If you want ethical AI, regulate data collection. The algorithms are just gussied-up linear algebra and cannot function without immense data collections derived from nonconsensual commercial surveillance.
-
-
Who knew in 2021 the Matrix would turn out to be just a stupid matrix.
Show this thread -
This is what I mean when I say working on AI ethics at places like Google is like doing climatology at Exxon-Mobil, or heading the coyote safety division at ACME. It removes an entire class of effective regulatory interventions from discussion.
Show this thread
End of conversation
New conversation -
-
-
This is something I have been thinking about for a few years. You can generally find a series of data points that proxy for the “protected characteristic” you trying to ignore. Many ML techniques will inadvertently learn the associations you tried to prevent.
-
There's a really good literature on this. Last time I checked the consensus was this proxy-finding is robust and makes attempts at debiasing ineffective.
- Show replies
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.