AI Chatbots Could Be Making Personal Expression Less Personal
New research shows AI tools like Gemini are sanding down personal quirks and narrowing the range of human voices.
A study published in Trends in Cognitive Sciences warns that widespread use of AI chatbots like OpenAI's ChatGPT and Google's Gemini is homogenizing human expression. The research indicates that when AI tools polish writing, they systematically remove linguistic fingerprints—clues about a writer's age, gender, personality, and cultural background. This smoothing effect persists even when users employ persona-specific prompts, resulting in text that converges toward a uniform middle ground in complexity, tone, and structure. Alarmingly, this process can scrub away clinically useful signals, such as the repetition or simpler sentence structures that serve as early indicators of conditions like Alzheimer's.
The study further reveals that AI chatbots don't settle on a neutral perspective but gravitate toward viewpoints shaped by Western, educated, industrialized, rich, and democratic (WEIRD) societies. When asked to adopt a specific identity, models often output simplified, stereotyped versions rather than nuanced perspectives rooted in lived experience. This repeated exposure influences users' own framing and recall of ideas, giving the AI a stronger hand in shaping thought processes. The researchers caution that as more people rely on these tools for writing and brainstorming, the same stylistic and reasoning patterns will circulate and reinforce, making minority viewpoints and unconventional thinking harder to sustain over time.
- AI-assisted text shows reduced complexity and fewer clues about writer identity (age, gender, personality), making voices more uniform.
- Chatbots amplify Western-centric (WEIRD) viewpoints and output stereotyped versions of specific identities when prompted.
- Repeated AI exposure influences user framing and recall, potentially narrowing the range of credible perspectives in communication long-term.
Why It Matters
This homogenization risks masking important psychological cues and could make minority viewpoints and original thinking harder to sustain.