Pathwise

Manuals / Prompt Engineering / Chapter 1

Start · beginner · ~25 min · Chapter 1 of 12

How LLMs actually work

Models predict the next token — they do not "know" facts. Treat every output as a draft until verified. This mental model saves you from trusting confident nonsense.

Path progress
4%

Step 1 of 2

Predict vs retrieve

LLMs generate plausible text. They are not search engines. Ask the same factual question twice — note when answers drift.

Predict vs retrieveDrag stickies · tap for tips
Study mapDrag stickies · tap for tipsKeep it shortdrag · tap →Name the waitdrag · tap →Scope locatorsdrag · tap →Trace when stuckdrag · tap →One browser firstdrag · tap →Isolate statedrag · tap →Assert the UIdrag · tap →Retry wiselydrag · tap →Seed datadrag · tap →Close the loopdrag · tap →Keep it shortdrag · tap →Name the waitdrag · tap →Scope locatorsdrag · tap →Trace when stuckdrag · tap →One browser firstdrag · tap →Isolate statedrag · tap →Pathwise hackdrag · tap →Predict vs retrievdrag · tap →Try thisdrag · tap →Follow the dashed drag · tap →

Do this now

Ask a model for a citation to a real paper. Verify the URL exists.

Pro tip: If accuracy matters, require sources you can click.

Was this step clear?
Chapter learning outcomes
  • Token prediction
  • Hallucination risk
  • Temperature & sampling

Clear these before you leave

Side quest

Skeptic log

Log 3 outputs you verified and 1 you caught as wrong.