Does brainstorming with AI make a whole group's ideas more alike, even when each writer's own output improves?

This explores whether brainstorming with AI makes writers' ideas more alike across a group, even if each writer individually gets more or better ideas. The short answer from the corpus: yes, but it depends on when the AI comes in. The clearest evidence is a preregistered experiment showing that AI-generated ideas reduced collective diversity for every group of writers, while AI that only refined ideas the writers had already come up with left diversity intact Does AI assistance homogenize or preserve creative diversity?. The narrowing comes from handing over the starting point, not from using AI as such.

The same study has a less obvious finding. Non-native English speakers contributed more diversity to the group than native speakers did, and AI ideation was the one condition that erased that advantage Does AI assistance homogenize or preserve creative diversity?. That matches a separate experiment in which GPT-4o autocomplete pulled Indian writers' essays toward Western phrasing and cultural references, while giving American writers bigger productivity gains Do AI writing assistants push non-Western writers toward Western styles?. The writers whose perspectives sit furthest from the model's training data are the ones who lose the most distinctiveness, and those are often the perspectives a group most needs.

This matters because ideation is where writers lean on AI the most. An interview study found writers use LLMs most heavily at the idea-generation stage and go back to it whenever they get stuck How do writers use AI through different creative stages?. Be careful with studies that report only per-person gains, though. One paper concluded that AI help narrows diversity, but it only measured how many ideas writers produced and how detailed they were, especially for less experienced writers. It never measured diversity at all Does AI assistance actually narrow the diversity of ideas?. More and richer ideas per writer can still add up to a more uniform pool of ideas across writers.

The narrowing also reaches past ideas into how writers come across to readers. A study of nearly 3,000 writers found that AI-assisted text reduced the variation in how readers perceived authors on 22 of 29 traits, so writers converged on a confident, polished persona Does AI writing make all writers sound the same?. That persona skews toward a particular profile. Readers judged AI-assisted writers as more educated, wealthier, and more likely to be native English speakers Does AI writing make authors seem more privileged than they are?. The shift showed up on every dimension tested and always in the same direction, so it isn't random noise Does AI writing assistance change how readers perceive the writer?. In fiction, the sameness goes deeper than style. AI stories can be told apart from human ones by narrative choices alone, such as how much agency characters have and how the timeline is structured, with no help from word choice Can AI stories be detected without analyzing writing style?. A model's default creative instincts are structural, and they spread into the work of the writers who borrow them.

The practical lesson is about order, not avoidance. Come up with your own ideas first and use AI to develop them. Research on multi-agent brainstorming points the same way: diverse teams only beat a single agent when members bring real domain expertise, and diversity without that grounding made results worse Does cognitive diversity alone improve multi-agent ideation quality?. Writers can also set up AI partners in advance with defined roles and limits on how proactive they are Can writers benefit from configuring AI writing partners in advance?, which gives you a way to keep the AI out of the first-idea seat. The corpus does not yet test whether this kind of setup actually protects diversity. That question is still open.