“The result of this is the model’s not just going to be better than a physicist. It’s going to be better than a superposition of somebody who’s at the top in physics, computer science and data science.”

That’s how Jonathan Siddharth, CEO of Turing AI, describes the shift happening in advanced AI model training. Instead of relying on low-cost clickworkers, companies like Scale AI, Toloka, and Turing are now turning to highly skilled experts in fields like physics, biology, software engineering, and finance. The goal is to create complex, domain-specific tasks that mirror real human thought processes – whether that means writing code, validating physical theories, or analyzing simulations. These specialized datasets are then used to train the new reasoning models at major companies like OpenAI.