AI Brainstorming Narrows Student Ideas, Research Finds
College application essays written with ChatGPT assistance contain more polished language but fewer original ideas than those written by hand, according to an eight-year study of over 370,000 student submissions.
Researchers at Georgetown University tracked how student writing changed after ChatGPT became available. The shift was striking: essays suddenly displayed richer vocabulary and smoother transitions, yet the underlying ideas converged into a narrow range of similar concepts.
In one analysis, human-written essays contained up to eight times more novel ideas than AI-assisted versions. A separate study of short stories found the same pattern-AI-written work used more interesting vocabulary but recycled homogenous plot lines, pushing distinctive or unusual ideas to the margins.
How AI Predicts Rather Than Creates
The problem runs deeper than surface-level polish. When a chatbot generates text, it predicts the next word most likely to produce a "good" sentence based on its training data. This means it can identify sophisticated language patterns independently of whether those words express genuinely new thinking.
Human writing works differently. When someone writes, they're pulling from neural networks that connect ideas in unique ways. A student linking a personal swimming memory to kites flying creates a distinctive association-a signal of actual creative thinking.
Suggestion Creates Conformity
The conversational nature of chatbots introduces another problem. Once a bot suggests a direction, humans tend to lock onto it. Users struggle to distinguish where their own thinking ends and the AI's begins, making it easy to adopt AI-generated perspectives as original work.
The impact falls heaviest on students with distinctive viewpoints. Research shows AI homogenizes ideas most aggressively among neurodivergent students and those from racial and linguistic minorities-precisely the voices that generate novel solutions.
What This Means for Creative Work
For professionals in creative fields, the implications are clear. Brainstorming-the work that precedes writing-is not a preliminary task. It's the fundamental work itself.
AI can serve a real purpose: specialists with deep knowledge can use these tools to handle technical or administrative work, freeing time for actual creative problem-solving. But treating brainstorming as something AI can handle while humans handle "real work" inverts the actual process.
A shrinking pool of original ideas affects more than individual careers. It shapes how organizations solve problems, how societies adapt to uncertainty, and what solutions emerge for complex challenges. The loss of distinctive thinking-especially from people with unconventional perspectives-is a loss for everyone.
Your membership also unlocks:
