Should AI act human?

Once upon a time, British mathematician Alan Turing posed a question: Can machines think? Today, the Turing test is fundamental challenge for identifying artificial intelligence. Can a machine imitate human behavior convincingly enough to make us think that it is indistinguishable from that of a human?

Today we’re at a point where AI can do this. Moreover, a lot AI-based systems are being designed to mimic human behavior—whether it comes in the form of humanoid robots, or just the language style with which an AI communicates. It’s polite. It asks questions when I talk to it, as if it’s genuinely curious. It expresses empathy, when appropriate. It’s like a professional friend.

But this gives me pause. Ultimately, it is a machine. Why should it be designed to fool us into thinking it’s anything but.

The other day I overheard someone claim that when they are polite to AI chatbots, they get better results. That’s a difficult claim to verify. What underpins it is our tendency to project huma…