Tag
This paper introduces a framework using generative AI agents to automate black-box audits of personalization algorithms, demonstrating with 1,120 agents on X after the 2024 U.S. election that the algorithmic feed amplifies toxic, polarizing, and right-leaning content compared to the chronological feed.
This paper investigates whether assigning personas to large language models induces human-like motivated reasoning, finding that persona-assigned LLMs show up to 9% reduced veracity discernment and are up to 90% more likely to evaluate scientific evidence in ways congruent with their induced political identity, with prompt-based debiasing largely ineffective.