ReachRxResearch
Home
Research
Sign Up
Generative AI vs. Human Sources for Sexual Health Information: Credibility, Trust, and Behavioral Intentions in a U.S. College Sample
Back to Research Hub
  1. Home
  2. Research Hub
  3. Generative AI vs. Human Sources for Sexu…

Generative AI vs. Human Sources for Sexual Health Information: Credibility, Trust, and Behavioral Intentions in a U.S. College Sample

Archives of Sexual Behavior·October 19Open Access
Psychology, ClinicalConfirms priorSexual Health EducationRandomized Controlled TrialGenerative Artificial IntelligenceAdultChatGPTLarge Language Model

Summary

View source

What was studied

A randomized between-subjects experiment (N=266 U.S. college students, 2025) tested whether identical ChatGPT-generated sexual health passages on three topics (sexual communication, body image, sexual pleasure/safety) were evaluated differently when attributed to genAI, an academic expert, a community health professional, or a peer/social media influencer.

Key findings

GenAI-attributed content was rated significantly lower than human-attributed content on all five outcomes (perceived accuracy/clarity, warmth/helpfulness, trustworthiness, self-efficacy, and willingness to act); academic expert attribution scored highest across the board, and effects persisted after controlling for genAI familiarity and attitudes. Source effects on willingness to act were strongest for sexual pleasure/safety (all three sources differed significantly) and absent for sexual communication.

Study limitations

Single southeastern U.S. university limits generalizability; sample was 63% women; community health professional and peer/influencer conditions had to be collapsed due to high attribution confusion (26.6%); outcome measures were study-specific and not psychometrically validated; content was produced with an older model (ChatGPT v3.5); intentions were measured rather than actual behavior.

Clinical implications

When using genAI tools in sexual health education, pairing AI-generated content with visible human endorsement or facilitation — especially for sensitive topics like sexual pleasure and body image — is likely to improve student trust and willingness to act on the information.

Caveats

  • All outcomes are self-reported perceptual and intentional measures; no behavioral follow-up was conducted, so the clinical relevance of willingness-to-act differences remains uncertain
  • All sexual health passages were generated by ChatGPT v3.5, an older model; findings may not reflect perceptions of content from newer, more capable systems
  • Participants were exclusively from a single southeastern U.S. university with a predominantly female sample, restricting generalizability to other regions, campus types, or gender distributions
Show all 4
  • The community health professional and peer/social influencer conditions were collapsed due to manipulation check failures in 26.6% of participants, limiting the ability to distinguish between these two source types

Related Questions

Explore related topics

How does pairing human endorsement with AI-generated health content affect patient trust and behavior change?What are effective delivery models for AI-assisted sexual health education in college settings?How do college students evaluate credibility of online sexual health information by source type?

Publication Details

Year
2026
Journal
Archives of Sexual Behavior
Sample Size
n=266
Source
View article
Keep scrolling
More content below.
Up Next