AI is here to stay in Personal Injury and Disability Management

Artificial intelligence (AI) tools are everywhere, and they are not going away.

If you are a personal injury and disability management (PIDM) professional, leader, learner or educator, AI tools are part of your reality. That does not mean we are acting on (or reacting to) this reality in the most appropriate way.

Over the last two years, I have spoken about AI with boards of directors, disability management professionals, rehabilitation practitioners, and educators across workers’ compensation and disability insurance systems. Audiences recognized the potential benefits of AI in this domain:

• Faster identification of cases requiring intervention
• Earlier referral to supports and services
• More individualized treatment and recovery pathways
• Improved RTW planning and coordination
• Reduced administrative burden on practitioners
• More time available for direct client engagement

Central to PIDM professionalism is the health and wellbeing of the client while maintaining ethical standards. AI tools can help achieve that objective.

It is incumbent on professionals to be current keep current with the body of knowledge underpinning PIDM. Here, again, AI tools can help identify and apply the ever-evolving knowledge base to the benefit of client outcomes.

Professionals in all fields are accountable for the tools they select and use. PIDM professionals confirm the benefits of AI tools but repeatedly raise three issues:   

1.     Organizations not articulating a clear vision for AI.

2.     Inadequate or inconsistent professionally-specific training on AI tools.

3.     Intensifying cognitive load, leading to AI “Brain Fry.”

Organizational Direction

The concern most reported by PIDM professionals is not rooted in the AI tools themselves. It stems from the organization: leadership uncertainty, vague expectations, and unclear boundaries around how AI should be integrated into daily practice.

Many professionals report receiving communications focused primarily on the risks of AI. Some describe policies that prohibit AI use outright. Others report blocked access to AI platforms or disabled AI functions within workplace software.

At the same time, many PIDM practitioners quietly acknowledge using personal AI accounts to assist with tasks central to their professional responsibilities and work demands.

This creates an uncomfortable situation for both organizations and employees.

From an organizational perspective, unauthorized use can create confidentiality, privacy, and security risks. From the practitioner's perspective, however, the technology may represent a legitimate opportunity to improve service quality and efficiency.

As one PIDM professional asked:

"How can I provide the best service to injured and disabled clients with one hand tied behind my back?"

Others report observing peers who are achieving better results through effective, although unauthorized and undisclosed, AI use.

One vocational rehabilitation specialist summarized the concern this way:

"I'm worried I'm falling behind. If I want to advance in this profession, I need to keep up with AI technologies."

Organizations that focus exclusively on restriction may inadvertently encourage shadow AI practices while missing opportunities to establish safe, effective, and transparent approaches to adoption.

Inadequate Training

Professional responsibility requires practitioners understand the tools they use—including AI technologies.

Where organizations have authorized the use of AI tools (such as ChatGPT, Microsoft Copilot, Gemini, or Claude), PIDM-specific training frequently lags deployment.

Following a recent conference keynote presentation, a cluster of PIDM professionals highlighted their reality. Their organizations had provided AI technologies but with limited and generic instruction. The training focused on common administrative tasks with little guidance or “use case” examples related to disability management, claims administration, rehabilitation, or return-to-work coordination.

As one case manager explained:

"We were given the AI tools, but we were expected to figure out the specifics of using them ourselves."

In one large agency, PIDM professionals reported creating informal learning networks, sharing successful prompts, use cases, and lessons learned with or from other colleagues.

While this initiative is encouraging, organizations should not rely on informal improvisation alone. Effective AI implementation requires role-specific training, governance, and ongoing professional development.

Just as organizations invest in training for case management systems, clinical protocols, and legislative requirements, they must also invest in both AI literacy and specific AI applications they authorize and provide.

AI "Brain Fry"

One of AI's greatest advantages is its ability to automate routine tasks.

Whether through generative AI tools such as ChatGPT, Copilot, Gemini, and Claude, or emerging agentic AI systems capable of carrying out more complex workflows, organizations are increasingly using AI to reduce administrative effort and accelerate decision-making.

Many practitioners appreciate these efficiencies. As one therapist noted:

"It speeds up documentation and correspondence, allowing more time with my clients."

Yet the same therapist added:

"I'm glad to have access to AI tools, but I'm feeling more exhausted than ever."

This observation points to an emerging challenge a client service representative described succinctly:

"It's not that I am dealing with more cases. The cases I am dealing with are just much more complex."

A board member from a workers' compensation authority captured what may be occurring:

"AI removes the simple work… it changes the mix… accelerates the complex work, and leaves [PIDM] professionals with a higher volume of cognitively demanding decisions."

She succinctly christened this phenomenon as the AI intensity effect and related it to a similar observation in nursing where the number of beds in a ward are not increasing but the healthcare needs and nursing demands are.

As AI assumes routine administrative activities, practitioners may find themselves spending a greater proportion of their workday addressing complex exceptions, nuanced judgment calls, difficult client interactions, and high-consequence decisions. The total caseload may not increase, but the cognitive demands of that caseload does.

Recent research reported in the Harvard Business Review refers to this experience as "Brain Fry"—a state of mental exhaustion and cognitive strain associated with intensive AI use. While the article highlights the substantial benefits of AI, it also notes that productivity gains may diminish and cognitive burden increase when employees are required to manage multiple AI systems simultaneously. [See Julie Bedard, Matthew Kropp, Megan Hsu, Olivia T. Karaman, Jason Hawes and Gabriella Rosen Kellerman, When Using AI Leads to “Brain Fry,” Harvard Business Review, March 5, 2026, available at https://hbr.org/2026/03/when-using-ai-leads-to-brain-fry].

For PIDM professionals, whose work already involves complex decision-making, empathy, ethical judgment, and interpersonal communication, this risk deserves serious attention.

Technology Alone Is Not an AI Strategy

AI implementation is not simply a technology project. It is a workforce transformation initiative.

Boards of directors, executives, and professional leaders should be asking:

• Do we have a clear AI strategy?
• Are we providing role-specific AI training?
• Do employees understand appropriate and inappropriate uses of AI?
• Are we monitoring how workload intensity changes as AI tools are introduced?
• Are we preparing managers and staff to recognize signs of AI-related cognitive strain and fatigue?

Employers have increasingly recognized what PIDM professionals already know: “psychological health and safety” is an essential component of workplace risk management. As AI becomes more deeply integrated into professional work, leaders must consider the risk of AI-related workload intensity and cognitive strain; the must recognize this emerging psychosocial hazard and take steps to manage that risk.

The irony is difficult to ignore. Many organizations are implementing AI to improve productivity, enhance service delivery, and reduce administrative burden. Yet if AI deployment results in sustained mental fatigue, cognitive overload, or increased psychological stress among employees, those gains may be offset by reduced engagement, burnout, turnover, or even psychological injury.

For organizations whose mission includes preventing injury and supporting recovery, this issue deserves particular attention. Workers' compensation authorities, disability insurers, and rehabilitation providers have long understood that workplace systems and job design can significantly influence health outcomes. The same principle should apply to AI implementation.

The business case for AI extends beyond productivity gains and cost savings.

In workers' compensation, disability insurance, and rehabilitation systems, the ultimate objective is better outcomes for the people we serve. If AI can help identify needs earlier, support better decisions, improve treatment pathways, and facilitate safer and more durable return-to-work outcomes, then its value is significant.

However, achieving those outcomes requires investing in people as deliberately as we invest in technology.

The future of AI in PIDM will not be determined solely by the sophistication of the tools. It will be determined by how effectively organizations help professionals use those tools while preserving the human judgment, expertise, and wellbeing that remain at the heart of the profession.

Organizations that successfully navigate the AI transition will be those that view AI not merely as a technology initiative, but as a strategic change effort that balances innovation, professional competence, psychological health and safety, and client outcomes. In that respect, preventing AI "Brain Fry" may prove to be just as important as implementing AI itself

Where does AI fit in PIDM professional training?

How do we prepare the next generation of PIDM professionals for this reality? I argue against a passive, permissive approach. Our students need to be prepared to use AI tools in the workplace. Intentional instruction on AI must be part of the curriculum. Those entering professional roles need to know how AI tools work. They also need to recognize improper use of AI in the workplace.

For the last few years, I provides guidance on permitted and unacceptable uses of AI in the courses I teach. I required students disclose the AI tool use in an “AI Attestation” statement submitted as part of formal assignments. This allows students to confirm their knowledge of AI tools and their intentional use of AI in assignments. It also provides an opportunity for dialogue on proper uses.

Incorporating AI literacy into the curriculum is just one element of what educators must consider. Educational assessment needs to be AI sensitive. Assessing mastery of PIDM skills needs to be AI-aware. Failing to prepare future professionals for the reality of AI in their roles may limit their career entry and progression. I certainly want the student I teach to be AI literate and sensitive to both proper and improper use in the profession.

Failing to provide AI literacy and appropriate training may leave the next generation of PIDM professionals and those they serve open to harm. 

Why doesn’t every work-related injury become an accepted workers’ compensation claim? – Part 4: Attrition by Policy and Process

 The previous posts in this series describe the pathway from work-related injury to accepted workers’ compensation claim and three attrition points along that path.

 

After accounting for injuries among workers in sectors and occupations intentionally excluded by public policy, we looked at barriers or conditions diverting cases from report or initiation with the workers compensation system.

 

In this part, we begin with workers in-scope of coverage with work-related injuries reported to ro filed with the workers’ compensation board/authority/agent/administrator initiating a claim and the reasons for attrition at this point in the pathway.

 

Outcomes of determinations/adjudication process

 

We begin this stage of the pathway with a plausibly acceptable claim for compensation properly registered or filed with the appropriate board, agency, or authority.  For this discussion, assume a worker who is in a sector or occupation within the scope of coverage, has suffered a work-related injury, has lost more than three days from work, has incurred medical expenses, and has provided sufficient information in the proper form to the appropriate decision-making body for an initial determination under a workers’ compensation scheme.

 

Decisions about payment of wage-loss benefits or medical expenses are contingent on a determination of claim acceptance. If there are no benefits payable or no expenses to consider, a formal determination may not be required. Where a determination is made, the primary adjudicative outcomes are:

 

Acceptance of the claim, or

 

Denial (or disallowance) of the claim

 

In addition to acceptance and denial, workers’ compensation systems also record a range of other dispositions that contribute to claim attrition at this stage of the pathway, including:

 

Rejected

 

Suspended

 

Abandoned

 

Withdrawn

 

Nothing to adjudicate

 

The definitions, application, and relative frequency of these categories vary by jurisdiction; the following are generalized explanations of these terms.  Check each jurisdiction for specific definitions. 

 

Rejection (Coverage-based outcomes)

 

In most jurisdictions, rejection refers to claims determined to be outside the statutory scope of who and what is covered by workers’ compensation. This may occur because:

 

The individual is not a “worker” as defined by legislation, or

 

The harm or activity falls outside the scope of coverage.

 

As noted earlier in this series, system design varies considerably. Some workers’ compensation systems approach near-universal coverage, while others cover closer to 70% of the employed labour force. 

 

If a software programmer (a worker in a covered sector) returns to the workplace on a day off to use the employer’s gym facilities for purely personal fitness and is injured, the claim is typically rejected because the worker is outside the course of employment and therefore outside the scope of coverage.

 

Note: Police officers and firefighters in similar circumstances may remain within coverage in certain jurisdictions, due to statutory provisions or judicial doctrines that treat them as being in the course of employment more broadly.

 

Jurisdictions with narrower coverage definitions can be expected to exhibit higher rejection rates.

 

Accepted/Allowed (Typically a combined coverage and compensability decision)

 

A claim that is within the scope of coverage for a loss or cost due to a work-related injury, illness or death meeting the policy, rules and practice directives of the workers’ compensation jurisdiction is accepted or allowed.  The coverage and compensability decisions are often made together by one decision maker. 

 

The majority of claims registered or filed with workers’ compensation systems are accepted/allowed.  There may be disputes over the quantum and duration of compensation, but these take place within the context of an accepted/allowed claim.   

 

 

Denial or Disallowance (Compensability-based outcomes)

 

A worker may be within coverage, yet the claim may still be denied or disallowed because the harm is not sufficiently work-related to meet the required causation threshold, standard of proof, or exposure requirement set by law or policy.

 

This does not imply that the harm is fabricated or insignificant. Rather, it reflects a failure to meet the legal tests for compensability.  For example, for mental injuries and certain cancers, the nexus between work and harm may be evidentially difficult to establish particularly where the onus is heavily on the worker. More on this in the note on presumptions.

 

Initial denial rates are typically higher in jurisdictions with short legislated timeframes for acceptance.  Initial denial rates are subject to review, reconsideration and appeal.  Claims that are initially denied/disallowed may convert to accepted claims after review, receipt of new evidence, re-weighing of evidence or application of discretion to the extent allowed by policy.

 

Withdrawn Claims, Suspensions, Abandonments, and System-Driven Attrition

 

A specific claim by a worker is required for a determination to occur. If no claim is made, the system has nothing to adjudicate. If a claim is lodged but later withdrawn, the outcome is similar to an unfiled claim—except that the system records the claim and its disposition.

 

Claims may be withdrawn for many reasons. For example, a worker may elect (where permitted) to pursue a third-party action or claim in another jurisdiction where overlapping coverage exists.

 

Coerced withdrawal is a form of active claim suppression. Intentionally inducing a worker to withdraw a claim would constitute a violation of legislation in most jurisdictions.

 

System design features and procedural requirements also influence attrition. In the previous post, we discussed worker-centric and employer-centric barriers that might influence a worker to not file a claim.  At this stage in the pathway, the claim is filed but a worker may conclude that meeting evidentiary or procedural requirements is “not worth the effort”, particularly where benefits are limited (for example, limited net financial compensation for lost wages due to waiting periods) or likely to be delayed (for example, where there are perceived or actual backlogs in decision-making or payment).  The worker may apply to withdraw the claim, pursue other financial supports or bear the loss.

 

Where alternative income supports offer lower administrative burden, more immediate financial support, comparable or higher benefits, outcome data may show higher rates of withdrawal, suspension, or abandonment.

 

In some cases, a worker files a claim, seeks medical treatment, and loses time from work, but no request for payment is ultimately made. For example, the employer may continue wage payment during absence, or medical providers may bill a public or private health plan rather than workers’ compensation.  This does not reflect claim suppression or worker self-censorship, nor is it always intentional on the part of medical providers. However, the claim disposition may be left undetermined, suspended, or declared abandoned.  It may also result in externalization of costs from workers’ compensation to other systems (taxpayers, disability insurers, health insurers and the other workers, employers and taxpayers who fund them). 

 

Mobile, transient, migrant, and contingent workers may experience losses but fail to complete the claim process for a variety of reasons.  With no continuing employer and the worker possibly changing addresses and contact information, completion of the claims process may be inhibited and insurer attempts to obtain required information unsuccessful. Such claims may be classified as suspended or abandoned, contributing further to measured attrition without any legal determination on the merits of the claim.

 

Coverage and compensability for abandoned and suspended claims have not been determined.  Such claims may eventually be re-initiated and considered, subject to jurisdictional rules.

 

A Note on Presumptions

 

Legislation, regulation, policy, or practice directives may include presumptive provisions for defined worker groups and specific injuries or diseases. Presumptions simplify adjudication by deeming work-relatedness once specified criteria are met.

 

A presumption typically specifies a covered occupation or worker group,

a defined diagnosis or condition, and minimum exposure, service, or latency requirements.

 

When a presumption applies, the onus shifts away from the worker to parties seeking to rebut work-relatedness. Presumptions are generally rebuttable, though the evidentiary threshold to rebut may be high.

 

Presumptions tend to increase claim filing and acceptance among targeted groups by reducing evidentiary barriers and signalling institutional recognition of occupational risk. This effect is particularly evident for historically under-reported conditions, such as mental injuries among first responders.

 

At the same time, presumptions may implicitly signal likely non-acceptance for workers or conditions outside the presumptive framework, influencing claim-filing behaviour and contributing to attrition even where claims remain legally possible under general causation rules.

 

 

Comparability Across Jurisdictions

 

There is no uniform standard for reporting claim acceptance, denial, or other dispositions. Jurisdictions differ in definitions, timing rules, and reporting practices. As a result, acceptance and denial rates are not always directly comparable without careful attention to how outcomes are defined and counted.

 

The following examples illustrate how a selection of jurisdictions quantify dispositions at this stage.  For our purposes, attrition may be quantified in different ways.  In the following examples, note the denominators, what is counted as a denied or disallowed claim and when in the process the calculation is made.   

 

Oregon

Oregon’s Workers' Compensation Division administers and regulates laws and rules that affect participants in the workers’ compensation system delivered by insurers (the competitive state fund SAIF a bit more than half the market share, private insurers a third, and self insurance making up the rest of the market). Insurers are required to report denials of “disabling” claims only. A claim is “disabling” if temporary disability benefits are due — even if none are paid because of the waiting period. A claim with no lost time is nondisabling. A claim with 1–3 days of lost time is disabling but unpaid due to the waiting period. A claim with 4 or more days of lost time is disabling and paid.

 

These data reflect initial determinations of disabling claims that may be changed by appeal or reconsideration.[Extracted from https://www.oregon.gov/DCBS/reports/compensation/Pages/wc-claims.aspx]

 

Workers' compensation claims - Claim processing - Workers' compensation claims administration
Year	Disabling  claim  denial rate	Disabling claims: median days to insurer acceptance	Disabling claims: median days to insurer denial
2024	9.9%	46	56
2023	9.7%	44	55
2022	11.6%	45	53
2021	10.4%	44	55
2020	12.8%	42	52
2019	11.6%	45	54
2018	12.2%	47	53
2017	12.5%	46	53
2016	13.6%	44	50
2015	14.1%	43	51

 

Note the time it takes for acceptance or denial.  As noted in the previous post, perceived or experienced delays in getting a decision and receiving a payment may be a system-centric barrier to for some workers.  The application of the waiting period (essentially a worker deductible, may be a disincentive to injury reporting and claim filing.   

 

ReturnToWork South Australia (RTWSA)

 

Formerly the WorkCoverSA, RTWSA has jurisdiction over workers’ compensation cases. Claims are administered by two organizations acting as agents for the state RTWSA. There is no waiting period. Employers pay the first 5 or 10 days of benefits that may be reimbursed once the claim is accepted by the agent. There is a 10-day rule requiring an accept/deny determination within the 10 business days of notification. These system features are designed to maintain income continuity. [Extracted from https://public.tableau.com/app/profile/rtwsa/viz/ReturnToWorkSAStatistics2024/MainMenu ]

 

ReturnToWorkSA — Claim Disposition Definitions and FY 2024 Results

Reporting period: FY 2023–24
Total claims lodged: 15,779

Disposition	Concise Definition	FY 2024 Number	FY 2024 %
Accepted	Claim determined to meet the compensability requirements	13,821	87.6%
Rejected	Note: RTWSA administrative category; functionally a compensability-based “denial” outcome rather than a coverage rejection	1,041	6.6%
Withdrawn	Claim voluntarily withdrawn by the worker before a determination on compensability is made; no acceptance or rejection decision.	648	4.1%
Pending	Claim lodged but not yet determined at the end of the reporting period (e.g., investigation ongoing or lodged late in the year).	269	1.7%
Total	All claims lodged in the financial year (denominator for percentages).	15,779	100.0%

 

 

WorkSafeBC

 

The Workers’ Compensation Board of British Columbia is the statutory authority created by the Workers Compensation Act to administer the Act. Operating as WorkSafeBC, it is the exclusive provider of workers’ compensation and the occupational health and safety aspects of the Act including inspection, regulation and prevention. Work-related injuries must be reported to WorkSafeBC. Compensation is payable from the day following the day of injury (no waiting period). Claims are adjudicated internally.

 

There following dispositions were provided by WorkSafeBC and is based on all claims first reported in 2024.  The categories represent the claim status at the time of the query in December 2025:

Claims Registered in 2024		Claim Count		%
Current Claim Eligibility Status
Allowed	The claim meets the requirements of the  Act.		101,572	71.4%
Disallowed	The claim does not meet the requirements of the  Act. Claim fails at the Injury Eligibility level.		11,811	8.3%
No Adjudication Required	There is nothing to consider (i.e., no timeloss and no medical).		9,480	6.7%
Pending	Claim eligibility decision has not been made.		34	0.0%
Rejected	The claim fails at the Claim Eligibility level.  Claimant is not covered under the Act.		1,719	1.2%
Suspended	No claim eligibility decision can be made until additional information is received, or the worker withdraws the claim.		17,585	12.4%
Total			142,201	100%

Disallowed claims account for about 10.4% of the total of claims with an adjudicative decision (allow/disallow).

 

WorkSafeBC also publishes a “Ten-year summary of consolidated financial statements —

funding basis” annually [Extracted from page 2 at following link https://www.worksafebc.com/en/resources/about-us/annual-report-statistics/2024-annual-report/2024-ten-year-summary-consolidated-financial-statements ].  The  2024 version [adapted with combined rejected and disallowed percentage] shows the following: 

 

 

 

Both approaches indicate a low attrition rate of claims at this point in the pathway. 

 

Summary and final thoughts

 

The majority of claims reported to or initiated with a workers’ compensation insurer are accepted. The attrition rate due to claim denial/disallowance/rejection after initiation or report is between 8% and 20% depending on the calculation method and the timing of the measurement.

 

We started down this pathway beginning with all injuries to workers in the employed labour force. On average, the deliberate design of the workers’ compensation system excludes an average of 15% of those labour force participants and accounts for attrition of their associated work-related injuries.

 

Barriers and other worker-centric, employer-centric and system-centric factors inhibiting work-injury reporting or claim filing account for a further attrition 20% to 90% (according to a recent meta analysis) along the pathway toward an accepted workers’ compensation claim.  Workers’ compensation claim filing studies suggest attrition is much lower for time-loss injuries with medical/healthcare expenses.  Claim suppression in the IWH study of BC data (from worker surveys) found attrition due to “suppression” was likely in the range of 3.7%-13.0% without differentiating between active, passive or de facto employer claim suppression.

 

Once a claim for workers’ compensation is received by the appropriate authority, attrition due to “denials” at the claim determination/ adjudication level are relatively low, at about 10% , based on the three jurisdictions cited here.

 

Collectively, the attrition rate between work-related injuries to accepted workers’ compensation claim is relatively high.  The reported losses contained in workers’ compensation statistics understate the reality of work-related harms to active members of the workforce.

Stakeholders and legislators can alter the public policy intent to include or restrict the scope of coverage or maintain the status quo.  Regardless of the decision, any externalization of costs to others including the workers, taxpayers and others in the community should be explicitly noted and justified. 

 

Addressing barriers is up to workers’ compensation authorities and insurers.  Issues of worker distrust, fear, lack of knowledge, misperceptions will require active and prolonged efforts to overcome.

 

Similarly, passive and de facto claim suppression require focussed educational efforts and audits.  Active claim suppression is a more difficult barrier to address.  Advanced analytics, on-site and record audits, follow-up probing of claim withdrawals and greater efforts to track down abandoned or suspended claims can help identify suppression and reduce attrition.

 

Root cause analysis always requires significant investment in time, resources, and system design. Whether looking for motivations of workers choosing not report injury or identifying the incentives behind employer actions that suppress claim filing, investing in identifying root causes is essential to future reforms.

 

More immediately, perceived or actual delays in decisions and payments are likely factors to under-reporting by workers and functional claim suppression by employers. Reputation management can help but this is not just a public relations exercise; investments in staff, training and systems to achieve and maintain transparent service standards that meet stakeholder expectations are necessary.

 

Experience rating or modification is often singled out as the root cause intent behind active and passive employer claim suppression.  Improving the understanding and removing the incentives motivating suppressive behaviour may require innovations.  Suggestions such as removing short duration, low-cost claims (under four to six weeks) from claims cost calculations for experience rating in combination with employers paying a refundable compensation payment to the worker could address claim suppression and worker apprehension. 

 

Thanks to the students who asked the questions that started this discussion.  Thanks also to the jurisdictions who published their acceptance/denial data and offered additional insights for this series.