Unlock the Editor’s Digest for free
Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.
Time is not on the side of the billable hour. Making lawyers account for every chunk of their working lives is not entirely healthy. But advances in generative artificial intelligence look set to put pressure on the legal profession’s traditional charging structure anyway. The disruption could be painful.
For now, most lawyers say they are optimistic about the impact on margins and revenues of adopting technology that can summarise, search, draft and review. Many are hoping for productivity gains that will help them do more — and sell more — billable work.
But there will be losers too. A third of UK law firms polled by PwC think generative AI will have a negative impact on profits and margins. AI might reduce one of the biggest firms’ competitive advantages, namely having large numbers of junior staff who can be deployed on large, complex lawsuits and transactions.
A bigger problem arises from law firms’ traditional reliance on junior staff to generate disproportionate profits. The gap between their pay and partners’ is bigger than the differential in charge-out rates. Given that AI is likely to reduce non-partners’ workload, it is also likely to erode profit margins.
In response, firms might change their pricing models. Levying flat fees will become more common, even if the most complex work is still billed by the hour. Fees focused on value could be stickier than those based on time. But realistically, clients will expect at least some of the AI savings to be passed on.
The providers of legal AI tools also stand to benefit, especially those that train their models on authoritative legal sources. Companies like Thomson Reuters and Relx have invested heavily in AI-assisted legal research. Relx has gained more pricing power after spending more than $1bn on its tools over the past decade. AI is the main driver of faster underlying growth in the legal division, which analysts expect to increase by 1 percentage point to nearly 7 per cent this year.
Not everyone is convinced. Researchers at Stanford University found that even bespoke legal AI tools hallucinated an “alarming amount of the time”. As well as factual errors and inaccurate sourcing, the answers tended to agree with the user’s incorrect assumptions. Sycophancy poses unique risk in legal settings, the researchers said.
Such concerns might reinforce the instincts of technologically-conservative partners. But there are risks to sitting on the sidelines. In principle, few industries are more exposed to AI than legal services. Firms that make it work stand to win business from those that do not.
vanessa.houlder@ft.com
