The question cuts to the core of how generative AI tools—ChatGPT, Midjourney, Claude, Gemini—reshape creative work. Does leaning on them sharpen individual output or dull the human spark over time? Recent studies from 2024-2026 paint a nuanced picture: AI often boosts personal creativity scores and productivity, especially for less experienced creators, but it frequently narrows idea diversity, homogenizes outputs, and risks long-term skill erosion when overused.
The evidence splits along individual vs. collective lines, short-term gains vs. sustained effects, and how deliberately people engage with the tool.
Individual Gains: AI Lifts Baseline Creativity for Many
Multiple experiments show generative AI raises the perceived quality of creative work.
In storytelling tasks, participants with access to AI-generated ideas produced stories rated higher for creativity, writing quality, and enjoyment—particularly among lower-creativity writers who saw 10-26% improvements. AI acts as a scaffold: it supplies prompts, structures, or starting points that reduce initial friction and free mental resources for refinement.
Field studies in workplaces echo this. Employees using ChatGPT in real consulting projects earned higher creativity ratings from supervisors and external evaluators, but only when they possessed strong metacognitive skills—planning tasks, self-monitoring progress, revising strategies. Without those habits, AI delivered little creative uplift.
In specialized contexts like sports education, moderate dependence on generative AI positively influenced creative performance by lowering cognitive load on routine analysis, allowing focus on tactical innovation or movement design.
For less skilled or blocked creators, AI functions like a co-pilot that expands what they can produce quickly.
The Collective Cost: Reduced Diversity and Homogenization
The clearest downside emerges at group or population level.
Wharton researchers reanalyzed ideation experiments and found ChatGPT users generated higher-quality individual ideas but far less variety. In one toy-invention task using a fan and brick, AI-assisted participants clustered around near-identical concepts (“Build-a-Breeze Castle”), while human-only groups produced fully unique solutions. Only 6% of AI-influenced ideas ranked unique compared with 100% in the human condition.
Storytelling studies confirm the pattern: AI-enabled stories scored better individually but showed greater similarity to each other than purely human ones. The dynamic resembles a social dilemma—each user benefits, yet the overall pool of novel ideas shrinks.
Exposure to AI ideas also homogenizes thinking over time. One experiment found repeated use led to “vanilla” outputs and fewer bold innovations, with the narrowing effect persisting even after stopping AI. Algorithms trained on vast but finite datasets favor patterns that already exist, pulling outputs toward averages rather than outliers.
Risks of Overreliance: Cognitive Atrophy & Skill Erosion
Several lines of evidence point to potential long-term decline.
Brain-activity studies using EEG during writing tasks revealed ChatGPT users showed lower engagement in regions linked to creative ideation, semantic processing, and attention. Over months, reliance increased, with participants resorting to copy-paste and reduced effort.
Critics argue this mirrors historical shifts: calculators changed arithmetic fluency; constant GPS use weakens spatial reasoning. If AI handles ideation or first drafts routinely, the “use it or lose it” principle applies—divergent thinking and originality may atrophy.
Public perception aligns: surveys indicate many Americans believe AI will erode creative thinking more than improve it. Self-reported creative confidence often drops when people attribute work to AI, even if output quality rises.
When AI Supports Rather Than Substitutes Creativity
The difference lies in usage.
Metacognitive users treat AI as a sparring partner—prompting for alternatives, critiquing outputs, iterating manually. This expands exploration while preserving ownership.
Overreliance—letting AI write full drafts or generate final concepts—tends to flatten results. Narrow AI lacks lived experience, intuition, or true novelty; it recombines existing patterns efficiently but rarely breaks paradigms.
Studies on text-to-image tools show productivity gains (up to 25% in some art workflows) and higher peak novelty in select cases, but average content novelty declines as users converge on similar aesthetics.
Practical Takeaways for Creators and Teams
- Use AI to overcome blocks or accelerate iteration, not replace core ideation.
- Actively diverge: generate your own ideas first, then compare with AI suggestions.
- In teams, mix AI-assisted and human-only brainstorming to preserve diversity.
- Build metacognition: reflect on prompts, evaluate outputs critically, revise independently.
- Monitor personal habits—if you feel less ownership or struggle without AI, scale back.
AI does not inherently reduce creativity; passive dependence does. Deliberate, skilled integration tends to amplify it for individuals while demanding safeguards for collective originality. The tools evolve rapidly, so the balance shifts with how we choose to wield them.
For deeper dives, review primary studies from Science Advances (2024-2025 storytelling experiments), Wharton (2025 homogenization findings), MIT Media Lab (brain-activity research), or the Journal of Applied Psychology (2025 workplace metacognition study). The evidence remains emergent—ongoing work will clarify long-term trajectories.
[…] Does Using AI Reduce Creativity? […]