"Bias" in the context of AI can mean many things.
At its core, bias is just a statistical characteristic of an algorithm.
What companies should worry about is bias that:
- Damages their reputation (e.g. non-white Nazis)
- Doesn't meet business requirements
- Breaks the law (e.g. SB-205)
Check out this post for an actionable guide to thinking about it: https://blog.stackaware.com/p/ai-bias-iso-42001-colorado-sb-205-nyc-ll-144-audits-chatbots
At its core, bias is just a statistical characteristic of an algorithm.
What companies should worry about is bias that:
– Damages their reputation (e.g. non-white Nazis)
– Doesn’t meet business requirements
– Breaks the law (e.g. SB-205)
Check out this post for an actionable guide to thinking about it: https://blog.stackaware.com/p/ai-bias-iso-42001-colorado-sb-205-nyc-ll-144-audits-chatbots