The Law Society told us it holds no independent measure of what AI saves solicitors. It now plans research of its own

On 2 September this paper wrote to the Law Society of England and Wales with one question. Did it hold or publish any independent measurement of time saved by AI tools in legal practice, as distinct from figures published by software suppliers or surveys of their customers?

It answered within the hour.

"Unfortunately this is all we have," a spokesperson wrote. What it had was research on mid-sized firms looking at how they expect things to change over the next five years, and a survey of Law Gazette readers on how they use AI and what they think of it. The Law Society added, without being asked, that the survey had been sponsored by Canon.

On 9 September it wrote again. The two studies, it said, "were not designed to provide an independent measurement of time savings generated by AI tools in legal practice."

It also told us something we had not asked about. "The Law Society plans to carry out research over the coming year to better understand how AI is used across different parts of the profession."

This paper should have printed both answers before now. On 14 September our editor's note reported that eleven organisations had not replied to our letters. The Law Society was not one of the eleven, and by then it had answered twice. A reader of that note could fairly have come away thinking nobody in legal services had engaged with us. The professional body for solicitors had, promptly and on the record.

What the answers establish is narrow. The body representing solicitors in England and Wales holds no independent figure for how much time AI saves them, and the two pieces of research it does hold were not built to produce one. That fits what this paper reported on 1 September: we had looked for an independent measure of what AI does to the speed of legal work and had not found one, and the three widely quoted figures all came from companies that sell the software.

The Law Society also pushed back on how we framed the question, and it was right to. Its interest, it said, "extends beyond time-saving." It listed quality, risk management, client service and professional practice, and called efficiency "only one aspect of a wider picture." A question about hours invites an answer about hours. A client buying legal work cares about more than that.

Nothing suggests our letter had any bearing on the research plan, and this paper makes no such claim. We asked on 2 September. The plan was mentioned to us on 9 September. That is all the dates show.

What matters now is what the research measures. If it asks firms what they believe AI does, it will join the perception studies already in circulation. If it measures what happens to a piece of legal work, in hours and in cost to the client, it would be the first independent figure of its kind this paper has found in Britain.