Artificial intelligence and machine learning are finding more and more applications lately. They are usually associated with monotonous tasks and the analysis of a huge amount of data. Now this field is on a new ground, which is art. It turns out that artificial intelligence can be quite creative. The image created (or rather generated) by one of the algorithms was sold at Paris’s Christie’s auction house for $432,000.
It might seem that computers are heartless, and the algorithms always provide the same results. However, this approach has no application in the case of machine learning and artificial intelligence algorithms. Here, running the same program several times, with the same input data, returns different results. Therefore, artificial intelligence may work well in art-related issues. Especially that we have quite a lot of algorithms for image processing. An interesting project of this type has recently been organized by the Christie’s auction house, which put up a series of portraits painted by artificial intelligence to bid. One of them (the portrait named Edmond de Belamy) was sold for $432,000.
Will artificial intelligence replace artists?
Probably in the network you can already find statements of “specialists” who foretell the imminent end of many professions. However, they forget about one very important aspect. All machine learning algorithms require a huge amount of learning data. That was also the case here. The team behind the artistic artificial intelligence in the first step chose a set of images that had many common features. This activity was performed by people who know the art. Then the discriminator stated whether the image ‘painted’ by the generating algorithm is similar to the real one or not. As you can see, there is quite a lot of place for real artists. First of all, they have to prepare data (i.e. paintings) that could be used to teach artificial intelligence.
Artificial intelligence will not replace people who create new trends and set directions for the development of painting. It’s the same with music and other artistic creations. Machine learning algorithms can at most generate new works that will be variances the ones previously created by people.