What is AI slop, and how can I identify it?
You’ve probably heard the phrase ‘AI slop’ – but what does it mean, and how can you avoid it?
New technology is usually intended to improve our lives or advance society in some way, and yet after an initial period of giddy optimism, it frequently achieves the opposite.
Social media was hailed as the ultimate democratisation of the individual’s voice, yet some platforms have devolved into a politicised bearpit of lying, boasting, bullying and hypocrisy.
Cryptocurrency was once seen as a borderless, frictionless alternative to discredited state-level currencies. Today, it’s a stock market plaything where fraud and theft are rampant.
Generative AI hasn’t yet achieved a similar level of disillusionment among the general public, in part because its abilities are rarely recognised.
Platforms like Microsoft Copilot offer an intriguing glimpse into tomorrow’s problem-solving – providing you avoid the free version.
However, other generative AI platforms are being used to churn out torrents of low-grade output for often questionable reasons.
This output is known as AI slop, and it’s flooding the internet with everything from cloned music to misleading information.
Oh AI?
Artificial intelligence has evolved far beyond its original 1950s definition as “the science and engineering of making intelligent machines”.
Today, AI is an all-encompassing term covering a multitude of services, solutions – and sins.
AI slop is a pejorative term applied to any content produced by generative AI tools like ChatGPT and which is demonstrably of a low standard, or lacking any quality control.
One obvious example is AI-written search results which are published above legitimate website links despite containing clearly inaccurate information.
Your correspondent was recently informed by Google that a make and model of car on sale in Britain since 2019 “was brought to the UK for the first time in autumn 2025”.
Another example is the endless churn of clickbait stories on websites or social media platforms, often basing a worthless article on a celebrity quote taken out of context.
In terms of AI images, we’ve had everything from Shrimp Jesus to Trump Jesus. Oddly, the figureheads of other religions have largely escaped such disrespectful misrepresentation.
Even music isn’t immune.
This summer, The Velvet Sundown released three albums in six weeks and briefly topped Spotify charts.
Everything about this AI “synthetic music project” (from AI photos of band members with fused fingers to the music’s overtly pastiche sound) is false – and an insult to any real musician.
Across all creative industries, from fiction novels to rock music, AI slop is to actual content what a plastic hamburger is to a real one.
Where does it all end?
The tidal wave of AI-generated content is only just starting to wash over us. Expect to see far more AI-generated Instagram influencers and fake online reviews in future.
Remarkably, there are now warnings that the algorithms producing this slop have already run out of worthy source material.
Having plagiarised almost everything which has ever been published online, in defiance of all copyright laws, AI engines are now incorporating their own flawed output into their models.
It’s hard not to foresee a downward spiral in quality, even as the algorithms are reprogrammed not to repeat their worst mistakes, like promoting non-existent events.
Since AI slop represents an existential threat to many of the services we rely on, from online reviews to search results, the tech giants may decide to target and downgrade AI output.
However, with incalculable quantities of AI slop already online, and more being generated every second, it could already be too late to put the generative genie back in the bottle.
We may have to accept that a great deal of what we see and hear online isn’t real – or worth seeing and hearing.
As consumers, that would require a level of ongoing vigilance and discernment that many people either can’t or won’t commit to.
The next time you see a photo, read an article or hear a recording of someone talking, ask yourself one question: could it be computer-generated?
(Incidentally, no generative AI engines were used in the production of this article, or any of the other content on BroadbandDeals.co.uk).
