Recently, the Université Libre de Bruxelles in Belgium unveiled a remarkable research discovery: commercial large language models (LLMs) have attained the capability to independently generate original mathematical proofs. Notably, OpenAI's ChatGPT-5.2 (Thinking) successfully tackled a mathematical conjecture put forth in 2024. This milestone signifies the formal foray of LLMs into the realm of theoretical mathematics, heralding a transformative shift in the landscape of scientific discovery. The research team has christened this novel scientific research paradigm 'Atmosphere Proof.'

Throughout the proof process, ChatGPT showcased exceptional prowess in original reasoning. While human verification remains indispensable, the AI's efficiency in deduction vastly surpassed initial expectations. This accomplishment decisively dismantles the misconception that 'AI lacks creativity,' and it is anticipated that the trajectory of theoretical mathematics research will experience a significant acceleration. Consequently, the primary bottleneck will transition towards ensuring logical correctness, rendering efficient automated verification a focal point of technological rivalry.