Michael Harris, «Solitude: In Pursuit of A Singular Life in a Crowded World»:

John McCarthy, the American computer scientist who coined the term artificial intelligence, had the gall to insist that pieces of technology could hold opinions and beliefs. This was 1979, and he was one of the middle-aged wizards of the computer boom; at the time many exaggerated claims were bouncing down the halls of MIT and Stanford (McCarthy taught at both). In a paper, he wrote: “Machines as simple as thermostats can be said to have beliefs.”

John Searle, the American philosopher, was less of an optimist and found the notion hard to swallow. One day he asked McCarthy to be precise: “What beliefs does your thermostat have?” And he was surprised by the ready answer: “My thermostat has three beliefs,” said McCarthy. “It’s too hot in here, it’s too cold in here, and it’s just right in here.”

Can a piece of technology have a preference for one thing or another? The distinction between human and computer intentions is growing hazy. That much was obvious when my friend showed off his new Nest thermostat, which was patched into Google’s cloud and had been tracking his whereabouts in order to optimize his house’s energy consumption. “Nest doesn’t like it when I come home early,” said my friend, tapping at the glowing black puck on his living room wall. “It really likes me to have a pattern. That way it can make better decisions.”

This tiny slip—it can make better decisions—is crucial, because if we begin by allowing a thermostat to have a belief, we start to allow more complex technologies to have more complex beliefs, and just as my friend feels swayed to regulate his patterns because a trumped-up thermostat has a certain preference, we can find ourselves swayed in all sorts of places where we once made more personal choices.