Artificial intelligence has already infiltrated workers' compensation systems through software updates and automated features, often without formal announcement or professional awareness. Unlike the dramatic rollouts of new technology, AI entered the industry quietly, embedding itself in claims software, electronic medical records, and automated documentation tools that professionals already trusted. Once integrated, these systems began shaping how information is presented and decisions are made, raising urgent questions about accountability that the industry has yet to answer.

How Is AI Actually Being Used in Workers' Compensation Today?

AI in workers' compensation does not look like the autonomous decision-making systems many people imagine. Instead, it appears in practical, everyday forms that professionals may not even recognize as artificial intelligence. These tools operate behind the scenes, subtly influencing professional judgment without making decisions outright.

• Automated Summaries: Systems generate condensed versions of medical records and claim information, shaping what professionals see first and what they focus on during review.
• Suggested Language: AI proposes specific wording for documentation and communications, influencing how decisions and findings are recorded and communicated.
• Prioritization and Sorting: Algorithms flag certain claims or issues as urgent, determining which cases receive immediate attention and which are deprioritized.
• Automated Reminders and Follow-ups: Systems generate prompts for next steps and administrative actions, guiding the workflow without explicit human direction.
• Documentation Assistance: Tools help draft notes and communications, reducing the burden of manual writing while subtly shaping the tone and content of professional records.

The critical issue is that these tools do not need to make the final decision to influence outcomes. A summary that emphasizes certain medical findings over others can shape what a professional focuses on, even though that professional still technically makes the final call. In workers' compensation, where decisions about medical care, benefits, and timelines affect injured workers at their most vulnerable moments, this indirect influence carries significant weight.

Why Does It Matter If Humans Still Make the Final Decision?

A widespread assumption in the industry is that AI does not really matter as long as a human approves the final decision. In practice, this assumption misses how decisions actually get made. Professionals do not evaluate claims in a vacuum; they work with information that has already been filtered, summarized, and organized by automated systems. The moment AI shapes what information appears first, what gets highlighted, and what gets buried, it has already influenced the outcome.

Workers' compensation decisions are also cumulative. One note influences the next professional's judgment. One summary informs another administrative action. One decision about medical care affects compensation, timelines, and the injured worker's trust in the system. When AI shapes information at any point in this chain, its impact travels far beyond the initial decision point.

This dynamic is not unique to workers' compensation, but the stakes are particularly high in this context. Injured workers are navigating health concerns, income disruption, and uncertainty. Decisions made quickly or based on incomplete or shaped information can have lasting consequences for their recovery and financial stability.

What Happened With the Last Generation of Algorithmic Claims Systems?

The current situation echoes a critical moment in insurance history. In the 1990s, a software program called Colossus was introduced to standardize bodily-injury claim valuations. The system was marketed as a neutral, data-driven solution that would reduce subjectivity and ensure consistency across settlements. Insurers adopted it enthusiastically, seeing it as a way to manage volume and control costs.

But Colossus had a fatal flaw: nobody could see how it actually worked. The system's internal logic, reportedly involving thousands of proprietary "value drivers," was hidden behind trade-secret protections. Adjusters could see the recommended settlement range, but not the reasoning behind it. Over time, regulators and claim professionals observed that a system designed to remove subjectivity had actually introduced new, systemic biases instead.

By the early 2000s, Colossus became the focal point of consumer litigation and regulatory scrutiny. Plaintiffs alleged that insurers had used the software to systematically undervalue bodily-injury claims by enforcing uniform settlement ranges and pressuring adjusters to conform to them. The central contention was not that Colossus malfunctioned technically, but that its design embodied a "one-size-fits-all" approach to inherently individualized losses. By converting subjective, claimant-specific evaluations into standardized algorithmic outputs, the software risked disregarding unique characteristics like occupation, pre-existing conditions, or quality-of-life impacts that traditionally informed damage valuations.

The most notable action culminated in a national class-action settlement in Hensley v. Computer Sciences Corporation, approved in 2005. While the software's developer denied wrongdoing, the settlement required modifications to marketing practices and greater transparency about the system's intended use. Subsequent market-conduct examinations in states including California and Michigan reached similar conclusions: insurers could not require adjusters to adhere rigidly to Colossus outputs, nor could they compensate personnel based on compliance with those valuations.

What Three Critical Lessons Does Colossus Teach About Today's AI Systems?

• Transparency Is Essential: Regulators and courts grew skeptical of Colossus not merely because it existed, but because it operated as a hidden arbiter of value. The lack of disclosure to claimants and, in some cases, to line adjusters transformed a management tool into a litigation risk. Modern AI systems face the same transparency challenge.
• Automation Amplifies Intent: A valuation system trained or tuned to achieve efficiency gains can easily be repurposed to accomplish cost-containment objectives inconsistent with fair-claims standards. Plaintiffs in Hensley alleged that Colossus was calibrated to produce systematic reductions in claim payouts, reportedly targeting decreases of up to fifteen percent. Whether or not that allegation could be empirically proven, the perception alone eroded confidence in algorithmic fairness.
• Human Oversight Is Indispensable: Following the Colossus settlements, several state insurance departments emphasized that adjusters must retain independent judgment and document reasons for either deviating from or adopting software recommendations. That principle, known as "human-in-the-loop" accountability, has since become the ethical baseline for any deployment of AI in claims handling.

These lessons from the 1990s and 2000s remain directly applicable today. Modern claim-analysis models, particularly those that evaluate images of property damage or generate text-based coverage explanations, echo the same dynamics of efficiency versus discretion that defined Colossus. The principal difference lies in scale and autonomy. Where Colossus applied deterministic rules to structured data, modern AI operates on probabilistic inference from vast, unstructured sources, millions of data points often processed without direct human review. The opacity has deepened, and with it, the potential for both error and abuse.

What Does Responsible AI Look Like in Workers' Compensation?

One framework gaining traction in the industry is the EEE Responsible AI Standard, which stands for Educate, Empower, and Elevate. The principle is straightforward: as artificial intelligence becomes embedded in professional systems, those systems must educate professionals about where AI is influencing judgment, empower them to remain actively responsible for decisions, and elevate expectations for oversight when technology shapes outcomes.

Education creates awareness about where AI may be shaping work. Empowerment reinforces human authority and ensures professionals retain the ability to override or question algorithmic recommendations. Elevation ensures that accountability rises alongside capability, meaning responsibility does not dilute when technology enters the workflow; it concentrates.

"Artificial intelligence is already part of workers' compensation, even if many professionals do not realize it," stated Nikki Mehrpoo, author of the analysis. "There was no announcement. No formal rollout. No meeting where someone said, 'We are now using AI.' Instead, artificial intelligence entered quietly through software updates, new features, and automated functions added to systems people already trusted."

Nikki Mehrpoo, MedLegal Professor

This approach does not treat AI as something to fear or blindly adopt. Instead, it treats AI as something to govern responsibly. At this stage, the goal is awareness, not immediate action. Understanding where AI may influence work is the first step before any changes are expected.

Why Does Responsibility Matter When Technology Shapes Decisions?

The fundamental principle underlying responsible AI in workers' compensation is simple but essential: responsibility does not disappear when technology enters the workflow. Efficiency does not excuse inattention. Automation does not remove duty. Convenience does not replace oversight.

When AI influences a judgment about an injured worker, even indirectly, responsibility still belongs to the people and organizations involved. The question is not whether responsibility exists; it is whether that responsibility is being recognized, reviewed, and owned. Professionals must ask themselves: Where is automation shaping our decisions? Who is responsible for checking its output? How do we know when something important is missing? Not asking these questions does not remove responsibility; it only delays accountability.

Technology can reduce administrative burden, help manage volume, and support consistency in a demanding system. Used responsibly, it can be valuable. But tools that influence professional judgment do not replace professional responsibility. The challenge for workers' compensation professionals and organizations is to harness the efficiency benefits of AI while maintaining the oversight, transparency, and human judgment that injured workers deserve.