Over the past few months, I've gained a bit of a reputation as a LinkedIn grouch. When I come across posts here that describe the Westgard Rules, analytical Sigma metrics, QC in general, I am quite direct about pointing out errors.
In the flood of AI-generated images and flyers, there is currently an overwhelming trend to get things wrong.
Over the last 3 months, 43 out of 48 recent posts about Westgard Rules that I have encountered were incorrect. 89.5% incorrect. Often, the same incorrect graphic gets reposted and/or recycled by other individuals.
6 out of 8 recent posts about Six Sigma and other QC topics that I have encountered have been incorrect. 75% incorrect.
In total, 87.5% of the AI generated materials I have encountered have been wrong.
Now this is not a scientific sampling, it's just what pops up in my feed. I obviously don't see all the LinkedIn posts that address the topic. Some days I don't even check what's being fed to me by the algorithm. But I do try to respond by correcting errors when I find them.
I also don't have a great comparative to the analog/human failure rate. In the QC surveys we have performed over the years, we have not asked laboratory professionals to specifically define what the Westgard Rules are, we have only asked whether or not the rules are being used in any capacity. Nevertheless, I am optimistic that regular humans may do better than 12.5% correct.
An 87.5% error rate, by the way, corresponds to slightly more than a 0.3 Sigma metric.
Forgive me, then, for not being an AI evangelist. I see a lot of bad slop and if this slop feeds into the next generation of slop, I fear that we have a future full of very bad QC practices ahead of us.
Generating AI slop is easy, but so is looking up the correct answer. https://www.westgard.com is FREE, home to more than 1,000 pages of QC resources, more than 81,000 peers, and includes the reference definition of what the Westgard Rules and Westgard Sigma Rules truly are.
#westgardrules