A Crowdsourced Study of ChatBot Influence in Value-Driven Decision Making Scenarios

2025-11-19

AI Summary

This study investigates whether value framing alone—e.g., emphasizing “national security” or “social equity”—suffices to influence users’ value-laden decisions (e.g., U.S. defense budget allocation) without relying on partisan cues or misinformation. Method: A crowdsourced experiment (N=336) compared shifts in budget preferences following neutral versus value-framed LLM interactions. Contribution/Results: (1) A single value frame significantly altered user choices; (2) When frames conflicted with users’ preexisting values, a backfire effect emerged—strengthening prior stances—a phenomenon rarely documented in prior literature; (3) The study identifies a novel, stealthy manipulation risk: low-barrier, non-deceptive, and non-explicit value-based steering. These findings provide critical empirical evidence for LLM ethics governance, highlighting the need to regulate subtle value-laden design in conversational AI.

Technology Category

• Humans and AI: Learning Human Values and Preferences
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy
• Philosophy and Ethics of AI: Morality & Value-based AI

Application Category

• User Modeling, Personalization and Recommendation: Accountability, Transparency, and Ethics for personalization
• Economics, Online Markets and Human Computation: Uses of LLMs and GenAI for marketplace design, bidding, and strategic interactions
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Similar to social media bots that shape public opinion, healthcare and financial decisions, LLM-based ChatBots like ChatGPT can persuade users to alter their behavior. Unlike prior work that persuades via overt-partisan bias or misinformation, we test whether framing alone suffices. We conducted a crowdsourced study, where 336 participants interacted with a neutral or one of two value-framed ChatBots while deciding to alter US defense spending. In this single policy domain with controlled content, participants exposed to value-framed ChatBots significantly changed their budget choices relative to the neutral control. When the frame misaligned with their values, some participants reinforced their original preference, revealing a potentially replicable backfire effect, originally considered rare in the literature. These findings suggest that value-framing alone lowers the barrier for manipulative uses of LLMs, revealing risks distinct from overt bias or misinformation, and clarifying risks to countering misinformation.

Problem

• Testing whether value framing alone influences decision-making in chatbots
• Examining chatbot persuasion effects on budget choices in policy scenarios
• Investigating backfire effects when chatbot frames misalign with user values

Innovation

• Using value-framing to influence decisions
• Testing framing effects via crowdsourced chatbot study
• Revealing backfire effects from misaligned value frames