September 25, 2025 | IT/AI Special Reporting Team
[Berlin, Germany] The definition of Artificial General Intelligence (AGI) has entered a new phase following a groundbreaking dialogue between OpenAI CEO Sam Altman and Oxford physicist David Deutsch. In a fireside chat held at Axel Springer on September 24, Altman proposed a radical benchmark:
“If a future AI model can understand quantum gravity — a theory humans have yet to fully grasp — and explain it to us, would that be enough to convince you it’s truly intelligent?”
Deutsch, a long-time skeptic of AGI and author of The Beginning of Infinity, responded after deep reflection:
“It might be.”
This moment marked a philosophical turning point in defining AGI — not by performance metrics or data processing, but by the ability to create new knowledge, akin to human discovery.
1. A Meeting of Minds: AI Skeptic vs. AI Visionary
The impromptu exchange occurred during a ceremony in Berlin, where Altman discussed Deutsch’s book — one of his personal favorites. Deutsch joined the conversation remotely, leading to a spontaneous debate on the nature of intelligence.
Deutsch has long argued that large language models (LLMs) merely recombine existing ideas and lack the creative reasoning needed to generate original knowledge. He defines true intelligence as the ability to identify problems, devise solutions, test hypotheses, and refine them — a uniquely human trait.
2. Quantum Gravity: The Ultimate AGI Test
Quantum gravity — the elusive theory that seeks to unify quantum mechanics with Einstein’s general relativity — remains one of physics’ greatest unsolved puzzles. Altman’s proposition reframes AGI as a system capable of transcending known knowledge and generating insights in domains humanity has yet to conquer.
This benchmark shifts AGI from imitation to innovation, demanding that future AI systems not only understand but explain the unknown.
3. Industry Implications: A New Milestone for AGI Development
🔬 A. From Benchmarks to Creativity
The Altman–Deutsch consensus signals a shift in AGI development goals. No longer is the focus on mimicking human intelligence or improving performance scores like Elo ratings. Instead, the emphasis is on scientific discovery and original knowledge creation — a qualitative leap for AI research.
🧩 B. Philosophical Closure to the AGI Debate
Deutsch’s agreement carries weight. As a leading critic of AGI, his concession provides a new philosophical foundation: AGI is not just about using knowledge like humans — it’s about expanding knowledge in ways humans cannot.
Industry experts view Altman’s challenge — decoding quantum gravity — as a bold declaration of OpenAI’s mission to build AI that pioneers new frontiers, not just automate existing ones.