Once it was that one could rest basic trust in the world through a complex web of interpersonal relationships that knit community together. A con-man lived in isolation because he had to keep moving to find fresh suckers. Ironically, the new global connections we feel through the www does quite the opposite. With the internet disease of misinformation, one’s only really sure of one’s close friends (and sometimes not even then, since sometimes dead people send me friend requests!) It feels like the umbrella of trust shrinks. Who can really be trusted and how can we tell?
I’ve been troubled over the new AI-generated apps that have become available to the public in the past few months. Like rubbernecking at a car crash, I’m both intrigued and horrified, but can’t look away. I love images and reviewing ones spat out by AI is like looking at a whole collection of new art. Granted, some apps are simple and stupid, while some, like DALL-E2 can create some pretty detailed and intriguingly believable images.
Among other subjects I’ve been exploring are generated images in the style of daguerreotypes. The imitation images, remarkably, have the distinguishing characteristics of “real” history. For instance they’re full of ‘artifacts’ of their production: cracks, wrinkles, dust and hair bits, torn or faded edges, each of which hint at the specific physical difficulties of their creation. The AI-generators convincingly replicate not only image details (like a 3D welt on the man’s side), but the bad image quality and artifacts from real photos, resulting in an image that reads like an authentic photo. Even the color bar here intimates that the ‘collector’ of the photo wanted to preserve its proper historic coloring.
For me, a lifetime fan of historic photography, this means that I read an invented image as if it were recording a real historic event. I know it’s fake, but my emotion is captured in any case. I can’t convince my heart to not participate, to not be alarmed at what looks like a tragedy. My mind can say “this phoney AI app can’t even do a proper face!” but a lifetime of interpreting photos as rooted in reality has trained my heart, which leaps.
Among hundreds of AI images I’ve “created”, this one continues to haunt me. I know it to be false. So why am I haunted? What in me has been triggered? It is that place in me–– and I would say in all of us–– that is threatened. Any con artist who presents me with this sort of image can capture my heart and I can do nothing about it, except perhaps destroy my compassion. To me this is terrifying. I’m a sitting duck. Not because of AI but because my weakness has been revealed and is now available. How many of my weaknesses are yet to be discovered? And will I know them?
What’s more, if anyone sees this image out of this context, how will they know it’s not real? There’s no material difference between this image and any other historic fact. In a world where a third of the populace believes Jewish space lasers set wildfires, who defends “truth” to whom? When reality cannot be distinguished from simulation the web of relationships that once knit us together in community is shattered. I fear that now, with anyone being able to mass-produce cheap virtual facts faster than the truth can get its boots laced, the umbrella of our trust will shrink to our immediate intimates. The danger is not in AI but in human nature, given increasingly powerful tools. Will we survive ourselves?
