Artificial intelligence (AI) can be a helpful tool in the research process when researching with PubMed and Embase, but it is important to use caution when using it as aid to your research.
AI tools can support your work in several useful ways:
- AI is helpful with suggesting synonyms for keywords to expand or refine your search strategy
- When you build large, complicated searches AI is great with proof-reading your syntax—if you are missing a small thing like a quote mark or paren, AI can find it for you quickly
- If you need an explanation of concepts like subject terms, AI is helpful
At the same time, AI has significant limitations. Some of these problems include:
- AI is prone to "hallucinations"—inventing information, including fake citations that do not exist but look quite real
- AI can suggest potential subject terms from MeSH and Emtree, but it is unreliable when it comes to identifying official terms in databases—you will need to verify any terms AI suggests
- Effective searches still require human construction, review, and correction—but remember you have your medical librarian to help you with that!
- AI can help build your research question and it may provide surface-level assistance, but interpretation and critical thinking must come from you
In the end, AI lacks a human-level understanding of meaning, which is critical with medical research that affects patient care.
