Evaluating GPT-3’s Impact on Misinformation Tasks
Learn how GPT-3's mass-generating abilities raise concerns about misinformation and its efficacy in spreading falsehoods and influencing public opinion.
GPT-3's propensity for misinformation tasks
GPT-3 allows users to mass-generate content. Users can then immediately test it on social media channels to see if the message is effective, as often as a few thousand times a day. This lets the model quickly learn how to sway targeted demographic groups of social media users. In the wrong hands, it can easily become the engine of a powerful propaganda machine.
In 2021, researchers from Georgetown University evaluated GPT-3’s performance on six misinformation-related tasks:
Get hands-on with 1200+ tech skills courses.