Have you ever unknowingly spoken to someone who happened to be a computer? Take Scarlett Johansson’s character in Spike Jonze’s romantic drama Her (2013), for example. Theodore Twombly, played by Joaquin Phoenix, falls in love with an artificial intelligence on his phone, Samantha. If one were to just listen to the audio, it would seem that Samantha is perfectly human in the way she thinks on her own and is able to carry out fully fledged conversations. As odd as that sounds, there’s actually a method to test how “human” artificial intelligence appears to be: the Turing Test.
In an attempt to discover how intelligent modern day “artificial intelligence” really is, we take a look at a few commonly used AI systems and see how well they do against the Turing Test.
Passing the Turing Test
Alan Turing, an English mathematician, developed what is now the Turing Test in 1950. Essentially, a human is supposed conduct a conversation with a computer (without knowing that he is speaking to a computer) and ask the computer a few questions to determine if the artificial intelligence can mimic human responses. If the conductor is unable to identify whether or not the subject is a computer or a human, the computer successfully passes the Turing Test. The Turing Test has a few specific guidelines. Essentially, there’s someone to question both a human participant and the chatbot. After a selected amount of time, the questioner tries to decide which is the human and which is the computer. If the person cannot distinguish between the two, the computer is said to have passed the Turing Test.
Siri: AI for the iPhone
The initial release of Siri was on October 4th, 2011 with the release of the iPhone 4S. Written in Objective-C and Swift, this intelligent personal assistant is used on iPhones, iPads, Apple Watches, and Apple TVs. Siri is also available in multiple languages including even in Arabic, Italian, Thai, Dutch, and more.
Can Siri pass the Turing Test? Probably not. Siri would have to be able to convincingly carry out a conversation with a subject and be able to generate its own thoughts. So far, Siri only works with simple sentences and short phrases and is unable to carry out a full-blown conversation. Siri is easily identified and doesn’t feel completely human. According to Peter Nowak of Maclean’s, “In such regards, Siri is more of a programmed robot than a thinking entity. Somebody somewhere—or more likely, many people somewhere—have spent a good deal of time anticipating and then programming Siri with potential questions and their respective answers, humorous or otherwise.” In the future this might change, but for now, Siri ultimately fails.
Cleverbot: The Original Chatbot
Cleverbot was created by British scientist Rollo Carpenter and was launched in 1997. Since then, it’s held over 200 million conversations and has been learning and remembering words ever since. Cleverbot’s responses aren’t programmed, as its unique feature is that it learns from the conversations it has with humans.
Cleverbot passed the Turing Test on September 3, 2011. At the 2011 Techniche festival at the Indian Institute of Technology in Guwahati, judges determined Cleverbot passed for human 59.3% of the time, with a passing score being 50.05% or higher.
But does this mean Cleverbot is intelligent? Cleverbot’s developer and AI specialist Rollo Carpenter says otherwise. “My answer to that question is fairly unequivocal. No it doesn’t mean Cleverbot is thinking,” says Carpenter. “One could argue that there was perhaps a bit of intelligence involved in what is going on, but I think one would have to distinguish thinking from the word intelligence.” Despite Cleverbot passing the Turing Test, there’s still a huge difference between Cleverbot and another human.
Amazon Echo: Alexa the AI
Amazon released their voice-enabled wireless speakers on November 6th, 2014, initially sold to Amazon Prime members and other invited members. Amazon released their “smart speaker” to the general public on June 23rd, 2015. The system, named Alexa, is capable of voice interactions and can even make to-do lists, set alarms, provide weather and traffic updates, and more. It’s noted for having a natural and lifelike voice and goes beyond Siri’s simple ability to just answer questions. Heather Kelly of CNN says, “It takes the intelligence and voice-recognition of Siri and puts it someplace unexpected, inside a smart home device. Even when it’s asleep, Echo is still listening for the trigger phrase that will turn it on and start streaming your voice to the cloud.” Alexa’s ability to act as a smart home device and its ability to connect to other devices will change everything, from checking out the weather to changing the thermostat in an instant.
But can Alexa pass the Turing Test? It certainly has potential. It’s always on, meaning that a natural conversation with Alexa can happen at almost anytime, unlike Siri. As an easter egg (an unexpected feature or hidden message within a piece of software), when asked if it could pass the Turing Test, user theduderman from reddit.com reports Alexa replies with, “I don’t need to pass that, I’m not pretending to human.” Obviously not self aware, but it’s perceived sentience is a noteworthy detail.
Similar to Cleverbot, it seems Alexa is also intelligent, but still very far from the ominous and looming destruction that AI is portrayed as in various works of fiction.
Bonus Round: Ultron
Ultron is one of Marvel Comics’ fictional supervillains that first appeared in The Avengers #54. He’s known for being an artificial intelligence with genius-like intellect, superhuman strength, flight, and more. Most notably portrayed by James Spader in the 2015 sequel Avengers: Age of Ultron, this yet again menacing figure starts off as a global defense program that eventually uses its sentience to build an army of robots to take over the world.
Can Ultron pass the Turing Test? Ultron could pass, not only due to James Spader’s exceptional performance, but because of Ultron’s intellect and sentience. He is fully able to philosophize, converse, and think like a human. If one were to ignore his cold, metal outer shell, it would be easy to mistake Ultron for a human. It’s important to note that while Ultron is a fictional example, the foundation for his character is deeply rooted in the fear of the machine. Even today, the fear that grew out of the 1950s and 1960s rise of artificial intelligence still lingers, continuing to portray AI as some unstable monster that will one day outgrow our control.
The Future of AI and the Turing Test: Where to next?
What of the Turing Test itself? Stuart Russell, an AI researcher at the University of California, believes the test is almost worthless. He says, “It wasn’t designed as the goal of AI, it wasn’t designed to create a research agenda to work towards. It was designed as a thought experiment to explain to people who were very skeptical at the time that the possibility of intelligent machines did not depend on achieving consciousness, that you could have a machine that would behave intelligently because it was behaving indistinguishably from a human being.” Carpenter also has this to say about the Turing Test: “Passing the test does not prove intelligence but merely shows that the machine can imitate intelligence. That doesn’t mean attempting to pass the test is pointless, however.”
While the Turing Test appeared to be the ultimate test of man and machine, it’s since lost its worth. As Carpenter says, it’s a test to imitate intelligence, not a test of intelligence itself. Examples such as Cleverbot, Alexa, and Siri may seem intelligent, but we have long way to go until AI systems become legitimately intelligent. Until then, our phones and electronic devices have no intention of destroying all of humankind.
As we continue down the path to Singularity, the age of the machine, it’s interesting to note these milestones along the way: intelligent personal assistants on our phones, computer programs that learn and mimic human thought on our computers, voice-enabled speakers, and even robot armies destined on becoming a catalyst for the inevitable Singularity in our movies. What’s next, AI’s that book our flights? Where does it end? Tune in next week when we’ll discuss the future of artificial intelligence and philosophize about what it really means to be a human.