Going bananas over artificial intelligence

In the highly polarising world of polarised opinions, there is a new entry that involves a robot and a banana. This is yet to create the ruckus that it deserves. But it is a matter of time till a dual-armed robot developed by researchers at the University of Tokyo will divide opinion among sentient bipeds - not about whether machines will take over our workplaces and lives, but whether this particular robot can be considered reliable in carrying out its function. This function happens to be peeling a banana without squishing it.

The robot, trained by a 'deep imitation learning' process, reportedly picks up and peels a kela in three minutes, which, to be fair for a deceptively delicate manoeuvre, is a decent start. But the quibble lies somewhere else. After 13 hours of training, the banana-peeling robot's success rate is reportedly 57%. From any majoritarian's point of view - optimistic and brimming with 'positivity', one would say - this spells success, and a future where we won't have to peel our own banana. For the less starry-eyed, however, the 43% failure rate spells doom for a mechanical fruit-handler. Does the future of robotics depend on which view comes up trumps? Perhaps. But to tweak the words of Francois de Charette, a Royalist during the French Revolution, one cannot make a robot without squishing a few bananas.