Post by @rickydata • Hey
With content moderation being such a controversial topic, I think AI probably presents a really unique opportunity to create an objective arbitrator to act
Comments
- I think the main thing here is the ability for the user to pick what is used and not, and if you can turn it all off, then I see you have control over how things are presented to you. AI's future is bright, and it is necessary to have platforms like this work well. That said, I believe no one answer fits all, and it can censor stuff without meaning to. As long as you can always opt out of it all, I see you have solid foundations. The problems occur when you can not get to the data before the ML code refactors the results set.
In terms of open algos, I think it does make sense, but as you said, at the same time, the ML built to hide bots then gives the bots the code to get around it. This then comes back to my point above if you can opt out of these models, then maybe that’s enough — a tough one.
- The main problem is the bias that could be introduced for underrepresented collectives. This is "dangerous" if it is not done properly and could create an even worse solution. As you said it should be open an audited by anyone that wants to so it is possible to flag bad cases and retrain it.
- DeFi minises dependencies on FS institutions. Hopefully DeSoc similarly can remove Big Tech social cos and, via an ecosystem of filtering tools and algo feeds, reduce possibilities for centralised regulatory overreach. Not a big ask then 😉
- Though I like the idea, the process seems painstakingly long. Imagine a governance type decision making process, except it involves everyone who remotely cares about the subject.
Training data has also been known to be sensitive depending on what it is, thus why companies such as OpenAI offshore testing data to third-world countries for ridiculously low rates of pay.
I'm sure many people would volunteer for this cause, but not many would for the sensitive content that no one wants to look at.
- I think you can set what you can and can't see according to your own preferences, but you shouldn't put restrictions on one person, otherwise why wouldn't I be on Twitter?
- wwww
- )