Computers become more and more intelligent. But how smart are they in reality? A research team around the key researcher Feng Liu at the Chinese Academy of Sciences in Peking developed now a test which shall answer this question. Main result was, that humans still by far smarter than their digital assistants.
With the test a ranking was made for existing digital assistants such as Google Assistant and Siri. According to this list Google Assistant is number one, Siri holds the last position. In summary all assistants are less intelligent than a 6 year old child.
(c) image by Apple
The test is based on a so called standard intelligence model which requires that the tested systems acquire data from external sources. Furthermore they must be able to transform these data so they can execute and analyze them and to use the generated knowledge in an innovative way. Last but not least the new generated knowledge must be returned to their surroundings. Briefly this can be summed up as: collect data, interpret data, creatively use data and produce an intelligent output. When a system offers these capabilities it can be defined as a standard intelligent system.
The by Feng and his colleagues developed test shall measure the skills of computers (but also humans). However, the researcher keep the test itself and how it exactly works a secret. This is understandable, because once the exact function of the test is known developers may construct their programs exactly in a way that the test is positive. On the other hand the results of a test which functionality is not known remain unclear.