New Preprint Shows GPT-5.2 Pro Can Derive Quantum Gravity Amplitudes

A new preprint demonstrates that GPT-5.2 Pro can help derive previously unknown results in quantum gravity—extending prior work on gluon scattering amplitudes to gravitons.

The paper, "Single-minus graviton tree amplitudes are nonzero," shows that a class of graviton interactions previously assumed to vanish can arise under specific kinematic conditions. The authors include Alfredo Guevara (Institute for Advanced Study), Alexandru Lupsasca (Vanderbilt University and OpenAI), David Skinner (University of Cambridge), Andrew Strominger (Harvard University), and Kevin Weil (OpenAI).

The key finding: in the "half-collinear regime," where particle momenta satisfy special alignment, the single-minus graviton amplitudes don't vanish—they exist as well-defined mathematical distributions.

The connection to AI: the earlier gluon paper was provided to GPT-5.2 Pro as context, and the model was asked to construct corresponding amplitudes in quantum gravity. According to the authors, "GPT-5.2 Pro not only solved this problem using a beautiful and surprising technique (the directed matrix-tree theorem), it also produced an excellent preliminary draft of the paper."

The result relates to an infinite-dimensional "w-(1+∞)" symmetry first studied by Roger Penrose in classical gravity, expected to play a role in quantizing the gravitational field.

The preprint is available on OpenAI's site, along with a transcript of the initial GPT-5.2 Pro exchange.