Chapter 3 · AI & Holistic Management

How LLMs Actually Work

You don't need the math. You do need the right mental model — because it explains everything else in this guide.

What they actually are

  • Prediction engines for the next word in a sequence.
  • Not sentient. Not thinking. Sophisticated pattern matchers trained on enormous amounts of text.
  • This framing explains both their strengths and their failure modes.

From prediction to reasoning

The basic prediction mechanism has been extended into outputs that look like reasoning. This is what makes these tools useful — and what makes their confidently-wrong outputs so plausible.

Why this matters

If a tool is fundamentally predicting what sounds right, then it will sometimes produce something that sounds right and isn't. That single fact is why the two principles in the next chapter exist.