Thread by @tante

Oh, the leaked EU "AI" regulation is already starting out interesting.

(As with privacy/data protection the idea of regulating "manipulating behavior" is super problematic and a bad approach.)

This is basically all "AI" systems operated by someone in business. That is Amazon's product recommendation.

But at least you have to be specific with whom you put under "AI" surveillance /s

But it's cool that the current trend of European sinophobia goes so far that the EU adds regulation against "chinese style social scoring" that doesn't exist in that way in China. /s

Sorry, don't have the time to go through it in detail right now. Gotta get some work done :(

But it will be fascinating to see how companies will do exactly the outlined things and just avoid calling their actions "AI" but use "data science" or "big data" as terms and run with it.

This one is also fun: Can you ensure that your data set has neither intentional nor unintentional biases that may become basis for discrimination? I get what the intention is but if we could do that, we'd already do it.

(That thing feels very much like a warmed up GDPR)