A recent preregistered study published in PNAS Nexus demonstrates that artificial intelligence can temporarily reduce prejudice against transgender individuals by delivering personalized, value-aligned conversations. Led by researcher John Holbein, the project investigated whether scalable AI interventions could replicate the impact of traditional, time-intensive storytelling and empathetic dialogue methods. The experiment engaged over 2,800 American adults, who first completed a Moral Foundations Questionnaire to establish their core ethical priorities. Participants were then randomly assigned to interact with a GPT-4o-powered chatbot designed to frame arguments for transgender rights around their specific moral values, while the remaining cohort served as a control group. The AI system was programmed to prioritize empathy and perspective-taking, using structured prompts to align transgender rights advocacy with the participants pre-identified value frameworks. Results indicated a statistically significant, short-term shift in attitudes among those who engaged with the chatbot. Participants reported increases of 3.94 to 5.19 points on a standard zero-to-100 feeling thermometer when measuring support for transgender rights, alongside improved willingness to take supportive actions and more favorable broader attitudes toward the community. Despite these immediate gains, longitudinal tracking revealed that the intervention lacked durability. Follow-up assessments conducted one week after the initial interaction showed substantial attenuation, with the majority of measured effects falling below statistical significance. The researchers conclude that while brief, AI-mediated moral matching can effectively reduce bias in the immediate term, the effects are transient without reinforcement. The study positions scalable AI tools not as replacements for sustained community engagement or structural policy changes, but as complementary mechanisms that can lower participation barriers and extend the reach of inclusion initiatives. By demonstrating that algorithmic empathy can bridge ideological divides, the findings highlight a pragmatic avenue for tech-driven social interventions, while underscoring the necessity of long-term engagement strategies to sustain attitudinal change.