For most of the internet’s history, making content was the expensive part.
A photograph needed a camera. A video needed production. A campaign needed some combination of time, talent and budget. Generative AI inverted that equation. Images, video, captions and entire content calendars now take minutes to produce, and often cost pennies.
That leaves the internet with a new problem. Content is becoming infinite while attention stays finite.
“AI slop” is the shorthand for what happens when those economics run to their conclusion — large volumes of cheap, algorithmically optimized content made to capture attention rather than deserve it. Merriam-Webster named “slop” its 2025 Word of the Year, which is roughly how fast the phrase escaped technology circles.
The economics point to more of it. Generation costs keep falling while the models keep improving. For anyone publishing at scale, the incentive is obvious: produce more, test more, keep whatever the algorithm rewards.
But there is a paradox inside that abundance. When everyone can make more content, making more content stops being an advantage.
The scarce resource becomes judgment.
From Creation to Intelligence
That is where start-ups like Aubrium are placing their bets. The Dubai-based social media intelligence platform, built for Instagram creators, brands and agencies, focuses on the decisions that happen before anything gets made. While conventional AI tools generate content, Aubrium is trying to answer a more fundamental question: what should you create in the first place?
Aubrium uses machine learning and AI to identify outlier content within specific niches. Its premise is simple: What if every post you made had a significantly better chance of performing well? The platform approaches that question by studying content across a creator’s niche, finding posts that dramatically outperform expectations, and decoding the patterns behind them.
The logic is straightforward. If an account with 2,300 followers produces a post with 1.7 million views, something happened that its follower count alone cannot explain. If another account with 17,000 followers reaches 23 million views, the anomaly becomes difficult to ignore. The interesting signal isn’t the size of the number — it’s the distance between what should have happened and what actually did.
Aubrium’s bet is that those anomalies can become intelligence. Instead of staring at a blank screen and guessing what might work, creators can build from a library of ideas, formats and patterns that have already demonstrated an unusual ability to travel beyond the audiences that produced them.
In an era where generating content is becoming virtually effortless, knowing what is worth generating may prove to be the more valuable technology.
AI as an assistant, not the strategy
Aubrium is not making an argument against AI. It incorporates AI throughout the platform, including as an assistant to help creators turn ideas into something usable. The distinction is more subtle: generation is a tool, not a strategy.
That distinction matters more as the cost of production approaches zero. If ten creators can each generate ten competent posts today, tomorrow they may be able to generate a thousand. Once everyone has access to roughly the same creative machinery, the ability to produce more of something stops being much of an advantage.
Audiences may already sense the difference. Research published in the Journal of Consumer Research, covering more than a million TikTok posts alongside eight preregistered experiments, found that posts disclosed as AI-made received 7 to 8 percent fewer likes. The researchers found that the penalty was not simply hostility toward AI. It was associated with a decline in perceived effort — and, in turn, the sense of connection that effort creates.
There is, of course, an obvious challenge to Aubrium’s approach. A system built to identify what has already worked could just as easily encourage creators to make more of what has already worked. Social media was converging on the same crops, fonts, hooks and formats long before generative AI arrived.
The difference may come down to how intelligence is used. There is a meaningful gap between copying yesterday’s winner and understanding why something unexpectedly broke through.
And that may ultimately be the more important divide. The future of social media probably will not be AI content versus human content. It will be content made simply because it could be made, versus content made because somebody had a reason to make it.
Generative AI has solved the first problem remarkably well. The second remains stubbornly human.
And as the supply of things to scroll past approaches infinity, knowing what is worth making may ultimately become more valuable than the ability to make it.
