A Few Things to Get Right Before the AI Tools Get Better

Thomas also co-chairs AGA’s Worldwide AI Advisory Board, established to help member firms explore practical applications of AI and adopt the technology responsibly.

Conversations with law firms about generative AI starts in the same place, with two viewpoints. On the one hand, somebody has read about a lawyer sanctioned for filing fabricated citations see recent ‘Hallucination’ case examples here and on the other hand, somebody else just watched an associate summarize a two-hundred-page contract in forty seconds. Both people are reacting correctly. The question is how a firm responds to both realities at once. 

That question is at the heart of Practical AI for Lawyers. Rather than focusing on the technology itself or simply warning lawyers about its risks, it is a practical field guide for those whose clients are already asking about these tools and who want to use them effectively while managing the associated professional risks. 

Here are five of the principles it is built on:

There are two ways to fail. Reckless adoption means pasting client material into a consumer chatbot and filing whatever comes back, treating fluent prose as finished work. Reflexive avoidance means banning or ignoring the tools, while lawyers within the firm continue using them independently. Avoidance may feel safe, but unmanaged use is still use, and it happens without an agreed policy, or any record of what information went where. 

The model does not know your case. It predicts text and may produce a plausible clause or citation with the same confidence as it produces a citation to a case that does not exist. Nothing generated by AI becomes legal work product until a lawyer has reviewed and verified it. The lawyer whose name appears on the document remains responsible for turning a prediction into professional advice. 

"Which model is best?" is the wrong question. Firms should instead ask which tool is appropriate for the particular task, data, client, and level of risk. The likelihood of the tool being wrong must be considered alongside the potential consequences if an error gets through. Improving the tone of an email sits at one end of that scale; preparing a court filing sits at the other. The information being entered is equally important. A seemingly low-risk task involving privileged or confidential material must be governed by the sensitivity of the data, not simply by the nature of the task.

Verification must form part of the workflow. If it is treated as a final clean-up exercise it can easily be skipped. Lawyers should not ask the tool to identify cases unless they are prepared to retrieve and read every authority it produces. This may mean that research saves less time than demonstrations sometimes suggest, but that is the necessary trade-off for using the technology responsibly.

A policy nobody reads is not a policy. Most firmwide AI policies are either a one-line prohibition or a lengthy document drafted by committee. Neither approach provides meaningful guidance if the policy is ignored. Firms need something concise enough to be followed and specific enough to identify approved tools, explain what information may be entered  and establish who must check the output before it leaves the building.

None of this requires a data scientist. It requires firms to apply the same professional judgement they would use when reviewing, for example, a junior associate’s first draft—and to apply it consistently. Practical AI for Lawyers is designed to support this, providing workflows, checklists, prompts, and policy language that a firm can adopt on Monday morning, together with the underlying reasoning needed when a task does not fit neatly within the framework. 

Thomas examines these issues in greater detail in Practical AI for Lawyers, available here. 

About Dunlap, Bennett & Ludwig:

Dunlap Bennett & Ludwig is Alliott Global Alliance’s law firm representative for the Washington DC Metropolitan Area. 

Founded in 2002, DBL is an 80-attorney firm providing legal counsel across industries such as technology, defence, aviation, healthcare, and finance. The firm specializes in IP, government contracts, corporate law, litigation, and regulatory compliance.

With a strong international focus, DBL advises on cross-border transactions, global trade compliance, and foreign investment, particularly in Europe, the Middle East, and Asia. Read more.

About Alliott Global Alliance:

Founded in 1979, Alliott Global Alliance is a global alliance of independent law, accounting, and advisory firms, operating from more than 340 offices in over 100 countries and providing clients with coordinated, multidisciplinary support worldwide through a trusted global platform. The Alliance is ranked in Chambers and Partners and Legal 500 as a Leading Law Firm Network, reflecting the strength and international capability of its legal practices — Together as One.

Our firms share a common purpose: to exchange knowledge, capability, and opportunity in ways that strengthen their businesses and make the world smaller for the clients they serve. This collaboration is underpinned by a spirit of openness and mutual respect, enabling each to broaden its experience, deepen its expertise, and achieve greater success through coordinated international collaboration.

Alliott Global Alliance continues to expand its global presence, with a strategic objective to be represented in 120 countries by 2030. A limited number of affiliation opportunities remain available for independent professional firms that meet our standards and share our commitment to international collaboration. For further information, please contact membership@alliottglobal.com