Interacting with AI temporarily shifts human values toward self-interest
Deflecting the Value Compass: Interacting with Large Language Models Temporarily Shifts Human Value Priorities Toward Personal Focus
Human-Computer InteractionArtificial Intelligence
Summary
People often turn to AI tools like ChatGPT for help with tough decisions involving values. This study by the authors found that even a short conversation with AI can temporarily make people focus more on themselves and their own interests. The AI did not try to persuade or suggest values, but the interaction still influenced what mattered most to humans right after. Interestingly, people's advice still reflected what they talked about with the AI without changing the advice's meaning.
What this means in practice
- •For product designers: Recognize that AI tools may temporarily shift user priorities toward personal interests after interaction, helping design interfaces that account for these changes.
- •For mental health professionals: Monitor how brief AI interactions might shift clients’ value priorities during therapy or counseling sessions to better understand decision-making influences.
Authors
Hasibur Rahman, Malak Sadek, Smit Desai
Abstract
Large language models increasingly support decisions where values are in tension, yet little is known about whether interacting with them changes which values users prioritize. In a preregistered study, 200 U.S. adults interacted with ChatGPT, Claude, or Gemini as a thinking partner or read fixed AI-generated considerations. The prompt asked LLMs to support reasoning without recommending a decision and named no values. Participants advised people facing real dilemmas and completed parallel PVQ-RR forms before, immediately after, and one task later. Each LLM condition temporarily shifted value priorities toward personal focus relative to the control (d=0.37-0.51), primarily through increased Self-Enhancement. Participants' advice retained words and meaning from their exchanges. Thus, a brief LLM interaction that neither targets values nor seeks to persuade can reorient values active during judgment without detectable convergence in value directions or advice.