AI can be more persuasive than humans in debates, scientists find
GuardianMAY 5
Study: GPT-4 More Persuasive Than Humans in Online Debate
Above: The welcome screen for the OpenAI "ChatGPT" app is displayed on a laptop screen on Feb. 3, 2023 in London, England. Image copyright: Leon Neal/Getty Images via Getty Images
The Spin
Techno-optimist narrative
AI chatbots, when designed to be persuasive and personalized, can be powerful tools for good. Instead of passively presenting facts, they can engage users in tailored conversations that gently challenge harmful beliefs like conspiracy theories. By mirroring individual reasoning and offering relevant, respectful counterpoints, AI can promote reflection, reduce misinformation, and encourage critical thinking.
Techno-skeptic narrative
The persuasive power of AI poses serious risks for democracy and public discourse. These tools could be exploited for mass manipulation, enabling sophisticated disinformation campaigns targeting vulnerable populations with personalized propaganda. The ability to deploy AI bots at scale for microtargeting voters threatens election integrity and could exacerbate social divisions.