Eye to Eye with AI

Trevor Paglen and Barry Hurewitz on machine learning and the human factor

Share this page

Trevor Paglen has produced some of the most thought-provoking photographic images of the last two decades. When it comes to art he has very clear ideas. “I want to see what the world looks like, and understand how my own perception works”.

His earliest works seem to flirt with the romantic aesthetics of American landscape painting. At a closer look, minute details reveal that the artist is, in fact, training his customized telescopic lenses on top-secret facilities, classified satellites, drones and aircrafts.

Paglen’s interest in the geography (and socio-political implications) of such controversial no-go zones relies on the latest photographic and printing techniques. Perhaps it was only natural that, in more recent years, the artist turned his attention on technology itself.

A good example of this interest is given by ‘Four Clouds: Scale Invariant Feature Transform; Maximally Stable Extremal Regions; Skimage Region Adjacency Graph; Watershed’ a 2017 work by Trevor Paglen in the UBS Art Collection. In it, four skyscapes are overlaid with computer vision diagrams, generated by algorithms scanning the sky and clouds for unique points of reference.

Artificial Intelligence -generated images and imageries will soon complement, perhaps even surpass our notions of what it means to “see”. In Paglen’s words: “Images have almost necessitated a human to look at them in order to be complete. What does it look like when you have machines making images for other machines?” Trying to answer this question can lead into uncharted territory.

For Barry Hurewitz, Global Head of UBS Evidence Lab Innovations, Paglen’s creative process bears clear similarities with his work at Evidence Lab, a team of experts, independent of UBS Research, that work across 12 practice areas and 45 specialized labs creating insight -ready datasets. They both start with the question: “How do machines see?” Evidence Lab is constantly looking at all types of data, deriving meaning from it to help a client make a decision. “One of the biggest misnomers around AI is the fact that it is going to eliminate the human”, Hurewitz says, “Machines can be incredibly creative, but it takes a human to point this creativity to a question”.

Like what you just read?