As disputes lawyers at Trowers & Hamlins we are steadily integrating artificial intelligence (AI) into our workflows, combining cautious optimism over these technological advancements with the discipline expected of legal practice. The goal is not to replace human judgment, but to augment it - leveraging AI where it can add efficiency without compromising ethical or procedural standards.
AI has become increasingly embedded in the broader legal landscape, and dispute resolution is seeing its own wave of adoption. From document review and grammar checks to drafting assistance and legal research, the tools are being tested, increasingly in live environments. However, integration of AI in what we do much remain incremental, reflecting the sector’s sensitivity to due process, risk and client confidentiality.
Enhancing efficiency—not replacing expertise
AI’s value proposition in dispute resolution is clear: faster review of voluminous data, better organisation of case files, and accelerated synthesis of information. This allows us to focus more on complex analysis, legal argument and case strategy.
AI has been in play in dispute resolution in different forms for many years; Predictive Coding / TAR being examples of this. The leap from traditional data-analysis tools to generative AI introduces, however, new possibilities - and also new risks. Unlike prior tools that mined existing material, generative AI can now create content: summaries, outlines, even first-draft documents. That functionality, while powerful, requires stringent oversight. Lawyers must continue to verify AI-generated output. The responsibility for what is submitted remains fully with the practitioner. AI is a tool not the master.
We have all seen the recent horror stories of fabricated citations and legal analysis – these examples underscore the dangers of relying on AI without scrutiny. These incidents are not technological failures; they are failures in judgment and due diligence by the lawyers using the technology.
Indeed, as perverse as it may sound, AI (at least currently) has the potential to increase legal cost to clients if it is not adopted correctly. Clients are now quick to seek to obtain their own 'legal answer' using AI tools and then put these conclusions to their legal counsel. While sometimes useful, AI tools often lack understanding of nuance and the specific complexities of issues in play. Taking time to respond to or debate with clients these quick fix 'AI legal answers' has a knock-on time and cost where clients may be better served in enabling practitioners to focus on using AI tools their the advantage and focus on the legal issues in play.
Three tiers of AI application in dispute resolution
As with any emerging technology, AI use has differing and evolving levels of acceptability for dispute resolution practitioners.
Arguably a useful framework for current AI use in dispute resolution is as follows (although this is open to debate and interpretation):
1. Lower risk / established usage
Administrative or non-substantive applications that support, but do not replace, legal judgment. Examples include:
- Technology-assisted document review
- Chronology building from disclosed documents
- Legal research acceleration
- Calendar coordination and task automation
These uses are widely accepted, provided data security protocols are observed. They increase efficiency without impacting case integrity.
2. Growing usage – caution required
These are applications that may assist with legal drafting or evidence preparation but raise more nuanced ethical or procedural concerns. Examples include:
- Drafting support for witness statements
- Generating preliminary arguments or award/judgment structures
- Analysing procedural history to model similar outcomes
These tools should be used only under close supervision. The practitioner must remain firmly in control of content, reasoning, and compliance with confidentiality and privilege obligations.
3. High-risk
These are applications that go beyond mere assistance and venture into delegating legal reasoning or decision-making to an AI system. Examples include:
- Allowing AI to draft substantive parts of an arbitral award, judgment or determination
- Using open (non-secure) AI systems to producing material containing confidential or client-sensitive materials
- In extremis, delegating dispute resolution entirely to AI agents
These uses are generally considered inappropriate—particularly so in complex commercial disputes.
As the AI landscape changes and develops the framework above will no doubt shift. It is incumbent on practitioners to change and develop as well. Luddites will be left behind.
Client expectations and the road ahead
The pace of technological change is rapid—and accelerating. As AI becomes embedded in legal software and case management tools, its presence will feel less novel and more routine. Clients are already asking how AI can reduce cost and delay. But they are also rightly concerned with confidentiality, reliability, and human oversight.
The future of AI in dispute resolution lies in thoughtful and purposeful integration. We, as lawyers, must continue to own and control the legal narrative, regardless of the tools used to build it. My advice to clients and colleagues is to adopt AI incrementally, apply it responsibly, and ensure that any use supports - rather than supplants - sound legal judgment.
