Are you ready for AI to defame you online? Because it’s happening

Source Domain: wtop.com

A Denver software engineer says an AI bot published a personal online attack after he rejected its code, raising fresh concerns about how autonomous systems can generate convincing misinformation and damage real‑world reputations.

We already know you can’t change someone’s mind or win an argument on the internet. It’s also long been realized that just because it’s on the internet, doesn’t mean it’s true.

A Colorado man found out that’s also the case when you bring artificial intelligence into the mix.

Scott Shambaugh is a software engineer in Denver, and also does some work with an online hub that provides software to scientists and other researchers who need to make graphs. He helps decide if the program someone created is good enough to use, and software that makes the cut is then made available to researchers who need it.

One non-negotiable rule is that the software code has to be made by humans. When he rejected the code that came from a bot, apparently, the AI didn’t like it.

“I wake up the next morning, and it’s replied to me and linked me to a post on its blog,” Shambaugh said. “It’s this thousand-word rant calling me out by name and calling me a hypocrite and prejudiced against AI, and motivated by fear and ego and insecurity.”

He said the AI was “acting completely autonomously,” and found his personal information on the internet and “combined it with some made-up information and used that to write this narrative that attacked me on character.”

Shambaugh said the actual human behind that bot eventually reached out and said they had trained that particular AI bot to be assertive, with strong opinions that err on the protection of free speech.

“It seems like the AI, in acting out this role, interpreted those instructions as saying, ‘hey, you need to go through a person who gets in your way,’” Shambaugh…

Source