A 2024 analysis of the legal profession by legal services provider Clio found that roughly three-quarters of hourly billable tasks have the potential to be automated by AI.
Areas primed for automation include documenting and recording information (which scored a high 86% for its automation potential), information gathering (which scored 70%), and analyzing data or information (which scored 69%).
Messages about generative AI’s potential to boost law firm profitability by automating repetitive tasks and allowing lawyers to focus on strategic legal work aren’t new. So why are some law firms still hesitant to incorporate large language models (LLMs) into their workflows?
In a recent white paper, Standford Law researchers Aksh Garg and Dr. Megan Ma explored some areas of “friction” between law firms and generative AI developers, which may be impeding both broad adoption and the development of more effective tools.
Highly sensitive client data
To provide value, LLMs must be trained on relevant and quality data. However, law firms are cautious about sharing proprietary data protected by attorney-client privilege. This creates what Garg and Ma call the “chicken-or-egg problem.”
“Proprietary data is essential for building tools with real competitive value,” they wrote, and is also key to helping an LLM see the “intermediary stages in the pipeline” of legal work which is essential to its ability to build reasoning into the outputs it generates.
While some legal tech developers are working with design partners and law firms to try to “break this data wall,” Garg and Ma note that, for AI developers, “even data from generous and willing firms can be hard to directly utilize” as it must be filtered through redaction software to ensure client confidentiality.
The standard business model
In today’s legal profession, the billable hour reigns supreme. While some firms are experimenting with fixed-fee and other alternative business models, the billable hour largely remains the norm.
That means the urgency to reduce time spent on tasks may not be a priority for law firms that get paid by hours of work, Garg and Ma note.
To reduce some of the “friction” law firms face when weighing the value of generative AI versus the billable hour, Garg and Ma note that legal tech developers are focusing on building tools that use generative AI to help firms address “monotonous” tasks that are often done by junior associates.
Some forward-thinking law firms, like Ashurst, Baker McKenzie, and Wilson Sonsini, have seen this create value for clients, they wrote.
Design roadblocks
Microsoft Office suite is ingrained into the workflows of lawyers. Getting lawyers to transition to a new platform isn’t easy and may require buy-in for a larger change management program at the firm.
Due to this reluctance, legal tech developers are having to build tools for firms “on top of Microsoft’s relatively clucky feature suite,” Garg and Ma wrote, creating “bottlenecks” and inefficiencies in their design, such as making it difficult for users to do things like “process and identify changes” in documents.
An “ideal interface,” Garg and Ma write, would be a generative AI tool that resembles a “legal copilot, with text autocomplete, and section insertion enabled.” However, this functionality isn’t currently possible, given the constraints of Word.
In addition, Garg and Ma note that LLMs “are bad pointwise editors,” which can be a challenge for lawyers who are “surgical with their edits on large documents.” Instead of editing out bits of information in a document as instructed by a lawyer, an LLM would regenerate an entirely new document, which creates the potential for hallucinations, they wrote.
‘Promising areas of innovation’
While Garg and Ma noted that there are still challenges to the development and broad adoption of generative AI for the legal profession, they identified some promising areas to keep an eye on, like personalized AI personas and AI models for computational and patent law.
“Carefully navigating these opportunities and challenges could help make the legal industry faster, more efficient, and more accessible for all,” they wrote.
Our Chief Counsel Mark C. Palmer writes frequently on why lawyers and law firms should embrace generative AI, as well as the ethical considerations when using these tools. His articles can be found here.
Staying up to date on issues impacting the legal profession is vital to your success. Subscribe here to get the Commission’s weekly news delivered to your inbox.
The AI Revolution in Legal and the Billable Hour: Is the End Near?
Crafting Influence: The Role of Language and AI in Legal Advocacy