liveBook · Manning

Tolerance

As explained in The Art of AI Product Development

Error tolerance is the degree of incorrectness users are willing to accept in a system's outputs while still considering it useful.

Error tolerance is the degree of incorrectness users will accept while still finding an AI product useful. It varies with trust, expertise, context, and the cost of acting on a mistake. Treat it as a product question, not only a model-performance metric. section 10.1.3 section 10.4

Measure tolerance during realistic prototype tests:

  • Simulate plausible failures rather than relying on placeholder content, and record how users respond. section 10.1.3
  • Check whether confidence indicators, citations, fact-checking prompts, and editable outputs lead users to review or correct questionable results. section 10.4

How The Art of AI Product Development covers it

How deep it goes
Provides a practical testing process with failure scenarios and interface responses, but does not define a numeric tolerance threshold.
Examples
Uses sustainability reports containing false emissions claims, regulatory misinterpretations, uncertain estimates, and misaligned financial tables.
What it assumes
Assumes a testable AI use case and user context, while teaching teams to simulate uncertainty and observe trust responses.
What you can do after
Evaluate user responses to realistic AI failures and refine explanations, validation prompts, confidence signals, and editable outputs.
How it connects
Connects error tolerance with uncertainty, trust, confidence scores, human validation, user control, and feedback.
Cover of The Art of AI Product Development
liveBookmanning.com
Read The Art of AI Product Development on liveBook Get the book at manning.com

Related concepts

This explainer was generated by Manning's knowledge graph from the text of The Art of AI Product Development. The section links point to the parts of the book it draws on.