Milei's government will monitor social media with AI to 'predict future crimes'

> Hell, isn't this essentially what Palantir does?

Palantir is (was, 10 years ago, when I interviewed for them) a graph database with a UI that allowed easy viewing of large datasets. Then they sold consulting hours to build custom plugins. I like graph databases, but to paint Palantir as a pre-crime system is a bit over the top. It would be a great tool to e.g. find and model relationships between Bitcoin addresses.

Happy to receive an update on that, but I suspect this reputation that they're doing some magic data processing is probably great for sales, so unlikely to happen.

Not to mention encouraging people to commit crimes...

AI here is likely vaporware speak for a simpler algorithm or statistical analysis.

It's not really an improvement if they incompetently arrest the wrong people for precrimes.

that presupposes many factors to come to that conclusion

That necessity is the the driver of crime. That systemic discrimination is the driver of necessity.

The part I think this fails is that these factors are not sufficient for crime.

Not everyone who is poor becomes a criminal. In fact, it is an incredibly poor predictor by the numbers. A poor person might be more likely than a rich one to commit grand theft, but this ignores the the fact that 99.99% of poor people don't commit grand theft.

Overall, I think overstating the power of the correlation does a great disservice to all the poor people who are not criminals.

Think less about crimes committed by individuals and more about group crime:

eg:

Plans and goals for January 6th were available on social media before the event happened.

Street takeover("sideshow") plans are commonly available on social media.

Looters commonly pass around where they will be looting to other would be looters.

Outside of just directly talking about crime, there is combining overlapping red flags an individual displays that would paint them as higher risk of committing crime. Some combination of taking part in, and frequency of posting of, various topics present a predictor pattern. The most "obvious" ones are usually called extremist views/behavior; however there exists more subtle views that are fairly innocuous when viewed individually, but together form a higher risk of crime category.

It can in some cases either directly or indirectly cause it: i.e. the Southport riots in the North of the UK were caused by someone on social media incorrectly attributing (at least so far) the blame of the stabbing of three girls and a teacher to "a Muslim", and then English Defence League supporters staged riots based off this.

You're misunderstanding. They're going to accuse people of (basically) thought crimes based on their social media.

You'd be surprised. Here's a 6-second clip that manages to squeeze in grand theft auto, driving while intoxicated and driving without a license, complete with clear footage of the perps' faces.

https://x.com/actionforalice/status/1816216433010889167

You would be surprised. Criminals are not the smartest of people.

You can find people deliberately posting video of themselves (with their full, unobscured faces) committing crimes. We're not talking about the smartest bunch of people.

A lot of people. Some even live stream as they're doing it