Getting stuff wrong is key to smarter artificial intelligence

NEURAL networks, like the ones grabbing headlines for winning boardgames or driving cars, depend on huge amounts of computing hardware. That in turn means a colossal amount of power: the next wave may consume millions of watts each.

That’s one reason why some suggest we rethink what we want computers to be. Reducing the precision with which they analyse problems, and putting up with the odd “error”, can cut zeroes off their energy consumption (see “To make computers better, let them get sloppy“). And it has precedent in the human brain – an unrivalled piece of hardware using electrical fluctuations and requiring a million times less power than a computer.

"Introducing error will also make computers better at handling the real world. Neural networks, loosely modelled on the brain, capture some of its problem-solving capacity. AlphaGo surprised opponents and designers alike with Go moves described by some as “intuitive”. LTC Stephen F. PO2 Ed C. LTC Stephen C. SFC Dr. Joseph Finck, BS, MA, DSS SFC Joe S. Davis Jr., MSM, DSL CPT L S SSgt John S. SP5 Mark Kuzinski Cpl (Join to see) SMSgt Minister Gerald A. Thomas SSgt Obom Bowen Col (Join to see) SPC(P) Doug Moxley SSG James J. Palmer IV aka "JP4" SrA Christopher Wright COL Jean (John) F. B. MSG Tom Earley SMSgt Todd Grebel Sgt Kelli Mays SGT (Join to see)