Artificial Intelligence and Visual Art

In recent years, numerous attempts have been made to use digital computers to produce visual artworks [1] and some to understand their characteristics [l(a)-3]. This may seem rather odd if one thinks of computers as devices that merely do arithmetic. However, computers are widely believed today to be universal symbol-processing systems, and, if this is so, then they should be capable of analyzing artworks. I share with some of my colleagues in computer science a strong feeling of optimism about what can be learned by trying to produce and to analyze artworks by computer, just as I share a strong feeling of optimism about the capability of computers to deal with other processes carried out by humans. I also believe that the way computers are used today to deal with human thinking can be applied to thoughts of aestheticians and art critics and to the thinking that lies behind at least some kinds of artworks. But those applying computers to the understanding of artworks have imposed on themselves an unnecessary limitation that must be removed or else the understanding will lead to artworks computers produce that are flat and lifeless.