These days you see an image post and instantly there's people in the comments bemoaning that the image is LLM generated, and sure, that could be the case, but does it matter?
People today are extremely vocal about what they think is real or not, and in doing so I think they're missing the trees for the forest. I'm not actually getting that phrase around the wrong way, I've deliberately reordered it. They're being outraged by the mundane generated images whilst becoming desensitised to the harmful mass produced content that aims to mislead, misinform and manipulate, whether that content is generated by LLMs or not.
If we're constantly angry about things that realistically cause negligible harm, then by the time we see actual harm we've exhausted our energy fighting against things that didn't really matter.
Recently I reblogged something where I just thought “that's a pretty imaginary place”, and I thought nothing more of it until someone yelled in my comments “ThIs is unfortunately LLM trash. Check the carpet.” and I was just like... who cares? Looking at it you knew the image wasn't real. You knew it was generated, and all it was was someone sharing a comfy place that they'd created with the help of generative AI. The kinda image that before just wouldn't have existed unless the person also had the artistic skills and inclination to make it, or the money to hire people to make it look like a reality, something you definitely couldn't have afforded.
This image was seen on Tumblr, it's clearly generative content, and the blog that posted it almost certainly didn't create it, it was taken from somewhere else on the internet without attribution because that's the cultural norm on the platform.
It's just like how many people lusted over fame and fortune because it gave them an escape from their mostly boring and pretty predictable and bland lives.
I think today people generally have this nostalgia about the web that somehow it was all real and that was never the case. The web was never real. It was always a version of what we wanted to portray to the world. It was always curated. It was always a lens on someone's life. It was never your 100% true self and that goes for everyone. That goes for every social platform that's ever existed.
If you look at Pinterest pre-LLMs, what did it contain? It generally contained a lot of staged photos:
the bedroom with the dirty laundry moved just out of frame
the bookshelf that had books from other rooms put on it just so that it wasn't empty
the bed with the pillows that nobody would actually sleep with
It was always about creating images that looked more appealing because that received more reposts of your content boosting your profile. The mountains of AI generated content on Pinterest today are the ultimate end state for a platform that was always about selling you something to fantasize about, it was never about selling you something real because real is boring.
If we look at magazines and advertising, that was always about selling you something perfect, something that was only attainable if you bought this thing. I remember growing up as a teenager with all the ads for how you must have the perfect skin, the perfect clothes, the happiest family, the most friends, the best music taste, all were this idealised view of someone's life.
There was a hot moment when that trend reversed, where authenticity and ugly photos were the name of the game, the whole "BeReal era", where people tried so hard to show how real they were that they ended up performing realness. I think this connects to something deeper which is that in times of great uncertainty people reach for the imaginary because it's too hard to deal with reality.
We probably shouldn't be exhausting ourselves on the mundane generated images, because those always existed in some form or another; people love to aggrandise and embellish things to appeal to others, they love to escape with realities that don't exist. When I say we're missing the trees for the forest: there's content that's actually harmful, that's designed to mislead you, to misinform you, or to manipulate you, that's what we should be worried about.
The bar should never be "was this content generated with AI", because that's measuring the wrong thing, instead we need to be assessing all content we consume with the bar of "was this designed to harm, abuse, mislead, misinform or manipulate you or others?" and that's a much harder test to perform, because it requires actual cognitive reasoning about what we're seeing, and deciding what harms are being done. Harmful content was created long before generative AI, and it'll exist long afterwards too.