Peter Thiel's Physics Department (philippdubach.com)

Peter Thiel recently highlighted the stagnation in physics since the 1970s, attributing it to a lack of significant breakthroughs despite advancements in digital technology. However, the introduction of AI into the field has begun to change this narrative. On February 13, GPT-5.2 made headlines by deriving and proving a new theoretical physics result related to gluon scattering amplitudes, a significant achievement noted for its simplicity and potential to push the boundaries of the field. This event marks a crucial demonstration of how AI can contribute to scientific discovery, echoing the success of previous AI applications like AlphaFold in predicting protein structures. The implications for the AI and ML community are profound, suggesting that with increased investment in scientific AI—currently dwarfed by commercial funding—a new era of discovery could emerge. Compelling examples include AI's role in advancing drug discovery, materials science, and even fusion energy. However, challenges remain, with concerns that AI, as a pattern-matching tool, may not be able to drive the conceptual revolutions necessary for breakthroughs in physics beyond current data capabilities. The widening "compute gap" between commercial and scientific AI spending presents a pivotal opportunity—addressing this could expedite scientific advancements, but also raises ethical considerations regarding the dual-use nature of such technologies.