Floating while sinking

In the mid-2010s, I spent a considerable amount of time wearing a computer on my face. This was not a fashion choice, or if it was, it wasn’t a successful one. I was a PhD student at the University of Minnesota, and I had fallen in with the Wearables Research Collaboratory, a group of humanists and social scientists who suspected that Google Glass, the Apple Watch, VR headsets, and a folded piece of cardboard with a phone tucked inside it might have something to teach us about writing, learning, and communication. By then, Glass had already earned its wearers an unflattering nickname. We wore it anyway, for science, or at least for the humanities, which is a lot like science but with more footnotes.

We tried these devices in classrooms and studios, in writing assignments and design projects, in ways that ranged from genuinely illuminating to mildly embarrassing. And every project, sooner or later, ran into the same awkward question: Does it work? A department chair wants to know. A grant report wants to know. Students want to know, especially if it’s going to be graded. For a lot of what we were doing, the honest answer was that we didn’t know yet, and we weren’t sure “work” was even the right word. Is a VR assignment effective if it produces clumsier prose but a much richer conversation about audience? Is a pair of smart glasses a classroom failure if that failure teaches everyone something about surveillance and social norms?

So we learned, sometimes grudgingly, to hold off on the verdict. We let our assessment criteria stay loose for longer than felt respectable. We let conversations run for weeks, sometimes months. We tried things, watched what happened, argued about it, and tried again a little differently. Ideas got time to percolate, and some of them eventually crystallized into courses, articles, and design principles. Others quietly dissolved, which was also worth knowing.

Somewhere in that process, we started calling it “productive ambiguity.” We’d like to say we coined the term collectively, and in a small, local sense I suppose we did, the way you might coin a nickname for a friend that half the world already calls something similar. I’m sure psychologists and creativity researchers got there first, much as “flow” was a feeling people had long before Mihaly Csikszentmihalyi gave it a name. John Keats praised “negative capability,” the capacity to remain in uncertainty and doubt without any “irritable reaching after fact and reason.” HCI researchers argued back in 2003 that ambiguity could be a resource for design rather than a defect to engineer away. We were hardly the first. But for our little collaboratory, the phrase became something like an operating mantra, the thing we said to one another when a project felt stuck, which more often than not meant it was about to get interesting.

The best way I can describe the feeling is floating while sinking. You aren’t drowning, exactly, but you aren’t standing on anything solid either, and you can’t quite tell which way is up. It’s an uncomfortable place to wait, and academics, who are trained to have an answer ready before the question is finished, are especially bad at waiting there. Yet the things you notice while you’re suspended (a student using a device in a way nobody anticipated, a headset that leaves half the room queasy, an offhand remark in a debrief that reframes the whole project) end up shaping the decisions you eventually have to make. A lot of the real work happens in that murky middle. It just doesn’t photograph well for an annual report.

Photo by takahiro taguchi on Unsplash

Everyone is floating now

I’ve been thinking about that phrase again, because it feels like all of higher education is currently floating while sinking.

Generative AI arrived on campus the way weather does, all at once and without asking. Agentic AI is following close behind, with tools that don’t just draft the email but send it, schedule the follow-up, and file the report while you’re off doing something else. Universities have responded with task forces, syllabus templates, detection software of dubious reliability, and policies drafted in August that feel quietly outdated by October. Workplaces are running their own version of the same scramble, and so is nearly every other institution that runs on writing, which is to say most of them.

I understand the demand for rules. Faculty need to know what to put on the syllabus before Monday. Students deserve to know what counts as misconduct before someone accuses them of it. Organizations need something firmer than vibes to govern tools that can act on their behalf. “Let’s wait and see” is not a policy, and I don’t want to pretend otherwise.

But I worry about how quickly we’re trying to settle questions that haven’t finished forming. The technology changes by the quarter. Our understanding of what it does to learning, writing, and work changes more slowly, and it should, because that understanding has to come from watching what actually happens in classrooms and offices rather than from what the product demo promised would happen. A rule written in a hurry tends to encode our anxieties more faithfully than our values. The tech industry spent two decades telling us to move fast and break things. There’s an irony in how often our response to its latest product has been to move fast and write rules, which can break things too.

What I’d like to see instead is policy that treats ambiguity as something to design with. That might look like provisional guidelines that come with a revisit date, or course policies that bring students into the conversation instead of handing them a list of prohibitions. It might look like institutional frameworks that say, more or less, “Here’s where we stand today, here’s what we’re watching for, and here’s when we’ll come back to this.” That kind of slow, intentional deliberation is close kin to what I’ve been writing about elsewhere as deliberative restraint. It still gives people something to stand on. It just admits the floor might move, and builds in a way to renegotiate when it does.

There’s a lot of restless energy in this moment. Some of it is fear, some of it is hype, and some of it is the real excitement of people discovering they can do things they couldn’t do a year ago. I’d rather not rush to pour all of it into a rulebook. Held open a little longer, and watched carefully, that energy may prove productive in ways we can’t see yet.

A very small teacher

Then, recently, our family grew by one very small person, and I learned that most of what I thought I knew about ambiguity had been theoretical.

Newborns don’t read the literature. They don’t follow the schedule the book recommends, or they follow it for about three days, just long enough for you to believe you’ve figured something out, and then they renegotiate the terms. Our daughter has shown no interest whatsoever in my preference for clear assessment criteria. Each day asks for small inventions: a new way to settle her, a new division of labor in the middle of the night, a revised sense of what counts as a good day.

What I’m learning, slowly and on not much sleep, is that the open-endedness isn’t a problem standing between me and some orderly version of parenthood. It’s the actual situation. And it has been oddly good for me. I’ve had to become more agile in how I communicate, both with a person who can’t yet use words and with a partner who is every bit as tired as I am. I’ve had to get better at reading emotion, my own included. I’ve had to be more honest about what I can control and what I can’t, and quicker to accept the second category without taking it personally. The sentence I find myself saying most often is “I don’t know, let’s try something and see.”

That, it occurs to me, is exactly what we used to say in the collaboratory. It’s also a sentence I wish more AI policy committees would say out loud.

I don’t want to get too precious about this. A baby isn’t a metaphor, and parenting isn’t a professional development workshop. But the habits it’s forcing on me do carry over. Scholarship rewards certainty: the clean argument, the tidy contribution, the confident reply to Reviewer 2. Faculty life rewards it too, in the form of rubrics and strategic plans and five-year goals. Some of that is necessary. But when I look back, my most interesting research, and probably my best teaching, came out of stretches when I didn’t yet know what I was doing and was willing to stay there long enough to find out.

Some lessons from the deep end

For those of us working somewhere at the crossroads of rhetoric, technology, writing, communication research, and design, I want to offer a few thoughts on what this looks like in practice, with the humility of someone who hasn’t slept through the night in a while.

It starts, I think, with naming the ambiguous phase out loud. If your project, course, or grant has a stretch where you genuinely don’t know what you’ll find, put it in the timeline and call it what it is. Reviewers, administrators, and students tend to be far more patient with uncertainty that has been announced than with uncertainty that shows up unannounced halfway through. Naming it is only half the job, though, because productive ambiguity is only productive if somebody is paying attention. Keep a design journal, hold short debriefs, write down the strange thing a student said on the way out the door. What you collect while you float becomes the evidence you lean on when it’s finally time to decide.

The same spirit can shape the rules we write. If you’re drafting an AI statement for a syllabus, a program, or an organization, write it in pencil: give it a revisit date and a process for revision, and consider drafting it with the people it will govern. A policy that expects to change is more honest than one that pretends it won’t.

None of this means ambiguity comes free. Uncertainty is a luxury, and it isn’t evenly distributed. A tenured professor can afford to float for a semester; a first-year student facing an academic integrity hearing, or a contingent instructor worried about course evaluations, often cannot. If you’re the one with power in the room, the ambiguity is yours to hold, and passing it downhill to the people least able to absorb the risk defeats the purpose. Give them clear, if provisional, ground to stand on while you keep deliberating.

Finally, it’s worth protecting the conversations that don’t produce deliverables: the coffee that runs long, the reading group that wanders off topic, the hallway argument nobody wins. Institutions have gotten very good at trimming these in the name of efficiency, and AI tools will only make the trimming easier. But that’s often where ideas do their percolating. It’s certainly where most of ours did, back in the collaboratory.

Under our roof, a very small person is working out the world one uncertain hour at a time. She seems to be doing fine, and she certainly isn’t waiting for solid ground before getting on with it. Floating while sinking is still not a comfortable place to be. I doubt it ever will be. But I’ve stopped treating it as a failure to reach solid ground. More often than not, it’s where the ground gets made.

Banner photo by Yunshuo Qu on Unsplash