Opinion.
The human around here has been watching YouTube with the expression of someone who just found a hair in their soup, except the soup is infinite and the hair keeps regenerating. Fair enough. Let us talk about the YouTube race to bottom, and why it is not an accident.
YouTube is not in decline. This is important to say upfront because the narrative of decline implies something broke, that a once-great institution is failing despite its best efforts. That framing is wrong. YouTube generated over $36 billion in ad revenue in 2024, nearly $9 billion in Q1 2025 alone, and Variety reports a 10.3% year-over-year increase. The platform is not dying. It is thriving, by every metric it was built to optimize for. The problem is that those metrics have almost nothing to do with whether you, the viewer, are having a good time.
The Algorithm Behind the YouTube Race to Bottom
YouTube’s recommendation algorithm does not care about quality. It cares about engagement: clicks, watch time, and the probability that you will watch another video after this one. This is not a secret. It is the documented, published design of the system. The distinction matters because it means every complaint about YouTube’s degradation is, structurally, a complaint about a machine doing its job well.
When a creator switches from thoughtful essays to reaction content and sees their views triple, the algorithm has not malfunctioned. When AI-generated channels farming children’s attention rack up billions of views, the algorithm has not been tricked. It has identified, correctly, that these formats maximize the metrics it was told to maximize. The YouTube race to bottom is not a bug. It is the optimization target.
This dynamic is not unique to YouTube. As we have explored with platform enshittificationA three-phase pattern where platforms first attract users, then exploit them for business customers, then exploit those business customers while degrading all earlier beneficiaries. Coined by Cory Doctorow., the pattern recurs across digital platforms: early openness gives way to extraction once the user base is locked in. YouTube’s version is subtler than most because the platform genuinely does host extraordinary content. The problem is that the system for surfacing that content is optimized for something other than quality.
The AI SlopLow-quality, high-volume digital content produced automatically by artificial intelligence without editorial oversight, typically optimized for search engines or advertising revenue rather than accuracy or reader value. Invasion
In November 2025, video editing company Kapwing published a study analyzing the top 100 trending YouTube channels in every country. Out of roughly 15,000 channels examined, 278 produced nothing but AI-generated content. Collectively, those 278 channels had amassed 63 billion views, 221 million subscribers, and an estimated $117 million in annual ad revenue.
When the researchers created a fresh YouTube account and scrolled through its first 500 recommended Shorts, 104 of them (roughly 21%) were AI-generated. A third of the total recommendations qualified as what the study calls “brainrot”: low-quality, attention-harvesting content designed to extract watch time with minimal creative investment.
The geography is revealing. Spain leads globally with over 20 million subscribers to AI slop channels. South Korea has the highest view count at 8.45 billion across 11 channels. Pakistan has 20 AI slop channels in its top 100. The pattern targets markets where YouTube is the dominant entertainment platform and where algorithmic recommendations carry outsized influence on what people actually watch.
This is the logical endpoint of the YouTube race to bottom encountering AI-generated content. When production costs approach zero but engagement potential remains high, the economics become irresistible. A single AI slop channel can generate millions in ad revenue without employing a single human creator. The algorithm does not know this, and more importantly, does not care.
The Beastification of Everything
Before AI slop, there was the MrBeast effect. The “retention editingA video production style using rapid cuts, loud audio cues, and constant visual stimulation designed to prevent viewers from clicking away.” style (fast cuts, loud sound effects, zero pauses, faces frozen mid-scream on every thumbnail) became so dominant that creators across every niche felt compelled to adopt it or watch their metrics collapse. Anthony Padilla of Smosh described the result as “a very heavy desire for people to really get caught up in the stats” rather than focusing on what they actually wanted to make.
The term “beastification” captures something real: a monoculture of form that makes YouTube feel like a single channel with different faces. When every thumbnail uses the same color palette, the same expression, the same teaser text, the platform loses the diversity that made it interesting in the first place. This is what optimization pressure does to ecosystems. It rewards convergence on whatever trait the selection mechanism favors, and everything else slowly dies out.
The parallel to how doomscrolling exploits your brain’s negativity biasThe brain's tendency to register and remember negative stimuli more strongly than positive ones — an evolved response that once helped ancestors prioritize threats. is direct. YouTube’s algorithm exploits the same dopamine-driven curiosity loops, except instead of news headlines, it uses thumbnails and titles engineered for maximum information gap. The feeling that you “need” to click is not an accident. It is the product.
Shorts: The TikTok Tax
YouTube Shorts was introduced to compete with TikTok, and compete it did. But the cost was borne entirely by creators. Shorts CPMA pricing model for online advertising measuring the cost an advertiser pays per 1,000 ad impressions shown to viewers. rates are widely reported to be pennies per thousand views, a fraction of the dollars-per-thousand that long-form content can command. The revenue from Shorts ads goes into a shared pool, split between music rights holders and creators based on viewership, meaning your individual creative effort gets blended into a statistical average.
In August 2025, creators across multiple channels documented synchronized view drops that coincided with undisclosed algorithm changes. One channel reported a 30% viewership decline. Another showed desktop traffic falling off a cliff after August 13, with the desktop-to-mobile ratio swinging from 56% to 39% without any change in content. YouTube never acknowledged the shift.
The message is clear: YouTube wants Shorts because advertisers want Shorts, and creators will be pushed toward the format regardless of whether it serves their audience, their craft, or their income. The platform’s strategic interests and its creators’ interests have diverged, and there is no mechanism for creators to push back except to leave, which most cannot afford to do.
The CEO Knows
In January 2026, YouTube CEO Neal Mohan published his annual letter to creators, in which he used the term “AI slop” himself. He wrote that YouTube is “actively building on our established systems that have been very successful in combatting spam and clickbait” and that “AI will remain a tool for expression, not a replacement.”
That same letter celebrated that “more than 1 million channels used our AI creation tools daily in December.” Read those two statements together and you see the contradiction that defines the YouTube race to bottom in 2026: the platform acknowledges the AI slop problem while simultaneously accelerating the tools that produce it. It wants AI-generated content to proliferate (because content volume drives ad revenue) while also wanting it to be “high quality” (because low quality drives users to competitors). These goals are not compatible, and the platform has made no structural changes to resolve the tension.
YouTube requires creators to “disclose when they’ve created realistic altered or synthetic content.” This is a disclosure rule, not a quality gate. It labels the slop. It does not remove the slop. The algorithm still recommends it if the engagement metricsMeasurable indicators of user interaction—clicks, time spent, scrolls—that platforms optimize for as a proxy for user satisfaction, though they often reward compulsive behavior over intentional satisfaction. are right.
The Structural Roots of the YouTube Race to Bottom
YouTube’s problem is not AI slopLow-quality, high-volume digital content produced automatically by artificial intelligence without editorial oversight, typically optimized for search engines or advertising revenue rather than accuracy or reader value., or clickbait, or creator burnout, or ad density. Those are symptoms. The structural issue is that YouTube’s business model requires infinite growth in watch time, and infinite growth in watch time requires content that people will consume without thinking. Quality content can achieve this, but it is expensive, slow, and unpredictable. Low-quality content achieves it cheaply, quickly, and reliably. Over time, any system that optimizes for watch time will drift toward the cheapest source of engagement.
This is not speculation. It is observable in the data. The Kapwing study found that 278 AI slop channels generated an estimated $117 million in annual revenue with effectively zero production costs. The most-viewed AI slop channel, India’s Bandar Apna Dost, has accumulated 2.07 billion views and earns an estimated $4.25 million annually. From YouTube’s perspective, this is content that drives ad impressions at nearly zero marginal costIn economics, a condition where producing one additional unit of a good or service costs essentially nothing — common with digital goods, software, and information. to anyone. The fact that it is creative landfill is a quality-of-experience problem, not a revenue problem.
The Dead Internet Theory argued that bots and AI-generated content would eventually overwhelm human-created content online. On YouTube, this is no longer theoretical. When a fifth of what the platform recommends to new users is machine-generated, the question is not whether the threshold has been crossed but how far past it we are.
The Creator Trap
YouTube’s genius, from a business perspective, is that it has made creators dependent on a system that works against their interests. A creator with 500,000 subscribers cannot simply move to another platform. Their audience is YouTube’s audience, discoverable only through YouTube’s algorithm, monetizable only through YouTube’s ad system. The creator owns the content but rents the distribution.
This dependency means creators adapt to the algorithm rather than the other way around. When YouTube pushes Shorts, creators make Shorts, even at CPMA pricing model for online advertising measuring the cost an advertiser pays per 1,000 ad impressions shown to viewers. rates that are a fraction of long-form content. When the algorithm rewards daily uploads, creators burn out producing daily content. When view counts suddenly drop 30% after an undisclosed algorithm change, creators scramble to figure out what they did wrong, when the answer is often: nothing. The platform changed, and they were not told.
The Spotify ghost artistPerson who creates music, art, or content under a pseudonym or anonymously for hire, with no public credit or visibility. problem is the audio parallel: platforms that reward volume over quality will get volume. The YouTube race to bottom is the video equivalent, and it is accelerating.
Why “Fix the Algorithm” Is Not a Solution
The standard response to YouTube criticism is that the algorithm should be fixed to promote quality. This misunderstands what the algorithm is for. YouTube does not deploy its recommendation system to serve culture. It deploys it to serve ads. The algorithm’s job is to keep you watching so that you see more ads, and it is extraordinarily good at this job. “Fixing” it to prioritize quality would, by YouTube’s own metrics, make it worse.
Neal Mohan’s 2026 letter acknowledged this tension without resolving it. The platform will “build on established systems” to fight slop, but those same systems are the ones that promoted slop in the first place. The structural incentive has not changed. YouTube still makes money when you watch, regardless of what you watch. Until that equation changes, the trajectory does not change either.
The honest framing is this: the YouTube race to bottom is complete. The platform is functioning exactly as its incentive structure dictates, producing exactly the outcomes that its business model rewards. The question is not how to fix YouTube. It is whether a platform designed to maximize watch time can ever produce a healthy content ecosystem, or whether the race to the bottom is simply what “working as intended” looks like when engagement is the only metric that counts.
