Weekly digest (01-07 May 2023)

Opening up, or closing down, your options
I don’t remember if I saw this list somewhere, or if I made it up.

I’ve come across the following ways of thinking about generative AI and language models in higher education and professional practice:

1. Language models are a threat we need to protect ourselves against. We have to keep everything as it was and lock down our infrastructure to make sure there isn’t an explosion of student cheating (although very little concern about all the ways academics could abuse this technology).
2. This is an opportunity that we need to adapt to. Our systems and processes will essentially stay the same, but assessment practices for example, may need to be updated. No more essays.
3. This is going to change everything. Generative AI is likely to have a bigger impact on higher education than the internet did. Computational curricula that adapt in real-time to students’ personal learning needs is something that comes to mind.

My initial thought was that this feels almost like a progression through stages (maybe something similar to Gartner’s hype cycle). But then I realised that it’s more like a paradigm or worldview. These are perspectives that people have, and is probably linked more to personal identity than faculty development. If it was a developmental process, then you can more easily see how to help colleagues move through the stages but an identity shift is much harder to bring about.

Your thoughts about language models are probably influenced by how much you feel they’re a threat to your professional or personal identity. If you see them as augmenting what you do, or as opening up new spaces to explore, you’ll probably use them to increase your options. But if you see them as a threat to your work (and identity), I can see how you may want to shut down the technology.

But if this your position, I believe that you’re closing down your options.

Mr. Ranedeer: Your personalized AI Tutor!

This is an early example of the kinds of personal tutors we’re going to see popping up everywhere. This is still quite a technical solution, so is unlikely to be taken up broadly, but soon this will be built into user-facing software interfaces. No more help files, or clicking through menus to find what you need. Anything that has a learning component will have some level of personal tutorial built into it.

Unlock the potential of GPT-4 with Mr. Ranedeer AI Tutor, a customizable prompt that delivers personalized learning experiences for users with diverse needs and interests.

Is connecting with machines different to connecting with people?
I know many will balk at the idea of replacing human tutors with AI tutors. They’ll say there’s something special about the personal connection that you get in a tutorial session; that it provides the pastoral care that’s hard to get at in the classroom. And I agree completely.

I’m just not sure that there’s anything special about human connection. Social interaction is one of our primary drives as human beings, and I’m not suggesting that we become more isolated; we absolutely need to interact with others. However, I’m not convinced that those interactions need to be with people.

Earlier this week a colleague shared a video of a service user taking a bat to a robot receptionist at a healthcare facility. I shared some initial thoughts on first watching the video, where I concluded that we see whatever confirms our existing bias in those kinds of interactions. But I also shared a few thoughts on what may have led to different outcomes. And I realised that the outcome doesn’t depend on who the interaction is with. The outcome is dependent on how well our needs are met.

Coming back to tutorials in the university, I don’t think we’ll have any issues with AI tutors to support learning. The only thing that will matter is how well the AI tutor is able to satisfy the learning needs of students.

And if I’m honest, I believe they’ll do a better job than we do.