The pace of legal technology continues to surprise us. Led by rapid developments in artificial intelligence (AI), we continue to see changes that impact the future of practicing law. Our Future Proofing column is, by nature, forward looking. We discuss emerging trends, interview lawyers leading change and share ideas to help lawyers evolve to meet the market changes while maintaining successful practices.
In this column, we look at some breaking news on the trends and changes we have covered in recent columns and cover some new trends we know we’ll be addressing in future columns.
As January ended, the AI community was buzzing with talk of DeepSeek-R1, a language reasoning model incorporating a novel approach to reinforcement learning, and with a compelling user interface that gave a running internal monologue showing how it "thought" and "reasoned" its way through queries as it formulated and reworked strategies for answering questions and generating reports. Adding to the excitement and intrigue, DeepSeek was developed in China where it was assumed U.S. export embargoes on latest generation chips and processing units would have prevented such a thing from happening. Yet not only did DeepSeek deliver performance against AI benchmarks that rivaled, and in some respects exceeded, the best and most expensive models on offer from OpenAI and its commercial peers, it was reported they did so on a less than $6 million model training budget. The real kicker? DeepSeek gave away their secret sauce as an open-source release with a detailed paper explaining their methods, effectively authorizing global, irrevocable, permissionless and royalty-free rights to use, reproduce and modify anything and everything about the R1 model.
Within days, the DeepSeek iPhone app was at the top of the App Store charts. Amazon, Microsoft and other established cloud computing and AI services companies incorporated the technology into platforms alongside other leading open-source and commercial models. And the technology and investment press chatter reached a fever pitch as people wondered how this model would tip the balance of the AI arms race between the United States and China. Moreover, the possibility that advanced language reasoning models could conceivably be developed so cheaply and without access to latest generation chips and processors wiped almost a trillion dollars (albeit temporarily) from the balance sheets of leading chip and technology manufacturers.
In our May 2023 column, we interviewed legal AI consultant Colin Lachance, who is now the Ontario Bar Association’s Innovator in Residence, about using ChatGPT for research and writing. A week after DeepSeek’s release we spoke to him again.
His first observation, “DeepSeek’s release was another AI moment like we experienced in November 2022 with the release of ChatGPT. When ChatGPT was released the technology advance, public awareness and developer rush to use/adopt/adapt in their own apps/servers hit simultaneously. With DeepSeek a whole new scope of potential was revealed all at once rather than incrementally or through glimpses and partially obscured views.”
So, we then asked Colin what this all meant. “Are we ditching ChatGPT, Claude, Gemini and the others for something better?” After acknowledging he had had a week to ponder the huge changes that the release DeepSeek will potentially bring, Colin said “the short answer is no, we won’t be ditching the established tools just yet, but the big impact of DeepSeek's release is that all AI models will get better and cheaper at a pace even faster than before and thanks to the open-source element we should expect the potential to run incredibly powerful AI models in fully private environments to be on the horizon.”
Expanding on this line of thought Colin said “while there will be some AI users unconcerned by the prospect of sending their queries to China-based servers for processing, large business and even large law firms have the opportunity and ability to create their own, high performance and fully private versions of DeepSeek on their own servers if they are prepared to spend around $10,000 to 50,000 on sufficiently powerful hardware. Indeed, even home hobbyists are finding ways to build scaled-down versions on ‘systems’ as basic as a high-end laptop.”
James Chesser, a South Carolina–based immigration lawyer, engineer and AI developer, sees very similar and important implications for the legal profession from the release of DeepSeek:
“This open-source, innovative, cost-efficient reasoning model rivals OpenAI’s capital-intensive GPT-o1 counterpart, yet it’s free to train and download from HuggingFace, and can be hosted offline on secure equipment that may soon be within reach of average businesses. What sets DeepSeek-R1 apart isn’t just its performance, it’s the power it places in the hands of small firms. Imagine law practices being able to fine-tune and securely self-host small domain-specific language models tailored to individual clients, whether locally or even on a smartphone. Imagine having the AI equivalent of legal assistants, researchers, archivists, focus groups, mock jurors and even judges in your pocket!”
Look for a more detailed article from James in the upcoming Futures issue of Law Practice Today exploring how DeepSeek-R1 could redefine the future of legal work—pros, cons and all.
Case Western Reserve University School of Law is requiring all first-year law students to earn a certification in legal AI. They recently launched Introduction to AI and the Law, a two-day seminar focused on understanding key concepts, introduction to AI-enhanced legal research, document summarization and generation and ethical considerations. The seminar is mandatory for all first-year students and requires passing an examination at the end of the seminar.
This development comes on the heels of our last column where we chatted with Dennis Kennedy, director of the Michigan State University Center for Law, Technology & Innovation about changing law school curriculum in light of AI. So, we asked Dennis to share his perspective on the Case Western development:
"Case Western's forward-looking move to require AI certification for 1Ls is a recognition of the fundamental shift happening inside and outside the legal profession. This isn't about 'if' AI will impact legal practice, it's about 'how' and 'how well' lawyers will adapt. Other law schools need to take note––are you preparing your students for the future, or are you letting them fall behind? With the speed of AI change, there will be a challenge in keeping this training focused and relevant––what might be part of certification for one set of students might be vastly different for another set. This is a fascinating first step, but it's also the start of a much larger conversation we need to be having right now about the role of technology in shaping the future of law and legal education. I can’t wait to learn from their outcomes."
This also raises the question: How much AI training do licensed lawyers have to take to show competence? If two days is the floor for 1Ls, how much do lawyers need? To paraphrase Dennis “are you preparing your lawyers for the future, or are you letting them fall behind?”
As technology rapidly evolves, states continue to review their legal regulatory structures to expand legal services to more consumers. The trend continues to allow nonlawyers to provide legal services, although under strict oversight.
The State of Washington, which once created and then closed their Limited License Legal Technician Program has returned with a new program seeking to expand access to justice in that state. This new program approved by the Washington State Bar Association (WSBA) and ordered by the Washington State Supreme Court changes Rule 5.4 to allow legal service providers who are not lawyers. According to the implementing order, it is a 10-year pilot program “to test entity regulation under the proposed Framework will assist the Board, the WSBA, and this Court in determining whether entity regulation will increase access to justice by enhancing access to affordable and reliable legal and law-related services consistent with protection of the public, and whether entity regulation will create risks of consumer harm, regulatory challenges, or other risks that would serve as barriers to implementing reform.” Each entity will have to appoint a compliance officer to ensure the rules and procedures of the pilot are followed.
Global consulting firm KPMG is the first Big 4 accounting firm to apply to the Arizona Supreme Court for permission to practice law under their Alternative Business Structure program. As of this writing, the Arizona Supreme Court has not ruled on this application, having requested additional information on how the firm would provide legal services in other states without violating unauthorized practice of law and ethics rules.
Many believe that KPMG is clearing the way for the other Big 4 players to join the Arizona ABS program. The full scope of their legal services, both geographically and by practice area are not known.
By revisiting several evolving issues in our recent columns, we hope you have a better appreciation of the pace of change external to and within the profession. Understanding these changes will help firms prepare for their future, which is much more quickly becoming the present.
