Idea of AI judges ‘intuitively troubling’, says AG
Attorney General Rossa Fanning SC has told a conference in Dublin that the idea of AI judges is “intuitively troubling”.
He was speaking at the annual dispute-resolution conference hosted by business-law firm Mason Hayes & Curran in Dublin today (30 September).
The AG cited a report from the Law Society of England and Wales earlier this year that outlined three possible scenarios for agentic AI and the legal profession:
- A. The technology remains confined to administrative tasks,
- B. It takes care of most routine client interactions, and
- C. Agentic AI systems are recognised as formal decision-makers.
Fanning described scenarios A and B as “reasonable”, but said that the third scenario was “alarming”.
Principles
While there might be room for AI to support judges’ decision-making, he told the event, “an AI system cannot easily be held accountable for the decisions it makes”.
“The general allocation of judicial authority to AI would compromise the principles of justice and fair procedures that form the bedrock of our legal system,” he stated.
The AG referred to last weekend’s judicial decision in Belfast on a parade on the Garvaghy Road in Portadown, saying that it was always going to be a contentious decision.
“But it is hard to see how a decision of that nature can ever simply be made by inputting data into a machine and awaiting an answer; difficult cases like that often require calibrated judgment, rather than the rigid application of rules favoured by an algorithm,” he said.
While acknowledging that not every dispute must be decided by a judge, he pointed out that alternative methods of dispute resolution all depended on “a wider framework of public confidence in the rule of law and in the fairness, independence, and authority of the courts”.
‘Legal reasoning without legal thought’
The AG told the conference that AI’s “fluency should not be mistaken for understanding”, adding that the technology could “reproduce the appearance of legal reasoning without engaging in legal thought”.
“AI lacks situated social intelligence; it possesses neither intention nor professional responsibility; and it cannot be held accountable for the consequences of its actions,” he stated.
Fanning told the event that lawyers and decision-makers should not resist changes in dispute resolution – such as the automation of routine processes and the integration of AI into their work – but he warned that “innovation is not an end in itself”.
‘Redefine justice’ risk
The AG told the conference that disputes were inevitable, but added that the institutions and procedures through which we resolved them were not.
“But however radically the procedures change, the underlying principles that grant them legitimacy – fairness, independence, accountability, and human judgment – must endure.
“The greatest risk posed by AI may not be that it will get the law wrong; the greatest risk is that it may tempt us to redefine justice,” the AG concluded.