It took me longer than it should have to get to the University of Mississippi.
Forty-eight states came and went before it. Mississippi stayed somewhere on the list, not avoided, just never urgent enough to move to the top. Until last week, when I finally arrived at Ole Miss for their International Education Week.
State number forty-nine. Campus number “I’ve lost count”.
After a while, campuses begin to resemble one another. Students start to look alike. Conversations fall into patterns you can anticipate before they unfold. You begin to recognize the rhythm before anything has actually happened. And once you recognize the rhythm, you stop expecting to be surprised.
That was the mindset I arrived with.
Familiarity has a way of convincing you that you already understand something—when what you really understand is the pattern.
It didn’t take long to realize that.
Amongst other things on the agenda, I was there to deliver a keynote on navigating the age of AI, a version of a talk I’ve given many times over the past year. The argument is familiar by now. The world that most of us were prepared for is changing faster than we expected. Work is becoming less predictable. Knowledge is no longer scarce. And the value of what we learn is shifting—from what we know to how we think.
If the value we offer is built on repeating what already exists, that shift should concern us. If value is built on imagining what does not yet exist, it opens a different kind of door.
That was the argument I made in the room.
When the talk ended, no one moved.
There was no polite exit, no gradual thinning of the room. Instead, people stayed. The conversation that unfolded was unscripted and, at times, unresolved. Students, faculty, and staff remained in the room, engaging with each other.
One faculty member spoke about empathy—not in the abstract, but in the context of what it means to connect in a world where interaction can be simulated. A student pushed on the environmental cost of AI, questioning the trade-offs we rarely acknowledge. Someone else raised the idea of cognitive offloading—not as a convenience, but as something that might be reshaping what we are capable of doing ourselves.
The questions weren’t what made it different. It was the willingness to stay with them long enough for them to matter. No one rushed to resolve anything. No one tried to sound finished. People listened, reconsidered, responded, and then paused again. There was no pressure to arrive at a conclusion, only a shared commitment to remain inside the conversation long enough.
That kind of engagement is rarer than we’re willing to admit—and it deserves to be recognized when it shows up.
Somewhere in the middle of that exchange, I began to realize that I had been paying attention to the wrong thing—not just in that room, but in most rooms like it. I had been drawn to the answers themselves, to how clearly they were expressed, how well they were structured, how quickly they arrived. It had become easy to assume that clarity meant understanding, that fluency was a reliable signal of thought.
What I walked into that evening with was familiarity. I expected the evening to follow a pattern I’ve come to recognize over time—but it didn’t. And once that became visible, it became difficult to ignore how much of what we now accept as normal is simply the result of having seen it too many times.
If students are capable of that level of engagement, without a grade attached, without an assignment driving it, without any external incentive to stay, then what exactly are we asking them to do the rest of the time?
Much of the conversation around AI in education has settled into something predictable. It focuses on misuse, on boundaries, on whether the work being submitted still belongs to the student. Those concerns are real, but they assume something we have not examined closely enough—that the work itself still measures what we think it does.
What I saw in that room challenged that assumption.
If a student can complete an assignment using AI without losing anything essential in the process, then the issue is not only that the student used AI. It reveals that the assignment may not have required what we believed it did in the first place.
For a long time, we have treated output as evidence of understanding. A well-structured essay, a coherent argument, a polished submission has been taken as proof that thinking has occurred. That belief made sense when producing that output required time, effort, and iteration. As that effort becomes optional, the reliability of that signal begins to erode.
The instinct, understandably, is to regain control—to detect, to prevent, to reinforce boundaries. But control does not strengthen what is being measured. It only protects it.
The more difficult question is whether we have been measuring the right thing all along.
Because if output is no longer a reliable signal, then the structure around it has to change as well. Not toward more control, but toward different kinds of evidence—what can only be demonstrated over time, in conversation, under pressure, in response to something that cannot be predicted in advance. Not just answers, but the ability to stay with a question long enough to understand it, defend it, and revise it in real time.
The kind of thinking I saw in that room cannot be downloaded, submitted, or outsourced.
It has to be lived through.
And that is where the tension begins to extend beyond education.
It’s not just students. The same shift is already happening everywhere else—in how quickly we arrive at answers, how fluently we express ideas, and how easily we accept outputs that sound complete even when the thinking underneath them is still forming.
The space between encountering an idea and expressing it used to demand something from us—uncertainty, friction, the patience to stay with something before it made sense. As that space narrows, nothing immediately appears broken. The work still gets done. The output still arrives. But we spend less time inside the process that produces it.
It shows up in subtler ways. There is a certain fluency that feels convincing, but something about it feels interchangeable, as if the work has been assembled efficiently without being fully lived through. We recognize it more often than we admit—and yet we rarely question it, because it still works.
That distinction is easy to ignore. But it accumulates, precisely because nothing appears broken. And as it accumulates, it begins to shape something deeper than output. It shapes how we think, how we express, and how we understand what we are capable of.
And perhaps that is why I keep thinking about how I arrived at Ole Miss in the first place.
It felt familiar. And familiarity, I’m beginning to realize, is often just repetition—something we’ve seen often enough that we stop questioning it, and eventually mistake for understanding.
That same pattern shows up elsewhere. In how quickly we assemble ideas that feel complete. In how readily we accept them as our own. In how rarely we pause to ask whether we’ve actually done the work required to understand them.
The shift is already underway, and what makes it difficult to recognize is precisely how reasonable it feels as it happens. Nothing is forcing us to give these things up. And yet, in small and justifiable ways, we begin to do it ourselves—choosing efficiency where we once chose depth, accepting clarity where we once would have questioned it.
There is no clear breaking point. No moment where something visibly fails. From the outside, everything continues to function as expected, and in many cases, it even appears to improve.
But over time, something changes in how much of that work is still ours.
We begin to lose familiarity with the very processes that once shaped our thinking—the ability to stay with an idea long enough for it to become our own, to follow a thought beyond the point where it first sounds right, to remain in the discomfort of not knowing until something real begins to take form. These do not disappear all at once. They fade from lack of use.
And when that happens, it is not immediately obvious. We will continue to sound thoughtful. We will continue to produce work that feels complete. We will continue to move through our responsibilities with a sense of progress that is difficult to question.
Until we are asked to do something without assistance.
To think something through without a prompt guiding us.
To make a decision without something shaping it for us.
To stand behind an idea that is entirely our own.
And realize, in that moment, how much of that work we have already handed over—not in a single decision, but in a series of small ones that felt efficient, reasonable, even necessary at the time.
Perhaps more unsettling than that is the possibility that we may not fully recognize what has been lost. Only that something feels different. That the connection between what we produce and what we understand is no longer as direct as it once was. That the work, even when it is correct, no longer feels entirely ours.
Because the real risk is not that these systems will replace our ability to think.
It is that, over time, we begin to expect less of ourselves.
And call it progress.
Ex Cogitatione, Progressus
Girish
