AI in arbitration and its effect on juniors in law firms: Lawyers reflect in SIAC symposium panel

Artificial Intelligence (AI) can assist arbitration tribunals with administrative and other tasks, but should not be used to make the ultimate decision in a dispute said Oliver Armas, the Global Head of International Arbitration at Hogan Lovells.

Armas said that the use of AI by tribunals was already becoming a reality and that the profession needed clearer boundaries around what could appropriately be delegated to technology.

“I do think it's absolutely appropriate for a tribunal to use AI for everything except literally the decision-making. I mean, at the end of the day, I think that's what we're paid for as arbitrators,” he said.

Armas was speaking at Young Singapore International Arbitration Centre (YSIAC) session titled Who Moved My Cheese? How Legal Tech is Changing Your Arbitration Practice.

The moderator for the session, Member of the YSIAC Council and White & Case Singapore Partner Aditya Singh, asked Armas where he would draw the line between permissible administrative assistance and substantive decision-making.

Armas said that tribunals were increasingly addressing AI use through procedural orders.

"Our procedural orders have AI provisions in them in terms of what isn't, what's allowed, what's not allowed, what has to be disclosed, what can't be disclosed,” explained Armas.

Michele Sonen, Head of APAC & MENA at Jus Mundi, said that tribunals should disclose their use of AI. However, she was also cautious about making such disclosure a mandatory institutional rule. She said parties should be able to discuss permissible AI use with the tribunal, with institutions potentially providing model clauses or guidance.

“I would want parties to be aware of the data security. That's the more critical thing. If you're going to write down in the PO (procedural order) and you're talking about AI, you have to make sure you include provisions to make sure that the data that you're providing, the submissions, the evidence, the emails, and all of that remains confidential,” she said.

Alexander Fenner, Regional Counsel for Southeast Asia and India at the Boeing Company, took a stronger view, saying he saw little benefit in a tribunal's use of AI without informing the parties.

Ng Jern-Fei KC, Advocate at Duxton Hill Chambers, said that the underlying issue was one of delegation: whether a task properly reserved for an arbitrator had been handed to someone or something else.

“The real concern here is one of agency. In other words, have you delegated tasks which are properly reserved to an arbitrator to someone else, whether artificial or human, to perform? And so what are the limits of the delegation, if at all there are any?”

The panel also highlighted growing pressure from clients on law firms to demonstrate tangible benefits from AI rather than simply adopting new technology. Fenner pointed out that clients were effectively “pushing and pulling” law firms on AI.

“We want legal services to cost less. But we also want work to be conducted responsibly. We are inherently conservative. We're not in a position internally where if something goes wrong on a matter and the problem originates with the use of AI, that's a good excuse for us in the same way that it's not a good excuse for outside counsel.”

Fenner stated that he would judge a law firm's use of AI primarily by the result, rather than by the sophistication of the technology it employed.

Armas observed that some clients were already making AI adoption an explicit requirement when selecting external counsel. He also noted that some clients were using AI to test the work produced by their external lawyers, creating a new dynamic in the lawyer-client relationship.

Aditya Singh then asked Armas who would bear responsibility if AI-generated work went wrong while lawyers remained accountable for it.

Armas said,

“If you (the client) want me to use AI for this aspect of the work and you don't want any attorney to review any of it, okay, who is going to be held accountable for that? Who carries that malpractice risk?”

He said that firms were not being held back from AI simply by the traditional billable-hour model. Rather, risk and the technology's newness remained significant obstacles.

Sonen noted that AI was already changing arbitration research. She said it was allowing lawyers to search across jurisdictions and languages and conduct tasks more quickly. She also identified arbitrator research as an area where AI could significantly alter established practice by allowing lawyers to examine an arbitrator's past cases and track record rather than relying primarily on word of mouth and personal experience.

Jern-Fei observed that technology could particularly help smaller practices compete with larger firms by reducing the resources required for labour-intensive work.

The panel also examined whether AI's ability to perform routine legal tasks could deprive junior lawyers of opportunities to develop core advocacy skills.

Singh asked Jern-Fei what junior advocates stood to gain or lose if AI began performing the “first cut” of tasks such as reading the record, constructing chronologies, testing evidence and preparing cross-examination.

Jern-Fei opined,

“What the advocate gains is time. What they lose is the opportunity to do perhaps tasks which are more mundane. But I don't think it's a dichotomy that means the more we use AI, the fewer opportunities there are for lawyers, in particular younger lawyers, to sharpen their skills. It just means that they can elevate to do things which are more meaningful, which are different, more substantive-driven."

Armas warned that firms would have to compensate for the loss of learning that traditionally came from routine tasks.

Fenner similarly cautioned,

“I think there's a risk that people, if they don't use AI the right way when they're a junior, could short-circuit that process and not have that experience of taking ownership for an issue as the person who understands it."