Most people do not realize when a personal message they receive was written by AI, study finds

The study investigated how people’s perceptions of a message’s author affect their judgment, focusing on AI-generated messages.
Researchers recruited over 1,300 participants, ages 18 to 84, and presented them with AI-generated messages in different contexts regarding authorship.
A significant “AI disclosure penalty” was found: disclosing AI authorship resulted in much more negative judgments of the sender compared to messages believed to be from humans.
Surprisingly, participants not informed about the authorship formed similarly positive impressions as those who knew the messages were human-written, demonstrating a lack of inherent skepticism about AI authorship.
Even frequent users of AI did not form inherently negative views about messages unless AI authorship was disclosed, indicating little baseline skepticism.
This lack of skepticism and negative impressions can have significant social and professional implications, affecting perceptions in areas like friendships, dating, and job applications.
The findings highlight a moral dilemma where secret use of AI can be beneficial without risks, but transparent use incurs reputational penalties.
Current research and future steps aim to understand how and when skepticism towards AI-generated content naturally develops in real-world settings.

You may have missed