A computer that can develop its language

The next generation of computers will probably be able to learn new things. As a Wallenberg Academy Fellow, Chris Biemann will produce computers that can develop their language, just as we humans do. By analyzing large amounts of text, the computer will, for example, be able to conclude that “stool” and “chair” are similar things. 

When computers were invented, we used them as calculating machines. They now fulfill many more functions, but only those decided by the programmer. Their function is static and does not change. To further widen computers’ areas of use, researchers are now developing a cognitive generation of computers, which should be able to learn things and themselves refine their abilities. 

An embryo of this already exists in, for instance, search engines, which take account of the internet pages you previously visited when they present search results. Dr. Chris Biemann at Technische Universität Darmstadt is one of the researchers who want to take this much further. He is developing computers’ linguistic ability. Just like a child, a computer should gradually learn a language. The more text the system scans, the better it will be at understanding the meaning of different words and how they relate to each other. 

If computers have a functioning language, we can start to interact with computers in an entirely new way. Perhaps the talking robots of science fiction films will become reality. 

As a Wallenberg Academy Fellow, Chris Biemann will be based at the University of Gothenburg.

Photo: Markus Marcetic