AI Should Be More Human, Not More Complex

This paper challenges a prevailing trend in AI development: the assumption that more verbose and lexically complex responses improve user experience. Through a large-scale comparative study involving approximately 10,000 participants and five major AI-powered search systems, we find that users overwhelmingly prefer concise, source-attributed responses that mirror effective human communication. Our findings indicate that AI systems optimized for artificial sophistication (lengthy, confident, but often opaque answers) create higher cognitive load and reduce user trust and engagement. By analyzing linguistic patterns, source attribution, and usability metrics, we show that users value clarity, brevity, and transparency over perceived intelligence. These results suggest a shift in AI design philosophy: from maximizing information density to maximizing communicative utility. Effective AI, we argue, should not sound more complex, it should sound more human.