The primary definition of artificial intelligence (AI), specifically strong AI, is the power of a machine to exhibit intelligence indistinguishable from that of a human. But researchers have long been divided on the best way to measure this quality. For several decades, they have resorted to some version of the Turing Test, named after the famed mathematician and World War II codebreaker, Alan Turing.
However, the Turing Test is not without critics. Originally called the Imitation Game, the test requires a machine and human to interact via some neutral interface. If the person is sufficiently fooled into thinking the machine is also human, then the test is said to be indicative of artificial intelligence.
The main contention is that with a mixture of clever coding and brute force processing, it is fairly easy to trick some people some of the time into thinking their interlocutor is carbon- rather than silicon-based.
Also, Turing himself never intended for the test to be used as an ultimate decider.
As AI comes back into popular focus with the advance of big data and machine learning, researchers are looking for alternatives to this inadequate device.
This had led Mark Riedl, an associate professor in the School of Interactive Computing at Georgia Tech, to develop the Lovelace 2.0 Test of Artificial Creativity and Intelligence, based on creative capacity rather than the ability to converse or deceive.
The Lovelace 2.0 Test continues along the lines of the original 2001 Lovelace Test, which required that a non-human agent produce a creative item in a novel way. But while the original Lovelace test does not include clear testing criteria, according to Riedl, Lovelace 2.0 gives the evaluator defined constraints distinct from value judgments.
“It’s important to note that Turing never meant for his test to be the official benchmark as to whether a machine or computer program can actually think like a human,” Riedl said. “And yet it has, and it has proven to be a weak measure because it relies on deception. This proposal suggests that a better measure would be a test that asks an artificial agent to create an artifact requiring a wide range of human-level intelligent capabilities.”
Riedl’s paper will be presented at Beyond the Turing Test, an Association for the Advancement of Artificial Intelligence (AAAI) workshop to be held January 25 – 29, 2015, in Austin, Texas.