When Speed Becomes the Enemy of Knowledge: Academia's Quiet Crisis
A recent piece in Science. "Why I may ‘hire’ AI instead of a graduate student", by Ariel Rosenfeld, a researcher I know well and whose early career enthusiasm and earnestness I remember with real appreciation, contains one sentence that stopped me cold: "The issue is not whether my students are valuable. In the long run, they are invaluable. The issue is that their value emerges slowly, whereas AI delivers immediate returns."
I want to sit with that sentence for a moment. Because Ariel is right about the diagnosis. And deeply wrong, I think, about where it leads.
The Irony at the Heart of the Argument
Ariel describes his younger self as a novice: earnest, enthusiastic, uncertain. Someone who needed time, guidance, and the permission to stumble before becoming capable. He is now a researcher whose work is worth publishing in Science. This trajectory, from eager beginner to genuine contributor, is not incidental to his value. It is his value.
And yet the logic he is now tempted by, would have made hiring that younger version of himself look like a poor return on investment.
This is not a criticism of Ariel. I really value his honesty in this piece, as I value him as a researcher. He is naming a real tension that must of us are encoutering, but are reluctant to say out loud. The academic incentive structure has shifted so decisively toward immediate, measurable output that the slow, uncertain, ultimately transformative development of a young researcher now looks like a liability on a career timeline.
AI did not create this problem. It has simply made it impossible to ignore.
What We Have Decided to Reward
The pressure Ariel describes, on the calculation that AI delivers now, versus what a student delivers eventually, is the natural endpoint of decades of (ill fated) choices about how to evaluate academic work. Publication counts, impact factors, grant cycle deliverables, institutional rankings: all of these reward throughput. All of them penalise the kind of investment that produces a researcher rather than a result.
As I have argued before, we are confusing speed and quantity with quality and relevance. This confusion is not new, but AI sharpens it to a fine point. When a tool exists that can produce something that at first glance looks like a competent literature review, draft a methods section, or summarise a dataset in minutes, the question of why one would spend months nurturing a student toward those same capabilities becomes genuinely uncomfortable.
The answer, that the student is not being trained to produce those outputs, but to develop the judgment, curiosity, and critical thinking that eventually produces something no model could anticipate, gets lost in a system that measures outputs and not the minds that generate them.
The temptation to use AI is a symptom, not the disease. The disease is an academic system that has quietly decided that the slow work of human intellectual development does not count unless it produces measurable outputs on a predictable schedule.
The graduate student who argues with their supervisor, fails in their first experiment, revises their assumptions, and slowly develops scientific judgment is becoming exactly the kind of researcher who can situate AI within the world it enters, and the world it helps to shape. If we replace that formation with tools that produce immediate returns, we will have trained a generation that knows how to use AI and does not know very much else. When that happens, something more fundamental breaks down. The purpose of academia is not to produce outputs. It is to cultivate the people capable of generating knowledge. That is what will have been traded away for efficiency.
Fixing this is not a matter of individual virtue. Individual researchers cannot unilaterally resist incentive structures without cost to their careers and their students. What is needed is a deliberate institutional reckoning: research assessment that values mentorship explicitly, funding horizons long enough to accommodate genuine development, and a willingness to ask what universities are actually for.
Are they engines for producing publications? Or are they places where the next generation of thinkers is formed?
The answer shapes everything, including whether the earnest, enthusiastic novice that Ariel once was gets a chance to become the researcher he is today.
That chance is worth protecting. And it will not protect itself.
Comments
Post a Comment