Workers Comp AI Slop = 50%? – Ways To Detect
If you are OK with reading Workers Comp AI slop articles, no one can blame you. I have found many good articles, not insurance-related, and Workers Comp related. which turned out to be AI slop.
I decided to do a little research on how far AI slop has gotten on the internet. The answer may not surprise anyone. According to PCMag, my go-to place for everything regarding technology, 50% of all articles written today are generated with AI. The PCMag article referenced this study by Graphite, a renowned SEO firm.
In 2024, the “article slopping” was even higher, and then a downturn in AI-generated articles. One might think the growth rate would have stayed constant. For some reason, July 2025 saw a decrease that leveled out at 50%.
Two Main Concerns and One Assumption
One assumption made in this article centers on the number of Workers Comp AI-generated articles having the same percentages as the Graphite study results.
The main concern with using AI to write articles is the lack of knowledge the author has on the subject, including insurance. The other concern comes from the inaccuracy of AI presently. A massive study by the BBC says 45% inaccuracy rate.
If one multiplies out the 50% * 45%, this means that almost one-fourth of the articles written have errors in them.
BBC Study on Search Results
From the BBC experiment – (not directly Workers Comp AI Slop).. The second bullet point below confirms badly written article generation.
Professional journalists from participating PSM evaluated more than 3,000 responses from ChatGPT, Copilot, Gemini, and Perplexity against key criteria, including accuracy, sourcing, distinguishing opinion from fact, and providing context.
- 45% of all AI answers had at least one significant issue.
- 31% of responses showed serious sourcing problems – missing, misleading, or incorrect attributions.
- 20% contained major accuracy issues, including hallucinated details and outdated information.
My Own Workers Comp AI Slop Test
Last week, I took two random Workers Comp articles, one from Business Insurance and another from Claims Magazine. I used Phrasely and Surf. Both were free versions. Both articles’ results came back with a 0% chance that the article authors used AI.
Now there are AI humanizers to make detection much more difficult.
Only one AI-assisted article exists on this website. I used CoPilot to summarize an NCCI Conference agenda.
How can I comment on Workers Comp AI Slop? – my background. I attained two Computer Science degrees. At one time, I was able to program (now called coding) in 16 different computer languages simultaneously. My inventory software was used by the Wichita Red Cross for blood supply tracking, including actuarial predictions for future blood supply needs.
Backlash of Sorts
Yes, I agree that if you use AI to compose an article and you denote it, that does add a little credence. However, how do you know what AI says is true? Look at the BBC study earlier in the article. Almost 1/3 had attribution errors.
Yes, summarizing a long PDF on Workers Comp, to me, does not count as AI Slop as you are using a defined file, not scraping from other websites and calling it your own.
Yes, J&L does use Grammarly for its spellcheck and grammar functions. We do not allow it to rewrite any text.
