as possible. How can I convince students that they will (probably) waste more time using generative AI than working through the difficulties themselves? Or do I have to resign myself to allowing an extra week or more for students to learn for themselves that there really isn't an alternative to putting in the effort and doing the derivation oneself with pen and paper, no matter how hard it is?

I've seen a lot of discussion, mostly in social media and teaching fora, about students using generative AI instead of doing the reading and writing assignments that have been central to university learning. But I've seen little discussion of generative AI in physics or other natural sciences. How worried should I be that some of my students will put their effort into learning to use generative AI rather than learning physics? How good are large language models at physics? And how are things changing as the models are improving?