Mathematics is what computers do best, right? It is hard for us to divide the bill with friends in a restaurant while a modern computer can do millions of calculations in a single second. But human beings have an innate and intuitive numerical sense that helped us, among other things, to build computers capable of doing that.
Unlike a computer, a human knows when he looks at four cats, four apples and the symbol 4 that everyone has one thing in common, the abstract concept of “four,” without even having to count them. This illustrates the difference between the human mind and the machine, and helps explain why we are not even close to developing AI with the broad intelligence that humans possess. But now a new study, published in Science Advances, reports that an AI has spontaneously developed a numerical sense similar to human.
For a computer to count, we must clearly define what we want to know. Once we allocate some memory to keep the counter, we can set it to zero and then add an item every time we find something we want to record. This means that computers can count time (signals from an electronic clock), words (if stored in computer memory) and even objects in a digital image.
This last task, however, is a bit challenging, since we have to tell the computer exactly how the objects look before we can count them. But objects do not always look the same: variation in lighting, position and posture have an impact, as do any difference in construction between individual examples.
Modern artificial intelligence systems automatically begin to detect objects when they are provided with millions of training images of any kind, just as humans do.