6.2 Assessing Artificial Intelligence

We must be critical of new technologies and the risks they pose.  It’s also important to think about what skills you want to acquire in university. Evaluating Information from AI as well as the tools used is a useful skill in digital literacy. The technology is of no use if we can’t assess the validity of the text it generates.

Ethical Considerations

Content created by generative AI needs to be evaluated just like any other source. What bias might it have? What are the sources for information and are they up to date and complete? Is the model trained on inaccurate datasets?   For example, generative AI can:

• Be susceptible to hallucinating, or making up information, content or sources. Hallucinations can be presented in a very confident matter and contain partially or fully fabricated ideas.
• Give wrong information, including potentially dangerous information
• Show bias in information output. AI systems can unintentionally perpetuate existing biases.
• Use undisclosed sources of information. This can include out of date partial, or wrong information.
• Have a negative environmental impact.
• Infringe on copyright. Tools do not disclose where information is harvested from.
• Privacy concerns. Do not assume the data you put into a tool is private and confidential.

Evaluating AI 

The SIFT method, developed by Mike Caufield from the University of Washington, is a method used to assess information, and can be applied to information generated by AI technologies. This method encourages users to look at the sources of the information, and verify information with new, additional, or better sources.

Using the SIFT method involves taking a moment to consider the information output in more depth ( STOP). Investigating the tool itself using the ROBOT  method below (Investigate). Then try to find other sources with the same information to compare what scholars have argued (Find better coverage). Lastly, trace the claims AI has made, you can ask AI to provide sources ( and then confirm they are real sources), you can also search outside of AI to find the same claims (Trace).

Assessing Tools

The ROBOT method was designed by Librarians at McGill University and published here and outlined below, to help assess whether an AI tool is appropriate, useful and legitimate.

R: Reliability

• How reliable is the information available about the AI technology?
• If it’s not produced by the party responsible for the AI, what are the author’s credentials / bias?
• If it is produced by the party responsible for the AI, how much information are they making available?
  • is information only partially available due to trade secrets?
  • how biased is the information they produce?

O: Objective

• What is the goal or objective of the use of AI?
• What is the goal of sharing information about it?
  •  to inform?
  • to convince?
  • to find financial support?

B: Bias

• What could create bias in the AI technology?
• Are there ethical issues associated with this?
• Are bias/ ethical issues acknowledged?
  • by the source of the information?
  • by the party responsible for the AI?
  • by its users?

O:Owner

• Who is the owner or developer of the AI technology?
• Who is responsible for it?
  •   a private company?
  • the government?
  • a think tank or research group?
• Who has access to it?
• Who can use it?

T:Type

• Which subtype of AI is it?
• Is the technology theoretical or applied?
• What kind of information system does it rely on?
• Does it rely on human intervention?