Chatbot passes Turing Test? Some aren't so sure

Critics raise questions about news that a conversation bot passed an AI test

A computer-powered chatbot that supposedly passed the Turing Test for artificial intelligence may not be all it's cracked up to be.

The conversation program, dubbed Eugene Goostman, poses as a 13-year-old Ukrainian boy and supposedly tricked a third of the human testers into believing the software was the adolescent during an event Saturday at the Royal Society in London. The announcement came from Kevin Warwick, a visiting professor of cybernetics at the University of Reading, an AI and cybernetics promoter with a history of making bold statements about the intelligence of machines.

Back in 2000, Warwick called himself the first cyborg after implanting a computer chip in his arm. On Monday, tech blog TechDirt recounted Warwick's history with the press and questioned the latest Turing Test results.

Warwick, in an email, defended the tests run on Goostman. TechDirt's critique of the test "picked certain bits of the press release and 'changed' some of the words to suit," he said.

The AI tests last Friday and Saturday were open to the public and "under constant independent scrutiny," from a team led by AI professor John Barnden of the University of Birmingham, Warwick said. The Goostman chat program achieved the 30 percent pass mark as set by Turing Test creator, the late Alan Turing, he noted.

"It is quite a tough challenge and that is why the Turing test has taken so long to pass," Warwick said. "There is no doubt on Turing's meaning."

While transcripts of the weekend's tests aren't immediately available, academic papers based on the recent tests will likely include conversations between Goostman and the human testers, Warwick said.

Several AI experts didn't return messages seeking comment on the tests, but Goostman's responses, on an online version of the program, were rather unconvincing in tests run by Time.com and by IDG News Service. It's not clear if the version tested here is the same one as was tested in London.

This is the text of a chat between IDG News Service and the Goostman chatbot:

IDG: Hello. How are you today?

Goostman: I'm in a great mood today.

IDG: Why are you in a great mood?

Goostman: Nice to hear that! Do you like the weather?

IDG: No. It's too hot. Do you like the weather?

Goostman: In my native town people says that meteorologists are like sappers: they make a mistake only once, but the difference is that sappers do that once in a lifetime and meteorologists -- daily. Do you like weather today?

IDG: I think I already answered that.

Goostman: Is this all you can come up with? I know you are supposed to trick me.

IDG: It appears I'm doing a good job tricking you. No?

Goostman: Yeah, it's lovely day!

IDG: What makes it lovely?

Goostman: ... wait

IDG: What am I waiting for?

Goostman: Could you formulate your question in more precious way? Could you tell me what are you? I mean your profession.

IDG: I'm a technology journalist.

Goostman: Journalist is not so bad profession -- especially for inquisitive people.

Grant Gross covers technology and telecom policy in the U.S. government for The IDG News Service. Follow Grant on Twitter at GrantGross. Grant's email address is grant_gross@idg.com.

Copyright © 2014 IDG Communications, Inc.

It’s time to break the ChatGPT habit
Shop Tech Products at Amazon