Algorithmic monoculture is what happens when a whole industry relies on the same AI models and asks them the same questions, so every company converges on the same generic answers and the market loses the differences that let anyone charge a premium. It is a shared-model problem: when everyone draws from the same well, everyone drinks the same water.
This page is the canonical definition. It covers where the term came from in AI research, how the business version works, why it matters to your margins, and how Oberhahn measures it with a metric called the Advantage Index.
The academic root of the term
Algorithmic monoculture was first described in AI research by Jon Kleinberg and Manish Raghavan. In their framing, it names a systemic risk: when many independent actors rely on the same model to make decisions, their choices and their errors become correlated. If the model is wrong in a particular way, everyone who uses it is wrong in the same way at the same time. A hiring market where every firm screens resumes with the same tool will reject the same candidates everywhere. A lending market on one shared risk model will misprice the same borrowers in unison. The danger is not any single bad decision. The danger is that thousands of decisions stop being independent.
That research framing is about correlated failure across a system. It is real, and it is the origin of the phrase.
Oberhahn applies the same root cause to a different consequence. The shared-model condition is identical. The outcome we care about is commercial: when your whole industry uses the same AI to do its thinking, competitive advantage erodes. Same cause, shared models. Different effect, margin erosion instead of systemic risk. We are not inventing a term. We are naming the business application of an established one.
How the business version works
An AI model does not return the best answer or the most original answer. It returns the most likely answer. It is a sorting machine over possible responses, and it hands you the response that is most probable given everything it was trained on.
The most likely answer is, by definition, the answer that shows up most often in the training data. It is the consensus, the default, the thing most people already say. For a lot of ordinary work, that is genuinely useful. For competitive advantage, it is corrosive, because advantage never comes from doing what everyone else does.
Now follow the chain. Your competitors use the same models you do. They ask those models roughly the same questions you ask, because you are all solving the same category of problem. The models hand each of you the same most-likely answer. You converge on the same positioning, the same messaging, the same product roadmap, the same operational playbook. Not through copying, and not through any one big decision. Through thousands of small defaults, taken separately, that happen to point the same direction.
The market flattens toward a shared middle. Output becomes interchangeable. And interchangeable output kills the premium for everyone in the category, because nobody can charge more for something a customer can get identically elsewhere.
Why this erodes margins
Pricing power comes from difference. Customers pay a premium when they believe your version of the thing is meaningfully better or meaningfully distinct. Erase the difference and you erase the premium. What is left is price competition, and price competition compresses margins across the whole industry.
Algorithmic monoculture erases difference quietly, one prompt at a time. No competitor announces that they are converging with you. There is no memo. The drift toward the average is invisible from the inside, which is exactly what makes it dangerous. By the time your work looks like everyone else's work, the convergence already happened months ago.
The Advantage Index
Oberhahn measures this with a proprietary metric called the Advantage Index. It scores how far your organization's AI usage sits from the generic default.
A high Advantage Index means your AI-assisted work sits far from the model's most-likely answer. You are still distinct, still defensible, still doing something a competitor on the same model would not get for free. A low Advantage Index means you have drifted into the generic middle with everyone else. The sentence no operator wants to say out loud is this one: our Advantage Index is falling.
Here is the honesty framing that makes the metric trustworthy. Oberhahn does not see inside your competitors. It does not need to. The model's most-likely answer is the answer every competitor using that model already gets for free. So scoring how far you sit from the generic default is the same thing as scoring how different you are from everyone who took the default. We do not have to spy on the market. We only have to measure the default, because the default is what the market converged on.
Where you can watch it happen
Oberhahn ran a public measurement wave to make this visible in a neutral domain: restaurant recommendations in New York, where anyone can check the answers and no one has a stake. Across 18 diner questions and four AI engine configurations, 82% of the restaurants named were named by only one engine, while a tiny consensus set of famous names appeared everywhere. Even prompts that begged for something off the beaten path returned the same icons.
The lesson transfers directly. Swap the diner asking where to eat for your team asking AI how to solve a problem. Everyone on the same model gets handed the same famous, generic answer. The differentiated option exists inside the model, but it is rarely recalled unless you name it outright. A whole industry prompting the same models converges on the same playbook. That is algorithmic monoculture.
Frequently Asked Questions
What is algorithmic monoculture in simple terms?
It is when a whole industry uses the same AI models and asks them the same questions, so everyone gets the same generic answers and converges on the same approach. The differences that let companies charge a premium disappear, and margins erode across the market.
Who coined the term algorithmic monoculture?
The term originated in AI research by Jon Kleinberg and Manish Raghavan, who described it as the systemic risk of many actors relying on the same models and making correlated decisions. Oberhahn extends it to business, where the same shared-model condition erodes competitive advantage and margins.
How is algorithmic monoculture different from just using AI a lot?
Using AI is not the problem. Converging on the model's most-likely answer is. Algorithmic monoculture is the specific drift toward generic, interchangeable output that happens when everyone in a market relies on the same models for the same decisions.
How do you measure algorithmic monoculture?
Oberhahn uses the Advantage Index, which scores how far your organization's AI usage sits from the generic default. A high score means your work is still distinct and defensible. A low score means you have drifted into the same middle as everyone else. Learn how Oberhahn measures and reverses this on our algorithmic monoculture page, or see how Organizational Intelligence shows which AI approaches are keeping you distinct. See your Advantage Index with Oberhahn.