3 ms·

This article seems mostly concerned about Facebook and YouTube's opaque filtering systems, but the authors are looking in the wrong place. These platforms aren'

This article seems mostly concerned about Facebook and YouTube's opaque filtering systems, but the authors are looking in the wrong place. These platforms aren't and shouldn't be responsible for hosting violent and disturbing content - regardless of its purpose or utility. The authors themselves discuss the can of worms that open when you start accommodating the interests of some groups and not others; it isn't sustainable and it isn't possible. Instead, this article should have focused less on the decisions of AI systems and neural networks, which can be difficult to decipher and interpret, and more on the decisions of the humans behind these tech companies. The latter is much easier to scrutinize and dissect. I think the truly concerning initiatives are projects like Google's Dragonfly. There's absolutely nothing opaque about a search engine that will actively assist the Chinese government with whitewashing history for a billion individuals. These are the decisions that should be examined.

djakjxnanjak 7y ago
>Instead, this article should have focused less on the decisions of AI systems and neural networks, which can be difficult to decipher and interpret, and more on the decisions of the humans behind these tech companies. The latter is much easier to scrutinize and dissect. I agree with your comment overall but this common assumption bothers me. We can put a debugger in a neural net, we can’t put one in a human brain. People get angry when a policy decision is made that targets them, and are told by that the people responsible that the decision was implemented with a neural net. Then they get mad at the neural net, which shifts blame away from the humans who control it.