How does artificial intelligence compare to human? Currently – and for 65 years – the method used by scientists to evaluate and define whether artificial intelligence is capable of being compared with that of humans is the Turing test (in honor of the scientist Alan Turing). Now, a professor at the Georgia Institute of Technology (USA), Mark Riedl, has put on the table a new way to carry out this evaluation.
The new test, known as Lovelace 2.0 (named after the mathematician Ada Lovelace) and which is an update of the one presented in 2001, is based on the fact that the machine in question must be able to write a fictional story, create a poem or elaborate a painting to pass the test .
“To pass this test, the artificial agent must develop a creative artifact from a series of artistic genres that require a minimum development of intelligence. In addition, the artifact must meet certain limitations that are imposed by the human evaluator . Creativity is not exclusive to human intelligence, but it is one of its hallmarks of identity ”, explains Riedl.
At the moment no machine has been able to pass the Lovelace 2.0 test and the reactions to this alternative to the Turing test are found: “There is a popular belief that humans are fundamentally different from artificial intelligence, because humans are creative and AI can only give rational answers, ”says Alan Woodward, a computer technology expert at the University of Surrey (UK).
