Human or Machine?
Where we learn that humans are passing Turing Tests every day, but that computers are now also getting there.
Recently, when logging into an account on a website, I was asked to solve a little visual puzzle, like decoding a few letters in various shapes and orientations over some confusing background of lines and dots. I’m sure you’ve encountered these so-called CAPTCHAs, too. And like me, with a little bit of visual care you probably guessed the right letters.
A CAPTCHA is used to tell if you are a human, since it is difficult for computers to solve. So when you correctly pass a CAPTCHA test, you are actually succeeding in a fundamental test of computer science, which is called the Turing test, named after the “father” of computer science, Alan Turing. But where does the idea of the Turing test come from?
Shortly after the invention of computers in the 1950s, researcher immediately started to think about the meaning of “artificial intelligence”. It quickly became clear that computers would be able to play chess (and one day even win), and that they were good at pattern recognition, which in turn leads to image and speech recognition. This thus immediately led to the question of how far computers could go in imitating human intelligence.
To address this, Alan Turing devised a theoretical test, which would tell apart humans from computers. Its simplest form goes as follows: assume you ask questions to a computer over an interface (for example via a website), and from the responses you get, you cannot say if it was a human being or a computer that answered. If this is the case, then the computer passed the Turing test, according to Turing. And in fact, on many topics, computer programs do indeed pass the Turing test, and often they actually even surpass human performance. An example is the game of chess, where world champion Garry Kasparov was beaten by a computer called Deep Blue by IBM in 1996.
A funny episode occurred also in the 1960s, when a chat software created by Joseph Weizenbaum from the Massachusetts Institute of Technology tricked many users into believing that it was a person chatting with them — even if it was a program! For many users, the chat software thus did pass the Turing test.
So when you solve a CAPTCHA question to log into a website, the provider is asking you to pass a Turing test to make sure that it is you, a human, that is entering the website, and not a robot — and so far, you win! This makes your account safe. However, in 2017 a few computer scientists from the United States managed to develop a program that was able to solve CAPTCHAs. This means for example that hackers can now again use programs to automatically enter websites, and cause trouble. The battle between man and machine seems ongoing… to be continued.
