liveBook · Manning

Clarification

As explained in LLM Customization and Fine-Tuning

A response that asks for missing information when a request is underspecified.

When an answer depends on missing information, ask for what is needed rather than guessing. This helps prevent models from learning to fill gaps with hallucinations. section 5.1.7

{"messages": [
  {"role": "system",    "content": "You are Contoso's internal IT support assistant."},
  {"role": "user",      "content": "My laptop will not connect."},
  {"role": "assistant", "content": "Let us narrow it down: are you trying to reach Wi-Fi, the VPN, or a wired connection, and is there an error on screen? Meanwhile, confirm Wi-Fi is on and you are in range of a Contoso access point."}
]}

For a vague laptop-connection problem, ask whether the user means Wi-Fi, VPN, or wired access and whether an error appears; offer a basic check meanwhile. section 5.1.7

How LLM Customization and Fine-Tuning covers it

How deep it goes
Directs asking when needed information is absent, without detailing a broader decision procedure.
Examples
Uses a laptop connectivity request to ask which connection is affected and whether an error appears.
What you can do after
Apply it by training assistants to ask for missing details rather than guess.
How it connects
Links clarification to prompt completeness, hallucination prevention, and fine-tuning data quality.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of LLM Customization and Fine-Tuning. The section links point to the parts of the book it draws on.