AI Informed Consent

This semester, I’m collaborating with a colleague from another university. We both teach third-year Japanese at our respective institutions, so we’re doing a project together to consider the utility of generative AI in foreign language classes. We’re having our students do a series of targeted AI assignments and we're planning to do a presentation about the project at an upcoming conference.

Recently, we were on Zoom working on a draft of an abstract for the project. One of the things we hadn’t decided yet was the title. We had a shared Google doc open and I suggested we do some brainstorming to come up with a list of possible titles. I wrote two candidate titles, and suddenly he dropped about ten different titles into the document. At first I was surprised that he’d come up with all those titles so quickly, but then I realized he had used ChatGPT.

I found the experience rather disorienting and I had several reactions in quick succession.

I don't want to collaborate with AI, I want to collaborate with you.

My first thought was, “I don’t want to collaborate with AI, I want to collaborate with you.”

Then I looked through the titles and thought that most of them were junk. Cheesy, trite, yucky.

I also thought about how I don’t want to farm out what is arguably one of the most important important tasks — picking a title — to AI. And I worried that seeing all those titles would poison the mind and make it harder to focus on what we think the actual presentation should be about. But now those titles were in our document, and I would have to read them and critique them.

In thinking about all of this, I was reminded that brainstorming and thinking work best when they follow a human timeline. I was still trying to come up with more ideas, having written down only two, but ChatGPT spit out four or five times that number before I could come up with idea number three. It felt a little like having a blockade thrown up in front of my own thinking process. Why keep thinking when AI is so much faster and appears to be so much ”better” at quickly coming up with ideas? This is a rhetorical question of course, although there are many answers to the question of “why keeping thinking,” not the least of which is that all of AI’s titles were junk.

An additional question is at what point in “using AI” can one no longer say the work is one’s own? This is a question a lot of people are asking right now, and certainly there are different ideas about where to draw the line. I hadn’t thought deeply about this question before because I haven’t been using AI for my own writing, so there wasn’t even a line to be drawn.

But now, even if I didn’t want to include AI in my own work, I was forced to. It’s like the expression about there being things that we cannot “unsee.” Those titles entered my eyeballs and my brain and now I couldn’t unsee them. And that meant I had to deal with them.

In the end, I talked with my colleague and he was happy to agree that we wouldn’t use AI for the work we do to present and write about this project. Which, granted, is a bit ironic, since the project is about guiding our students in using AI. But I’m ok with that.

The question of whether or not to use AI is something our students struggle with as well, especially in group work situations. Even if one student is committed to doing the task without AI assistance, they may find themselves using AI against their will because their partners are using it.

So not only do instructors need to be clear and explicit in their expectations for AI use (for example, in the syllabus), but all of us need to do this whenever we collaborate. The spread of AI suggests the need for a new kind of informed consent. We need to inform each other of our intentions and preferences for using AI, and we need to seek the consent of collaborators before using AI in projects involving others. If we establish some ground rules at the outset about whether or how AI might be used during the project, it will prevent surprises like the one I experienced and help make sure everyone is operating from a shared understanding.