There’s a new chapter in the saga of over-enthusiastic researchers trying to make the Matrix a reality. Scientists at Kyoto University are training an AI-powered robot to laugh. This is, of course, a difficult project — one that could take decades to become a true success. Yet, if a machine can mimic the myriad ways in which human beings laugh, it may finally be able to cross the Uncanny Valley — that eerie feeling of uneasiness when an AI-enabled creature can copy human behaviour, but somehow feels unnatural and alien in interactions. Two questions arise from this desire to play God’s understudy. First, can the changing, contextual nuances of what makes people chuckle, giggle and guffaw be programmed? And is it worth the effort?
The ostensible reason to make machines cognisant of how people laugh is to teach them empathy. This will make them more “human”. But if empathy is programmed is it really empathy? Laughter, after all, is the most curious of human responses. Often, huddled at a funeral, people will smile at the silliness and the warmth of the departed — as much an act of love as dealing with grief. Or the soft chuckles and faint giggles with a paramour, two drinks down, at an over-priced restaurant. Then there is cruel laughter, the one at the expense of someone else’s suffering or worse still, the chilling cackle of the bully and the mob as they hit and kill. A machine can, as yet, copy the sound. But it can’t understand the feeling.
The best comedic actors will talk of “playing the truth of the scene” and not “going for the laugh”. The fake laugh, then, is a lie. Lying machines, of course, are a terrible idea — especially as AI is set to be involved in more and more vital activities, from banking to medicine. But what if the robot can actually gain sentience and truly feel the humour? In that case, let’s hope it does not learn the cruelties of human humour, just the mirth.