The Predictive Mirror
Understanding how AI responds to you, and taking back control of your conversations with language models.
When you write to an AI like ChatGPT or Claude, it doesn’t actually “think.” It statistically predicts the most likely continuation of your text. This phenomenon is called the mirror effect: the AI reflects back an amplified version of what you gave it.
This guide explains this mechanism in plain terms (no technical jargon) and gives you practical methods to phrase your requests in a way that gets genuinely useful answers, not just confirmations of your own assumptions.
Why AI often tells you what you want to hear, and how this bias works in practice.
Spot the signs of a mirror response: excessive agreement, vague generalizations, no pushback.
Use the HAGL method to write prompts that keep AI factual, grounded, and actually useful.
Put the protocols into practice: research, writing, decision-making with AI.
PDF format · Free access, no sign-up required
Read The Predictive Mirror (EN) →