A little over a week ago, OpenAI, a leading artificial intelligence company, unveiled ten notable advances in mathematics and computer science facilitated by their upcoming model, Astra. These discoveries span various fields, including geometry, cryptography, and coding theory, highlighting the powerful potential of AI in driving mathematical exploration and innovation.
This announcement marks just the latest phase in a burgeoning trend of mathematical breakthroughs powered by generative AI systems. As these technologies evolve, the traditional landscape of mathematical research is being reshaped, prompting new questions about the role and implications of AI in this discipline.
AI is Raising Big Questions
The impact of large language models (LLMs) like ChatGPT and Claude is far-reaching, creating ripples throughout the mathematical community. University departments are currently grappling with the ethical ramifications these AI tools introduce in both research and educational settings. Peer-reviewed journals are inundated with AI-generated submissions and must navigate the complexities of assessing their quality and authenticity. Moreover, platforms like arXiv have experienced a surge in mathematical papers being submitted, often produced with the assistance of AI technologies.
Funding agencies are also stepping into this conversation, developing policies that govern how AI can be ethically employed in research processes. Alongside these practical implications, broader philosophical inquiries about the essence of mathematical discovery arise. If an LLM can derive a proof or generate a novel question, what does it mean for human mathematical creativity? The pivotal queries revolve around whether mathematics is merely about the generation of new theorems or if it also encompasses a deeper quest for understanding. Additionally, when AI contributes to discoveries, who deserves credit?
Growing Concern – But Little Consensus
The mathematical community is rife with divergent opinions regarding these developments. Many mathematicians express genuine concerns about the direction of their field, with one voice even describing the rise of AI-assisted mathematics as triggering a “profound spiritual crisis.” In contrast, others are actively striving to articulate principles for responsibly integrating AI into mathematical research. For instance, the recent Leiden Declaration, signed by thousands of mathematicians worldwide, asserts that AI should augment, not replace, human creativity. It emphasizes the critical importance of transparency, accountability, and proper attribution in the collaborative environment of mathematics.
At the International Congress of Mathematicians, Terence Tao, a distinguished Fields medalist, encouraged attendees to consider the future, highlighting the importance of reflecting on how AI systems can reinforce the cultural and ethical values fundamental to mathematics.
Two Attitudes to AI
The evolving landscape of mathematical research is mirrored in contrasting attitudes toward the use of AI. In my collaborative work on the “semiregularity problem,” my colleague Saul Freedman made it explicitly clear that he preferred to proceed without the aid of AI, citing environmental and social concerns associated with these technologies. Our research, which focused on highly symmetric networks, was accomplished entirely without AI inputs. Interestingly, after our breakthrough, I learned that two other colleagues had independently attempted using LLMs to tackle the same problem, but with no success.
In stark contrast, shortly after, my colleague Aluna Rizzoli reached out to share an incredible discovery involving an object that had eluded my collaborators and me for over two years. Aluna utilized an OpenAI model alongside a supercomputing cluster, completing computations in just 43 hours—a task that would have taken us months to accomplish using traditional methods. Notably, rather than claiming sole credit for the discovery, Aluna invited us to co-author a paper, demonstrating a collaborative spirit despite the use of AI tools.
AI-Powered Discovery – With Human Connection
Both collaborations yield significant mathematical insights, but they embody radically different philosophies regarding AI’s role. One pair of researchers outright rejected the technology over ethical considerations, while the other embraced it as a transformative collaborator without sacrificing the etiquette inherent in mathematical research. Both perspectives are valid and indicate that there’s no singular path to producing outstanding mathematical work.
This striking diversity of opinion encapsulates the current moment for mathematicians. The focus has shifted from whether LLMs can contribute meaningfully to the discipline. Now, discussions concentrate on how the integration of these powerful tools aligns with the core values of collaboration, transparency, and intellectual integrity that have long defined mathematics.
Inspired by: Source
