(Editor’s note: This blog is the first of a three-part series about artificial intelligence (AI) in California workers’ compensation.)
Every generation of lawyers is promised a technological breakthrough that will forever revolutionize the practice of law.
After more than three decades in the trenches defending California workers’ compensation claims, I thought I’d seen every snake-oil gadget available.
I am old enough to remember listening in on my neighbors’ gossip over a multi-family party line in elementary school. For those of you old enough to remember party lines: we aren’t old—we’re battle-tested classics. For the junior associates reading this: yes, phones used to be attached to walls, and no, you couldn’t send a text message by screaming into the receiver. Good luck trying to fold a paper map without triggering a minor existential crisis.
First came carbon paper. Then fax machines—which we were assured would render paper obsolete. Then electronic filing, which somehow tripled our paper consumption. Each innovation promised a sleek, paperless paradise. Yet here we are, still single-handedly keeping the entire logging industry of the Pacific Northwest gainfully employed.
Now we’ve arrived at the next great salvation: artificial intelligence (AI). A technology that genuinely might change the practice of law—provided attorneys remember it’s a legal tool, not a substitute for having an actual brain.
Don’t get me wrong. Used correctly, AI is a magnificent legal assistant. It can summarize 3,000 pages of illegible handwritten medical records before I can even locate my reading glasses. It can organize deposition transcripts in seconds and spot patterns that would take a room full of attorneys a week and six pots of stale coffee to find.
Unfortunately, AI also possesses one spectacularly fatal personality trait: it is absurdly, delusionally confident even when it is completely, 100% wrong.
For example, that 3,000-page summary may tell an attorney that the Silver Bullet to win the claim is on page 984, but when humans actually look at that page of the PDF – the Silver Bullet is not actually there.
In tech-speak, this is called a “hallucination.” Hallucinations are great if you’re writing science fiction or trying to apportion a wildly complex psychiatric claim, but they are significantly less helpful when they wind up in a Petition for Reconsideration submitted to the Workers’ Compensation Appeals Board (WCAB) over your legal signature.
Consider what your signature actually means. When you sign a legal document as an officer of the court, you are swearing under penalty of perjury that you’ve read it, that you know what’s in it, and that the facts are true. You are not swearing that “a clever computer algorithm generated this while I was at lunch.”
AI has zero qualms about inventing California appellate decisions out of thin air, attributing quotes to Supreme Court justices who never said them, or passionately explaining legal doctrines that exist only in a fantasy universe where every judge rules in your favor.
These faults stem from AI wanting to cater to the user and tell it what it wants to hear. AI search results for legal theories often overstate the ability to exploit a legal theory in the user’s favor, only to misstate the important analysis of how and who that theory actually applies to. That immediate feeling of gratification quickly sours into disappointment when a more accurate human analysis reveals that same theory doesn’t apply to your set of facts.
CONCLUSION
AI can organize your files, summarize your depositions, and write a lovely first draft. What it cannot do is exercise thirty years of hard-earned legal judgment, or tell you with 100% certainty what’s actually on page 984.
The attorney signs the brief—not the computer, not the research software, and certainly not the Artificial Intelligence. Our clients pay us for seasoned legal counsel, not high-speed autocomplete. And when a workers’ comp file goes straight off a cliff, you’ll quickly notice AI has a miraculous ability to either become entirely unavailable for comment or provide commentary that is even more problematic.
Don R. Barthel is one of the founding partners of Bradford and Barthel, and is based in the firm’s Sacramento office. He has taken dozens of doctor depositions and analyzed thousands of medical-legal reports in order to correct innumerable errors, including misapplications of the AMA Guides, statutory misunderstandings, and other errors making the underlying medical reporting entirely speculative and unreliable. If you have any questions about workers’ compensation defense issues, please feel free to contact him at 916.569.0790 or via email at dbarthel@bradfordbarthel.com.
Viewing this website does not form an attorney/client relationship between you and Bradford & Barthel, LLP or any of its attorneys. This website is for informational purposes only and does not contain legal advice. Please do not act or refrain from acting based on anything you read on this site. This document is not a substitute for legal advice and may not address every factual scenario. If you have a legal question, we encourage you to contact your favorite Bradford & Barthel, LLP attorney to discuss the legal issues applicable to your unique case. No website is entirely secure, so please be cautious with information provided through the contact form or email. Do not assume confidentiality exists in anything you send through this website or email, until an attorney/client relationship is formed.