TL;DR

Speaking to the ICC (International Chamber of Commerce), Master of the Rolls Sir Geoffrey Vos warned that countries’ different views on whether AI is acceptable could undermine recognition of arbitral awards across borders. The Law Society Gazette reported the speech on 16 October. Vos, who leaves the post at the end of October, said many practitioners are underestimating how far AI will change dispute resolution.

Courts and arbitrators on different tracks

Vos drew a sharp line between litigation and arbitration. In his view, state courts cannot pass decisions to AI unless legislation allows it, whereas arbitrators may use machines in their decision-making in whatever way the parties agree. He expects rising cost and complexity in disputes to push parties towards that option.

He also predicted that litigants will use AI to forecast results. Once both sides have been told what a model expects, he said, they will feel “economic pressure to settle” at that figure instead of waiting for a ruling.

The enforcement problem

The difficulty comes when an award shaped by AI must be enforced abroad. Vos said nations are adopting the technology “at very different speeds”. He pointed to the Shenzhen Intermediate People’s Court, which he visited in September, where AI drafts proposed judgments and reasons for human judges to use, and named Kazakhstan, South Korea, Argentina and Brazil as other countries moving quickly.

The New York Convention, the treaty under which arbitral awards are enforced across borders, lets a state refuse enforcement on public policy grounds. Vos warned that parties in countries regulating AI harshly might argue that enforcing an award where AI had played a significant part would offend that public policy. He called for “an international debate on good practice in the use of AI in judicial and arbitral decision-making”.

Looking forward

The Gazette describes this as Vos’s latest speech on AI and justice, and we have covered his earlier remarks on small claims. In our view, this one carries the argument from domestic courts into international commerce. For UK businesses that use arbitration in cross-border contracts, the practical risk he describes is not that AI produces a bad award, but that a winning award becomes harder to collect in a jurisdiction that distrusts the method. In our view, firms using AI-assisted arbitration should record how the tools were used, so they can answer that challenge if it comes.