A Turing Test for Generative Art

Aus de_evolutionary_art_org
Wechseln zu: Navigation, Suche


Kevin McGuire: A Turing Test for Generative Art. In: Generative Art 2011, 268-274.



When we produce generative algorithms whose output closely matches the style of a particular architect [1] or artist, what exactly is it that we think we are doing? Do we believe we are simply mimicking the architect, producing shallow results that have the surface qualities of the architect but no substance, much as a parrot might speak words [2] but there is no meaning" Is it "mannered" Or do we believe we are somehow capturing something deeper, a kind of partial working copy of the architect's brain and their creative essence?

How can we even begin to have this discussion?

In the area of Artificial Intelligence, there is the concept of the Turing Test [3] which allows us to discuss when we believe a computer program is behaving "intelligently". The Turing Test is not a real test, but rather a rhetorical framework in which one can disentangle the discussion around artificial intelligence.

We can apply a similar philosophical framework to the problem of generative art. By applying an equivalent to the Turing Test, we can discuss topics like mimicry, forgery, functional correctness, topological similarity, branding, signature style, whether a machine can be creative, and whether a machine can create art.

Extended Abstract


Used References

no c&p


Full Text


intern file

Sonstige Links