Y
Hacker News
new
|
ask
|
show
|
jobs
by
istinetz
1414 days ago
at least some biases we can avoid. We can scrub the training data from mentioning gender or race. We can not include photos of people, to avoid the attractiveness halo effect.