The goals of the AI companies and the goals of the mathematical community are severely misaligned. We see these as part of broader alignment issues impacting other scientific and creative professions, as well as the whole of society.A retired mathematician's comment suggests the use of algorithms might not affect the academic enterprise much in any event.
Research mathematics deals with understanding basic structures of shapes, numbers, and natural phenomena. Over the course of generations, it has built a large corpus of sophisticated ideas, methods, abstractions, and other tools to comprehend the mathematical landscape. In turn, modern technologies and sciences are based on mathematical tools.
I have literally never used AI in any of my research. My area is obscure enough that I doubt AI would have much interest any time soon in the research I do. Slightly more broadly, the big unsolved problems in my field (e.g. the Alperin weight conjecture) could maybe be finished by AI, but (best case scenario for AI) it would be pretty transparent that it was putting the finishing touches on a project that humans had been building up for decades (for instance, the McKay conjecture and the Height Zero conjecture were around in my field for decades, and have recently been solved by humans - humans who spent years putting the finishing touches on work that others had spent decades building up).His concluding remark, however, motivated my thesis. "And a person reading about the proof of Fermat's Last Theorem might be inspired to sharpen their analytical skills, learn some math, and that could develop into something useful. Like developing AI, oddly enough."
Perhaps that aspiring programmer can write a prompt to come up with a proof of that theorem that would fit the margin of a medieval book. The medalists' brief, however, illustrates the quest for solutions to those big unsolved problems being a way to generate Minimal Publishable Units.
Famous problems have often served as landmarks and lighthouses against which one can measure an improved understanding of this landscape. Solving one of these problems has been a certain sign of new insights and interesting methods, which would then be studied by a community of mathematicians, through a long and arduous process of talks, discussions, simplifications. At the end of this process, one will ideally find a textbook presentation of the results suitable for any graduate or even undergraduate student to study. Some of the mathematical ideas pursue their journey even further to become, decades or centuries after, tools that are understood and used by the whole population.Yes, in laboring away at the central problem, the researcher can produce supporting results that might be of use to other researchers. Heck, the proof of the Fermat theorem that does not fit the margin is all about generating such supporting results. Along the way came a proof that ran "If elliptic curves are modular, Diophantine equations of order greater than two have no nontrivial solutions in integers." That's the sort of academic work that drives people unfamiliar with the rules of logic and inference nuts, as the premise strikes such people as like the economist assuming a can opener. That proof is one of the landmarks, though, as it focuses subsequent mathematical attention on demonstrating that elliptic curves indeed are modular.
That propagation from cogitations in Pierre's library to distillations an undergraduate ought be capable of understanding, though, also produces a lot of Minimal Publishable Units. Disrupt that, warn the medalists, and troubles follow.
Indeed, the mass production at faster and faster pace of "true/false" statements could destroy fertile ground instead of breathing life into new ideas.The underlying problem, dear reader, might lie in the incentive structures of the academy. Consider a recent essay in Inside Higher Ed asserting "Our metrics for measuring faculty productivity are fueling a crisis in scientific integrity." It's the machine-enhanced "announced in a rush" situation, only the rush gives those deans citations to count.
Often these solutions are announced in a rush, leaving no time for a proper writeup, the isolation of new methods and ideas, and citing relevant previous work of others. As in all creative professions, this raises severe attribution and plagiarism questions. Moreover, without the willing mathematicians who must take care of their development and integration into the mathematical canon, AI-conceived ideas would never become fully alive and the crucial human transmission chain between mathematicians would be lost.
As generative AI and paper mills flood academic publishing with industrial-scale fraud, higher education faces a reckoning: The metrics built to evaluate faculty productivity are now accelerating the decline of scientific integrity. Once a quirky incentive problem, it has become a governance and reputational crisis. Paper mills have evolved into AI-powered factories producing thousands of automated manuscripts and fake peer reviews, overwhelming journals and making fraud harder to detect.It's fitting, I suppose, that computer-enhanced writing is now overwhelming computer-enhanced vanity journals — it's likely that in allowing my Northern Illinois University electronic mailbox to die rather than take a membership in the Democrat super-PAC calling itself the annuitants' association I've been spared lots of solicitations to submit or to review — but what the medalists see as a system failure, the Inside Higher Ed contributors, themselves senior scholars at institutions you might have heard of, see as a predictable feature of publication for publication's sake.
This is not about isolated bad actors; these are not bugs in the system. Academic leaders often express shock when fraud is uncovered. They should not. When careers depend on publication count and count itself becomes the defining success metric, some will cheat to seek prestige.Recognizing that publish-or-perish is a Glass Bead Game is good. Recognizing that the Glass Bead Game is rigged is better. Recognizing that market tests matter? Knock me over with a feather.
This is the outcome of the publish-or-perish culture that pervades academia, where an overwhelming majority of papers are published, simply to perish in obscurity. This is not cynicism. It is the predictable result of a broken academic system and flawed ranking agencies that confuse publication with contribution and count with impact.That passage puts more faith in the grant-issuing agencies, frequently governmental enterprises, than it deserves (where is Bill Proxmire when we need him?) The authors aren't done surprising me.
Business schools have perfected this pathology. Rankings that focus on counts of top journal publications and school-funded summer research support reward quantity rather than impact. Unlike engineering, public health or medical programs—where faculty must raise grant support—business schools often operate without a market mechanism to discipline research priorities.
We have engineered a system where survival depends on feeding the count meter, not necessarily on advancing knowledge. Merit reviews have become count reviews.Better, I suppose, to call out, as senior scholars, the system they successfully navigated upon successfully navigating it, rather than take the we-suffered-with-it-and-so-should-you stance. They even call out the citation cartels and the fakery.
We recently heard a business school dean boast about a faculty member who boosted the school’s research ranking with dozens of elite journal publications. As one of our mentors once warned, that sounds “too prolific to be profound.” When asked what this scholar was actually known for, the dean could not answer. That silence says everything.
As junior faculty members, we were repeatedly told that tenure required being known for something: an idea, a field or a contribution, a unique intellectual signature. Today, what matters is not being known for an idea but being known for count. The count meter ticks relentlessly, tracking how many papers you can place in a (very) short list of journals and how many citations your work is garnering. In fact, tenure requirements are reduced to five or some expected papers in top journals rather than a discussion of impact. It is a mockery that count has replaced impact and relevance for a lifetime appointment.
This fixation on numbers produces predictable and corrosive behaviors. Authors strategically cite editors to curry favor. Faculty participate in citation cartels under the guise of collegiality and talk of informal networks to trade reviews, acceptances and references like currency. These practices are whispered about, joked about and quietly tolerated because they deliver what the system demands: higher counts.Shortly after I took my pension, I lodged a protest against the landmarks and lighthouses, as making for boring job meetings.
In the academic world what matters is creating a minimal-publishable-unit that is also a contribution before someone else does, and going on to create sufficient such units to earn tenure or promotion. And everyone recognizes what the Hot Ideas are: there was one session of interviews I participated in, years ago, when my department was searching for a macroeconomist, and I heard a lot of "Lucas critique" and more than a few "unit roots" in those thesis summaries. Empirical failures in the policy world might have been secondary. Failure to acknowledge the state of the art, though, is a career-killer. "Rightly or wrongly, all subsequent models had to have these two elements present within them (RE and microfoundations), or they would be dismissed."The requisite elements are still there in any technical paper, and they might not be productive.
The journal list has become a curse masquerading as quality control. Nobel laureate James Heckman moderated an influential panel at the American Economic Association annual meeting, warning of the curse of the top five journals in economics. He noted that more than 70 percent of the most cited papers in economics over the past 10 years were published elsewhere and that more than 80 percent of the papers among top 50 scholars appear outside of the top five journals. Another Nobel laureate, George Akerlof, offered an analogy drawn from entomologists’ study of beetles to illustrate how gatekeepers reward similarity over insight where conformity triumphs over innovation.Somewhere, Charlie "Prof Scam" Sykes is nodding.
So, what is research impact? Advancing knowledge within the discipline—making a distinctive, novel and important contribution—is essential. But this cannot be determined solely through where the research was published and how many citations it garnered. Faculty advancement processes must require the evaluators to read the papers rather than rely on the CV. And impact doesn’t end there: Influencing policy, being featured in prominent media, testifying as an expert, translating research findings for organizations and educating executives are all important indicators of a scholar’s impact. And, sadly, they are rarely considered during promotion and tenure discussions.
Meanwhile, teaching—the very reason universities exist—has been relegated to an inconvenience. As academic leaders, we persistently field demands for teaching relief, not due to work overload from an important assignment, but because faculty simply do not want to teach and/or because their count of research papers matters more. Highly paid professors lobby to teach one to five students indefinitely while administrators look away to avoid conflict. Faculty governance is weaponized to preserve privilege, not to protect and advance the noble mission of discovery.

No comments:
Post a Comment