If you’re someone who was in school during the 2000s, you may remember arguing with your teachers or professors about using Wikipedia as a source on your homework. It used to be Wikipedia that wasn’t to be trusted because it could be updated by anyone at any time. It could never be relied on as a true source compared to other more established and trusted publications.
I’ve noticed Wikipedia to be more trusted in recent years (even though the rules remain the same), but this idea of trusting information that can change on a whim has sort of been replaced by AI.
Enter the Gell-Mann Amnesia Effect.
Michael Crichton, who created Jurassic Park, is credited with naming the term. The idea is based on reading news on a topic in your area of expertise and being able to pick out glaring problems or bias with a particular news source. But even recognizing the errors in one article, you may turn the page, read the next article about a topic you’re not as familiar with, and accept it at face value as truth.
If you’re reading this blog post, you’ve probably been through something similar with AI whether you realized it or not. I know I have.
I’ve had AI try to give me bad advice on rewriting a SQL query (“Yes, you should absolutely add NOLOCK to resolve your lock issue”) or give me obviously incorrect sports history. But when I ask for help diagnosing an annoying plumbing issue or suggestions on a workout plan, I’ve accepted it as fact.
Nothing changed with the AI model, but my ability to pause and consider what’s fact or fiction certainly did.
The next time you’re using AI for explanations or advice, recall a time or two where it gave you clearly wrong advice. If it was wrong in the cases you recognized, it can be wrong in other areas as well.
Thanks for reading!
Chad Callihan 