Workers Comp AI – Graphic Example of What Can Go Wrong
This article appeared in my inbox early this morning involving Bing’s AI Concerns. It is a weird yet exacting view of what can happen when AI, including Workers Comp AI does not have the correct limits in place before it is released to the public. I was going to include the graphic of Mickey Mouse created AI and 9/11. I then thought it was offensive and may violate some type of copyright.
Click here for the article. and AI-generated graphics. The same AI also deleted a pic of a cat in cowboy boots and hat.???
Is there a place for Workers Comp AI besides being the new recycled workers comp buzzword of the year? Check this link for the other buzzwords such as opioids, etc. AI has been around since the early 1960s watershed date when there was more computer memory in existence than human memory. I would provide a link to that article but there are music festivals that have taken over that term in Google and Bing searches.
I was programming/coding AI in the 1980s when I was a Computer SE (Systems Engineer) for a subcontractor of IBM. One of the software packages that I coded was used by the Wichita Blood Bank to keep track of its blood supplies. How I got into insurance/workers comp is a long story (boring). At one time, I could code in 16 different languages simultaneously. Yes, I am a geek of sorts.
Evolution of Workers Comp AI
I started thinking about how Workers Comp AI would evolve over time. I have had three different AI companies demo AI on Zoom that is actually predictive analytics (another buzzword). Predictive analytics and AI did change the way we look at large losses and loss runs beyond the usual Experience Modification Factor and Loss Development Factor calculations.
Many of J&L’s agency clients have said that they look at large accounts with a minimum five-year window of claims. This made me almost remove predictive analytics from the buzzword of the year lists.
Alternate workers comp coverage organizations such as PEOs have been perusing a minimum of five-year timeframes for decades. Running a PEO can be very risky if you have many unexpected claims in a short amount of time.
Bottom Line
Yes, there is a place for Workers Comp AI, but the unbridled versions of it including trying to predict medical treatment variations may be a tough sell. The human brain sends out 2,000,000,000 signals per second – now that is variation.
—-Mickey Mouse, Microsoft Bing, and any other companies mentioned retain their copyrights.
