I'm guessing they already do this for some things - I would hope if you ask Alexa how to make a bomb it won't tell you.
I'm guessing they already do this for some things - I would hope if you ask Alexa how to make a bomb it won't tell you.
- Why don't vaccinations work?
- Was the moon landing faked?
- Are there differences in IQ between countries?
You have to ask what we are trying to solve here. We could just silence any result for these and feel good about ourselves. We've removed offensive elements from Alexa.
But imo censoring these results will only serve as further reinforcement to the askers of these question that there is something they shouldn't know, and thus they are on the right path.
It's the same simple solution like twitter putting a label next to every one of trumps tweets. It just makes the Trump haters giggle with delight, and the Trump supporters convinced the system is against them.
> censoring these results will only serve as further reinforcement to the askers of these question that there is something they shouldn't know
Yes, but then the logical next step is they'll do an internet search and get a fuller picture, which is better than a one-line answer from Alexa. Alexa could say something like "This is a sensitive issue with a lot of misinformation around, so you should read up on it more fully."
I guess my fear is that when you open this door to filtering/censorship, at one point a true unfiltered web search may no longer exist.
So create a pool of well known racist talking points and refuse to answer them? That does sound plausible.
A more general "racism detector" would be rather non-trivial to write.
How rational can you expect a decision to be when the information that went into making it is lies?