|
|
|
|
|
by alpaca128
1904 days ago
|
|
That is a problem, but with the algorithm or its flawed application. I don't think it's too far-fetched to expect one of the largest and richest tech companies worldwide to be able to tell with decent accuracy how the word "asian" was used. The sophistication of their blocking method is basically on par with a basic word filter that doesn't even attempt to avoid the Scunthorpe problem. I mean, we're not talking about reaching 99% accuracy or really advanced linguistic detection. When a company sells smart home voice assistants surely they can be expected to tell "asian food" from more lewd content apart. Keep a list of naughty websites, track what web searches regularly return those as result and then target those specific queries. Or throw some Machine Learning at it, might as well put their enthusiastically marketed AI processing unit to use. Accurate content filtering is hard. But completely blocking anything using one of the most common words in the language just because it frequently links to porn sites? Come on. |
|