The decade isn’t over yet, but we’ve seen some remarkable advancements in the field of artificial intelligence. We’ve marveled at the invention of the first self-driving car in 1995. We witnessed Deep Blue beat Garry Kasparov in 1997. Lastly and more recently, we enjoy the company of Apple’s Siri, Google’s the Assistant, Microsoft’s Cortana, and Amazon’s Alexa. The advancements don’t end there.
While much advancement in artificial intelligence came about relatively recently, the idea of a machine-based artificial intelligence actually existed even before the computer. Its theoretical basis came about in the 1950s, introduced by British mathematician Alan Turing.
Alan Turing and The Imitation Game
In 1951, Alan Turing proposed a test in a paper called The Imitation Game. The test aimed to finally settle the issue of machine intelligence. Its first version did not involve computer intelligence and instead involved three people: a man, a woman, and a judge. The objective was for the man to help the judge decipher his manhood without any physical cues while the woman attempted to deceive the man and erroneously identify her as a male.
Turing then proposed a modification of the game, in which a computer was involved and the judge would attempt to decipher which of the contestants was human and which one was a machine. Under these conditions, if the judge were less than 50% accurate, i.e. if the judge were as likely to pick either the human or computer, then the computer can be considered intelligent since it passes as a simulation of a human being. This process now came to be known as the Turing Test.
The Turing Test
In computer science, the Turing Test is done to demonstrate the ability of a computer or machine to simulate human intelligence and be able to pass as “human”. The test can be described as a test on the quality of artificial intelligence displayed by a machine, carried out through a text-only conversation with a user. Essentially, the machine is able to pass if the user cannot tell the difference between the artificial intelligence and a human being chatting the same way.
This was demonstrated somewhat in 2014 when Reading University claimed that a team of Russian scientists had successfully devised an algorithm that could pass a Turing Test. They claimed that they had managed to persuade one in three people on a team of judges that it was a 13-year-old Ukrainian boy named Eugene Goostman through a series of online chats.
While the Turing Test doesn’t really hold much value in the human world yet, it does mean serious business in the world of science fiction. In a world full androids, humanoids, and high-level artificial intelligence, the Turing Test is perhaps one of the best tools for deciphering between metal and silicone and flesh and bones.
Artificial Intelligence and the Turing Test in Science Fiction
The first idea of artificial intelligence in science fiction can perhaps be traced back to Mary Shelley’s Frankenstein, which gives a look at Dr. Frankenstein’s successful attempt to create life – but not intelligence. The created being was treated as a monster and goes on a rampage, blaming the doctor’s treatment of him as a monster as the reason for his monstrous behavior – a trope widely depicted in future depictions of artificial intelligence.
The AI-killing-humans-because-they-are-becoming-a-threat trope has been portrayed countless times in movies such as HAL 9000 in Stanley Kubrick’s 2001: A Space Odyssey, SkyNet and its robotic assassins in Terminator, Gort from The Day the Earth Stood Still, and many more. All of these portrayals serve as a reminder of the precautions that, even though they ultimately do not succeed, AI takes due to humanity’s actions. They point to a future that Turing describes as when machines have over humanity.
But where does the Turing Test fit in all of this?
In futuristic time periods where machines have humanoid forms and possess superior artificial intelligence, a robot or piece of machinery’s ability to pass a Turing Test in its original form is highly probable, considering the original’s basis for passing the test. The results of highly-advanced, sentient, super computers such as HAL 9000 and SkyNet attempting to fool humans through a text-based conversation for a short period of time is certainly never in questions. This begs the question of whether the Turing Test can be a true test for distinguishing humans and artificial intelligence.
One of the flaws of the Turing Test is highlighted in Ex Machina where the judge is already aware that he is testing a machine. Another flaw is when the test guidelines are relatively too simple that even a Chatbot such as Eugene Goostman can pass the test. So how will highly-intelligent supercomputers be tested then?
Perhaps the answer lies in modifying the actual test itself. Several modifications to the Turing Test have been proposed in order to accurately distinguish humans and artificial intelligence. In Blade Runner, a much more accurate Voight-Kampf test was used to distinguish replicants from humans. In the movie, the fictionalized twist on the Turing Test is designed to catch replicants, highly-intelligent humanoids, posing as humans.
Why Does the Turing Test Matter?
In the world of science fiction, passing the Turing Test, or any modified form, might mean the difference between keeping a robot slave and destroy all metal so humans can continue to exist. However, in the real world, the threat of killer robots and humanoid assassins is but a far cry from the true situation as we already live in a world of cheeky chatbots and relatively intelligent voice assistants right in our phones.
In the real world, passing the Turing Test doesn’t really hold any value. In fact, it should be considered more as a test for humans and not for robots. Moreover, it is a testament to a programmer’s ability to create complex algorithms capable of mimicking human behavior. It is a way for humankind to gauge this ability and its progress in creating more intelligent machines that could later hold much value in our ever-changing world.