Forget ChatGPT—Law Firms Are Lau nching Their Own Gen AI Chatbots

Each week, the Law.com Barometer newsletter, powered by the ALM Global Newsroom and Legalweek brings you the trends, disruptions, and shifts our reporters and editors are tracking through coverage spanning every beat and region across the ALM Global Newsroom. The micro-topic coverage will not only help you navigate the changing legal landscape but also prepare you to discuss these shifts with thousands of legal leaders at Legalweek 2024, taking place from January 29 to February 1, 2024, in New York City. Learn more and register today:

The Shift: Forget ChatGPT—Law Firms Are Launching Their Own Gen AI Chatbots

This summer, the pitfalls of lawyers leveraging ChatGPT took center stage when a New York attorney used the chatbot to draft an error-ridden court brief that he filed in federal court. Such a mistake was exactly what one insurance company meant when they warned their law firm clients that ChatGPT is “not ready for primetime.” In response, some firms have banned the use of the chatbot altogether.

But far from inhibiting law firms' embrace of generative AI, this ChatGPT backlash may actually be accelerating it. The chatbot from AI developer OpenAI was never going to be viable for lawyers in the first place. Because the data users input into it goes back to OpenAI, the danger of exposing client or sensitive information is far too great. By reining in ChatGPT, however, law firms could more confidently figure out how to leverage the underlying generative AI models powering the software.

“What [our ChatGPT] block also afforded us is the breathing room to establish a safe sandbox where we can leverage the capabilities of ChatGPT and other GPT models without the added risk of the data going back into [public access] models,” said Sarah Alt, the first ever chief process and AI officer at Michael Best & Friedrich.

What a growing number of law firms have realized is that they can safely allow their lawyers to experiment with the underlying generative AI models—so long as it is through their own proprietary chatbots. This summer has already seen a handful of law firms release generative AI chatbots, including Dentons, Troutman Pepper Hamilton Sanders, Davis Wright Tremaine, Gunderson Dettmer Stough Villeneuve Franklin & Hachigian, U.K.-firm Travers Smith, and Australian law firm Allens.

Firms’ ability to launch generative AI chatbots is in part due to the launch of Microsoft Azure OpenAI Service, which allows users to plug their own tech into OpenAI’s generative AI models without data being sent back to OpenAI. But as many found out, mitigating the risks of generative AI isn’t just a technology question. Launching a generative AI chatbot also requires changes to firm policies, culture and training, as well as ongoing oversight.

While it will take time to discern the long-term impacts of firms using generative AI, these chatbots will likely lead to some immediate changes for law firms, from increased efficiencies to new skills lawyers need to learn.

The Conversation

Law firms created generative AI chatbots primarily to allow their lawyers to experiment with the technology within specific use cases. These use cases vary by firm, but include tasks ranging from assigning stakeholders to projects and extracting data from documents, to drafting meeting summaries and marketing content, among other things.

While the ability for programs such as Microsoft Azure OpenAI Service to create a closed system was key in allowing firms to develop their own generative AI chatbots, it wasn’t just the technology that needed to change.

To mitigate the risk of using generative AI, Troutman Pepper, for instance, had to mandate ethics courses for its attorneys and draft a new AI policy incorporating guidance from their insurance provider.

Like many other firms, Troutman Pepper restricts staff from inputting client or sensitive data into the firm's generative AI chatbot, which is called Athena, or using the chatbot for legal research purposes.

But even though it trained its attorneys on what not to do with the chatbot, the firm is taking no chances. “[We are] monitoring the logs [of attorney prompts], making sure that the use cases are consistent with [our policies] and communicating that over and over to our attorneys,” said Troutman Pepper’s director of innovation solutions Andrew Medeiros.

Davis Wright is also monitoring how attorneys use the firm's generative AI chatbot for similar purposes. Dan Szabo, the director of enterprise projects at Davis Wright, noted that the firm gets “a summary of the hottest topics that are asked daily, and we're scanning to look for opportunities to train our people on how to use the tool better, or ways to guide them...on ways they shouldn’t be using it."

Davis Wright also ensures that its lawyers don't place client data in its chatbot by essentially isolating it from such information in the first place. Davis Wright partner and chair of its AI steering committee Vidhya Prabhakara noted that “one of our most important baseline prohibitions right now is we are not using any client data or anything that's confidential as part of the sort of datasets that [the chabot] can pull from. So, at this moment, we don't actually have a required training module that our attorneys and staff must use because we already have sort of sandboxed our chatbot."

The Significance

It’s still early days for generative AI in the legal market, and even though law firms are embracing the technology head-on, it will likely take years to see its full impact on legal market jobs, operations and the billable hour. But law firms’ generative AI chatbots will have more immediate impacts in the meantime.

Many law firms, for instance, expect to soon see their chatbots bring even greater levels of efficiency to their operations.

Davis Wright’s Prabhakaran said that in addition to allowing their lawyers to experiment with the technology, the firm also created the chatbot so attorneys can “learn how to use it to make non-legal tasks involved in lawyering more efficient, in a safe way.”

One major way such law firm chatbots can elicit greater efficiencies is through automating first drafts of certain content, such as blog or social media posts. “You still edit that and communicate from your expertise. … but getting [to] that first draft, that blank page is gone. You don't have to start on the blank page anymore,” Troutman Pepper’s Medeiros said.

Some law firm chatbots are also looking to harness the power of generative AI to more efficiently extract key data points from documents or databases. Not only could this technology speed up e-discovery and research tasks, but also add significant capabilities to law firms’ knowledge management efforts as well.

Still, law firms' chatbots will only lead to greater efficiencies if lawyers know how to properly use them. So it’s little surprise that law firms are stressing the need for prompt engineering skills—the ability to craft specific questions for generative AI chatbots in order to get more accurate and relevant outputs.

Prabhakaran noted that Davis Wright is looking into prompt engineering training for its staff—and it’s far from the only one. In fact, even firms that haven’t released generative AI chatbots still see this as a necessary skill as generative AI becomes more widespread. This summer, for instance, Orrick, Herrington & Sutcliffe partnered with legal education provider AltaClaro to create a prompt engineering learning module for its summer associate class. Orrick’s global head of practice innovation Kate Orr noted that the firm is “very focused on generative AI” and like others, is “in the active testing phase.”

The Information

Want to know more? Here's what we've discovered in the ALM Global Newsroom:

The Forecast

There’s little doubt that these law firm chatbots will evolve and expand in the near future, covering more use cases and becoming more integrated into attorneys’ workflows. “Right now, you have to choose to interact with the AI assistant. The goal is for AI to seamlessly plug into the legal tasks we are already planning to do. That is the next evolution—the real promise of this,” Prabhakaran said.

Of course, these chatbots aren’t the only ways lawyers will be exposed to generative AI. A host of legal tech tools, from e-discovery and contract review platforms to legal research offerings, are integrating generative AI into their functionality. These products, however, often use generative AI in specifically targeted ways. In contrast, law firms’ generative AI chatbots are much broader, allowing lawyers more flexibility in experimenting with the technology.

As a learning aide—a sort of a crash course into generative AI—such chatbots are invaluable, especially because law firms themselves can exert direct control over them, mitigating many security and privacy concerns.

It’s therefore likely more law firms will release their own generative AI chatbots in the months to come. Whether these tools evolve into mainstay products within firms, or just act as a stepping stone to prepare attorneys for more advanced generative AI-powered offerings in the future, is yet to be seen. But one thing is certain: fewer lawyers will be talking about ChatGPT in the near future.

Rhys Dipshan is the senior editor on the legal technology news desk. Contact him at [email protected] and follow him on Twitter: @R_Dipshan