Most people are unaware when a personal message they receive was written by AI, a study shows

Two new experiments show that most people don’t even consider that a personal message could be AI-generated, even if they themselves use artificial intelligence to write it.

To see how people judge someone based on their texts in the age of ChatGPT, my colleague Jiaqi Zhu and I recruited more than 1,300 U.S.-based participants ages 18 to 84 and showed them AI-generated messages like an apology sent in an email. We divided our volunteers into four groups: Some people saw the news without having any information about who or what wrote it, as in everyday life. Others were told that the messages were definitely written by a human, definitely AI generated, or that the source could be both.

We found a clear penalty for AI disclosure. When people knew that a message…