Communications of the ACM
Georgia Tech Professor Proposes Alternative to 'turing Test'
A Georgia Institute of Technology professor has developed a way to evaluate the intelligence of a computer that he says is more rigorous than the Turing Test.
Credit: Juliana Jimenez Jaramillo; Creative Commons
Georgia Institute of Technology professor Mark Riedl has developed what he describes as a more rigorous way to evaluate the potential intelligence of a machine or computer than the famous Turing Test.
Riedl notes the Turing Test, developed by renowned computer pioneer Alan Turing in 1950, was never meant to act as an official test of machine intelligence.
Riedl's new test, the Lovelace 2.0 Test of Artificial Creativity and Intelligence, is based on the Lovelace Test from 2001 and involves the artificial agent being tested having to produce a piece of art or other creative work.
In the original test, the artificial agent had to create work the production of which the agent's designer could not explain and that was valuable, novel, and surprising. Riedl believes the original test was too subjective and has created specific parameters for his test that must be followed by a human evaluator. This allows the evaluator to assess a machine's creative work without making subjective value judgments.
Riedl will be presenting his new test at the Association for the Advancement of Artificial Intelligence's Beyond the Turing Test workshop in Austin, Texas in January.
From Georgia Tech News Center
View Full Article
Abstracts Copyright © 2014 Information Inc., Bethesda, Maryland, USA
No entries found