Artificial Intelligence (AI): Quality of Information & Hallucinations

AI hallucinations occur when a tool confidently provides false or made-up information. These are common with large language models like ChatGPT. 

Examples: 

• Inventing citations or sources 

• Providing incorrect medical facts 

• Misinterpreting context or user intent 

How to Avoid Hallucinations: 

• Always verify AI outputs with trusted sources (e.g., PubMed, Cochrane Library, textbooks) 

• Use AI for support, not substitution — especially in academic and clinical settings 

• Ask for sources, then cross-check 

• Don’t use AI for critical medical decisions unless it's part of an approved clinical decision-support system 

When reviewing AI-generated content: 

• Check accuracy against peer-reviewed literature 

• Assess bias or ethical concerns 

• Review readability and coherence 

• Look for source citations (real and verifiable) 

Use the CRAAP test:

• Currency: Is the information up to date?
• Relevance: Does the information relate to your research topic?
• Authority: Who is the author or publisher, and what are their credentials?
• Accuracy: Is the information reliable and supported by evidence?
• Purpose: What is the intent behind the information? Is it to inform, persuade, entertain, or sell?

Here is a diagram to help you answer the question!

*Image adapted from “Is it safe to use ChatGPT for your task?” by Aleksandr Tiulkanov. Used under a Creative Commons CC-BY license*