"The Turing test is the most famous benchmark of artificial intelligence, but it is flawed. Now an addition that gauges a machine's visual skills has been proposed.
Devised by 20th-century mathematician Alan Turing, the test pits the conversational abilities of chatbots against humans. To pass, judges must be tricked into believing that a bot is human, based only on a typed exchange. But many researchers believe the test is sorely in need of an upgrade.
"It has served its purpose. Now we need Turing Test 2.0," says Aladdin Ayesh, who organised a symposium entitled Towards a Comprehensive Intelligence Test at the AI and Simulated Biology conference in York, UK, in April.
That's why Michael Barclay and Antony Galton at the University of Exeter, UK, and colleagues have created a test that asks machines to mimic some of our visual abilities.
Click below to continue to New Scientists interactive feature to try the test yourself."
New Scientist Turing Test
Published on September 17, 2011 00:40