Президент MIT: почему фундаментальная наука важнее хайпа вокруг ИИ

While Silicon Valley is obsessed with the race for the number of parameters in neural networks, MIT President Sally Cornbluth went on air at Boston Public Radio with an important reminder: technological leadership is not just software. It's deep, often invisible, and expensive work by scientists that lays the foundation for decades to come. In this conversation, Cornbluth essentially became an advocate for the entire American scientific ecosystem, which today faces serious pressure.

The context here is transparent and somewhat troubling. We're used to the USA being an undisputed leader in innovation, but this status rests on a research enterprise that was built over decades after World War II. Today, when budget discussions in Washington are becoming increasingly heated, and private capital is flowing into quick AI-based solutions, fundamental science risks being left with crumbs. Cornbluth understands: if we don't protect laboratories today, tomorrow we'll have nothing to automate.

What's particularly interesting is how MIT is trying to demystify science. The university's faculties discuss everything: from mathematics in football to the physics of Olympic figure skating. This isn't just an attempt to entertain listeners. It's a demonstration that AI is just one tool in a vast arsenal of human knowledge. When we talk about finding a cure for ovarian cancer, we mean a synergy of biology, chemistry, and computational power. Without deep expertise in each of these areas, no algorithm will produce results.

Why is this important for the AI industry right now? Because we're approaching a moment when extensive model growth — simply adding data and GPUs — is starting to slow down. Further progress will require qualitative breakthroughs in understanding how the human brain works, how information is structured, and how to use energy efficiently. These answers don't lie on the surface, they can't simply be pulled from the internet. They are found by people in the laboratories of MIT and other leading universities.

Sally Cornbluth emphasizes that modern science has become a team sport. The days of lonely geniuses are gone. Today, success is when a mathematician helps an oncologist, and a robotics specialist consults with a physicist. AI here acts as a catalyst, accelerating the testing of hypotheses that used to take years. But for this catalyst to work, there must be an environment for generating these hypotheses in the first place.

The main risk that MIT's president warns about is a loss of focus. In the pursuit of the next bright announcement in the world of generative AI, society may forget that real breakthroughs often happen where there are no cameras and enthusiastic tweets. We need to learn to value the research process as much as we value its end product in the form of a convenient smartphone app.

The future of AI is inextricably linked to the survival of the academic environment. If universities become appendages of corporate development departments, we will lose that very freedom of inquiry that gave birth to the internet, GPS, and those very neural networks. Cornbluth insists that supporting science is not charity, but the most profitable investment in national security and economic prosperity.

The key point: Whether AI becomes a tool for saving humanity or remains an expensive toy for generating memes depends on whether we maintain support for fundamental science. Will we have the patience to invest in something that will bear fruit only in twenty years?