Today I'd add to the list the discussion on Israel using AI to decide who to bomb. AI isn't smart enough to understand that people who just bought a washing machine shouldn't be shown ads for washing machines, but apparently it's good enough for deciding who's a terrorist.
Having said that, I personally know someone who was arrested and mistreated by police forces for being related to a criminal, and yet I've never managed to convince them that maybe the police shouldn't have that kind of power. In those cases I typically consider that the person discussing probably won't change their view, but maybe I can still reach those listening to the conversation.
I doubt ad targeting actually uses AI, at least not in the quite heavy form we currently associate with the term, but the rest of your point stands.
However, I am not sure if using examples of something currently considered crime will work well, especially as people will actively deny anything other than petty violations like speeding. It is valid of course, in particular when combined with inhumane regimes and authorities, or even regular ones that do not see lightly on people airing their dirty laundry, but convincing others takes more than just having a correct argument...
Ad targeting has long offered ML based targeting models. Usually with product names like "audience extension", "predictive segments", "likely X buyers". LLMs have taken over the "AI" term in general discussion, but the strengths and weaknesses are not too different here (just more computationally efficient).
https://news.ycombinator.com/item?id=37114262
Today I'd add to the list the discussion on Israel using AI to decide who to bomb. AI isn't smart enough to understand that people who just bought a washing machine shouldn't be shown ads for washing machines, but apparently it's good enough for deciding who's a terrorist.
https://news.ycombinator.com/item?id=39918245
Having said that, I personally know someone who was arrested and mistreated by police forces for being related to a criminal, and yet I've never managed to convince them that maybe the police shouldn't have that kind of power. In those cases I typically consider that the person discussing probably won't change their view, but maybe I can still reach those listening to the conversation.