What physics offers for artificial intelligence? Lessons from the brain's inner time and its dynamics.
Philosophical transactions. Series A, Mathematical, physical, and engineering sciences May 14, 2026 Georg Northoff, Yasir Çatal, Samira Abbasi
Computing can learn from physics by conceiving time in terms of dynamics—changing patterns of activity over time. The brain's intrinsic neural dynamics, specifically its spontaneous activity, scale-free activity, and variability, enable it to actively participate in the world's physical time through processes like entrainment, where neural activity follows external rhythms such as music. This active participation shapes experience and consciousness. Current computing devices, whether classical or natural, lack spontaneous activity and an 'inner time' that can actively influence processing. Consequently, they cannot actively encode input dynamics or participate in the world's physical time, leaving them 'locked out of time and world' and unable to acquire tacit knowledge or behave flexibly in a changing world.