Please or Register to create posts and topics.

Polegonardo Charly

Museum Link:

Source Link:

Date Minted: January 4, 2020

Artist Description: Generative art made by DADAGAN, a model trained by Alexander Reben with Dada's unique drawings dataset.

CohentheWriter’s Commentary:

Generative art which rushes madly towards the uncanny valley, Polegonardo Charly, is the product of an AI (designed by Alexander Reben) trained on god-knows-how-many drawings from DADA’s legendary visual conversations. That an AI could construct a piece of this emotional complexity is wildly impressive (even if it’s the product of an abject accident; who’s to say?), but whether that says more about Reben’s algorithm or the quality of DADA’s dataset is TBD. But stop and think of what we’re looking at here: an AI, even one trained on totally hand-drawn data, is without personal knowledge of emotion, humanity, or communication even, and yet it constructs this haunting, owl-eyed image of a man. There’s much to unpack here. And what better entry-point than the image’s title, Polegonardo Charly itself. Confirmed after a bit of research, no, this is not a real name, which seems obvious in hindsight…it’s just a tad too ridiculous. But it seems juuuuuust close enough to keep you unsure. And this piece unfurls itself in just the same way. 

Polegenardo Charly is a face. Charly’s, his crestfallen expression and his hair, and his shoulders puffing out against a brown shirt. In that bizarre way characteristic of so much AI-generated art, there is a strange pattern of lines and colors composing the face itself, overlapping strings that appear almost like brushstrokes, and rainbow colors coalescing pixel-by-pixel into darker, more stolid hues. The physical face itself, poor Charly, looks deeply distressed. If he was meant to have whites in his eyes, they’re barely here, his corneas instead colored by browns and reds, with pupils that can’t completely give themselves over to blackness. He looks troubled, wide-eyed, smooth-lipped. He could just as easily be about slipping a switchblade into your kidney as he might be readying to light a bong. That ambiguousness is central. As is the concerning level of barely-restrained passivity we can read on his face, emotions unshown but slithering under the surface. And all of this in shades of white, black, and brown; it breeds unconscious images of disease, feces, crackhouses, tweakers.

Think about what you’re seeing: This is not a human as seen by a machine, but a human as seen by many humans as then subsequently seen by a machine. Not a human as captured by a machine, but a drawing of a human face, informed by so many drawings of human faces beforehand, that this machine believes is as near to the real thing as possible, at least according to the inputs fed it. However realistic this angst-addled man seems to us, Reben’s AI has deemed it acutely human. A best effort. To a machine, this is a triumph, a perfection. Look into those dire white eyes. Look at that melty, veiny skin. Look at the wavy hair, clumped like kelp, hair like beams of light, hair that looks like roils of filthy water crashing on beach sand. That’s what the computer thinks we are.

Polegonardo Charly makes me deeply uncomfortable, and I can’t be certain why. After all, this is just a face, and all it’s doing is staring forward from the confines of its frame, not betraying any particular emotion, not constructed in any particularly demented ways —though perhaps a bit oblong and off-color— but by all reasonable metrics a seemingly-acurate, if not highly-sylistic, portrayal of a human face. So why does it seem so wrong? 

If you haven’t already, open Charly to full-screen and give its many facets each a moment of your time. Really focus on the hair, the texture of the skin. Make sure to look carefully at the eyes and the mouth, and the way the chin descends cliff-like onto the chest. Think for a moment about the colors used, salmon lips and whatever greenish-brown color you’d like to use to describe the shirt our dear Charly wears. If you’re squeamish, don’t spend too much time examining his ears. If you’re antsy about interlocking lines, avoid this man’s eyebrows.

Here is the thematic fulcrum of Generative Art, and why, upon further reflection, it’s so strikingly odd: none of what you see here is legitimate; it is all an attempt at imitation. Imitation of brushstrokes and color palettes and stylistic artistic techniques. As mentioned, this is a machine’s attempt to capture an artist’s attempt to capture a person. That’s three levels of mimesis we’re witnessing! 

If we identify a deep mental anguish in Charly’s eyes, then, well, where does that come from? If an artist paints a picture of a man, and the man has a deep mental anguish, and the subsequent portrait’s eyes communicate that anguish as well, we can glean what the artist is trying to represent. But this was a piece made of many other pieces, an imitation of many many previous imitations. How much mental anguish did this AI have to see, and in how many different forms? How much mud-dauber hair was it exposed to before deciding, somewhere in its code, to cover Charly’s forehead in dense thickets of stringy locks?

Every time I look into this beautifully, artistically-composed, AI-generated face, built by something that understands us only aesthetically, I find myself asking questions like: Are we humans so easily captured? Are our innermost emotions so easily put through a processor and understood? And is that good or bad?