"AI slop" is in the eye of the beholder
Add Axios as your preferred source to
see more of our stories on Google.
Illustration: Sarah Grillo/Axios
Social media users complain that there's way too much "AI slop" in their feeds today, but one person's slop is another's cool new meme or funny post.
Why it matters: Blanket disdain for AI content is less and less useful in a world where AI is part of every digital tool and system.
- Instead, some experts say, we need to learn to separate useful or creative AI output from potentially harmful and annoying spam that's clogging the internet.
State of play: Some people want to throw everything that's AI-generated into the AI slop bucket, while others reserve the term for anything they think is overwhelming human signal in AI-generated noise.
- A viral video of an AI baby interviewing an AI dog on a podcast could be slop to some and simply something fun to forward for others.
Yes, but: Soon, nearly all online content will involve AI in some way.
- While there are pure AI tools like ChatGPT, Gemini and Copilot, developers are also building AI capabilities into all manner of software and even the underlying operating systems of phones and PCs.
- In Photoshop, you can use AI to expand a scene or change the background, while Apple, Microsoft, Google and others have added options to get AI assistance for writing.
What they're saying: Some see the dividing line not in the tool, but in the intention behind its use.
- For former Evernote CEO Phil Libin, it comes down to whether the human directing the AI is looking to make something better or to make it cheaper.
- "When AI is used to produce mediocre things with less effort than it would have taken without AI, it's slop," said Libin, who these days runs Airtime (formerly Mmhmm), a service designed to make humans look better on video. "When it's used to make something better than it could have been made without AI, it's a positive augmentation."
The intrigue: For all the criticism of AI-made content, it's clear a lot of people appreciate being able to easily hop on the latest meme bandwagon.
Zoom in: Users' embrace of OpenAI's latest image generator showed exactly what can go wrong and right when you give everyone the power to turn brief prompts into images.
- The first wave of users turned themselves into Studio Ghibli animations.
- Then came Muppets, "Simpsons" characters and those from other fictional worlds, followed by AI-generated action figures from photos, complete with accessories.
- And now that image generator will be even more places, with OpenAI announcing yesterday that developers can integrate it into their apps using an API.
Zoom out: In addition to the unsettled intellectual property issues around copying an artist's or studio's style, the ability to create memes at scale also could overload content moderation systems.
- AI memes are already playing a role in politics, including elections, as well as in the spread of misinformation.
- This innovation comes at a significant (though yet to be fully quantified) environmental cost, given the massive amount of compute capacity needed each time one of these memes takes off.
Our thought bubble: While AI content is starting to flood our feeds, it will flourish or fail largely for the same reasons as other types of content — because we engage with it, or don't.
- Meta is among the companies betting big on AI-generated content. The social giant is putting its assistant everywhere and experimenting with all manner of synthetic content, from suggested images and prompts to AI-created comments and posts.
What's next: Memes always have a limited shelf life.
- AI memes might be even shorter-lived than human-created viral content, given how easy it is to make them and how quickly we might tire of seeing everyone and their mother doing the same tricks.
- But by the time that fatigue sets in, it may be too late. AI content — through its sheer volume — could end up flooding the internet and crowding out human creations.
