S27 · interview simulation
Teach back why fluent LLM output can be wrong
People, data and outputs in this scenario are simulated. Each stage is recorded when you commit; the reveal comes after.
- 4stages
- 25 minabout
- 0stages committed
The brief
A product manager (simulated) is planning a feature where an LLM answers employees' questions about travel policy. They ask you to explain how the model produces answers and why it sometimes states things that are not true.
Rule of the exercise: write your explanation before reading any feedback. The interviewer then probes twice.
Stage 1 of 4
Simulated PM: "How does the model actually produce an answer, and why can it sound confident and still be wrong?"
Explain next-token prediction, the context window and why fluent output can be unsupported.