Let’s filter out the decline of AI, you cowards

It’s almost impossible to avoid seeing AI-generated content online, but it doesn’t have to be that way. YouTube, Instagram, TikTok and others have ramped up their content authentication efforts over the past year, with many now automatically applying labels to distinguish AI-generated photos, videos and music from those created by real human creators.

This is all well and good if we’re finding random content, but do you know what’s better? Allowing us to filter the AI ​​regression.

Current labeling efforts have not meaningfully changed how content is presented online. You may notice that some Tik Tok or YouTube videos Your feeds now have AI disclosures in the description, or information labels overlaid on the clip itself. Meta takes a similar approach by applying “AI information” labels to Pictures on Facebook and Instagram That carry specific AI metadata or voluntary disclosures from the creators.

But if you actually want to Avoids Seeing anything bearing such labels — which is justifiable, given the brain rot they cause, as well as the ethical and environmental concerns around generative AI — is actually very difficult to do. A filter would solve this easily. All we need is the “AI” checkbox to toggle.

I reached out to Meta, Google, TikTok, and Spotify to ask if they have plans to let users filter the various content they’ve authenticated using AI rating systems. TikTok and Spotify never responded, and Google said it had nothing to share. Meta did not provide an attributed comment. But in short, none of these companies said “yes.”

One of the only online platforms I’ve seen with an AI content filter is DeviantArt, and their implementation is very straightforward. For one thing, you can’t access it through the DeviantArt feeds or store page, so it appears to be somewhat hidden. Instead, you have to create an account and then hover over your user icon at the top right of the page to find the “AI Content Settings” menu. From there, you only have two options: the default “Show AI” setting, or the “Suppress AI” option which claims you’ll see “fewer instances” of images generated or processed by the AI.

After trying both options, unfortunately, I don’t see a noticeable difference. I have a pretty good eye for spotting AI-generated “digital illustrations” at this point, but I didn’t have to rely on my suspicions alone — nearly every questionable image I selected included a creator disclosure in the description that confirmed the work had been spat out by a bot. DeviantArt does a poor job Automatically apply AI labels To images with metadata that clearly indicates the source of the AI.

Pinterest It has a similar system in place. Users logged into a Pinterest account can tap the Settings icon, select “Refine Your Recommendations,” and then tap the “AI Content” tab to toggle specific categories, including art, beauty, fashion, and home decor. Disabling any of these options will show “less AI-modified content” for that specific category, according to Pinterest, but in my experience, it’s far from perfect. It’s also arguably harder to find the setting than a built-in filter in your Pinterest feeds. I still saw a lot of photos with suspicious AI information (including strangely perfect photographic models and unexplained sharpening errors), even though the AI ​​filters were maxed out.

This is almost certainly what would happen if other platforms like YouTube or Instagram introduced an AI content filter: it wouldn’t work as well. But that’s okay because it would expose the ineffective “solutions” worn by AI moguls. They exist on paper to please regulators and critics, but they do little to address the actual problem of distinguishing between AI forgery, original photography and creative works.

And platforms He does I know it’s a problem. Instagram header Adam Mosseri said in December that “authenticity is becoming a scarce resource” amid the rise in AI-generated content. And now we have the CEO of Google Sundar Pichai’s confession Recently Decryption “There are a lot of things wrong with AI,” and that Internet users need to “adapt to it.” Well, give us the filters.

Source-based systems like C2PA and SynthID work by embedding invisible metadata or watermarks into the content at the point of creation. But there are plenty of open source AI models that don’t do this (especially if they’re designed for nefarious purposes), and even then, metadata can be stripped too easily to make them reliable. There are also detection-based methods that analyze patterns in digital content and then assess the likelihood that AI was used to create it, but these can provide false positives. None of this currently works effectively on a large scale.

However, companies, incl AI providers like OpenAIAI-powered labeling solutions are currently heralded as something that will help prevent people from being scammed by deepfakes and other misleading fakes. If regulators discover how ineffective they are, online platforms and AI providers may need to find an actual solution He does Action, rather than what currently seems like a smokescreen.

Platforms will argue that they risk misbranding original content if they push ratings initiatives too hard. both of them dead and YouTube This was discovered the hard way after AI labels were applied to photos and videos that creators said were produced without the help of such tools. If this is a concern with current labeling systems, Then find a better solution. Surely improving the user experience for millions of users is a worthwhile investment to stave off competition?

And while I’m asking, why can’t I report all the unnamed AI bugs I see every day? Given the scale of the problem – with A study conducted by Kaboeng last year Finding that more than 20 percent of YouTube videos shown to new users are low-quality videos, for example — I imagine a lot of human moderators will be needed to effectively sift through each report.

Maybe that’s the rub. At a time when big technology companies are replacing workers with artificial intelligence He’s supposed to outdo themCan she undo her carefully crafted narrative by hiring them again to fix AI problems? Humans tend to have annoying requirements e.g Salaries and benefitscompared to automated supervision systems that lack precise investigative skills.

An alternative to categorizing AI-generated content would be to start Rating verified human creators instead. This will not necessarily identify artificial content posted by These creators, but it can help us cut down on unverified content farms that produce low-quality content. This is the future that Instagram awaits Move fiercely For Meta’s photo-sharing platform and something Spotify already does Verified artists.

Of course, Meta, Spotify, and Google don’t just host AI-generated images, ads, and music; They are also responsible for making the tools you create. That’s why they insist not everyone The AI ​​content is sloppy, and it’s more about quality – if it becomes compelling enough, they hope you won’t notice and continue happily slurping from the gutter. Allowing users to filter them regardless would go against every effort these platforms have made to leverage AI: them He wants Yours to embrace the slop plant.

I’m happy to be proven wrong. I’m actually honest Pleading For online platforms to prove that AI labeling efforts were not a waste of time. But for now, they hold all the cards and we can only hope that their efforts to moderate AI will be enough. So, give us a basic filter of “no AI” or “verified human creator” and we’ll judge how well that actually works.