7 comments

[ 4.8 ms ] story [ 33.6 ms ] thread
As I understand it, this was one of the issues that Timnit Gebru was trying address at Google AI.

A Prominent AI Ethics Researcher Says Google Fired Her https://www.wired.com/story/prominent-ai-ethics-researcher-s...

How is the Ai coming to this conclusion? Just curious.
I would tell you, but I would get banned.
When training for people, not enough training data of African-Americans.
no, this isn't coming from a lack of training data.
This is where I got my information:

Since algorithmic systems “learn” from the examples they’re fed, if they don’t get enough examples of, say, black women during the learning stage, they’ll have a harder time recognizing them when deployed.

A new study finds a potential risk with self-driving cars: failure to detect dark-skinned pedestrians

https://www.vox.com/future-perfect/2019/3/5/18251924/self-dr...

Well, it's not technically wrong. Humans _are_ primates, after all. But since I doubt that the algorithm has problems deciding on a species for white male homo sapiens, I guess that there's quite a bit of work still to be done.