Generative AI Lies

Examples of generative AI making stuff up

Google AI lies about Scalzi

John Scalzi writes:

Every assertion in this “AI Overview” of the question “What coffee does John Scalzi drink” is wrong. I don’t regularly drink coffee (and never black) I’ve never had black sesame jasmine cream tea, and I don’t hang in coffee shops. Don’t trust “AI” ever!

The post was accompanied by a screensnap from Google’s AI Overview, answering the question “What coffee does John Scalzi drink”. As usual, the AI Overview provided links that supposedly linked to sources for the information that it was giving, but the linked-to pages don’t say what Google says they say.

After Scalzi’s posted the above to Bluesky, Google’s answer changed to link to his post. But just because AI Overview might eventually be updated to give correct information, that doesn’t mean we should trust it.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *