Hacker News Reader

We can build algorithms to avoid racial bias

Can we? Those algorithms are built/trained by humans with biases - how can you be sure we can build/train AI to avoid those same biases?

View on HN

It's an excellent open question. I'd argue that relative to training it out of humans, we have a glimmer of hope of building algorithms that avoid racial bias because you can always bust an algorithm open and "examine its entrails," and the same can't be done with human thinking.

... but it's not categorically, definitely better, just more likely to be so.

Reply on news.ycombinator.com