ChatGPT’s Foolery: AI-Generated Scientific Abstracts Deceiving Scientists

Fake Scientific Abstracts written by ChatGPT fooled scientists, study finds

According to a recent study, fake scientific abstracts generated by OpenAI’s advanced chatbox ChatGPT were mistaken for real research papers in nearly one third of cases. The program, which is eerily similar to humans, raises questions about the future of AI.

Researchers from Northwestern University and University of Chicago asked ChatGPT for fake research abstracts that were based on ten real ones published in journals of medicine. They then fed the fakes to two detection programs to try to differentiate them from the real reports.

ChatGPT generated completely original abstracts using fake numbers and confused reviewers almost one third of the time.