People are more likely to believe AI-generated tweets than ones written by humans, study finds
2023-06-29
Business Insider
The information displayed in the AIM should not be reported as representing the official views of the OECD or of its member countries.
Researchers at the University of Zurich found that OpenAI's GPT-3 can generate disinformation in tweets that is more convincing and harder to detect than human-written content. The study highlights the risk that AI-generated falsehoods could be used in large-scale disinformation campaigns, potentially harming public opinion and trust.[AI generated]
Why's our monitor labelling this an incident or hazard?
The article describes a study showing GPT-3's ability to generate convincing disinformation that can mislead people, which is a credible risk for harm to communities and public health. While no actual harm is reported as having occurred, the AI system's use in this way could plausibly lead to incidents of misinformation causing societal harm. The event is not a direct incident but a credible hazard due to the AI's demonstrated capabilities and potential misuse. Hence, it fits the definition of an AI Hazard rather than an AI Incident or Complementary Information.[AI generated]