Two independent research teams recently utilized OpenAI’s GPT-5.6 Sol Ultra to solve a complex quantum cryptography problem, highlighting the transformative and disruptive role of AI in scientific discovery.

Key Points

• MIT student Seyoon Ragavan and a U.C.S.B.-U.C.L.A. team independently produced proofs for efficient unclonable encryption using the same AI model.
• Both research groups were inspired by an open question presented at the Simons Institute for the Theory of Computing at U.C. Berkeley.
• The AI model, GPT-5.6 Sol Ultra, coordinated multiple agents to generate core proof ideas and constructions that researchers then refined and verified.
• Neither paper has undergone peer review, and the researchers are currently discussing merging their findings into a single submission.
• The near-simultaneous discovery raises significant questions regarding academic credit, independent research, and the future of graduate-level training.

Why it Matters

The rapid convergence of these results demonstrates how AI can accelerate theoretical breakthroughs, potentially solving long-standing problems in record time. However, this shift challenges traditional academic hierarchies and creates uncertainty regarding the future role of graduate students in a field where complex proofs can be increasingly automated.