Big tech companies have had a tough year in PR, after being accused of distorting elections, automating jobs, fueling mental illness, and avoiding taxes. But they may face an even bigger challenge in a disruption still to come — ambitious plans to develop a fully fledged and autonomous machine intelligence, a safety risk some observers compare to nuclear weapons.
Killer robots are not exactly banging down the door, but entirely unregulated developments in artificial intelligence (AI) should give even the savviest early adopter pause, since these innovations bring us closer to a much-discussed but still unresolved policy problem: How will machines created to exceed human intelligence remain accountable to their creators?
Today, AI is mostly limited to software algorithms that can perform an increasing number of high- and low-skill tasks with moderate success. But Google, Facebook, and untold others are racing ahead to develop the data-absorbing “neural networks” of machine learning that designers hope will yield a machine intelligence capable of mimicking or surpassing human cognition.