söndag 26 oktober 2008

The creative computer

A new initiative:

The computer is often viewed as a deterministic rather than a creative entity, which follows its program in a strictly deterministic fashion. As computers have become more powerful, this view has become less obvious. When the internal state of a complex entity cannot be monitored in detail, its actions may appear genuinely creative to an observer. Aspects of creativity appear in several new computational paradigms, such as evolutionary algorithms. Active at Chalmers one of the leading research centers in Europe in areas such as genetic programming and artificial life, I would like to explore the boundary between this work, engineering design problems, visual design and artistic applications. Evolutionary algorithms in design may either evolve solutions automatically given a fixed quality criterion, or a human may partially guide the search for new designs (interactive evolution). Interactive evolution can play an important role in amplifying human creativity. As simulation techniques improve, automated techniques will become common tools in engineering, without any bias in favor of conventional solutions. In the project, principles of computer assisted design, and human-computer interactions in such systems will be studied, as well as methods for learning about user preferences through adaptive methods. A common theoretical framework for automatic design in terms of fitness landscapes in domains from biotechnology to mechanical engineering will be explored. New powerful simulators, e.g., for deformable media and fluid flow will be used in this framework. Evolution of 3-D objects (also with time dependence) will be studied, in particular using growth algorithms. Collaborative interactive design, and interactions between a learning user and an adaptive machine will also be studied. Collaborations with artists will be developed both in areas of image and sound. The project will also include studies of the concept of creativity itself, involving researchers from philosophy and cognitive science, and complex systems modelling.

Inga kommentarer: