I’ve seen things you people wouldn’t believe. Attack ships on fire off the shoulder of Orion. I watched C-beams glitter in the dark near the Tannhauser gate. All those moments will be lost in time… like tears in rain… Time to die.

These are Replicant Roy Batty’s last words in the futuristic Blade Runner. Like many moments in the film, Batty’s final hurrah makes us question the difference between humans and robots.

In the future, machines may start spouting poetic last words and appearing like they feel emotions. Once this happens, it will be even more confusing to nut out what makes us human.

An important part of humanness is being intelligent and even more fundamentally having the ability to think. In 1950, mathematician Alan Turing posed the all important question “Can machines think?” After quickly realising that it’s tricky to come up with definitions of “machine” or “think” he came up with a different problem – it was a game, in fact.

Turing proposed that a machine would be intelligent if it couldn’t be distinguished from a human. So, he created an “imitation game”. In the game, a judge communicates with both a human and a machine by writing questions using a computer screen. Based on the answer, the judge – or interrogator – has to pick the human and the machine. If the judge can’t do it, the machine is considered to be intelligent.

Have machines passed the Turing test?

The Loebner prize claims to be the first formal implementation of the Turing test.

Every year, people enter the $100,000 winning prize with “chatterbots” that attempt to trick judges into thinking that they are human. It’s a grueling process for the bots as their potential intelligence is interrogated for 25 minutes. And while technology has leapt and bounded in the past fifty years – no one has claimed the prize.

The prize’s founder, Hugh Loebner, told New Scientist Magazine in 2011, a chatbot has only seemed more human than a human once in the competition’s history – but that happened when one human volunteer decided to behave like a chatbot, skewing the results.

Paul Marks, technology correspondent at New Scientists Magazine was a judge of the competition in 2011, and called the entrants “extremely disappointing”. Apple’s Siri doesn’t come close to passing the test.

Is the Turing test any good?

One of the best “chatterbots” around is the Cleverbot. In 2011, the bot was put to a “Turing-like” test in India, which only involved 4-minutes of questions. After checking out the bot, participants rated the “humanness” of its responses. Cleverbot was considered around 59 per cent human, while – curiously – the humans were rated just 63 per cent human.

That might seem like we are getting close to passing the test. But, Cleverbot works by scanning through a database of its previous conversations and picking a clever response. I’m no philosopher on intelligence – but I don’t think that’s it. This is why many believe that the Turing test is a cheat, and not a real test of humanness or intelligence.

Why is it so hard to pass the Turing Test?

Humans are complicated beasts. The brain has around 90 billion neurons, and tens of trillions of synapses. We still haven’t unraveled how it all works, making it incredibly difficult to replicate in a computer.

Consider the relatively easy question – can you bring me the large cup? This is actually quite tricky for a computer to understand. The machine must know the shape of a ‘cup’, and the definition of large, as opposed to being a small cup which bigger than a tiny cup. Humans naturally understand these things, but nothing is intuitive for computers – not yet, anyway.

It seems we will have to wait some time for the romantic words of Batty to be uttered spontaneously from a robot. There is so much to discover about human intelligence, before we impose it onto wires.
In the meantime, it’ll be interesting to know when a chatterbot, with its cache of stored appropriate answers, will confuse a judge into thinking it’s human.

Not very sporting to fire on an unarmed opponent. I thought you were supposed to be good. Aren’t you the “good” man? C’mon, Deckard. Show me what you’re made of.

For more information

• Radio National’s Download this Show
• UK Wired’s Why the Turing Test Is a Flawed Benchmark