An empty chair
Imagine a chair beside a window. Nothing remarkable about it. A little light on the arm, a floor that needs sweeping, another chair close enough for a conversation. Now imagine someone waiting there. I have not said who they are waiting for, or how long they have been waiting. Already the room feels different. The furniture has not moved. The meaning has.
Put a machine in the other chair. Let it speak gently. Let it remember the right details, respond at the right moments, notice that the person has not moved in a while. It can do useful things. Important things. I do not want to make those things small. But I keep looking at that second chair and wondering what exactly has arrived.
Working with robotics makes this question uncomfortable in a useful way. I can think in terms of movement, sensing, control, the very concrete difficulty of getting a system to respond to the world. Those difficulties are real. A thing that looks effortless from the outside can require an absurd amount of work. Still, solving the movement does not solve the meaning of the movement. Coming closer is not always the same as coming to visit.
The right response
There is a temptation to describe care as a sequence of successful responses. Someone needs help, help arrives. Someone speaks, something answers. Someone is alone, the room becomes less quiet. From a distance this looks almost complete. The inputs were understood. The outputs were appropriate. What else, I find myself asking, should there be?
Then I think about the difference between being remembered and having information about me stored somewhere. The information might be identical. My name, what I like, the thing I mentioned last week. But being remembered means that I occupied some part of another person’s life while I was not in the room. I mattered in my absence. I do not know how to reduce that without losing the part that matters to me.
This does not mean a machine must become a person before it can help one. A handrail does not love you. A lamp does not understand the dark. Both can make a life easier to inhabit. Perhaps I create the wrong problem when I ask a useful thing to justify itself by becoming something else. Perhaps the machine can be allowed to be a machine, and the person can still be treated as a person.
What I want to build around
The worry is not that technology will fail to imitate us convincingly. Sometimes I worry about the opposite. A response can become convincing enough that the people around it stop asking what remains missing. The room sounds occupied, so we stop visiting. The system reports that a task is complete, so we decide the need has ended. I do not want to confuse the absence of a problem with the presence of a person.
I want to build things that leave room for us. Tools that take away some of the difficulty of caring without quietly taking away the responsibility. A machine might carry something, notice something, make a place easier to move through. These are worthwhile ambitions. They do not become less worthwhile because they cannot answer the entire question of loneliness.
And I have to be careful with my own certainty here. The person in that chair may feel comfort from a machine in a way I do not understand. It would be another kind of arrogance to stand outside their happiness and declare it invalid. I have made that mistake with other kinds of comfort. I would rather ask what their experience is than tell them what it ought to be.
So the second chair remains a question. A machine can sit there. It can be useful, perhaps even welcome. I only hope that when we make it good at arriving, we do not use that as an excuse to stop showing up ourselves.