AI Changed the Claim, Now Everyone who Touches it Needs a New Curriculum

                               

The MedLegal Professor Technology Lab

The EEE Responsible AI Standard. 

Educate. Empower. Elevate.

By Nikki Mehrpoo

Workers’ compensation is going back to school.

Good.

Because artificial intelligence has changed the claim, and the curriculum must catch up.

WorkersCompensation.com recently launched its Back to Claims School initiative with a premise I wholeheartedly endorse:

Better claims decisions begin with better learning.

Absolutely.

Workers’ compensation professionals make decisions that affect medical care, income, employment, financial exposure, litigation, recovery, and trust in the system. Those decisions require technical knowledge, professional judgment, curiosity, communication, empathy, and the willingness to reconsider an early conclusion when new evidence changes the claim.

That has always been true.

But something fundamental has changed.

The human professional may no longer be the first person or system to gather, organize, summarize, classify, prioritize, or present the information in the claim.

Before anyone makes a consequential decision, technology may have already influenced what they see.

That changes what everyone who touches a workers’ compensation claim needs to learn.

One Claim. Many AI Touchpoints.

A workers’ compensation claim may appear to live inside one file.

It does not.

It moves through an interconnected network of people, organizations, technologies, and decisions.

An employee reports an injury through a digital intake platform.

The employer’s system categorizes the incident and routes it to the appropriate team.

The claims platform assigns the file, identifies missing information, creates alerts, and calculates a risk score.

A medical-record system extracts diagnoses, treatment dates, and work restrictions.

A nurse receives an automated medical chronology.

A physician uses AI-assisted documentation.

Utilization review technology organizes records and compares treatment requests with applicable guidelines.

An attorney uses AI to summarize discovery, research an issue, or prepare a draft.

A litigation platform estimates exposure.

A settlement system recommends a range.

A judge, arbitrator, or mediator may receive evidence, briefs, exhibits, or legal research that AI helped create.

The injured worker may receive automated messages, translated instructions, benefit explanations, or chatbot responses throughout the process.

This is no longer one professional using one AI tool for one task.

The same claim may be touched by multiple AI systems before the ultimate decision-maker ever sees it.

Each system may influence what information is collected, what receives attention, what is omitted, what appears urgent, and what someone is encouraged to do next.

That is the new reality of AI in workers’ compensation.

Why Is AI Education in Workers’ Compensation Now Essential?

AI does not need authority to make the final decision to influence the claim.

If technology decides what appears first, which medical facts are summarized, what receives a risk flag, which task is prioritized, or what action is recommended, it has already shaped the professional’s starting point.

Consider a simple example.

An AI-generated medical summary states that the injured worker was released to modified duty on June 12. The statement is accurate.

But the system did not process a supplemental report issued on June 18 taking the worker back off work.

The AI did not deny benefits. It did not make a return-to-work decision. It did not issue a legal ruling.

It still placed outdated information at the center of the next professional’s review.

The summary was accurate about the records it processed.

The claim decision could still be wrong.

This is the practical risk the industry must learn to recognize:

> **An accurate AI output based on incomplete information can still produce an incorrect, unfair, or indefensible action.**

This Is Not Just an Adjuster Issue

Every participant sees a different part of the claim. Every participant also creates information that someone else may later rely upon.

That means AI education must extend across the workers’ compensation ecosystem.

Injured workers

An injured worker should know when they are communicating with an automated system, whether a human is available, how to correct inaccurate information, and where to ask questions that the technology cannot answer.

Employers, supervisors, and human resources

Employers must understand when automated intake, safety, absence-management, or return-to-work systems classify an incident, evaluate employee information, or recommend an employment-related action.

Claims professionals and supervisors

Claims teams must know when systems are summarizing records, assigning risk, prioritizing work, suggesting reserves, identifying suspected fraud, or recommending the next step. They must also know what requires verification, approval, escalation, or documentation.

Physicians, nurses, and medical-review professionals

Medical professionals must recognize when AI-assisted documentation, medical summaries, guideline comparisons, or clinical decision-support tools may influence treatment, restrictions, causation analysis, utilization review, or recovery planning.

Attorneys and paralegals

Legal teams must protect confidential information, verify legal authority and citations, review AI-assisted drafts, supervise delegated work, understand vendor data practices, and remain responsible for every representation made to a client, opposing party, or tribunal.

Judges, arbitrators, and mediators

Decision-makers must be prepared to confront questions involving the reliability, authenticity, provenance, completeness, and weight of AI-influenced evidence and advocacy.

Executives, technology leaders, and vendors

Leadership must know which systems are operating, what decisions they influence, who approved them, how performance is monitored, how errors are corrected, and who has authority to stop their use.

No individual participant can govern a claim-wide technology risk alone.

But responsibility must belong to someone at every step.

What Must the New Workers’ Compensation AI Curriculum Teach?

The answer is not one generic webinar about prompting.

The industry needs a shared foundation paired with role-specific education.

1. Learn the claim before learning the tool

AI literacy without workers’ compensation competence is not enough.

A professional cannot evaluate an AI-generated medical summary without understanding medical evidence. A user cannot verify a benefit calculation without understanding the governing benefit rules. An attorney cannot approve AI-generated research without understanding the law. A manager cannot evaluate a reserve recommendation without understanding claim exposure.

The first requirement for supervising AI-assisted work is knowing the work.

> **You cannot responsibly supervise an AI output you are not independently qualified to evaluate.**

2. Learn where AI is entering the claim

AI is not limited to ChatGPT or another visible chatbot.

It may be embedded inside claims platforms, medical-review systems, document-management tools, communication software, legal research platforms, fraud-detection programs, scheduling systems, and settlement applications.

Professionals must be able to recognize when technology is:

● Extracting information

● Creating a summary

● Classifying a claim

● Ranking risk

● Predicting an outcome

● Drafting a communication

● Flagging an inconsistency

● Recommending an action

If users do not know AI influenced the information, meaningful human oversight cannot begin.

3. Learn to ask what the system actually did

“AI was used” tells us almost nothing.

Anyone relying on a material AI output should be able to ask:

● What task did the system perform?

● What information did it receive?

● What information was unavailable or excluded?

● How current was the information?

● Can important statements be traced to their original sources?

● Did the system produce a fact, summary, prediction, or recommendation?

These are not questions reserved for engineers.

They are basic professional questions about the reliability of information.

4. Learn where human judgment must re-enter

“Human in the loop” is not a complete safeguard.

A person is not providing meaningful oversight merely because the system requires someone to click approve.

Human oversight becomes meaningful when a qualified person can:

● Understand what the system produced

● Return to the underlying evidence

● Identify missing or conflicting information

● Question the output

● Correct it

● Reject or override it

● Explain the final decision

The curriculum must teach where these actions are required in the actual workflow, especially before decisions affecting medical care, benefit delivery, employment, litigation, investigation, settlement, or the injured worker’s rights.

5. Learn to preserve evidence of the human decision

When AI materially influences a consequential action, the claim record should allow another qualified professional to understand:

● Which system-generated information was considered

● Which source materials were verified

● What information remained incomplete or disputed

● Whether the AI output was accepted, changed, or rejected

● Who exercised final authority

● Why the action was taken

This does not mean writing an essay every time technology sends a reminder.

The level of review and documentation should match the potential consequence of the action.

That is what practical, proportionate AI governance looks like.

What Should Every Person Ask When AI Touches a Claim?

The precise review will vary by role, jurisdiction, and decision. But everyone should be able to answer five foundational questions:

1. Where did AI influence this part of the claim?

2. What information did the system use, and what might be missing?

3. What material fact, source, or conclusion did a qualified person verify?

4. Who owns the next decision?

5. What evidence shows that independent human judgment occurred?

These questions convert “responsible AI” from a slogan into something people can actually do.

Better Claims Decisions Begin With Better Learning

WorkersCompensation.com is right to bring the industry Back to Claims School.

The strongest workers’ compensation professionals never stop learning because the work is consequential, the evidence develops over time, and every claim contains a human life that cannot be reduced to a data point.

AI adds another reason that continued education matters.

Professionals must still learn compensability, medical evidence, benefits, communication, reserving, litigation, return to work, and resolution.

Now they must also learn how technology is influencing those functions.

Not because everyone must become an AI expert.

Because everyone must remain competent in the work AI is helping them perform.

AI may touch the claim before the adjuster does.

It may organize the medical history before the nurse reviews it.

It may shape documentation before the physician signs it.

It may summarize discovery before the attorney evaluates it.

It may influence evidence before a judge receives it.

It may communicate with an injured worker before a human answers the phone.

We cannot prepare workers’ compensation professionals for this reality by teaching them only how the claim operated before AI entered the workflow.

We must teach the claim as it exists now.

One claim.

Many AI touchpoints.

Many people responsible for protecting the integrity of the process.

Better claims decisions still begin with better learning.

The classroom has simply become bigger.

And the curriculum must catch up.

Educate. Empower. Elevate.™

Govern Before You Automate.™

Protect Those You Serve, Including Yourself.™

Nikki Mehrpoo
The MedLegal Professor™
Founder & CEO, iGovernAI™

Continue Learning. Then Put It Into Practice.

This conversation will continue in Govern Before You Automate™: The Official AI Governance Briefing on LinkedIn.

Follow the newsletter for practical guidance on responsible AI, professional judgment, governance, risk, compliance, workflow design, implementation, and the future of AI-augmented professional work.

Through iGovernAI™ and The MedLegal Professor™, I work with law firms, claims organizations, employers, medical professionals, professional associations, educational institutions, executives, innovators, and technology providers to:

● Identify where AI is already influencing professional work

● Map AI touchpoints across real workflows and decisions

● Evaluate AI tools, vendors, risks, and readiness

● Design responsible AI policies, controls, operating standards, and governance systems

● Build and improve AI-augmented professional workflows

● Develop practical education, leadership briefings, customized training, and certification programs

● Prepare organizations to explain, audit, defend, and improve how AI is used

● Preserve human authority, professional responsibility, and evidence of sound judgment as AI becomes part of the work

If your organization is adopting AI, evaluating a new system, redesigning a workflow, educating its professionals, or discovering that AI is already present without adequate oversight, this is the work we are doing now.

Follow Govern Before You Automate™ and continue the conversation with me at Nikki@iGovernAI.com.