The Chinese search company Baidu said its Minwa supercomputer has set a new benchmark for artificial intelligence by accurately identifying all but 4.58 percent of a set of 1,000 pictures. Humans have an error rate of about 5 percent, while AI software from Microsoft has a 4.94 percent error rate and Google's is wrong on 4.8 percent of its guesses.
Baidu unleashed its Minwa supercomputer on ImageNet, a vast set of images that test computers' deep learning skills by asking them to differentiate between different dog species, for instance, or a french loaf and a meatloaf, according to the Wall Street Journal. Google previously used its own deep learning skills to improve Android voice recognition, though Baidu is using a more advanced supercomputer to sift through Chinese and English language search information.
“I am very excited about all the progress in computer vision that the whole computer community has made,” Andrew Ng, Baidu's chief scientist and a former Google project leader, told the Journal. “Computers can understand images so much better and do so many things that they couldn't just a year ago....It's interesting that the top three teams processing ImageNet all appear to be large tech companies with considerable computational resources.”