Big day today – for the first time a computer has passed the Turing Test. Here is a detailed news article with names and dates on Gizmodo.

This is big news indeed. Those who studied theoretical Computer Science may have heard of the Turing Test. It’s a test, (proposed by Alan Turing, the Godfather of Computer Science), designed to distinguish whether an entity is a human being or a mere machine with formidable artificial intelligence.

A typical Turing Test looks somewhat like this.

A human and a machine are separated by a non-transparent screen, so that the human can’t see whether the entity on the other side of the screen is another human or a machine.

A conversation takes place, the human asks the entity on the other side of the screen a series of questions, noting the other party’s responses. The questions do not have right or wrong answers, instead what matters is whether the obtained answers are “human-like” – the human operator can thus judge whether the conversation partner on the other side of the screen is a human being or a machine.

The idea is that even if eventually robots were to look and move just humans, they would still not be able to perfectly emulate human emotions and thought process and this would be betrayed by questions like “If you were a song, what song would you be?”, etc.

As an example, here is a Turing Test in Blade Runner.

If, on the other hand, a computer on the other end of the screen has managed to fool the human operators into believing that it’s a human being, the machine is said to have passed the Turing Test.

And this is what happened today, for the first time.  A computer has finally passed the Turing Test.  Few points to note though, before getting overexcited:

• Firstly, only 30 percent of human judges were fooled by Eugene (this particular machine’s name) – not quite 100% and not even the majority, but still a very significant number and the best result up to date by far.

• Also, the machine pretended to be a 13 year old boy for whom English was a second language thus “gaming” the test slightly and perhaps getting away with a few more linguistic errors and fact gaps.

But, nevertheless, this is still be best result so far and is a loud and clear indicator of similar things to come.