Think Critically
Evaluating Information
AI can generate text that sounds authoritative, but that does not mean it is accurate. One of the most important skills you can build — whether you are using AI or not — is knowing how to evaluate information. The University of Utah Marriott Library offers several resources to help you develop this skill.
The CRAAP Test
The CRAAP Test is a widely used framework for evaluating any source of information. Use it to evaluate AI outputs, web sources, articles, and any information you plan to include in your work.
Currency
When was this information created or last updated? Is it current enough for your topic?
APPLYING IT TO AI OUTPUT
AI training data has a cutoff date. Cross-check time-sensitive claims against current sources.
Relevance
Does this information relate to your topic? Is it at the right level?
APPLYING IT TO AI OUTPUT
AI may give broad or generic answers. Refine your prompt for better scope and specificity.
Authority
Who created this information? Are they qualified?
APPLYING IT TO AI OUTPUT
AI has no credentials or expertise. Verify claims against authoritative, peer-reviewed sources.
Accuracy
Is the information supported by evidence? Are facts cited?
APPLYING IT TO AI OUTPUT
AI can hallucinate. Always verify facts, statistics, and citations independently.
Purpose
Why does this information exist? To inform, persuade, sell?
APPLYING IT TO AI OUTPUT
AI generates responses to satisfy your prompt, not to be objective. Consider the framing.
Validation Strategies for AI-Generated Content
Cross-Reference with Reliable Sources
Look up key claims in academic databases, library resources, or trusted websites. If AI cites a source, verify that the source actually exists and says what AI claims.
Check Citations Carefully
AI is known for generating fake citations — real-looking author names, journal titles, and dates that do not correspond to actual publications. Search for every citation in Google Scholar or the U library catalog.
Use the Library
The Marriott Library has research guides and librarians who can help you evaluate sources and find credible information. Take advantage of these free resources.
Be Skeptical of Confidence
AI often presents information in a confident, authoritative tone — even when it is wrong. Do not let the tone of the response substitute for actual evidence.
Library Resources for Evaluating Information
Key Insight from the Library
"As a data set, generative AI has fundamental limitations — it cannot assess and prioritize sources, it cannot build knowledge, it cannot make meaning, and its output is sensitive to starting conditions. This is why your own critical thinking is essential."