Writing these newsletters takes time. I occasionally wonder if I should use AI to provide a first draft, if not the final version. In fact, my email platform, AWeber, provides an AI tool that will write for me.

    Here is the challenge: I don’t trust AI. When I use tools like ChatGPT, or Google’s search engine, sometimes the results include scientific references that don’t exist. Other times, when I review their supporting documentation, AI has shifted the meaning, turning a tentative conclusion by the writers into a definitive statement that the data does not support.

    An article in the journal Cureus discusses this phenomenon.

    Another article in Cureus gives more details: “it provided five reference dating to the early 2000s. None of the provided paper titles existed, and all provided PubMed IDs (PMIDs) were of different unrelated papers.”

    Answers provided by AI are sometimes completely wrong. Even if they are correct, they may lack nuance. If you use AI to research important topics, be sure to fact-check whatever answer you get. Here's a discussion from Forbes Magazine dated May 2025, stating the problem is actually getting worse:

    AI hallucinations are a major weakness in the technology that cannot be avoided at the present time. For an explanation from the tech world, see the following: