Categories
Latest
Popular

Artificial Intelligence Convincingly Pretends to Be a Real 13 YO Boy

By Alejandro Zorrilal Cruz [Public domain], via Wikimedia Commons

By Alejandro Zorrilal Cruz [Public domain], via Wikimedia Commons

This will either amaze you or creep you out. A supercomputer named “Eugene” is now considered as the first artificial intelligence to demonstrate the ability to think. It is said to be capable of simulating a real human’s ability to engage in a conversation. It has become the first machine to pass the Turing Test.

The Turing Test

Developed by Alan Turing in the 1950’s, the Turing Test is designed to measure a machine’s ability to demonstrate intelligence. This level of intelligence could be equal to or comparable to that of a human’s. It involves human judges who engage the machine in natural language conversations. A machine should be able to respond in a manner that is indistinct to the way a human acts or responds in natural language conversations. Conversations are done using a text only channel similar to this online Turing Test page. Obviously, voice-based conversations are not used since it provides a quick hint on which is human and which is not.

At present, the Turing Test is regarded as an essential concept in the philosophy of artificial intelligence but it has its share of critics. It is considered to have tractability, simplicity, and subject matter breadth as its strengths. On the other hand, it is criticized for its inability to properly distinguish human intelligence against intelligence in general, real intelligence against simulated intelligence, and the intellectual capacity of interrogators against anthropomorphic fallacy.

By Mushii (Own work) [GFDL (http://www.gnu.org/copyleft/fdl.html) or CC-BY-3.0 (http://creativecommons.org/licenses/by/3.0)], via Wikimedia Commons

By Mushii (Own work) [GFDL (http://www.gnu.org/copyleft/fdl.html) or CC-BY-3.0 (http://creativecommons.org/licenses/by/3.0)], via Wikimedia Commons

Passing the Test

Eugene passed the test by being able to make 33% of the human judges believe that it was a 13-year-old boy. The Turing Test requires only 30%. No other computer in the past has been able to achieve this feat. The specific test (Turing Test has a number of variations) Eugene was subjected to involved a series of five-minute text-based conversations.

The tests were conducted at the Royal Society in Central London. It was organized by the University of Reading. There were five machines tested. There were 30 human judges involved. Ten of the judges failed to properly identify which one was the human and which one was Eugene or a failure rate of 33%. According to Professor Kevin Warwick of the University of Reading, the test on Eugene is a milestone that “will go down in history as one of the most exciting.”

Eugene the ‘Intelligent’ Machine

Eugene Goostman or simply “Eugene” is a collaborative creation of Russian-born Vladimir Veselov who is based in the United States and Ukrainian Eugene Demchenko who resides in Russia. Conversing with Eugene means using a plain interface with a photo of Eugene on the left side, a text field and reply button at the center, and the name Eugene Goostman on the upper lefthand corner with the tag line “The Weirdest Creature in the World.”

Is the Turing Test a Real Measure of Artificial Intelligence?

Despite the supposed “breakthrough” in creating artificial intelligence, there are those who think that Eugene’s feat wasn’t really worth celebrating. Some even call it a fake victory for artificial intelligence. The following are a few of the criticisms over Eugene’s supposed breakthrough:

  • Eugene is not a supercomputer but merely a chatterbot that responds to questions or conversations using scripts. It can be compared to the likes of Siri and Cortana. It can even be considered significantly inferior since it cannot interact verbally with its lack of speech recognition technology.
  • The creators “cunningly” intended Eugene to be a 13-year-old Ukrainian boy who is not that well-versed with the English language. This is perceived as a form of cheating as it provides a convenient fallback in instances when Eugene does not know enough about a topic being discussed and is unable to correctly interpret idioms or understand differences between words with varying meanings.
  • According to Professor Stuart Russell of the University of California, Eugene did not pass the Turing Test because the test was not meant to be an applied test. Russell claims that the University of Reading test was being too literal in their interpretation of how Turing Test should be done. He claims that Turing had a more sophisticated idea of how the conversations should flow. Russell pointed out certain conversations with Eugene that show a pattern of repetition and the inability to sustain conversations without trying to bring up a new topic to avoid answering questions that appear ambiguous to the computer.
By en:User:CharlesGillingham, User:Stannered (en:Image:Weakness of Turing test 1.jpg) [Public domain], via Wikimedia Commons

By en:User:CharlesGillingham, User:Stannered (en:Image:Weakness of Turing test 1.jpg) [Public domain], via Wikimedia Commons

Whether or not Eugene has really set a new benchmark in the field of artificial intelligence, it’s good to know that discussions on artificial intelligence is pretty much alive. The differences in opinions is a good sign that there is interest in the technology and that real progress in AI development may soon be achieved. It’s good that people don’t quickly shun the idea of the possibility of being able to develop AI that can match the human level of intellect.