How Can AI Researchers Save Energy? By Going Backward. | Quanta Magazine
Reversible programs run backward as easily as they run forward, saving energy in theory. After decades of research, they may soon power AI.
Introduction
For Michael Frank, efficiency has always been a major preoccupation. As a student in the 1990s, he was originally interested in artificial intelligence. But once he realized how much energy the technology would use, he took his research in another direction. “I started getting interested in the physical limits of computation,” he said. “What’s t...
Read more at quantamagazine.org