The tiny lights on the child’s shoes start to blink. This is not a human child, but a robot. The moment this thought occurs to me I take a leap into the future. I’d just read a dystopian novel in which children were no longer being born and the human race was heading for extinction. People were starting to demand child robots. Although software enabling robots to converse with human beings had been perfected long ago, the development of “my-child-robots”—with the capacity to use outside data organically, mixing it with their programs to form their own personalities—was stuck at a rudimentary level without much hope for improvement. Company A’s robot, for instance, was very popular at first because it responded so sweetly to whatever its parents said, but in time the adults got sick of all that sweetness and stopped talking to it, leaving the child robot to gather dust in its own room, where it eventually stopped moving. Company B, on the other hand, came up with an intellectually curious type that peppered its parents with questions, storing up more and more knowledge. It was programmed with just the right amount of cheekiness and persistence, too. Unfortunately, it only asked things like “How many people live in this city?” and “What year was the French Revo­lution?” so the parents soon got bored and switched it off in irritation. There’d been something fascinating about the stuff real children had asked about long ago, but Company B failed to discover the source for questions like “Where is my dead uncle living now?” or “Why doesn’t our dog talk?”

One research group concluded that inputting knowledge through parent-child conversations alone would be too difficult, so they developed software that absorbed a little information each day through sensors. Parents were instructed to take their “my-child-robot” with them wherever they went. If they took it shopping, for example, it would learn not only about the products sold in human society but also the difference between days when its parents dragged their feet, gloomy and distracted, and days when they smiled as they briskly marched along. But a good mood didn’t guarantee that the parents would buy their child-robot whatever it wanted; sometimes, when they were too depressed to talk, they would buy it toys or chocolate cake. Little by little, the robot would have to learn how the weather affected people’s moods. For instance, at twilight on a winter’s day, when the sun, stubbornly hovering just above the horizon, suddenly drew a deep velvet curtain across itself, darkening the city, the parents’ moods darkened too, until they went outside and saw fresh, clean snow sparkling in the glimmer of streetlamps, shining on the reds and oranges of winter scarves, and they would be enveloped in a brightness and warmth that comes only in winter. This, too, was important information about human society that would be carved into the child-robot’s brain—information that couldn’t be translated into digits, which is why Company C decided to use visual images. People’s feelings, smells, the moisture in the air, the cold, and the light from streetlamps could all be captured in a single picture. But then they found that just putting a picture into a robot’s brain was useless. First they’d need to program the brain with the capacity to appreciate the picture. So about halfway through, this dystopian novel ran smack into the mystery of what it means to be human.