3 ms·
In my opinion, this is exactly how it should be done. People may have different preferences and have a complete right not to hear some versions or another. The
by g42gregory 6y ago
In my opinion, this is exactly how it should be done. People may have different preferences and have a complete right not to hear some versions or another. The default should be no filters though.
- spicybright 6y agoI agree with the concept, but I don't think it's a good idea for Intel alone to decide how to classify the speech. A better approach may be a few companies with their own AI's that vote on how to classify something. That would prevent things like classifying AMD as a swear word, or suppressing talk of industries Intel has a vested interest in. Usually in these situations, users are subjected to that behavior for months, until one person points it out on twitter. Then the company goes "Oops! It's AI so things like that happen! We'll patch it" and then they'll pull things again, continuing the cycle. Maybe I'm overthinking it as it's just for gaming, but this technology can and likely will be applied to other domains if it works out.
- g42gregory 6y agoI think these classifiers need to be open-sourced. Can you imagine Google, FB and Twitter AIs voting on the filters? Who would trust that? With open-sourced code, people can check it and see for themselves. Maybe there could be multiple versions of the filters? That way, if some people disagree with the words classifiers, they can use the versions that they like. I think the key here is to give people tools to protect themselves, if they want to. On the flip side, there should be no censorship of the content for the general public. The default = all filters are off. If something offends you, - go get the filter and don't bother other people.