A research team from the School of Electrical Engineering and Computer Science at Washington State University has received a grant of more than a million dollars from DARPA (a department of the United States Department of Defense) to prepare a test to test intelligence of AI systems ( AI = Artificial Intelligence = artificial intelligence ) can be measured.
In the past, considerations about the intelligence of AI systems were primarily theoretical. The researchers took into account neither the actual performance in new, until that unknown environments nor the complexity of the tasks to be performed. However, the research team is now working on a concrete test that assesses AI systems depending on the degree of difficulty of the problems to be solved – basically the same as with human IQ tests.
Important here is the ability to learn in unfamiliar environments – something that excels in. Many AI systems are optimized for very limited, specialized tasks such as recognizing faces and the like. Machines that learn and act intelligently in new environments are still the holy grail of AI research. The measurement of performance in such situations therefore forms the basis for the development of more flexible systems that, for example as assistant robots, can help people with many daily tasks. Interesting is the extent to which these AI systems can transfer what they have already learned to new assignments. So far, for example, learning drafts for learning to play chess has improved performance in the latter, or that previous video game training helps solve problems such as Rubik's cube.
The CASAS ( Center of Advanced Studies in Adaptive Systems ) department of the WSU also deals with robot assistants for the elderly. These could make a positive contribution to the safety, health, mobility and social life of older people. In that respect, their AIQ website also includes rankings.
Writer: https://www.elektormagazine.nl/authors/55/thomas-schererTags: AI, EQ, IQ, Kunstmatige intelligentie