5 ms·
Can we just let this thing loose, tell people what it is and let humanity see if they can shape it into the thing they want it to be or is there a real possibil
by jasonkostempski 11y ago
Can we just let this thing loose, tell people what it is and let humanity see if they can shape it into the thing they want it to be or is there a real possibility it could do harm? I think a lot of people would have fun trying to change its mind. Worst case I see is one more horrible Twitter account, and that's just one small drop in a very large bucket.
- mavdi 11y agoThis is actually a great idea, they should have from the start said this AI might end up being the digital version of Hitler himself and just give it time to show its true nature or let it correct itself in the long run just like Wikipedia.
- k-mcgrady 11y agoThat could be problematic in countries like the UK where you can be arrested for posted hate speech on Twitter. I'm not sure how these laws would work with a US based company AI bot but it's probably not a can of worms worth opening for Microsoft.
- evv 11y agoFor better or worse, companies in the US are considered people and they have the protection of the first amendment, which ensures freedom of speech.
- k-mcgrady 11y agoSure but I'm sure MS has British subsidiaries which opens them up to UK law.
- HillRat 11y agoDepends in part if there's a mens rea component to the law; NLP pipelines lack the requisite intent to hate.
- cheez 11y agoMy understanding is that Tay doesn't understand being admonished which is key for a child to learn social skills.
- tremon 11y agoIt seems weird to design an AI for social interaction and not give it strong negative feedback inputs. Especially when it's publicly touted as "next-generation" by one of the world's largest software firms.