The "Should Lawyers Use AI?" Debate Is Giving Way to a More Important Question

Here's the First Story 

On 14 May, at Wandsworth County Court, freelancer, Tamires Camal Taquidir won a £7,000 debt claim she'd nearly abandoned. She'd done some work and hadn't been paid, but like many people owed money through a contract dispute, she faced an uncomfortable reality: pursuing the debt could cost almost as much, if not more, in stress and fees as the debt itself.

She decided to use Garfield AI, the first SRA-authorised AI law firm. The platform helped prepare the claim, manage disclosure, organise witness statements and assemble the trial bundle.

The opposing side arrived with both a solicitor and a barrister. The dispute proceeded to a full three-hour trial.

Taquidir won. The counterclaim was dismissed, and her cost to recover the £7,000 was reportedly around £400.

But here's the important part. AI did not argue the case. Garfield AI instructed a human barrister to show up in court and do the advocacy. What the AI did was clear the paperwork mountain so that a qualified human could get on with the lawyering. The legal judgment, strategy and courtroom advocacy remained firmly in human hands.

Remember that.  

Now Consider the Other Story

Around the same time, in Rodney v Gee'z Micro Bar & Pitstop, a judge referred two solicitors to the Solicitors Regulation Authority (SRA), after court filings included case authorities that did not exist. The citations had apparently been generated by AI, and one of the documents containing those citations had been submitted under a signed statement of truth.

The judge was unequivocal: "That is inexcusable on the part of a professionally qualified lawyer."

And it wasn't an isolated incident.

Days later, international law firm Pinsent Masons self-referred to the SRA after identifying a similar issue involving AI-generated legal authorities.

Judges have increasingly warned that fabricated citations pose a genuine risk to the administration of justice.

One Success. One Failure. The Same Underlying Lesson. 

It's tempting to read these stories as opposites. AI appears to be the hero in one and clearly a villain in the other. But look more closely and they're really the same story.

In the successful case, AI handled the repetitive, procedural work, and a qualified human owned every judgment that reached the court. In the failed cases, AI-generated content appears to have been relied upon without adequate human verification before being presented as fact.

The difference wasn't the technology. It was the how.

That's the shift worth sitting with. For years, the legal profession has debated whether lawyers should use AI at all. But that question is becoming less useful. Litigants are using it. Opposing counsel are using it. Law firms are using it to review documents, draft content and summarise information.

Like it or not, the technology is already part of legal practice.

That doesn't mean the risks have been solved. AI can hallucinate. It can miss context. It can produce confident answers that happen to be completely wrong.

Which makes the real question quieter and harder: where does AI do the work, and where does a human take responsibility for it?

Get that balance right, and you get Tamires's story. Get it wrong, and you get a referral to the regulator.

If we're still asking whether lawyers should use AI, we're asking the wrong question.

The more important question is how.

And, more specifically, where responsibility remains unmistakably human.