Predictions for AI Use in Product Liability Litigation

Attorneys have watched with excitement, fear, or both as generative artificial intelligence (AI) products for the legal industry have become available over the past year. While attorneys must use this technology ethically and appropriately, AI has the potential to make litigation more efficient and accessible. And as I have tried some of the products, I share in the excitement and can visualize ways in which AI can benefit product liability and mass tort practices. This article shares a few of my hopes of what generative AI will be able to do in this practice along with how attorneys can prepare themselves for these use cases.

1. Medical Record Summarization. A single product liability case may require the analysis of thousands of pages of medical records. Attorneys often rely on medical record summarization services, in-house medical professionals, or junior attorneys to review and summarize these documents. At some point, generative AI will be able to produce summaries of these records that are as accurate as the aforementioned review options, greatly reducing costs and time. With these savings, attorneys might even consider obtaining more medical records than they previously would have, ultimately creating a more complete view of a plaintiff’s medical status.
   
2. Written Discovery. Many manufacturers are sued multiple times over a single product model. In each suit, however, the discovery requests can be different, requiring additional hours of work. At some point, generative AI will be able to use prior discovery responses to complete a first draft of objections and responses to slightly different requests. Depending on the product used, however, attorneys will need to carefully review any sources of authority generated by AI to ensure there are no “hallucinations,” which is a phenomenon where AI software generates false information.
   
3. PII/PHI Identification. Medical device and pharmaceutical manufacturers engaged in litigation have large amounts of data, including customer or patient personally identifiable information/ protected health information (PII/PHI). As part of document productions, companies may need to redact PII/PHI to ensure they are meeting the Health Insurance Portability and Accountability Act and other data privacy requirements. This can be a tedious process that requires manually looking through the entirety of numerous documents, including onerous native documents and unique database formats. Companies already advertise AI software that has the capacity to detect PII/PHI, allowing for redaction. Importantly, attorneys must carefully choose AI software based on many considerations, including data security and privacy, for documents with PII/PHI.
   
4. Claims Evaluation. In many cases, mass torts result in many claims with varying levels of potential exposure based on a variety of factors. As part of a global settlement, attorneys may struggle to appropriately value such claims. At some point, generative AI will be able to help to more efficiently, and potentially accurately, value these claims. Indeed, the insurance industry is already using AI to assist with claim evaluation. One such model allowed an insurance company to detect and classify car damage from images, translate that damage into the individual parts that were affected, and retrieved images of similar vehicles to help estimators evaluate whether the part at issue was actually damaged. There are numerous ways AI can be used in litigation claims evaluation, but a direct corollary to the given example would be to use images of medical devices at issue to value claims where such data is available. Attorneys should consider the characteristics important to the valuation of their claims that could be analyzed using AI software.
   
5. AI Professionals. As with any disruptive technology, some predict generative AI will replace certain jobs. I think it is more likely to change job responsibilities, however. For tasks that can now be made more efficient using AI, a professional will still need to determine how to use AI software to complete the task, in essence acting as a project manager over the software. Not only does the use of AI require human input and quality control, it also requires imagination and creativity to get the most out of it. Attorneys interested in providing value should learn AI prompting best practices through online resources, classes, and collaborating with other AI professionals. Eventually, implementing AI in litigation contexts will likely be its own legal specialty or will be a subset of existing e-Discovery specialty practices, requiring equivalent expertise.

In sum, generative AI has the potential to bring exciting changes in challenges to product liability and mass tort litigation. Attorneys need to stay abreast of AI software to meet professional responsibility obligations, especially understanding which tools can and cannot be used with proprietary client or patient information. Attorneys should use their human skills to determine how to leverage AI in their practices and implement these ideas by learning best practices for developing AI prompts.