Skip to Content

News & Knowledge

Artificial Intelligence – The Implications of Machine Learning in Workers’ Compensation

January 31, 2025

The use of artificial intelligence (AI) technology in the workers’ compensation arena is on the horizon. Given the scale of AI applications and the speed at which the technology is developing and being applied, AI has the potential to optimize efficiency, increase accuracy, and enhance compliance. On the other hand, AI also presents unique challenges and ethical considerations, which if applied without care and consideration, could have disastrous consequences.

Regardless of the oncoming challenges and ethical quandaries it poses, we anticipate it is inevitable that AI technology will be incorporated into many facets of worker’s compensation practice, from employers and healthcare providers, to legal practitioners, and in due course, the Workers’ Compensation Appeals Board.

It is therefore essential that all practitioners understand what AI is, what it can provide, and how best to apply it to our complex workers’ compensation system.

What Is AI Technology and How Does It Work?

AI is a field of science involving building computers or systems that simulate intelligent human behavior autonomously. It incorporates the three branches of science –social, natural, and formal sciences—through interdisciplinary fields like data analytics, computer science, natural language processing, software engineering, and neuroscience. AI uses theories and methods from multiple disciplines to develop systems that reason, learn, adapt, and make predictions or solve problems.

AI can be non-generative or generative. For non-generative AI, machine learning programs consume data to make predictions, including probabilities or classifications. First, the AI program is trained from unsupervised and/or supervised data that can make decisions or predictions. Unsupervised data has no labels. Supervised data can include information such as names, treatment therapies, numbers, body parts, types of injuries, settlement categories, and values. Then, the AI program solves problems and develops predictions by interpreting data. These predictions are non-generative AI. In the world of workers’ compensation, non-generative AI could make predictions, including an injured worker’s likely treatment, settlement valuation, and permanent disability.

In contrast, deep learning, including generative AI (GenAI), generates new data. Deep learning is a subset of machine learning that uses artificial neural networks, modeled like neurons in the human brain, to perform complex decision-making. Deep learning requires a higher volume of data consumption and more layers of modeling to train the system. GenAI learns patterns to generate new data including natural language, audio, images, text, video, audio, and other data. For example, currently the most common form of generative AI is to generate images or art, or create written documents based on a prompt which can be honed and edited through additional prompts (for example, Chat GPT).

What Are Some Applications for AI in Workers’ Compensation?

Based on what we presently know about AI, it is possible that it can be trained on data and models to aid in pattern recognition to identify trends, fraud, and anomalies associated with a workers’ compensation injury claim.

On one hand, there are many potential benefits from AI:

  • For insurers, AI could make cases more efficient by reducing the life of a claim, identifying fraudulent claims, increasing automation, and making more accurate settlement valuations.
  • Employers could analyze safety record data and workplace injury reports to make predictions about accidents; thereby allowing an employer to diminish their exposure with data driven mitigation.
  • Health systems and medical providers can deploy AI to interact with injured workers, recognize anomalies with treatment, identify the overprescription of medications, standardize care, and improve recovery outcomes.
  • Attorneys can benefit from the use of AI to analyze qualified medical examiners, enhance compliance, and reduce litigation costs.  Every case, of course, is fact specific.  AI, however, can provide basic outlines for questioning and cross-examination.
  • The Workers’ Compensation Appeals Board may benefit from leveraging AI to increase equity and consistency with its decision making.

Largely, the automation, prediction, and efficiencies resulting from AI may diminish human error and distill workloads so that more attention can be focused on complex claims.

On the other hand, the difficulty in incorporating AI systems into workers’ compensation practices include the expense of the technology and poor modeling which can result in less accurate outputs. We have all seen the horror stories of the misapplication of AI in the practice of law, with one such attorney being sanctioned for using AI to draft a brief which cited nonexistent case law. The large volume of varied data necessary to properly train AI, biased or outdated data resulting in inaccurate predictions, the lack of oversight regarding the use and accuracy of the data, and reluctance to learn a new technology, can all exacerbate the errors and issues with AI, and could lead to inconsistent or inaccurate application.

Ethical Considerations of AI

The use of AI technology also draws ethical concerns regarding its privacy, control, transparency, bias and accuracy. AI systems are trained on datasets and raise questions as to whether sharing private information with a system is a breach of confidentiality. Additionally, the limited control and ownership of the data used to train an AI system has implications for reduced security. Furthermore, confidentiality may be breached when a third party accesses the data used in an AI training model.

The complex modeling required for some AI systems, such as machine learning and generative AI, makes transparency challenging, especially when trying to explain the rationale associated with an outcome or decision when challenged in an adversarial system such as Workers’ Compensation.
AI systems outputs and predictions are only as good as the data inputs. Machine learning is conditioned on datasets which can include human bias and outdated information. Therefore, poor modeling for an AI system can result in less accurate or unlawful results.

Does the Law Offer Guidance on AI Usage?

Regulation of AI technology is on the legal horizon. As with newer technologies, the law seeks to adapt to the changing landscape. In 2023, the California State Bar released guidance on the use of generative AI in the practice of law that conforms with an attorney’s obligations of professional responsibility. The Standing Committee on Professional Responsibility and Conduct noted the attorney’s responsibility to understand the benefits and risks of AI technology as it relates to duties such as confidentiality, diligence, candor, and communication.

In an effort to ensure transparency and prohibit discrimination, California introduced Assembly Bill 2930 to prevent AI discrimination based on protected class characteristics and regulate the employer’s use of AI as a decision-making tool by developers of the technology and users. The bill proposed non-compliance penalties of administrative fines and civil penalties. Currently, the legislature has not passed the bill.

In an unsuccessful attempt to regulate AI safety, California introduced Senate Bill 1047 which sought to mandate generative AI transparency. The bill was vetoed but is likely to return in another form as California makes proactive efforts to ensure the safety of AI technology.

In summary, as AI becomes more robust in its applications and capacity to streamline the workers’ compensation field, we can expect future laws governing its use and safety. As the technology is furthered by its development and usage, there is also the potential for cross-system integration and overall increased efficiency. Everyone working in the workers’ compensation system will need to be prepared to deal with the implications of AI’s continued expansion.

Written By:

Sherri L. Bridgeforth, Esq. of our LFLM – Sacramento Office

Laughlin, Falbo, Levy & Moresi, LLP.

www.lflm.com