Terence Tao: AI Revolutionizes Mathematical Ideas

⚡

Key Takeaways

1Terence Tao compares the impact of AI on mathematics to that of the automobile on urban planning, highlighting unexpected parallels.

2AI facilitates idea generation but shifts the challenge to verification, requiring a new approach to mathematical infrastructures.

3Tao proposes a discipline of "AI planning" to seamlessly integrate AI into mathematical practices without losing the benefits of traditional methods.

💡Why it matters — AI is transforming scientific research, but without suitable infrastructures, the potential of these innovations could be limited.

Terence Tao and the Impact of AI on Mathematics

The renowned mathematician Terence Tao discusses the growing influence of artificial intelligence (AI) on the field of mathematics, comparing it to the historical impact of the automobile on urban development. According to him, this analogy could also extend to other sectors, such as programming, where AI is beginning to play a central role.

The advent of cars revolutionized transportation by offering unprecedented speed compared to previous modes of travel. However, this advancement also brought challenges, such as the congestion of roads designed for pedestrians and horses. For Tao, only thoughtful urban planning could have harmonized these two worlds by building infrastructure suited to the new demands.

The Current Mathematical Infrastructure: A Hindrance to AI

In the realm of mathematics, existing structures such as scientific journals, conferences, and mentorship are comparable to those old narrow roads. While traditional mathematical proofs are often slow to produce, they provide essential secondary benefits, such as the development of expertise and the discovery of new research avenues.

Tao emphasizes that AI-assisted proofs, although effective in moving from a hypothesis to a result, often lack these secondary benefits. They do not easily fit into traditional publication formats, as they omit the detailed narrative of the journey to the proof. He compares efforts to adapt AI to current publication models to trying to drive cars on roads designed for pedestrians.

Towards a New Mathematical Infrastructure

Rather than forcing AI to adapt to existing structures, Tao proposes creating a new mathematical infrastructure specifically designed for machines. This new approach would not replace human methods but would complement them. He cites the example of major mathematical challenges where solutions are verified by formal proof assistants or by automatically generated proof libraries, which humans then refine.

Tao also suggests the emergence of a new discipline he calls "AI planning," inspired by urban planning, to maintain the accessibility of mathematics while integrating AI capabilities.

AI: A Tool to Enrich Research

In a discussion with Dwarkesh Patel, Tao elaborated on this idea, explaining that AI enriches and broadens his work, particularly through more graphs, code, and in-depth literary research. However, he continues to perform the core of his mathematical work in a traditional manner, with pen and paper.

Despite the new possibilities offered by AI, Tao notes that the speed of composing a paper has not significantly increased. AI has not so much accelerated mathematical work as it has opened new perspectives.

The New Challenge: Verifying Ideas

Tao concludes by stating that AI has drastically reduced the cost of generating ideas, much like how the Internet has reduced the cost of communication. However, this abundance of ideas does not automatically translate into an abundance of verified results. The real challenge now lies in the verification and evaluation of these ideas, a process that requires renewed attention and appropriate resources.