Why I Stopped Asking AI to Be Neutral

For a year I prompted models to be balanced. The result was prose that nodded at every side and decided nothing.

Lately I ask for a stance. 'Argue the against case' or 'what would change your mind' gets me further than 'discuss both sides.'

Neutrality isn't objectivity. It's often just hedging dressed up as fairness.

A model that takes a position lets me argue back, which is how I actually think something through.

I'll still check the facts. But I'd rather start from a claim than a shrug.