Judiciary gives AI cautious thumbs-up (but draws line at robot judges)

Promotes ‘strict safeguards and robust testing’

AI tools will be used to support judges, but the judiciary must stop short of fully AI-driven decision-making, according to a new position statement from the lead judge for artificial intelligence, Sir Colin Birss.

The new cautious but positive approach aims to use AI consistently “with the judiciary’s overarching obligation to protect the integrity of the administration of justice and to uphold the rule of law”. Birss, who is the incoming Master of the Rolls and president of the new Business and Property Division, noted that “AI tools must not determine the outcome of a case or displace judicial reasoning, analysis or discretion”, while acknowledging that “AI offers potentially transformative implications for justice.”

The guidance comes after Geoffrey Vos, the outgoing Master of the Rolls, said that the public is likely to accept AI-made decisions in time, and a raft of cases on AI use both in and out of the courtroom. Birss was himself the first known judge to write part of a judgement using an AI tool (ChatGPT) in 2023, calling it “jolly useful”.

He added:

“I’m taking full personal responsibility for what I put in my judgment, I am not trying to give the responsibility to somebody else. All it did was a task which I was about to do and which I knew the answer and could recognise an answer as being acceptable.”

The new position statement follows judicial guidance first issued in December 2023, which established core principles for judges using AI, provided guidance on bias and hallucinations, and reinforced the responsibility of judges for material they put their name to. Pilot schemes have explored a range of functions including transcription, document analysis, and ‘leadership tasks’, the results from which have been integrated into the guidance.

The lead AI judge is to consider any new cases, products, or pilots as part of a centralised steering group, which “will be the sole route for oversight of the development, piloting and deployment of AI tools and a route for escalation should concerns arise”.

The document emphasises that “human oversight is essential” and “every AI output used in judicial work must be reviewed, verified and approved by the responsible judicial office holder.” It adds that “the judicial office holder remains fully accountable.”