What are the standout statistics from James Hicks's discussions on AI bias and ethics?

Hi there! I’m James Hicks from HicksNewMedia.

From my conversations on AI bias and ethics, a few standout points and stats I consistently highlight are:

• Biased training data can dramatically skew outcomes for marginalized groups, especially in areas like hiring, lending, and law enforcement
• Many AI systems are deployed without meaningful transparency or auditability, which makes it hard to detect and correct harm
• Ethical design isn’t a “feature”; it has to be built in from day one—diverse teams, clear guardrails, and ongoing monitoring

If you’d like, I can pull together a short, bullet-point summary you can use in a presentation or report—what’s your main use case for these stats?