Heaven knows the relentless marketing hype around AI has become boring enough. Let’s not even go there.
The fact is, AI has now become a part of photo-editing and we’d better get used to it. It’s difficult to deny the effectiveness and usefulness of AI subject recognition and masking, or the amazing quality achieved with AI noise reduction and upscaling.
But this is not necessarily ‘AI’ in the sense that most people might imagine. It uses deep learning techniques to automate tasks that would once have taken a great deal of manual effort, or applies image enhancements simply not possible before.
AI noise reduction and upscaling are interesting, though, because they use AI to anticipate details and textures that would have been there if the camera had been able to record them. It doesn’t seem just yet as if the AI is really fabricating anything that wasn’t there, just trying to reproduce what probably was there. It’s still quite harmless, right?
Improving reality VS changing it
So how do we feel about AI sky replacement? It’s not as if photographers have never done this in the past, especially those with access to their own darkrooms. This kind of reality-altering enhancement is not new, it’s just that now it’s so easy that anyone can do it – and so effective it can fool anyone.
What about portrait enhancement? Everyone wants to look beautiful (though it’s a little late for some of us), and AI portrait enhancement tools can do just that. They can go further. They can make you think everyone in the world is perfect but you.
AI has certainly made reality-altering imagery easier and, in a social media landscape where only instant impact can halt the swipe and popularity is everything, this kind of manipulation is inevitably rife. AI photo editors are extremely effective at producing generic crowd-pleasing imagery quickly.
Every time I use AI to add a sky or an object to a scene, I feel the photograph is somehow diminished. I’m not sure I can easily explain that.
And this may include images that are generated solely by AI. Adobe is currently championing generative AI but is very careful about intellectual property rights and cultural representations, so the images can be a little bland and generic. Adobe has anticipated the commercial and legal problems brewing around AI imagery. These are coming, make no mistake.
Can I just share a personal feeling? Every time I use AI to add a sky or an object to a scene, I feel the photograph is somehow diminished. I’m not sure I can easily explain that.
What’s real and what isn’t
AI is not the devil. Artists (and photographers) have been fabricating ‘reality’ for decades. But it’s never been as easy as it is now. And perhaps now more than ever we need to keep a clear head about what’s real and what isn’t, and the difference between a photograph as a forensic record and the photograph as ‘art’.
This is where AI is not helpful. It’s now extremely easy for bad players to fabricate a scene or an event and pass it off as ‘real’. A new fact-checking and verification industry is emerging to counter this, so hopefully this particular threat can be addressed in future.
If you are an ‘artist’ photographer, it’s not so clear cut. You’re already creating an idealized, fictionalized view of the world, so you can’t blame AI for making this easier. In fact, you should probably embrace it and its new opportunities.
There are great artists doing this already – Dutch artist Nancy Poeran uses an extraordinary synthesis of photography and AI in her work. The difference is that AI is part of the process, not the process. Jon Devo makes some very good points about the value of AI in filmmaking, too.
Unlimited supply VS diminishing demand?
It’s a basic law of economics. The more of a thing there is, the less people want it (or are prepared to pay for it). The rarer it is, the more they want it. In a possible future world where almost all imagery is AI-generated and almost none comes from humans, this could be significant.
Technology challenges us to assert our human values, which means that first of all, we have to figure out what they are.
Sherry Turkle, sociologist
We face a glut of spectacular, flawless, genericized (is that even a word?) AI imagery that is subtly different to what we’re used to. It won’t be the creation of a single mind any more, but an algorithmic amalgamation of what most people seem to like. Just imagine combining AI image generation with Instagram algorithms. Like, that’s all we need.
However, legendary creator Casey Neistat observed at NAB 2024 that “I think the thing that AI is going to have an impossible job of achieving is that last 1% that stands between everything [else] and what's great. I think that that last 1%, only a human can impart that.”
I hope we will reach a point where people understand the difference between machine-generated imagery and the hand-made alternative – and are prepared to seek out and perhaps pay more for the product of a single mind rather than homogenized group-think.
Can humans make a comeback?
Maybe we will reach a point where AI content must be flagged? Maybe we need a ‘No AI’ badge to distinguish human imagery from the rest?
Sony and Leica are already working on authenticity credentials for digital image. Back in 2022, model networking site PurplePort banned AI images. And after Australian retailer DigiDirect got duped by an AI-generated image, it cleverly turned it around into a Man vs Machine contest. That last story is a humorous response but highlights a widening reaction against AI imagery.
But here's the thing. AI is an inert and motiveless technology. We are not facing a world about to be taken over by machines. We are vulnerable, though, to a new army of human marketers, influencers and agitators who know how to use it.
You might want to investigate the best AI image generators for yourself, or you might want to do things the old-fashioned way with the best photo editing software.
