IEEE Spectrum AI linked the spread of machine-generated content on social networks to a trust issue. In the author-described description, roughly half of the feed consisted of materials that careful users recognized as AI-generated.
According to this description, some such publications were spread by people who did not notice their origin, and some deliberately, for entertainment or to sow confusion. The material also mentions GPT-3 as an example of a chatbot that could confidently give plausible but false answers.
The source is presented in the format of a brief metadata synopsis rather than the full text, so the information cannot be considered independent confirmation of broader conclusions. It is unclear how the share of content in the feed was measured and how representative the observation is.
editorial commentary
Why it matters
A likely consequence is that users will increasingly need to verify the provenance of posts, but the source does not confirm either the scale of the trend or its dynamics. The next observable signal will be independent measurements of the share of machine-generated content and data on the accuracy of recognition. Substantial uncertainty is tied to the conclusion being based on a single author's observation and a metadata synopsis.