People want the human internet back.
Pinterest’s new AI filter could be the next step in the evolution of the internet. It’s the first major platform to actually adjust to what a lot of people have been feeling. Our feeds across a broad spectrum of apps are being flooded with low-quality AI slop. The effect is a degradation of trust and an increasing desire to have more control over what we actually see. Pinterest’s move is the first real signal that platforms are recognizing there’s a growing problem instead of pretending that AI generated content and design is the future.
While I’m not proud to say it, I do occasionally scroll TikTok to kill a few brain cells and a few minutes in my day. My partner also sends me TikToks which she says is her love language, which I in turn use to justify the habit. But these days I’m seeing a trend, there is a stark increase in comments trying to figure out whether a video is AI. We’re literally becoming reality police in the comments of these feeds. Sometimes it’s obvious, there’s that monotoned, programmatically friendly voice playing over a grainy, intentionally retro aesthetic designed to hide AI distortions. But even when it isn’t obvious, people are suspicious. There’s a growing paranoia around what to believe and not believe when scrolling through your feeds on Facebook, Instagram, Tiktok and more.
It’s a painfully predictable side effect of releasing an unregulated video-generation tool that tries its hardest to pass for reality. There are many tools racing towards that meaningless end but the most notable example is Sora. There’s the famous bouncing bunnies video or the progressive pastor video, that went mainstream and gave Sora some early hype. But to me it seems like the hype has evaporated. I don’t know a single person who actually uses Sora. Not one. I barely know anyone who even downloaded it! And the people who did opened it once, generated something vaguely cursed, laughed nervously, and then abandoned it.
To me it’s unsurprising, as it seemed like the main use case for downloading the app was making creepy videos of Sam Altman being mocked in increasingly bizarre and slightly more humiliating situations. Along with a string of racially insensitive videos of MLK making fake speeches, which were eventually disallowed after the very busy ethicist of open AI Olivia Gambelin told the BBC OpenAI that limiting further use of Dr King’s image was “a good step forward”. Not only were there weird historical recontextualized videos but also blatant copyright infringements of famous characters. Which we all know is a favourite of OpenAI. Videos on Sora are considered a “success” when they reach escape velocity and seep into mainstream content, which is a strange metric for anything, let alone for a CEO who keeps insisting this tech is “inevitable.”
But all of this “inevitability” talk from AI tech CEOs is starting to wear thin. Sure, we exist within certain systems. Yes, capital moves the way capital moves. But companies only have power because users hand it to them. When people walk away and engagement drops the inevitability story starts to look flimsy.
It’s been a awhile now since the first big wave of AI models started generating shockingly polished visuals. And I’ll be honest: sometimes the results are impressive! But my standards for AI are different from my standards for humans. There is still a huge amount of rich, deeply human art being made every day as well as a huge historical trove to dig through that is fundamentally more interesting because it comes from people with taste and lived experience.
Also the hard truth for AI generated content is that the people who can make beautiful things already are the ones best positioned to make beautiful AI-assisted things. And what are their incentives? Across the board creatives are getting fired because of the threat of AI replacement, why would they be the first to hop onboard to support a tool that’s literally trying to take their jobs. Right now it seems like lot of AI content online comes from people with no creative background dumping average outputs into the feed. People are trying their hand at it, making something new from nothing, and while I admire the spirit in that the harsh truth is it adds almost nothing.
It turns out making something good that people care about is extremely hard. And making something great is harder still. Great work in my view is inherently collaborative, political, relational. It requires conflict, iteration, and concession. You can’t flatten all of that into a prompt without losing what makes it resonate. Most recently OpenAI announced that they would be making an AI first movie called Critterz for under 30 million as an attempt to try and show the power of AI in film/animation production. But it does beg the question, why is it that OpenAI has to front the cash here? Maybe it’s because independent filmmakers aren’t interested in leveraging this tool? If it was truly a great tool for filmmakers, we would already be hearing about the new voices breaking through the cultural marketplace signalling the importance of this new medium. It ain’t happening!
Oftentimes when I tell this to AI boosters there’s this pushback that it’s only getting better and better, and while yes the fidelity and realism is increasing, I have yet to feel the rumblings of a creative AI revolution. I firmly believe you cannot collapse a lifetime of discipline and taste into a few paragraphs of prompting. You can’t try and remove collaboration and expect a better product. Yes, I’ve seen beautiful AI imagery, but they are the exceptions, not the rule. And more importantly: no one is asking for more of it, in fact they’re asking for less.
When people identify a video as AI, it’s usually to warn others not to trust it. That alone tells you what demand actually looks like. Most social platforms exist on human-generated content and I think the flood of AI slop is not likely to increase engagement. We all buy into this story that AI will optimize for our attention, stimulating the exact parts of our brain to create dopamine zombies and while I don’t deny that’s certainly possible, I do think that it’s actually quite unlikely. There is still no comparison between human content and AI content from a quality perspective.
Pinterest’s new filter lets users reduce or outright hide AI-generated content. TikTok is now testing a similar feature. And I know I’ll be turning it on the second it becomes available. No one begged for this influx of AI content. It doesn’t solve a problem that actually existed or if it does, I certainly can’t put a finger on it. It’s a novel toy that has overwhelmed entire corners of the internet, suffocating feeds and making us all paranoid and platforms are finally starting to admit it. While they’ll never say “our feeds were drowning in AI nonsense and users hated it.” features speak louder than words. When Pinterest gives you the option to filter out AI, and TikTok rushes to pilot something similar, the message is clear: people want the human internet back.
I have hope that creativity online won’t be defined by who has the most generative models pumping out piles of content. The novelty of AI is already wearing off and the fatigue is very very real. If platforms want engagement, if they want trust or if they want actual culture to keep happening on their watch, they’re going to have to give us control over our feeds, including the ability to filter out, reduce, or completely remove AI-generated content.
The internet needs an AI off switch was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.