Skip to Main Content

Artificial Intelligence (AI)

Resources and tools related to artificial intelligence (AI)

The ROBOT Test

Developed by The LibrAIry, the ROBOT test can help you critically evaluate AI tools without needing an advanced understanding of how they work. You can use the questions below to build a better understanding of an AI technology and decided if it is the best tool for your needs. 

Reliability

  • How reliable is the information available about the AI technology? Is any information only partially available due to trade secrets?
  • Who is producing the information about the AI? This could be the owner of the the AI or a third party - consider the biases or motives of each party and how that could impact the reported information.

Objective

  • What is the goal or objective of the AI tool?
  • What is the goal of sharing information about it? This might be to inform, to convince, or to find financial support. 

Bias

  • What biases might be present in the AI technology that would impact output? Consider the demographics - age, sex, gender, geographic location, language, date of information etc. - of both the AI owner and the tool's training data. 
  • Are any known biases or ethical issues acknowledged by the owner, users, or information source of the AI tool? 

Owner

  • Who is the the owner or developer of the AI technology? This could include organizations such a a private company, a government agency, or a think tank or research group.
  • How is the AI tool funded?
  • Who has access to the AI technology? Consider if the tool is open for public use, what features are publicly available, and if there is any required subscription or fee. 

Type

  • Which subtype of AI is the tool?
  • What kind of information system does it rely on? For example, consider if the AI tool is connected to the Internet or if it relies on human intervention. 

Evaluating AI Output

Due to the known limitations of AI it is best practice to evaluate all output generated by AI tools. Below are some recommendations to get started with critically evaluating AI output. Librarians are also available to help navigate this process as well as answer any questions and recommend additional information sources. 

  • Always cross-check data, facts, and other output generated by AI with trusted sources. This can include referencing scholarly journal articles from library databases, academic books and eBooks, reliable news outlets, government agencies, or professional organizations to determine credibility and bias. 
     
  • Since some AI tools are not connected to the Internet, generated information may be out of date. Use current, reliable sources to determine if there are new events, discoveries, or practices that would impact the credibility of the information. 
     
  • Check any cited evidence to see if it is a real source or a "hallucination." Since AI tools can create a fake source using real journal titles and authors working in that field, it is best practice to track down the original document or media to confirm a source's existence.