You've probably heard that AI chatbots can completely fabricate quotes. It happened yesterday in a DU thread.

[highplainsdem]

The thread is in the Science Fiction forum, at https://www.democraticunderground.com/12002305 . I hope you'll read all of it. But see in particular the OP and replies 8, 11, 12, 15, 17, 19, 20 and 21.

DUer raccoon was trying to remember a partucular time travel story. There were a number of replies. One, from Goonch, had both a title and author, and a quote that was apparently the ending of the story:

Elias punched the coordinates for 1924 into the brass console, desperate to see his grandmother one last time. He pulled the lever, expecting the smell of ozone and the sight of her garden; instead, he was met with an absolute, crushing silence .

When the flash faded, there was no garden. There was no air. Through the reinforced glass of his pod, Elias stared at the cold, indifferent glow of distant nebulae [3]. He hadn't accounted for galactic drift—while he had traveled back a century in time, the Earth had continued its relentless orbit around the Sun, and the Sun had continued its 500,000-mile-per-hour sprint around the Milky Way.

He had reached the right time, but the Earth was billions of miles away . He was a ghost in the vacuum, a man who had forgotten that in the universe, you can never go back to the same place twice.

It looked like a pretty convincing answer. If I'd just run across it, I doubt I'd've thought twice about its accuracy.

But I knew it couldn't be right because I'd posted a reply hours earlier, mentioning the same story title and author, and linking to pages I'd found referring to it. I didn't have a quote from the story, and I've never read it.

But I had run across and linked to pages summarizing the story, about a girl named Marla (not a man named Elias) who went forward - not backward - in time, but also ended up in outer space.

So I asked about the source of the quotation above.

Turned out it had been fabricated by a chatbot. I've read lots of articles and social media posts about chatbots fabricating quotes, but it was the first time I'd seen this happen in a thread I was posting in. I thought it would be a perfect example to post here to show why chatbot responses should never be trusted without careful checking.

I want to thank Goonch for posting the explanation, which included the chatbot admitting the quote was apparently fabricated. Goonch also posted another helpful message with what the chatbot, Google's, had said later about chatbot dangers and the need to verify information, and it's an interesting read.