AI is already making life-and-death calls through courtroom risk scores and military targeting, and that is bad news for anyone caught on the wrong side of a biased algorithm.

• Sentencing and bail tools already help decide who goes to prison, and they weigh against Black defendants.
• Drone warfare increasingly relies on software to pick targets, meaning kill decisions are partly automated today.
• Palantir is pushing into this space because automated targeting is where military spending is heading.
• The fear of a fully independent "AI god" misses the point — people are still steering these systems from behind the scenes.

Outlook: Expect more pressure on courts and militaries to explain who is accountable when the software gets it wrong.