I can't stop thinking about the day after the Uvalde massacre when a 17-year-old trans girl I interviewed was assaulted outside a public library in El Paso. Three men harassed her, and one violently manhandled her, repeating false information from the internet about how the shooter had been a transgender, undocumented immigrant. Because she appeared to them to be a transgender Latina woman, they went after her. She managed to get on her bicycle and flee, but she was terrified.

The false rumors about the shooter began on 4chan, apparently. But it didn't take long for Facebook and Twitter to blow up with them. It didn't take long for Republican political leaders to start repeating them either. The claims are still circulating, and some people still believe they're true.

This is just one concrete case about runaway misinformation on the Internet harming a person in the real world. The misinformation was apparently not deliberate, in the sense that it wasn't centrally coordinated, I mean. Some knuckleheads made it up, some other knuckleheads decided to believe it without verification, and then it started to spread like wildfire.

And that's without AI, although in a sense, I think runaway social media information brush fires are a sort of decentralized AI in a very broad manner of speaking.

The false rumors started raging because enough people out there hate trans folks. All it took was a spark to light a larger social media fire, which then spread into the real world and burned an innocent 17 year old who hadn't even heard the rumors.

The fire was helped along by prominent influencers and political leaders without access to actual AI.

Imagine how much worse if they did have powerful tools like that. Imagine if they could use AI to spread hateful misinformation and have much more confidence that it would take off.

That's something we had better all worry about.