Google’s computer-brain art draws noisy crowd

Anti-gentrification activists bewail the exodus of artists and their art from an increasingly pricey San Francisco overrun with big-data jockeys and software engineers.

Now Google may have hit on a method for getting San Franciscans to care about art: have computers make it. A recent exhibition of the firm’s algorithm-generated artworks attracted a sizable subset of the city’s new dominant life form.

However, the level of genuine interest in art among “the fleece pullover, sneaker-wearing art-show newbies” described in a Wall Street Journal article is unclear. The casually clad horde in the back of the Mission District historic movie theater where the art show was held Feb. 26 talked and whispered, and were repeatedly “shushed” before the auctioneer suggested that if they didn’t have cash for Google artificial-intelligence art, they could at least pay attention, the WSJ reported.

Google’s artworks were “hawked” for as much as $8,000, according to the paper. The pieces derive from “artificial neural networks” – computer simulations of a human brain, used in development of the company’s facial-recognition software. Facebook has also produced AI-created art, according to the tech website Engadget.

The Google system’s muse, however, appears to be one Timothy Leary. The imagery is “very much something that you’d imagine you’d get with psychedelics or during hallucinations,” Karl Friston, a professor of neuroscience at University College London told Vice. Google’s artificial neural networks cough up the kinds of patterns that arise from people’s lower-brain functions when the higher-processing areas of the brain are impaired by hallucinogens, according to the Vice article.

Google’s artificial brains approximate human neural functions – very approximately, Johns Hopkins cognitive neuroscientist Frederick Barrett told the online magazine. “The complexity of the brain is such that I’m not sure if you can model [it] with artificial neural networks. I just don’t know if we’ve gotten there yet, or even anywhere close,” Barrett said.

Last June, a Google software engineering team blogged on the AI art, explaining that the artificial networks can create works essentially from scratch. “If we apply the algorithm iteratively on its own outputs and apply some zooming after each iteration, we get an endless stream of new impressions, exploring the set of things the network knows about,” the team wrote. “We can even start this process from a random-noise image, so that the result becomes purely the result of the neural network.”


Photo: From painting by Georges Seurat, image processing by Google software engineer Matthew McNaughton  (Google Research Blog)


Tags: , , , , , , , ,


Share this Post