Computational aesthetics can be traced as far back as 1928, when American mathematician George David Birkhoff proposed the formula M = O/C where M is the “aesthetic measure,” O is order, and C is complexity. Birkhoff applied that formula to polygons and artworks as different as vases and poetry. In the 1950s German philosopher Max Bense and, independently, French engineer Abraham Moles combined Birkhoff’s work with American engineer Claude Shannon’s information theory to come up with a scientific means of attempting to understand aesthetics. The ideas of Bense, which he called information aesthetics, and Moles were influential on some of the first computer-generated art, but some artists objected that such art and its assessment using Bense and Moles’s work was not “natural.” In the 1970s American psychologist Daniel Berlyne introduced the “new experimental aesthetics,” which was based on measuring the qualities of an object and relating them to a viewer’s aesthetic perception and nonverbal responses. Berlyne also insisted on not treating aesthetic perception in isolation from other psychological factors.
In the early 1990s the International Society for Mathematical and Computational Aesthetics (IS-MCA) was founded, specializing in design with emphasis on functionality and aesthetics and attempting to be a bridge between science and art. By the beginning of the 21st century, computational aesthetics had become sufficiently established to sustain its own specialized conferences, workshops, and special issues of journals. Computational aesthetics attracts researchers from diverse backgrounds, particularly AI and computer graphics.
Computational aesthetics has been applied in a number of different fields for various purposes. For example, it has been used to automatically assess aesthetics in photographs (and thus improve the quality of photos taken by amateurs), to distinguish between videos shot by professionals and by amateurs, and to aid in vehicle design. In some cases, computational aesthetic systems have also been used to aid human judges. The final verdict in delicate aesthetic assessments, however, is usually left to a human or a panel of human experts.
Ultimately, the goal of computational aesthetics is the development of fully independent systems that have (or even exceed) the same aesthetic “sensitivity” and objectivity as human experts. Ideally, those systems should be able to explain their evaluations, challenge humans with new ideas, and generate new art that could lie beyond typical human imagination. Nevertheless, it is difficult to ascertain using present technology, from the standpoint of psychology and neuroscience, whether a system that performs on the same level as a human expert is actually using similar mechanisms as the human brain and, therefore, whether it reveals something about human intelligence. A notable objection to the field among philosophers of aesthetics is that computer scientists can never prove what is or is not “truly” aesthetic.
Computational aesthetics is usually classified as a subfield or branch of AI. However, computational aesthetics research is also of interest to mathematicians, engineers, psychologists, and even philosophers. A perhaps more closely related field is computational creativity (also a branch of AI), which addresses the issue of creativity exhibited by machines. Aesthetics, being an aspect through which creativity is manifested and can be assessed, therefore sometimes comes into play and blurs the distinction. In principle, computational creativity research need not necessarily involve the generation or assessment of aesthetics. Neither computational aesthetics nor computational creativity should necessarily be associated with the field of artificial consciousness (another branch of AI), because it has been demonstrated that machines need not be conscious (like humans) in order to evaluate aesthetics or exhibit creativity.