idle thought #594

Re: stuff like the Wattpadd purge. If you’re asking the question “is this AI lgbtq-phobic” etc...the answer is 99% yes, and it doesn’t even have to be on purpose. AI is trained on (potentially stolen) almost certainly racist, lgbtq-phobic, misogynist, etc data. In most cases, the techbros think that a) it’s not as big an influence as it is (spoiler alert, it 100%, look at the surveillance systems that have been put in place in retailers and the black man who got arrested because the AI system paired with it could not tell the difference between black men and is technologically racially profiling) and b) the way to fix that is with even more data, not curating your data or going slower. Because tech only works by moving fast and breaking things. Never mind if they break society, that’s okay, they’ll just fix that with more tech down the line.