AI can do a better job of persuading people than we do

Recent research highlights how large language models can subtly influence public opinion using minimal user data. In an experiment involving 900 participants, those who provided personal details like age and education appeared to tailor their arguments more effectively when facing AI opponents, making it harder for fact-checkers to detect discrepancies in real time. This suggests a growing risk of coordinated disinformation as technology advances, urging policymakers to consider these emerging risks.