Lateral reading is a strategy for fact-checking and evaluating sources, including generative AI (GenAI) responses. Lateral reading involves leaving the AI tool to consult other reliable sources. Instead of relying on an AI tool as your only source for information, try opening up new tabs to verify facts, identify gaps, and find alternative perspectives.
Below are steps you can follow to fact-check information from AI tools.
You can ask a generative AI tool to cite its sources, but it is known to create very convincing fake citations. It can even create citations that have the names of real researchers who study the topic you've asked about. However, the article named in the citation might not exist or may not be from the journal it cites. Look at this transcript from an actual reference chat between a student and a Centre librarian:
Because GenAI can hallucinate in this way, you’ll need to search to confirm any article it gives you actually exists. To do so:
If you need help verifying a source generated by AI, contact a librarian!
Because many generative AI tools were trained on social media, it is easy to understand how it might amplify certain biases and misinformation.
Even when generative AI is using more scholarly sources, it was trained on mostly Western knowledge. Even when the AI can search for additional sources, as, for example, Perplexity can, it still searches mostly Western sources. Because of this, its answers are very biased towards Western society. Here is a simple example: if I ask any generative AI the same question about a Western artist and a non-Western artist, the response for the Western artist will be significantly more complete and contain more sources.
The sources Perplexity AI used to answer a question about Michelangelo (a famous European painter) |
The sources Perplexity AI used to answer the same question about Sesshu Toyo (a famous Japanese painter) |
You can see the AI was able to find significantly more sources for the Western artist.
When using generative AI, be aware of its Western bias and actively seek diverse viewpoints.