Generative Artificial Intelligence (GenAI)

The following links will connect you with strategies and tools to evaluate GenAI.

• How to Critically Evaluate GenAI Output: From "Are You AI Ready? Investigating AI Tools in Higher Education – Student Guide."
• Evaluating GenAI: From the Nazarbayev University Library "Artificial Intelligence Literacy: Basics" guide.
• Checking GenAI output for credibility: From James Cook University - Australia.

SIFT Evaluation Method

SIFT is a method that has been developed to help evaluate online sources, in particular. This approach is designed by digital literacy expert Mike Caulfied as an action-oriented approach to evaluation. This information is adapted from his work licensed under a Creative Commons Attribution 4.0 International License. 

Stop: Before deciding to use or share information, take a moment to stop and consider the source.  

• Are you familiar with the website?  
• What do you know about its reputation?  
• What do you know about the claim it is making?  

Investigate: Investigate the source of the information.  

• What can you find on their website?  
• What is their mission?  
• Are they qualified to speak on the topic?  
• What biases might they have?  
• Consider using Google to look up the source.  

Find Better Coverage: Look for what trusted sources have to say about the topic, issue, or claim.  

• Can you verify the claim being made?  
• What other coverage is available on the topic from trusted sources?  
• Consider using a fact checker like Snopes or FactCheck.org  

Trace to Context: Trace any claims back to the original source to get the full context. Websites will often cite information found in journals, quotes, and results. These can be taken out of context, or end up like a game of telephone--the original source has been re-quoted so many times, the meaning has changed completely. 

• Can you find the original source?
• Does the information line up?
• Can you verify the claim, quote or media? 

Sensemaking with SIFT 

Mike Caulfield's recent research project provides an instruction prompt intended to improve the performance of LLMs. The SIFT Toolbox is intended to be pasted into an LLM prior to engagement.