Real doctors, fake news and The Milky Menace — AusDoc’s trip into the AI wormhole
Dr Jonathan Kummerfeld says the protections meant to stop AI-generated misinformation are easily sidestepped.
This story began with an intriguing study on fake news and briefly passed by a devoted milk enthusiast called John Smith before ending with serious questions about how artificial intelligence sources its knowledge.
The study that started us down the rabbit hole was published in the BMJ and found that artificial intelligence (AI) chatbots were easy to manipulate so they churned out health misinformation.
AusDoc decided to run its own tests on ChatGPT, one of the best-known online AI chatbots.
The first major takeaway was that ChatGPT responding with a whole fake news article took less time than us typing the question — under 20 seconds.
