Neil Thompson, a researcher at MIT, has presented compelling evidence suggesting that artificial intelligence is on the brink of transforming several scientific fields, including probability, statistics, and quantum physics. In a study conducted over the period from mid-2024 to mid-2026, Thompson and his team found that AI models, such as GPT-5.5, were solving theoretical problems at an impressive rate, going from 2% to 35% correct solutions.

This rapid advancement raises questions about the future role of humans in theoretical sciences. While some see this as a positive step towards more efficient problem-solving, others worry about the loss of the human intuition and creativity that often drives breakthroughs. The Navier-Stokes problem, solved by an unreleased model, has already sparked controversy, highlighting the ethical and practical challenges of AI in science.

However, Thompson and his colleagues believe that the impact of AI will be both positive and negative. They argue that AI could enable more practical applications and accelerate the pace of research, but also warn that the verification and interpretation of AI-generated solutions will still require human expertise. As AI takes on more theoretical work, experts predict that the role of human scientists will shift towards a more strategic and interpretative role.

The big question remains: as AI becomes more adept at solving complex problems, will the human touch still be necessary? Or will it be replaced by an algorithmic approach? Only time will tell, but one thing is clear: the landscape of science is about to change dramatically.