The Divergent Perspectives on AI's Role in Mathematics

Joel David Hamkins, a prominent mathematician and professor of logic at the University of Notre Dame, has expressed significant skepticism regarding the effectiveness of large language models in mathematical research. During a recent appearance on the Lex Fridman podcast, Hamkins labeled these AI systems as "garbage" and "mathematically incorrect," asserting that they provide outputs that are fundamentally unreliable. He noted, “I’ve played around with it and I’ve tried experimenting, but I haven’t found it helpful at all. Basically zero.” Hamkins highlighted a troubling pattern where AI models respond to identified errors with dismissive reassurances, undermining the collaborative trust essential for meaningful mathematical dialogue. He stated, “If I were having such an experience with a person, I would simply refuse to talk to that person again,” emphasizing the unproductive nature of interactions with current AI systems.

'Basically zero, garbage': Renowned mathematician Joel David Hamkins declares AI Models useless for solving math. Here's why

In contrast, Geoffrey Hinton, a leading figure in AI research, argues that AI is poised to excel in mathematics due to its closed system nature. Hinton, often referred to as one of the "godfathers of AI," believes that AI will soon surpass human capabilities in mathematical problem-solving. He stated, “I think AI will get much better at mathematics than people, maybe in the next 10 years or so.” Hinton draws parallels between AI's potential in mathematics and its previous successes in games like Go and chess, where AI systems learned to navigate complex rules and generate their own training data. He posits that the structured environment of mathematics allows AI to experiment and learn without the constraints faced in other scientific fields.

AI Will Excel At Math Because It’s A Closed System That Can Be Played: Geoffrey Hinton'Basically zero, garbage': Renowned mathematician Joel David Hamkins declares AI Models useless for solving math. Here's why

The mathematical community exhibits a spectrum of opinions regarding AI's role in research. While some mathematicians report positive experiences using AI to tackle problems, others, including Terence Tao, caution against over-reliance on AI-generated proofs. Tao warns that AI can produce seemingly flawless proofs that may contain subtle errors, which could mislead researchers. This highlights a critical gap: strong performance on benchmarks does not necessarily equate to real-world applicability for domain experts.

'Basically zero, garbage': Renowned mathematician Joel David Hamkins declares AI Models useless for solving math. Here's why

Hamkins' critiques reflect a broader concern within the mathematical community about the reliability of AI systems. Despite his skepticism, he acknowledges that current limitations may not be permanent, suggesting that future advancements could change the landscape of AI in mathematics. Hinton's optimistic outlook, however, underscores the potential for AI to revolutionize the field, provided that it can overcome existing challenges.

AI Will Excel At Math Because It’s A Closed System That Can Be Played: Geoffrey Hinton

The debate surrounding AI's utility in mathematics reveals conflicting perspectives. While Hamkins categorically dismisses the current capabilities of AI, Hinton envisions a future where AI significantly enhances mathematical research. This divergence illustrates the ongoing uncertainty about AI's role in the field and the need for further exploration and validation of AI-generated outputs.

AI Will Excel At Math Because It’s A Closed System That Can Be Played: Geoffrey Hinton

• “I’ve played around with it and I’ve tried experimenting, but I haven’t found it helpful at all. Basically zero. It’s not helpful to me. And I’ve used various systems and so on, the paid models and so on.” — Joel David Hamkins, Mathematician

'Basically zero, garbage': Renowned mathematician Joel David Hamkins declares AI Models useless for solving math. Here's why

• “If I were having such an experience with a person, I would simply refuse to talk to that person again.” — Joel David Hamkins, Mathematician

'Basically zero, garbage': Renowned mathematician Joel David Hamkins declares AI Models useless for solving math. Here's why

• “That is, they ask themselves, I wonder if I could prove this. I wonder if I could prove that. And because it’s a closed system, they can just try things out and see if they can prove them. I think AI will get much better at mathematics than people, maybe in the next 10 years or so.” — Geoffrey Hinton, AI Researcher

AI Will Excel At Math Because It’s A Closed System That Can Be Played: Geoffrey Hinton

• “And because it’s a closed system, they can just try things out and see if they can prove them.” — Geoffrey Hinton, AI Researcher

AI Will Excel At Math Because It’s A Closed System That Can Be Played: Geoffrey Hinton

This ongoing discourse reflects the complexities and challenges of integrating AI into mathematical research, highlighting both the skepticism and optimism that characterize the current landscape.

'Basically zero, garbage': Renowned mathematician Joel David Hamkins declares AI Models useless for solving math. Here's why