Essay: Science is attunement, not deduction — AI changes discovery

In a widely shared essay, the author challenges the conventional view that science progresses through deduction from established principles. They argue that while applied science—like calculating a bridge's load or a vaccine dose—does rely on deductive formalism, the essence of scientific discovery is intuition and attunement to the world. The formalism, they contend, is merely scaffolding; intuition is the real building, and reasoning exists to construct that intuition. The essay traces this idea through history, from Greek mythological explanations to Anaximander's revolutionary concept of a free-floating Earth, which marked the first move toward naturalistic explanations.

Galileo's mathematization of science is presented as the decisive step, but the author stresses that this did not make deduction the core method. Instead, mathematics became a tool for attuning to nature's patterns. The essay then connects this perspective to modern AI: machine learning and neural networks, which learn from data in ways that are not purely deductive, may represent a new form of scientific attunement. By embracing these tools, scientists could move beyond rigid axioms and let models reveal relationships humans might miss. The piece, edited with Anthropic's Claude Fable, offers a philosophical lens for understanding how AI might reshape the practice of science—making intuition, not deduction, the guiding principle.