The virtual witness stand: AI's new role in arbitration advocacy

From data-crunching to dialogue: AI's evolving role in arbitration

Artificial intelligence is no longer a futuristic concept in the legal world; it's a practical tool reshaping international arbitration. While headlines spotlight generative AI, the reality is that AI has supported legal teams for years, streamlining tasks and paving the way for the more advanced applications we see today, enhancing efficiency and allowing lawyers to focus on high-value strategic work.

AI has –for almost 20 years– powered e-discovery and translation. Before the generative AI boom, technology-assisted review (TAR) platforms helped lawyers sift through vast document sets, while specialized Neural Machine Translation (NMT) tools analyzed sentence patterns to translate complex legal texts. While powerful, these technologies were ultimately data-processing tools. They were designed to analyze existing text, not to generate new, dynamic conversations.

A leap in the development of AI has, however, now arrived with generative AI. Powered by large language models, this new wave of AI marks a significant shift from processing data to simulating human interaction. Unlike the narrow AI used in TAR and NMT, modern generative models understand context, generate human-like text, and engage in dynamic conversation. This evolution moves legal technology beyond document-review and translation tasks to offer new tools for advocates to sharpen their hearing preparation and performance.

A new tool for the advocate: AI-powered hearing preparation

A prime example of this shift is the AI-powered witness bot—a platform designed specifically to help lawyers prepare for hearings by simulating witness and expert interactions. These bots provide a structured, on-demand environment for practice, serving as a powerful supplement to traditional, in-person mock trials.

Under the hood: how AI witness bots work

An AI witness bot functions by adopting the persona of a witness or expert based on case-specific materials. The process is straightforward: lawyers upload relevant documents—such as witness statements, expert reports, and transcripts—to a secure, confidential cloud environment. This data forms the AI’s knowledge base, allowing it to construct responses consistent with the designated role. The interaction mimics a real hearing; lawyers can ask questions verbally using speech-to-text technology, and the AI delivers answers in both text and speech in near real-time, creating a dynamic session to test arguments and strategies.

Operational advantages of AI integration

Integrating AI witness bots into hearing preparation offers several key advantages over traditional methods:

• Scalability: An AI bot can be trained quickly, unlike a human participant who needs extensive time to get up to speed. System prompts can also be easily adjusted to match the style of a specific hearing, from the types of questions to the desired length and depth of responses.
• Time and Cost Efficiency: Automating mock examinations can reduce the time and resources spent arranging practice sessions. This frees up team members to focus on strategy and substance, especially when secure AI solutions are available at little to no extra cost.
• Unlimited Availability: AI bots are available 24/7, eliminating the scheduling challenges that come with coordinating multiple people for practice sessions.

While not a direct replacement for the iterative feedback of human mock trials, these AI tools can process large volumes of case materials with speed and precision.

The role of human oversight

Depending on the AI model and how it is customized, the system may occasionally have difficulty with highly technical content or complex arguments (although the same could also be true of a human preparing to be mock-crossed). 

As with other functions by AI, there is a risk that AI witness models can sometimes generate factually incorrect information — a phenomenon often called 'hallucination'. As a result, it is standard practice for legal professionals to verify all output. This can be done continuously by the lawyers most familiar with the case materials. This verification process mirrors what mock examiners do in traditional mock trials and adds to the training effect. To further test a lawyer's preparation, legal teams can also direct the AI bot to include deliberately inaccurate answers to see how well the examining lawyer responds.

The evolving landscape of AI in advocacy

The adoption of AI witness bots could significantly alter international arbitration and complex litigation. By making sophisticated training tools more accessible, these platforms could influence advocacy standards across the board, leading to clearer and more coherent examinations. Ultimately, by helping to level the playing field for hearing preparation, AI agents may influence case outcomes.