The potential – and possible pitfalls – of using artificial intelligence (AI) in arbitration and dispute resolution work is a hot topic. The subject was a talking point during Dubai Arbitration Week in November, where HKA hosted a panel discussion: Unleashing the AI Expert Superpower.

The theme highlighted how AI could assist legal and dispute experts, particularly in freeing their time by doing the hard work of scrutinising casework documents. During the discussion, a panel of technology and dispute resolution specialists provided a live demonstration that showcased the transformative power of AI and machine-learning Large Language Models (LLMs).

While AI can improve efficiency by handling massive volumes of data, the experts agreed that one thing is beyond its capabilities. It cannot replace human analysis and expert opinion. Rather, AI is a tool that needs to be tamed and molded, an enabler for better decision-making.

A superpower is born

Clare Lavin is a Partner in HKA’s Dubai-based EMEA forensic accounting and commercial damages team, which specialises in disputes and investigations work, and she moderated the panel discussion.

Setting the context of the debate, she said:

“AI is becoming a powerful tool that is reshaping many industry sectors, economies and societies. It has the potential to solve complex problems and automate tasks. The question is, how do we superpower our experts to bring the best of them to our projects? And how, in turn, can this assist lawyers in their strategies and case management?”

David Blayney’s dual roles as a KC at Serle Court and also CEO of Associo, a collaborative software platform, give him insight into both aspects of the challenge. His interest in AI is how it can assist practically in day-to-day casework through better data management.

“I’m looking at the problem of collecting evidence, managing teams and avoiding the duplication of work when preparing cases,” he said. “When ChatGPT arrived, it opened up exciting possibilities. One thing AI is fantastic at is translation from one format to another, so you can use your software tools together more effectively. We developed our own software platform, Associo, that acts as a central hub where analysis can be worked up.”

Use of AI in construction dispute arbitration

Arbitration in construction disputes is fertile ground for deploying AI, believes Alison Eslick, a Senior Associate at law firm Reed Smith. She focuses on arbitration in major infrastructure and construction projects and is a self-confessed “skeptical person and a traditional lawyer who never cuts corners.”

However, Alison is a cautious convert to AI, despite the challenges the fledgling technology poses to the legal profession.

“I have to admit that when I started using ChatGPT I realised that AI was a tremendous opportunity for us to manage our cases more efficiently and to focus on the matters that count.”

Her firm has completed trials and already is using the ‘public’ ChatGPT along with a private, secure AI product. A key point she raised during the discussion was about transparency in how AI is implemented and the importance of ensuring accuracy.

“If we use AI for clients, we need their consent and must record it on our timesheets, so we are transparent about it. We check every output and at this point we are being very cautious,” she said.

Transparency about using AI

Darren Mullins, a Partner at HKA who specialises in financial crimes, cyber investigations, digital evidence recovery and data analytics, agreed on the importance of transparency.

“You have to be able to say where AI is being used and what quality control processes you undertake – and explain this succinctly to an arbitrator or judge. The integrity of information has to be maintained,” he said.

Increasing the application of AI in arbitration work will probably mean a shift in traditional corporate team set-ups, he predicts, with tech experts working more closely alongside their lawyer colleagues.

“Tech teams are no longer the ‘nerds down the corridor’,” Darren pointed out. “A collaborative approach between teams helps to ensure the questioning of the information and the quality control methodology is correct. At the same time, lawyers and experts are going to have to upskill on how AI is utilised.”

Otherwise, as he pointed out, AI risks going the way of many other software tools and becoming unfit for purpose. “We need to put a leash on AI a little. If left to run, it could bolt and be hard to keep a check on.”

By helping to improve time efficiency and accuracy through rapid and detailed data analysis, AI is a new power for experts. A superpower, if you like. Dimitrios Tousiakis, a Partner at HKA who specialises in construction disputes relevant to delay, explained further:

“AI will assist experts in making more informed decisions, especially in construction disputes where the amount of data is huge. AI can also help to reduce human error and human bias. Having a tool to automate processes will save a lot of time, but you have to do it carefully and focus on value-added tasks.”

The positive powers of AI could be multiplied when combined with other technologies, as Dimitrios explained:

“The construction industry uses well-established Building Information Modelling (BIM) software to manage projects. Imagine how BIM and AI can combine – the potential is impressive.”

Finding the right tools for the job

During the session, the panel demonstrated AI platforms in action. One version was an AI provider Merlin being trialled by HKA. Darren explained:

“We can maintain data securely, with encryption keys and secure access. It brings generative AI and makes use of LLMs. This is enabling us to ‘superpower’ our experts and lawyers. We are already beginning to see the benefits.”

The panel demonstrated how Merlin’s “Discovery Partner” and Associo can combine to break down a case into various issues and provide an ‘end-to-end process’. The real-life example they chose to be scrutinised with AI involved a delayed high-rise residential building project.

“The project experienced multiple issues, including on design, approvals and with subcontractors,” said Dimitrios. “For this example, we uploaded about 1,000 documents from the project to carry out analysis knowing what the issues were, so we can see how AI could assist. One aim was to gain a brief overview of the claim.”

The audience was shown how the most relevant documents could be identified in answer to the question: ‘Summarise contractor extension of time claims and list delay of events claims.’

“You can find the impact of time delays on the contract. In addition, next to every sentence that AI generates are references, so you can trace the sources of the information,” continued Dimitrios.

However, he stressed that AI-generated answers won’t necessarily provide a complete picture.

“It is definitely not everything, but it is a good start. It allows you to ask detailed questions regarding certain issues and you can tell AI to focus on certain events.”

Alison considered whether there is an ethical requirement for arbitration lawyers to disclose their use of AI in preparing legal submissions. Referring to the Silicon Valley Guidelines on the Use of AI in Arbitration, Alison explained that legal counsel is generally not obliged to disclose their AI use to the arbitral tribunal or opposing counsel:

“If I used AI to prepare my submissions, just as I would use LexisNexis to look up cases or go into a physical library to look up a textbook. I wouldn’t necessarily have to disclose my methodology in preparing my submissions.”

Darren offered this analysis of AI capabilities: “If you have the depth of information, you will get a good quality answer. Otherwise, the answers you get back could be very generic.”

Can we trust what AI creates?

Darren answered this vital question: “It won’t give you everything. At the end of the day, a human must look at the evidence to make sure all the information is relevant. You could say AI currently is an ‘80%’ tool.”

Dimitrios concurred: “You should verify that it is checked in an independent and neutral way. Use AI like an assistant.”

David added: “Imagine a case where you have thousands or even millions of documents. You can search according to topic chronology and AI can form an understanding of what is in those documents in a way that’s relevant to the question being asked of it.”

Fine-tuning the AI focus

The live example highlighted how AI-enhanced processes can improve productivity and accuracy for Experts. As Dimitrios demonstrated, you can ask AI to focus only on documents that refer to one particular event, in this instance it was delay events related to design and procurement of the doors of the building project. 

This narrowed the information down to 200 documents and provided a detailed chronology.

An audience member asked: “I think it’s really important that as lawyers, we don’t miss anything. We would rather be over-inclusive than miss something. I’m wondering whether your live example is just for retrieval – is the mapping of claims handled using AI as well or does the user have to link documents to policies?”

David answered: “The mapping can potentially be done with AI. But we still see that as involving a human judgment as to what evidence is best.”

Output depends on the quality of the input

The panel emphasised that the output you can generate with AI is likely to be only as good as the input. To an extent, that comes down to the documents you scan in and how you interrogate them.

Darren said: “We’re currently running a case that has a significant number of date-stamp documents and hand-annotated notes. You have to understand where there are lines made by people through parts of documents. From our use of Merlin, we are getting strong analysis back. I know that the Merlin team is working on a proof of concept with a US-based health department, where documents contain doctor’s handwritten notes.”

He added that his approach to LLMs is to be open-minded as it is expected to handle multiple legal cases across many industry sectors, where there can be a huge number of different data sets.

The panel concluded by agreeing that the potential of AI in the disputes arena has to be tempered with careful control and transparency in its usage. There was a sense that now is the time to seize the initiative, so AI becomes a superpower for good in disputes and arbitration work.

Interested in finding out more about how experts are using AI in dispute resolution? Use the contact button below to get in touch.

---

Connect with Clare Lavin, Darren Mullins, and Dimitrios Tousiakis directly.

For further details of HKA’s claims and dispute resolution service worldwide please visit: https://www.hka.com/services/claims/