Raw LLM Responses

To be charitable, maybe a properly-trained algorithm could provide _corroborating_ evidence, like how lie detectors are used... We don't arrest people because of lie detector results alone though... The algorithms fail with black and asian americans (compared to white & latinx), because there are much fewer samples in the American population to learn from, and because the algorithms' initial training is on predominantly people of european descent. Like how if a white kid never met a black person for the first years of their life, any prejudice is already deep-rooted enough, that meeting black people might cause biases to be reinforced rather than corrected. In a country where most people believe having a weaponized military is justified for 'defense'... It needs to be up to the people how/when to use facial recognition technology. We can try to put a public ban on the technology but we have to acknowledge that today's government secretly spies on people unconstitutionally already. There is *no* excuse for releasing algorithms for real-world use before testing for prejudice results in any case... and the fact it's an afterthought, shows not just how undemocratic but how systemically