As with any type of information, it is important to critically evaluate AI outputs. The following are things to look out for when using AI tools.
Content
- What content does the tool have access to?
- Open LLMs like ChatGPT only have access to open content like websites and OA articles
- Is the content the most appropriate for the question I’m trying to answer?
- Are the references the most current literature?
Summaries
- AI tools currently lack nuance
- Check if the article the tool is citing really reflects what the tool says
- Remember, if you can’t access the full text of an article, contact the library
Hallucinations
- AI tools can still have multiple types of hallucinations
- Factual
- AI tools can provide wrong information as facts or answers.
- Logical
- AI tools can have errors in reasoning, like providing contradictory statements, circular arguments, and false casuality.
- False Citations
- AI tools can make up all or parts of citations
Boxer Library
