As artificial‑intelligence capabilities keep expanding, universities are confronted with increasingly complex and urgent challenges, as well as fresh opportunities. In a recent essay, Sasha Rakhlin, director of MIT’s Statistics and Data Science Center, outlines three pivotal questions focused on research and graduate education.
1. What is changing in research, and why is it happening so fast? Mathematics illustrates the speed of AI progress: a model won a gold medal at the International Mathematical Olympiad last year, and just a year later similar systems have proposed solutions to one of the Millennium Prize Problems. The key driver is the speed and reliability of verification—formal proofs can be checked automatically, programs can be run and tested. When evaluation is fast, systems can generate candidates, learn from outcomes, and iterate. This feedback loop also applies to AI research itself, accelerating model and tool development. For seasoned researchers, this opens questions that were previously out of reach, while sharpening the distinction between obtaining a solution and understanding why it works, how it generalizes, and what to ask next. We should not assume abstraction, judgment, or problem formulation will stay beyond AI’s reach.
2. How should academic credit and graduate training be rethought? In many fields a polished paper is becoming a weaker signal of individual expertise. Departments need to redefine what they reward—not merely by what AI cannot yet do. Asking good questions, replication, synthesis, informative negative results, and shared datasets deserve greater recognition. Evaluation should clarify a researcher’s actual contribution and intellectual responsibility, especially when AI performed substantial parts of the work; this will guide hiring, promotion, and funding and must be communicated to current and incoming PhD students. Training is harder: intellectual muscles develop through routine calculations, coding, failed attempts, and small discoveries that build intuition. Delegating all of this to AI risks stripping away formative experiences. We must separate unnecessary friction from work that cultivates expertise, teaching students to formulate problems, audit model outputs, reproduce results, and defend choices. Fundamentals may become even more valuable as the foundation for effective tool use.
3. What should MIT and other universities build now? The goal is clear: universities must be able to pursue long‑term questions, share results openly, and evaluate claims independently. Industry partnerships are essential, but we cannot rely solely on commercial priorities to cover the breadth of science, so a degree of technological independence is needed. The most strategic asset is the tacit knowledge accumulated in labs—failed experiments, abandoned directions, and the reasons behind them rarely appear in publications. Capturing these negative results and expert interpretations could dramatically improve AI models’ ability to explore science. Imagine MIT’s labs functioning as a single, living scientific organism linked by shared AI infrastructure. For instance, a neuroscience lab needing better neuron‑segmentation could have an AI agent identify a relevant advance from a computer‑vision group, connect researchers, propose benchmarks, and help iterate. The agent could surface open questions to students and faculty, and propagate lessons across labs. To enable this, we must build workflows that record hypotheses, interventions, outcomes, failures, and interpretations, and integrate them through secure data systems and substantial compute resources. Such provenance tracking would preserve idea lineage, make contributions—including those of graduate students—easier to recognize, and, with clear consent and credit rules, foster more open collaboration. AI already synthesizes information from more sources than any individual researcher; its growing ability to make useful connections should be harnessed to unite us around hard scientific and engineering problems.
Review: Universities need to invest now in both technical infrastructure and cultural policies that leverage AI’s acceleration while preserving human creativity and critical judgment, ensuring sustainable academic progress.