Age and generational differences in anthropomorphism and trust in large language models

Introduction As large language models (LLMs) increasingly mediate everyday information seeking, a fundamental question emerges: do people conceptualize these systems as social agents, and does this tendency vary by age? Methods We investigated how individuals anthropomorphize and trust a pseudo-LLM, examining English-speaking adults from the United States ( n = 1,485) across four generations: Gen Z (age 18–26), Millennials (age 27–42), Gen X (age 43–57), and Baby Boomers (age 58–77). We experimentally manipulated anthropomorphic cues (text versus text+speech; first-person “I” versus third-person “the system” framing). Results Age and generational differences were observed: Gen Z (the youngest age group) showed consistently lower anthropomorphism, reduced trust ratings of the system, and lower ratings of accuracy for responses generated by the LLM. Anthropomorphic voice cues increased perceived human-likeness and accuracy uniformly across age groups. Qualitative analyses further revealed age and generational differences in how participants conceptualized the system. Discussion Together, these findings suggest that age and generational differences shape how people attribute mind and agency to artificial systems.