The pace of machine-learning advancement is often underestimated, according to Massachusetts Institute of Technology principal research scientist Andrew McAfee. Massachusetts Institute of Technology principal research scientist Andrew McAfee says the pace of machine-learning advancement is often underestimated, and he cites data center management and the game of Go as several areas in which machines have lately beaten humans. McAfee notes Google DeepMind's AlphaGo algorithm has gained in both these fields, making a data center's energy efficiency 15% better on average, and beating world-class Go players even though experts predicted such a milestone would never occur. He says these examples illustrate the hazard of relying on the "highest-paid person's opinion," or following the views of experienced professionals' bosses, who are "chronically second-guessing the geek" and ignoring bold, rational, iterative, transparent, and experimental mindsets that are more likely to accurately predict smart-machine advances and cushion disruptions. McAfee also urges companies to experiment with advanced technology ideas, warning, "If an organization isn't ramping up the successful experiments they aren't doing their job."