Algorithmic Persuasion Resistance: How AI Detection Shapes Human Argument Processing

With Wojtowicz's (2024) work showing that persuasive messages are computationally complex to generate but easy to adopt, and the rise of AI persuasion tools, this research explores a fascinating blind spot: what happens when people KNOW they're being targeted by AI-generated persuasion? Drawing from Isaac and Calder's (2024) persuasion knowledge model, I hypothesize that awareness of AI involvement triggers a unique processing mode that's different from both traditional advertising skepticism and general media literacy. People might develop what I call "algorithmic source monitoring" - an overactive skepticism that causes them to reject genuinely persuasive arguments simply because they suspect AI involvement. This could be tested using a clever experimental design where participants receive identical arguments but with varying (sometimes false) information about whether they were human or AI-generated. The implications are huge for everything from political campaigns to mental health apps - we might need to rethink whether transparency about AI involvement is always the ethical choice if it undermines beneficial persuasion.

References:

1. When and Why is Persuasion Hard? A Computational Complexity Result. Zachary Wojtowicz (2024). AAAI/ACM Conference on AI, Ethics, and Society.
2. Thirty years of persuasion knowledge research: From demonstrating effects to building theory to increasing applicability. Mathew S. Isaac, Bobby J. Calder (2024). Consumer Psychology Review.