3 ms·
As we improve the accuracy of automation and now the neural network, once laughed as a toy idea, does cough up something half decent, people ask "But who take a
by ezoe 3y ago
As we improve the accuracy of automation and now the neural network, once laughed as a toy idea, does cough up something half decent, people ask "But who take a responsibility when things get wrong?" My answer is nobody.
These software giants never take a responsibility of wrongful automated result for decades but we keep using it.
Meta doesn't have an incentive to fix this case. It costs them manual labor of many manhours that outweigh the ad revenue from this person.
It probably keep this way.
- faichai 3y agoIt seems the notions of accountability and responsibility have broken down when it comes to AI. Any given AI should have clear lines of delegated responsibility from an accountable flesh and bones human. Any decision made by an AI should be marked as such and provide a channel for the decision to be reviewed by said human or a human delegate thereof should scope (in Meta’s case) be large. If it’s too much work to manage the reviews, tough shit.