Did ChatGPT Just Lie To Me? - The Scholarly Kitchen

To understand how Artificial Intelligence (AI) is affecting science publishing, we need to push these systems to their extremes, analyze how they perform, and expose their vulnerabilities. Only then can we discuss how they will transform our industry. Earlier this week, Todd Carpenter asked ChatGPT some generic questions about the potential role of AI in scientific communication and, as you can imagine, it generated some generic, hedged, inoffensive output. I wanted to see how ChatGPT would perform with scientific controversies -- situations in which the scientific community supported one belief and the public another. Or, in situations where there was no consensus in the scientific community.