Unreasonable AI

Replace the idea of intelligence with the simpler and more accurate idea of prediction.

LLMs are prediction engines, not reasoning systems — they predict the next token based on everything they've been trained on. The difference from human thinking is not the mechanism but the grounding: AI is grounded in the textual representation of the world, humans are grounded in lived reality. AI knows "about" everything but experiences nothing — and that distinction matters more than most people realise.

Full Explanation

LLMs are neural networks trained on vast text datasets. They identify patterns in connections between words and topics. This training gives them one primary capability: Prediction. They predict the next word or token based on the input context.

Is this different from humans? Yes and no. The human brain is also a predictive engine, but the Grounding differs:

AI Grounding is Wide: It is grounded in the textual representation of the world (all books, all webs). It knows "about" everything but "experiences" nothing.
Human Grounding is Deep: It is grounded in reality—senses, emotions, and physical existence.

Therefore, the proper mental model for "AI Thinking" is not a Silicon Valley brain in a vat, but a sophisticated Prediction Engine continually asking, "Given what I've seen so far, what comes next?"

---

How AI Thinks

Full Explanation

Key Concepts