The Algorithm Didn't Change

In 2016, digital advertising revenue in the United States surpassed television for the first time: $72 billion to $71.3 billion. By 2017, digital had overtaken TV, broadcast and cable combined. In 2019, social media alone exceeded print advertising globally. In 2025, social ad revenue overtook search in the US: $117.7 billion to $114.2 billion. This year, Meta is projected to exceed Google in total digital ad revenue — $243 billion to $239 billion — the first time in history Google has not held the top spot.

Through the 2010s, Google, Meta, and Amazon absorbed a growing share of the advertising market that traditional media lost. The money didn't disappear. It moved.

The conventional story is that social media democratised content. Anyone with a phone could publish to the world. The gatekeepers lost their monopoly on distribution.

That story is correct, but it describes the surface. The structural change was a bottleneck shift. Traditional media controlled distribution — printing presses, broadcast towers, cable networks. When social media removed that bottleneck, the scarce resource moved upstream: to the ability to create something worth distributing. The creator became the bottleneck. And for roughly a decade, that was the game.

The Subsidy That Built the Creator Economy

The creator economy is now estimated at roughly $250 billion, with Goldman Sachs projecting it could reach a $480 billion addressable market by 2027. More than 200 million creators are active globally. YouTube's Partner Program has paid more than $70 billion to creators, artists, and media companies in the past three years.

What powered this was not just the removal of traditional gatekeepers. It was a specific mechanism: major platforms increasingly shifted from follow-graph feeds to interest-graph recommendation models. Followers no longer guarantee reach. Content earns distribution on its own merits — watch time, engagement velocity, completion rate. On TikTok, the For You Page made recommendation-driven discovery the default consumption experience, not a side feature. Instagram's head Adam Mosseri has said plainly that Instagram is a recommendation engine, not a feed app.

This is the subsidy. Not the old kind, where you built an audience and the platform showed your posts to that audience. The new kind, where the platform distributes your content to people who have never heard of you, based on whether the content itself performs. A zero-follower account can reach millions. That was impossible in the television era.

The platforms didn't do this out of generosity. They did it because your content kept people scrolling, and people scrolling generated ad revenue. The free distribution was a subsidy — not unlike the ones I described in The Subsidy IS the Strategy and Google Killed the Click. Different market, same structure: subsidise distribution to capture attention, monetise the attention through advertising.

And for creators and businesses who could actually produce content worth watching, the economics were extraordinary.

The platform paid for your distribution. You paid for nothing.

The Supply Side Exploded

Then AI arrived — not as a replacement for the subsidy, but as a force that changes what the subsidy is worth.

An Ahrefs analysis of 900,000 newly detected English-language pages estimated that 74.2% contained AI-generated content. Nearly 94% of content marketers plan to use AI for content creation this year. Adobe's 2025 creator report found that 86% of global creators now use creative generative AI. Merriam-Webster chose "slop" as its 2025 Word of the Year, with AI-generated low-quality content central to the term; Australia's Macquarie Dictionary went further and chose "AI slop."

The average cost of producing a 2,000-word article has dropped 44% in two years — from $480 to $268. The exact numbers vary by source, but the direction is not ambiguous: content production is getting cheaper, faster, and more abundant, simultaneously.

The pipe still works. The algorithm still distributes content to whoever might engage with it. Nothing changed on the distribution side. What changed is that production — the thing that used to be the bottleneck, the thing that separated creators from everyone else — got cheap. And when it gets cheap for everyone, it stops being the advantage.

Last week, the New York Times Magazine published a profile of Tilly Norwood, an AI-generated actress. The writer, Taffy Brodesser-Akner, directed a scene with Tilly and received four passable options in four seconds. Four seconds. For most of the content we consume every day, that is more than enough.

The Tool Is Also the Threat

This is where the obvious objection lands, and it is correct: creators and businesses benefit from AI too. The productivity gains are real. Marketers report saving 2.5 hours per day. Production costs are down nearly half. A single person with the right tools can produce what used to require a team. I use AI in my own content pipeline. The economics are not debatable.

And this is not only about efficiency. AI gives access to people who were previously locked out of production entirely. The technology gave Tilly Norwood's creator a career the traditional system would not. A real benefit, for a real person.

AI is good for the creator. It is bad for creators.

The problem is not that AI helps you. The problem is arithmetic. The same tool that cuts your production cost in half does the same for every other account in your space. You produce more. They produce more. The platform distributes all of it through the same engagement signals it always did. The supply explodes. The distribution doesn't change. And the value of each individual piece of content — the thing you just made faster and cheaper — drops, because it's competing against a flood of content that was also made faster and cheaper.

This is not a temporary adjustment. Brands are testing virtual influencers because they offer control, availability, and significantly lower production costs. AI-generated influencers surged 40% in 2025, and Ogilvy projects that CMOs will allocate up to 30% of their influencer budgets to virtual personas this year. The performance claims vary and the economics are not uniformly better — results depend on audience, campaign type, and disclosure norms — but the direction of the budget is clear.

Meanwhile, the income reality for human creators was already fragile. Goldman Sachs estimates that roughly 4% of creators earn more than $100,000 — meaning 96% do not. For many monetising creators, brand deals remain the dominant income source, accounting for roughly 70% of creator revenue. Brand deals are also the first budget line to be reallocated when a cheaper alternative appears. And a cheaper alternative just appeared.

What Replaces the Moat

The pattern is now clear enough to state plainly. Traditional media had a monopoly on distribution. Social media broke it — creators could reach audiences without permission from an editor or a network executive. The bottleneck moved from distribution to production. For a decade, if you could create, you could compete.

AI is breaking that bottleneck now. Production cost is approaching zero. The tools are available to everyone. If anyone can create, creation is no longer the competitive edge. The bottleneck moves again.

Where does it move to?

The answer is not that you should stop producing. The algorithm still rewards frequency, and AI lets you maintain it. But more content, faster content, cheaper content — that is now the baseline, not the achievement. Every account in your space has access to the same acceleration. If volume were the moat, the race would already be over.

The bottleneck moves to the things AI cannot replicate and volume cannot substitute.

Identity — who you are, and why anyone should listen to you instead of the ten thousand other accounts in your space producing similar material with the same tools.

Judgment — what to make and what not to make, what matters to your audience and what is noise, including the willingness to say things that work against your apparent interest.

Trust — the accumulated credibility that comes from being consistently right, consistently independent, consistently present.

Owned relationships — the audience that follows you, not the feed you're in.

That is the diagnosis. Here is what to do with it.

The Shift

The first instinct most businesses have when content gets cheaper is to produce more of it. The instinct is not wrong — the algorithm rewards consistency and frequency, and reducing volume means losing reach. That is a real constraint, not a negotiable one. But the mistake is treating volume as the strategy rather than the baseline.

AI lets you maintain the posting cadence the algorithm demands without burning your team. Use it for that.

But if all you are doing is filling the pipe with AI-generated output, you are indistinguishable from the ten thousand other accounts filling the same pipe with the same tools.

The volume is table stakes. The differentiation comes from the editorial layer on top of it: which topics you choose, what position you take, what you decide not to publish. AI can generate a competent post on any subject.

It cannot decide which subject matters to your audience this week, or take a stance that risks alienating part of that audience because the stance is honest. That editorial judgment is the part that is not automatable. It is also the part most businesses skip.

Use AI in your production pipeline — but keep your hands on the steering wheel. AI is excellent at accelerating research, drafting structures, generating variations, handling translations, resizing assets across formats. It is not excellent at knowing what your business should say, or why, or to whom. The operators who will win this transition are the ones using AI to execute faster on decisions they made themselves — not the ones outsourcing the decisions to the same tools their competitors are using.

If your AI-generated content is indistinguishable from your competitor's AI-generated content, you have not built a content strategy. You have built a commodity.

And put a person in front of the brand. Brand pages are losing reach on every platform. Founder-led accounts, personal voices, faces and names attached to the work — these outperform corporate content by significant margins, and the gap is widening. This is not a trend. It is the algorithm responding to what audiences actually engage with: people, not logos.

This does not mean turning your company page into a personal vlog or posting founder selfies. It means the content your business publishes should carry a recognisable perspective — someone with domain expertise who takes positions, explains reasoning, and stands behind the work publicly. The difference is between a brand page posting a generic AI-generated carousel about industry trends and a named person from that company writing what they actually think about those trends and why it matters to their clients. The first is interchangeable. The second is not.

None of this is revolutionary advice. But the window between knowing it and acting on it is closing faster than most operators realise. When content was hard to produce, you could get away with mediocre strategy and good distribution. When content is free to produce, distribution is not the differentiator. The person is.

The algorithm still works. The distribution is still free. But the subsidy only matters if what you're distributing cannot be replicated by everyone else using the same tools you are.

If it can, you're not a creator. You're supply.

Being content is not enough anymore.

Sources

1. eMarketer / Social Media Examiner: Digital ad spending surpassed TV in 2016
2. IAB/PwC via Adweek: Digital surpassed TV + broadcast + cable combined in 2017
3. Zenith via Retail Dive: Social media overtook print advertising globally in 2019
4. IAB/PwC via The Keyword: Social ad revenue surpassed search in 2025 — $117.7B vs $114.2B
5. eMarketer via Marketing Dive: Meta projected to surpass Google in ad revenue in 2026 — $243B vs $239B
6. U.S. Census Bureau: Newspaper revenue dropped from $46.2B (2002) to $22.1B (2020)
7. The Hill / Census Bureau: Platform ad revenue growth through the 2010s
8. Grand View Research: Creator economy valued at ~$250B in 2025
9. SQ Magazine / Goldman Sachs: YouTube paid $70B+ to creators, artists, and media companies; $480B TAM by 2027
10. Digital Applied: Major platforms shifted toward interest-graph recommendation models
11. Ahrefs: Estimated 74.2% of new English-language web pages contain AI-generated content
12. Presenc AI: Article production costs down 44%; marketers saving 2.5 hours/day
13. Typeface via theStacc: ~94% of content marketers plan to use AI in 2026
14. SQ Magazine / Adobe: 86% of global creators use generative AI (2025 report)
15. Merriam-Webster / Macquarie Dictionary: "Slop" and "AI slop" — Words of the Year 2025
16. The New York Times Magazine / Taffy Brodesser-Akner: Tilly Norwood profile
17. Metapress: Virtual influencer market growth and brand adoption
18. Communicate Online / GroupM: AI-generated influencers surged 40% in 2025
19. AutoFaceless / Ogilvy: CMOs projected to allocate up to 30% of influencer budgets to virtual personas
20. DemandSage / Goldman Sachs: ~4% of creators earn over $100K; brand deals ~70% of creator revenue

Fabio Lauria,

CEO & Founder, ELECTE

Every week, we explore AI without the hype—using data, analysis, and an independent perspective.