Experts Explore the Boundaries of Artificial Intelligence at Cross-Disciplinary Dialogue

People's Daily reported on a cross-disciplinary dialogue in Shanghai where Bao Yungang discussed the boundaries of artificial intelligence capabilities. Bao observed that while artificial intelligence demonstrates notable efficiency in generating mathematical proofs, verifying their correctness remains a critical bottleneck. Dong Bin stated at the event that artificial intelligence is reshaping mathematical research by freeing researchers from burdensome proof workloads. He encouraged mathematicians to actively embrace artificial intelligence, learn how to manage agents, identify model hallucinations, and achieve co-evolution alongside artificial intelligence tools. Dong further noted that artificial intelligence currently excels at knowledge storage and reasoning, whereas formulating high-value mathematical conjectures remains an advantage for human mathematicians. He noted that mathematical proof results released by OpenAI contain errors, while mathematicians can provide evaluation benchmarks, long-chain reasoning data, and training environments that nurture artificial intelligence reasoning capabilities. Zhang Ya indicated that mathematics and artificial intelligence have established a mutually reinforcing relationship, with mathematics providing the foundation to comprehend artificial intelligence and artificial intelligence offering new exploratory tools. However, she emphasized that challenges remain regarding how artificial intelligence research results can be verified and applied within the physical world.