Alan Turing was a renowned computer scientist and mathematician and a pioneer of artificial intelligence and machine learning research. It was Turing who proposed a test for distinguishing between “true” artificial intelligence and the machines that can kind of replicate it.
The Turing Test
The Turing Test is rather simple: a human converses with a machine (either verbally written) without seeing it. If the human cannot tell whether the machine is human or not, it is said to has passed the Turing Test. Artificial intelligence and machine learning research has come a long way since Alan Turing’s death in 1954 but we still haven’t been able to create a machine capable of tricking a human into thinking it’s human. For a good example of how far we’ve come and how far we have to go, look no further than the growing popularity of chatbots.
What are chatbots
Chatbots are computer programs that can be deployed on SMS or other messaging platforms to communicate textually with humans. They’re a perfect real-world example of the Turing Test in action. The problem is that it doesn’t take humans very long to realize that the entity they’re chatting with online isn’t human.
What chatbots can and can’t do
Most chatbots are designed to do one or two simple tasks that humans ask of them or they might be programmed to select the appropriate response to a predetermined set of frequently asked questions. More advanced programs like Amazon’s Alexa or Apple’s Siri have been programmed to respond to a wider number of questions but they still fall short of passing the Turing Test because they aren’t actually able to think as humans do. They have access to vast amounts of information that they can draw from to answer questions or perform certain tasks. What they can’t do is infer meaning from language. They can select an appropriate response to a question when certain keywords are used or when it’s worded in a way they’re programmed to recognize. But they can’t parse language as humans and take into account the context of when it was said and by whom and with what tone to know what the person is meaning.
What are they good for
More intelligent chatbots that can fool humans and pass the infamously difficult Turing Test may yet be developed but in the meantime, the chatbots that are being developed serve a purpose. They automate customer service and free up human resources by fielding the easy questions and handling the more mundane parts of the job that require less critical thinking. And many consumers prefer dealing with chatbots than having to get on the phone and be put on hold.