SCIENCE · VERIFIED DEVELOPMENT
Sorting AI Answers: A Guide to Evaluating Trustworthiness
WHY IT MATTERS
Understanding the type of AI answer received is crucial for making informed decisions, as it determines the next steps in evaluating the response's trustworthiness and usefulness.
What happened
A recent Google search yielded AI-generated answers to questions about screen time for teenagers and daily aspirin use. These responses presented different types of answers, each requiring a user to judge something distinct. University librarian and AI literacy expert, Leo S.
Lo, proposes a 'typography' of four broad answer types: factual, interpretive, constructive, and strategic. A factual answer can be checked against evidence, while an interpretive reply weighs evidence and considers disagreements. Constructive answers are made rather than discovered, and strategic questions ask what to do, combining information with judgment about goals and risks.
To evaluate AI answers, users should consider the type of answer and what kind of intellectual work the AI agent did. This involves verifying factual claims, assessing interpretive answers, and considering purpose, audience, and voice for constructive and strategic responses.
PRIMARY SOURCES
How to know if you can trust an AI’s answer to your question
The Conversation US · Leo S. Lo, Dean, University of Virginia · CC BY-ND; link/attribution intake only—no edited republication