Alexander (Sasha) Rakhlin

Rakhlin's three questions together point to the same core contradiction: the pace of AI development far exceeds the pace of institutional adjustment in academic institutions, and the bottleneck is not the technology itself, but organizational structure and incentive mechanisms.

The first core question Rakhlin raised is whether academic institutions should position AI as a research tool or as an independent discipline.

Rakhlin leans toward the path of building AI as a discipline, but at the same time acknowledges that not every university has the conditions to take that path.

Rakhlin believes that statistics and data science should not passively accept changes in the landscape, but should actively participate in building AI methodology, for example by providing a statistical theoretical foundation for large-scale models or developing uncertainty assessment methods suitable for complex AI systems.

Rakhlin points out that if incentives at the department level are not adjusted, interdisciplinary AI research will be hard to truly implement; for example, unless issues such as who leads promotion reviews and how outcomes are distributed in collaborations between statistics professors and computer science departments are resolved, collaboration can only remain an individual spontaneous act.

Rakhlin suggests that when designing AI strategy, institutions should make institutional support one of their priorities, rather than being limited to building laboratories and hiring people.

Rakhlin's position is that academia should serve as the conscience and infrastructure of the AI field: probing fairness, interpretability, and social impact, and producing public datasets, benchmarks, and theoretical frameworks.

Rakhlin believes that the unique value of academia lies in exploring directions that show no commercial return in the short term but are crucial to long-term knowledge accumulation.

Rakhlin emphasizes that one of the core tasks of the Statistics and Data Science Center is to cultivate students' ability to understand the underlying principles of AI methods.

Rakhlin believes that genuine interdisciplinary integration requires AI researchers to understand the core problems of the application domain, and domain experts to grasp the basic logic of AI methods.

Rakhlin cautions that against the backdrop of AI-improved research efficiency, researchers' motivation for careful verification may decline.

Rakhlin suggests that while embracing AI efficiency tools, academic institutions should place greater emphasis on methodological training, including experimental design, assessment of the robustness of conclusions, and honesty under uncertainty.

Alexander (Sasha) Rakhlin, director of MIT's Statistics and Data Science Center, raised three core questions about the development path of AI in academia in an in-depth interview.