Explained derivatives to 14 yo on the way back from breakfast, but before I could do integrals he diverted the conversation by asking about the Denver Nuggets.
This is big. Really big.
Researchers from Google Quantum Al and Google DeepMind developed an Al system that continuously keeps Google's 105-qubit Willow quantum processor optimized while it is running.
"Instead of stopping for manual recalibration, the Al uses reinforcement learning to monitor quantum error correction and automatically adjust hardware control parameters in real time."
"The system improved logical stability by 3.5x under hardware drift and achieved a record logical error rate of 7.72 Γ 10β»β΄ per cycle for the surface code."
"This is an important step toward future fault-tolerant quantum computers that can operate autonomously for long periods with minimal human intervention."