A scientific study reported in Nature Human Behaviour reveals that AI, specifically GPT-4-powered systems, can be as persuasive as or even more so than humans in structured online debates. The research highlights that when provided with personal data, AI outperformed human opponents 64% of the time, especially on topics where participants had less entrenched views. The findings raise concerns regarding possible misuse during elections and public discourse, as AI can tailor arguments to individuals more effectively than people can. While some benefits are noted, such as the potential for promoting healthier behaviors or reducing polarization, experts warn of risks including misinformation, radicalization, and the need for proactive regulation.
Related article:
