4 ms·
A more practical way to deal with adversarial content would be a "filter" AI we can use like AdBlock to hide the nuisances. You can have your own preferences, n
by visarga 3y ago
A more practical way to deal with adversarial content would be a "filter" AI we can use like AdBlock to hide the nuisances. You can have your own preferences, not the ones prevailing on any platform you want to visit.
It is possible to do this with current day technology. Pinbot (was on HN 3 days ago) has a transformer model for embeddings. It could also filter any topic or type of content we want, and the kicker - it runs all in the browser, private and no internet connection required
We can just tick a box to avoid Elon, Bitcoin, or various kinds of hype and activism, it works by semantic matching, so it understands phrase variations. Or maybe we just want to skip the occasional cat videos. The solution to the current situation is end user empowerment, we can make our own AI filters.
- majormajor 3y agoI've seen this in some sci-fi stories, it makes a lot of sense, but I have a personal hunch that setting up a filtering solution on various feeds of mostly-crap is going to be too much effort for most people, and they'll just slowly abandon the crap-feeds for anything beyond the most trivial sorts of entertainment. And then you could see something like a move from commenting on Twitter about news articles linked out to going back to going directly to the source and discussing things with other "verified subscribers" (whether paid or otherwise verified) there. Even in a "free, but ad supported" world it wouldn't surprise me to see outlets push for more "I'm a real reader and commenter" verification to keep their ad rates up.