When AI meets tradition: How Japan's lawyers are rewriting their roles

As part of the Daily Journal Series on how legal professionals are using AI this is the third of three articles based on interviews personally conducted in Tokyo by Howard Miller, past President of the State Bar of California and Daily Journal Contributing Editor, and Marc Miller, Ralph W. Bilby Professor of Law and Dean Emeritus, James E. Rogers University of Arizona College of Law

The interviews were with Atsushi Okada, Mori Hamada & Matsumoto • Professor Ichiro Kobayashi • Tomohiro Okumura, LegalOn Technologies • Takahiro Homma, Renesas Electronics • and Toshimi Itakura, Sojitz Corporation, all of whom are further identified in the articles. All interviews were recorded, transcribed and sent for review by participants prior to publication.

Building the AI-Ready Lawyer

Japan's legal organizations are confronting a skills revolution--and developing surprisingly structured, human-centered approaches to an inherently uncertain challenge

In August 2024, Toshimi Itakura gathered the entire legal department at Sojitz Corporation--all 70 members--into a large room for what she had declared "AI Month." Attendance was mandatory. The consultant she had retained started gently, asking everyone to generate nicknames for themselves using the chat interface, then moving toward actual workflow applications. By the end of the month, she had given everyone a specific assignment: go back to your section, identify a process that AI could improve, build a prompt for it, see what happens, and come back ready to discuss.

Before the session, about half of the department's members were using the company's AI chat platform. After it, usage rose to roughly 90 percent. It has since settled at around 70 to 80 percent--a figure Itakura considers a meaningful baseline, not a ceiling.

"Familiarity is very important," she said. "To get the familiarity, you have to use it. Whether it works or not, you have to use it."

Across Tokyo's major legal organizations, versions of this challenge--how do you actually get lawyers to engage with AI and then develop genuine fluency with it--are being answered with approaches that are more deliberate, more structured, and more human-centered than many of their California counterparts might expect.

The Adoption Problem

The resistance is real and consistent. It does not sort cleanly by seniority, contrary to the conventional wisdom that senior lawyers are the holdouts and young associates are the enthusiasts. Takahiro Homma, General Counsel of Renesas Electronics, put it plainly: "It really depends on personality." Some senior lawyers have been early adopters. Some younger lawyers are skeptical. The common excuse, Homma noted, is that AI cannot match expert judgment--"Oh, I'm the expert, AI cannot do this." His response is a different framing: "Even if 70, 75 percent of what AI can do for me, legal professionals can do a much better job by just focusing on that 25 or 30 percent. Once they try, maybe they will understand."

Atsushi Okada, partner at Mori Hamada & Matsumoto and a member of the government committee on AI policy, made a similar observation about generational dynamics inside his firm. More than 80 percent of lawyers at Mori Hamada are now using generative AI--a figure he said compares favorably to many U.S. firms. But he was notably skeptical that formal training was the most effective mechanism for senior lawyers. "Senior attorneys should learn a lot more from younger associates rather than having training sessions," he said. The firm structures its training to accommodate both: mandatory basic training for all lawyers, more advanced sessions for the interested, and knowledge-sharing within specific practice groups so that conversations about AI are grounded in the specific work of M&A, litigation, or bankruptcy teams.

"I named the project 'Challenging Current AI Every Year.' Even if we couldn't do something last year, maybe this year we can."  -- Toshimi Itakura, Sojitz Corporation

Renesas is taking the competency question global. Homma is planning what he believes may be Japan's first global legal team AI hackathon--an online session bringing together lawyers from Japan, the United States, Germany, and other locations to share their best ideas for using Copilot in legal work, from optimal prompting to actual AI agent designs. The hypothesis is that useful ideas will come from unexpected places, and that the exercise itself will build familiarity across teams.

Itakura has extended the campaign past the initial AI Month. Her department now maintains a prompt library--a shared resource where its members can find tested prompts for common tasks--and publishes an internal AI newsletter featuring interviews with department members about how they are using the tools. Most recently, she launched what she has named "Challenging Current AI Every Year"--an initiative to revisit, annually, tasks that AI could not handle the year before. "Something that didn't work last year, maybe this year it can," she said. "Making it a story is very important."

A Framework for New Competencies

What distinguishes the most thoughtful of these conversations is not just the tactical question of how to get lawyers to use AI, but the deeper question of what it means to be a competent lawyer in a world where AI is embedded in the work. Itakura has developed a three-part framework that is worth examining closely.

The first competency is traditional legal knowledge--"necessary," she said, "to point out the hallucination risk when using AI, because the human has the final responsibility to make the decision." This is the competency that does not disappear; if anything, it becomes more important, because someone has to be able to catch the machine's errors.

The second is what she calls workflow design skill: the ability to break a legal process into components, identify which parts are suited to AI automation and which require human judgment, design the structure of that interaction, and take responsibility for the result. This is a new competency--one that did not appear in any traditional legal education curriculum and is not part of any bar exam. It is closer to systems thinking or process engineering than anything traditionally associated with law practice.

"Human beings are really good at sensing the environment. AI can handle textbook situations. It cannot read a room."  -- Toshimi Itakura, Sojitz Corporation

The third is human-centered skill: communication, empathy, situational judgment, the ability to sense the environment of a business and respond to what is not said as much as what is. "Human beings are really good at sensing those things," Itakura said. "AI can handle textbook situations. It cannot read a room." She runs a separate management training program within the legal department specifically to develop these skills--one year focused on empathy, another on feedback, a third on personal management style. All of it face-to-face. None of it AI-assisted.

The Law School Question

Missing from Japan's picture is meaningful engagement by law schools. Professor Ichiro Kobayashi, who heads a center for the Global Legal Innovation Education and Research Center at his university, is among the most active academics working on AI and law questions. But he acknowledged that no Japanese law school has yet made AI a required element of legal education--it remains optional. Some schools are opening new law-and-technology courses. His own institution is developing a curriculum. But the integration is nascent.

In the U.S., at least one law school--Suffolk University in Boston--now requires first-year students to build an AI agent, not merely to use AI tools. Kobayashi found the approach intriguing. The distinction matters: learning to prompt an AI system teaches familiarity, but building an agent develops the kind of deeper understanding--of capabilities, limitations and design logic--that Itakura's workflow design competency requires. Whether Japanese law schools can develop similar programs quickly enough to be relevant to the profession's current needs is an open question.

Okada, who lectures at law schools and tracks their AI engagement, was optimistic if measured. "There are some law schools trying to educate students about AI--how to use AI tools more effectively and more cautiously, and what should be the lawyer's role in the era of AI," he said. "I think more and more topics on AI will be discussed in the context of law school training."

What Remains Human

Across every conversation, one point recurred with the consistency of a refrain: the irreducibility of human judgment. Not as consolation--not as the defensive claim of a profession trying to protect its market--but as a working hypothesis about where value actually lies.

Itakura's management training. Okada's insistence that creative, high-judgment work is what law firms need to protect. Homma's observation that AI can handle 75 percent of any given task, which means a lawyer's value is concentrated in the 25 percent that AI cannot do better. Kobayashi's point that Japan's culture of informal dispute resolution depends on the kind of contextual human sensitivity no model can replicate.

None of them are naive about the trajectory. Okada sat with the question of what happens in the long run, when AI can do more and more autonomously, and chose honesty over reassurance. "If AI can automatically, autonomously do more and more things, and human creativity becomes not necessary, it is certainly a crisis for us," he said. But in the near and medium term, he concluded, the technology plays a more positive role than a negative one--augmenting, not replacing.

The lawyers building AI competency inside Japan's legal organizations are, in effect, placing a bet on that conclusion. They are training for a world in which the machine does more of the work--and the question is not whether that world is coming, but whether the people inside the profession will be ready to do the part that remains distinctly theirs.