Washington State University researchers are creating the first-ever ‘IQ test’ for artificial intelligence systems that scores systems on how well they learn and adapt to new, unknown environments.
Diane Cook, Regents Professor and Huie-Rogers’ Chair Professor and Larry Holder, Professor in the School of Electrical Engineering and Computer Science, received a grant of just over $1 million from the Defense Advanced Research Projects Agency (DARPA) to create the system.
Cook and Holder are designing a test that will grade the performance of AI systems based on several criteria, including the difficulty of problems they can solve; accuracy; time taken to solve the problems, and amount of data they require to perform well.
As most AI algorithms can only perform specialized tasks like making predictions or detecting faces, they hope that a credible scoring method will provide insights into developing machines that can learn and act intelligently under new circumstances. More specifically, the research team are focusing on how well the AI machines they test can transfer what they have learnt from one task and apply it to new and previously unperformed tasks.