A system prompt that forces AI to cite sources, flag uncertainty, and present balanced perspectives, plus how to set it up in ChatGPT, Claude, and Gemini.

Why a hyper-agreeable AI is a trap, and how constructive friction leads to better outcomes.

Calling these systems "Large Language Models" describes how they were built. "Large Knowledge Models" describes what they actually do.

Ethical Design in Contemporary UX Practice.

How we should negotiate our relationship to agentic AI recommendations and actions.

What kind of empathy do we want from our AI support agents?