𝕹𝖔𝖙 𝕰𝖛𝖊𝖓 𝕿𝖍𝖊 𝕯𝖊𝖆𝖉 𝕬𝖗𝖊 𝕾𝖆𝖋𝖊

Adversarial hallucinations feat. Trevor Paglen.

What can you see?

Trevor Paglen’s work collides with secrets.

The things we see, the things we can’t. The things we’re not meant to.

Paglen’s work blends investigative art with a critique of technology and its place in our world. It takes us from the desert to the cosmos, from ourselves to the machines always watching us.

His series, Limit Telephotography focuses on clandestine government and military installations. Through telescopic lenses, Paglen captures these obscured sites, portraits of the world’s unseen mechanisms of control.

But these places are also watching back.

ImageNet Roulette, a collaboration with AI researcher Kate Crawford, examined one of the most significant databases used to train AI systems, exposing the inherent biases and stereotypes embedded within AI. The same systems that increasingly find their place in police databases, immigration and employment.

Questioning the supposed objectivity of these technologies, the project revealed the often hilarious, biased, nonsensical categorisation of people by machine learning tools. (Paglen revealed that eventually, the data itself was deleted. No wonder: too visible.)

He takes these topics head-on. Adversarially Evolved Hallucinations was a series of experiments where the aim of the process wasn’t to create clean visuals, but explore the darker psyche of the machine. The training data input were folk tales, nightmares, predators.

As Paglen puts it ‘Even the dead are not safe’, pushing the GAN to reveal its faceprints, the algorithmic identifiers that enable facial recognition. Creating ghostly portraits of dead luminaries.

We will continue to push the limits, staring and staring back.

Finding every inch of our universe categorised and observed, in an attempt to see.

There is nothing we will not pore over.

But the mechanisms remains obscure. In a sense, the more we see, the more is hidden.

.

.

.

Go further:

https://delistraty.com/2020/09/13/trevor-paglen-is-putting-the-art-in-artificial-intelligence/