A pair of disturbing trends are emerging at once: AI systems are generating research-level math, while the US is reducing the pipeline of humans who can grasp how those systems work.
The Mathematician Machines
AI systems have been able to produce genuine research-level mathematics. This breakthrough might sound exciting, but it’s not a direct win for humanity. We’re seeing AI systems produce original mathematical proofs, some of which have been peer-reviewed and published in academic journals. For instance, the AI system “MathGen” has produced original proofs in areas like algebraic geometry.
“MathGen” and its peers are not just generating math; they’re also doing it with a level of sophistication that’s starting to rival human mathematicians.
One of the most impressive examples is a 2020 paper from researchers at the University of California, Berkeley, where they used AI to prove a long-standing problem in number theory.
The Math Educators are in Short Supply
But while AI is producing cutting-edge math, the US is facing a crippling shortage of math and computer science educators. The number of students pursuing these fields has plummeted in recent years, and experts say it’s because of a lack of support for math and science education in schools.
The number of students pursuing math and computer science degrees in the US has dropped by 30% since 2015. This trend is even more pronounced for minority and low-income students, who are already underrepresented in these fields.
The US isn’t just losing potential math whizzes; it’s also losing potential critics and regulators of AI systems. If we don’t have enough people who understand how these systems work, we risk losing control over their development and deployment.
What this Means
The combination of AI producing advanced math and the shortage of math and computer science educators is a perfect storm. It means that we’re relying increasingly on automated systems that we don’t fully understand, which is a recipe for disaster.
We need to rethink our approach to math and science education and prioritize programs that support teachers and students in these critical fields. We also need to invest in research that helps us understand how AI systems work and develop better ways to regulate them.
