Artificial intelligence is no longer a theoretical discussion within litigation. It is already being used across the lifecycle of a dispute, from initial case assessment through to disclosure, drafting and strategy. The pace of its development is such that the real issue is no longer whether it will be used, but how far its use should go.
That question was brought into sharp focus for me recently when I attended the AWS Summit hosted by Amazon Web Services. What stood out was not simply the scale of innovation, but how quickly AI tools are becoming embedded into professional services, including law.
The tools on display were, in many respects, remarkable. Generic AI systems are now capable of reviewing and summarising large volumes of documentation in seconds. More striking, however, were the tools specifically tailored for legal use. These were able to analyse legal documents, identify weaknesses in arguments, suggest alternative approaches, and highlight potential risks in a way that, until recently, would have required significant fee-earner time.
From a purely operational perspective, the attraction is obvious. The potential to reduce time spent on document-heavy tasks is significant. In large-scale litigation, particularly where disclosure runs into thousands of documents, the efficiency gains could be considerable.
However, my immediate reaction was not simply one of enthusiasm. It was a very practical question: what does this actually mean for the client?
Assuming that the use of such tools is compliant with data protection obligations- and it must be said that the systems demonstrated were positioned as fully GDPR-compliant, there remains a more fundamental issue. How would a client feel if they knew that parts of their case- documents, arguments, even analysis- were being handled by AI rather than by the solicitor they had instructed?
There is something about that which does not sit entirely comfortably.
If a task can be carried out by a human, should it be carried out by a computer? That question goes directly to the heart of what clients are paying for. Litigation is not simply a process of analysing documents and applying legal principles. It involves judgment, nuance and an understanding of context which clients expect to come from a human adviser.
This is not simply a theoretical concern. The courts themselves have already had to address the consequences of improper reliance on AI.
In R (on the application of Ayinde) v Haringey LBC, the court made clear that those using artificial intelligence for legal research carry a professional duty to verify its accuracy. The judge stated that anyone relying on AI-generated research must check it against authoritative sources. This serves as an important reminder that responsibility cannot be delegated, whether to technology or otherwise.
Importantly, this is not just a lesson for lawyers. It extends to members of the public who may seek to represent themselves using AI tools. The increasing accessibility of such tools creates a risk that individuals may place undue reliance on outputs which appear authoritative but are not, in fact, reliable.
At the same time, the direction of travel within the justice system itself is clear. Despite these risks, the judiciary and government are actively progressing the use of AI within the courts.
Evidence given by Baroness Carr to the House of Commons Justice Select Committee in November 2025 confirmed that every judicial office holder in England and Wales now has access to Microsoft Co-Pilot Chat. It was also noted that hundreds of senior judges have access to more advanced AI functionality, and that work is underway with His Majesty’s Courts and Tribunals Service to develop a bespoke, cross-jurisdictional AI tool for the judiciary.
The position is therefore not one of resistance, but of adoption, albeit with caution. This raises a wider and more difficult question. Are we entering a phase of “man versus machine” within the legal profession, and if so, who is likely to prevail? Or is this a contest that we are not realistically in a position to resist?
Based on my own experience with clients, the answer is not uniform. A significant proportion would, I suspect, take issue with the idea that AI is being used in place of human input, particularly in sensitive or high value matters. There is an expectation of personal engagement and careful consideration which may not be easily reconciled with automated processes.
At the same time, there is a growing minority of clients who would expect their advisers to be using such tools. For those clients, efficiency and cost are key considerations, and the failure to adopt available technology may be viewed as a weakness rather than a strength.
The reality is that AI is not limited to document review or drafting. Its application extends into the operational side of legal practice, where it may offer more immediate and less controversial benefits. For example, AI-driven prompts within case management systems can assist with billing, ensuring that time is recorded accurately and consistently. Automated communication tools can be used to send updates and chase outstanding information, improving efficiency without compromising the substance of legal advice.
In those areas, the benefit to the client is clearer. Improved systems, better organisation and more efficient processes should translate into a smoother experience and, potentially, lower costs.
However, when it comes to the substantive handling of a case, analysing evidence, developing arguments and advising on strategy, the position is less settled. The question is not whether AI can assist, but whether it should take a leading role.
For clients, this is where the distinction really matters. At GSP Law, our position is clear.
Owning the client journey means that your matter is handled, end to end, by experienced lawyers who take responsibility for every stage of it. It is not fragmented, and it is not delegated to automated systems where judgment is required.
We have lawyers within the firm whose experience is built over years of actually running cases, attending court, dealing with judges, and navigating the realities of litigation as it unfolds in practice rather than in theory. That experience is not something that can be easily replicated or input into an AI system.
There are countless factors in litigation which sit outside the confines of any document or dataset. The approach of a particular judge, the dynamics between parties, the subtle shifts in position that occur during a case, and the practical realities of how disputes are resolved in court all play a role. These are not variables that can be easily quantified or fed into a system.
Our experience is shaped not only by legal knowledge, but by an understanding of people, risk and context. It is developed through doing the work, not simulating it. That is not something that can be transferred to a machine.
The debate has now moved well beyond internal law firm discussions and into the civil justice system itself.
In 2026, the Civil Justice Council published its consultation on the use of AI in preparing court documents. The consultation considers whether formal procedural rules should be introduced governing the use of AI across key stages of litigation, including statements of case, witness statements, disclosure exercises and expert reports.
A central concern running through the consultation is transparency. The issue is no longer simply whether AI can be used, but whether its use should be disclosed to the court, to opponents, and potentially to clients. Particular concern arises in relation to witness evidence and expert reports, where authenticity, independence and accuracy are critical.
The response from the The Law Society is particularly notable. While recognising the role that AI can legitimately play in legal practice, it has advocated for mandatory but proportionate declarations of material AI use across statements of case, witness statements, disclosure lists and expert reports.
In practical terms, this signals a shift in approach. Lawyers may soon be required not only to supervise AI carefully, but to be transparent about its use in the preparation of documents placed before the court.
This approach also aligns with the position of the Solicitors Regulation Authority, which has made it clear that while the use of AI is not prohibited, firms remain fully accountable for the work produced. The SRA has emphasised that solicitors must understand any technology they use, ensure that outputs are accurate, and maintain confidentiality at all times.
Ultimately, the position in 2026 is one of cautious progression. AI is already influencing litigation, and its role will continue to expand, including within the courts themselves. However, its use brings with it a corresponding increase in responsibility.
At GSP Law, our view is measured. We will use technology where it genuinely improves the service we provide. However, we do not see it as a substitute for the skill, judgment and experience that define effective litigation practice. We have a remarkably talented team, and it is that depth of expertise, applied carefully and personally to each case. That remains at the core of what we do.
For now, the jury is still out. What is clear, however, is that clients are entitled to know how their case is being handled, and to expect that the people advising them are applying real judgment, not simply relying on technology.
