© FT Montage/Getty Images

It was the cautionary tale shared across the profession, and beyond. When a New York lawyer used ChatGPT to help with a brief for his case against the Colombian airline Avianca, it provided him with entirely fictitious case law and was immediately called out by opposing counsel and the judge. Hundreds of humiliating headlines later, the lawyer has been fined and the dangers of relying on generative AI have been thrust to the fore.

Such risks may explain why a recent survey by Moore Global, a network of accounting firms, found that law firms were notably less excited than accounting and finance groups to invest in AI, and more likely to say it was a threat to their business.

Human supervision of AI

In professional services — where lawyers, consultants and accountants charge handsomely for their human expertise — firms have, generally, been embracing the latest generation of AI technology but, at the same time, insisting it will be kept within strict guardrails, and under human supervision.

Consultants’ investments in AI


PwC spending on AI over three years


Value of KPMG AI contract with Microsoft over five years


Accenture aggregate spending on AI by 2026

“We are encouraging our people to play with ChatGPT but not to work in ChatGPT,” explains Joe Atkinson, chief products and technology officer at PwC US. For work, the firm is rolling out Microsoft’s enterprise version, which it is calling ChatPwC.

It is one of many consulting firms now competing over how much they are investing in AI, as they try to persuade businesses to hire them for advice on implementing the game-changing technology. PwC recently said it would spend $1bn over three years; KPMG has a $2bn, five-year contract with Microsoft; Accenture announced it will spend $3bn by 2026.

Applications of AI in professional services

AI has become increasingly prevalent across professional services firms over the past decade, in ways large and small.

Investment in machine learning has helped giant consultants more quickly find solutions to evolving business problems — such as supply chain snarl-ups or the impact of climate change.

Auditors have been able to supplement the historical sampling of a client’s financial transactions with analyses of full data sets that allow them to zero in on anomalies.

In law, a plethora of tools exist to speed up the grunt work of reviewing contracts, analysing case law, or hunting for change-of-control provisions in M&A diligence.

And there are everyday efficiencies to be gained from automating back-office tasks — such as IT support, as the accounting firm Grant Thornton has done with a chatbot named after its founder, Alexander Grant.

Gains from the use of AI

Accenture, which runs outsourced IT and customer support services for global clients, credits AI for meeting its targets for relentless efficiency improvement. “As we’re getting to the maturity of automation and [earlier versions of] AI . . . we see generative AI as our ability to continue to give at least that 10 per cent productivity year in and year out,” said Julie Sweet, chief executive, on a recent earnings call. She singled out new tools to speed up the writing of computer code as being of particular benefit.

Enzo Santilli, Grant Thornton’s chief transformation officer, predicted it could be possible to double the amount of business done by the firm over the next five or six years, while increasing staff by only 20 or 25 per cent.

Whether these efficiencies come to pass depends on the outcome of a frenzied period of experimentation inside these firms, that began this year with the rollout of Microsoft tools built on the technology behind ChatGPT.

“We’ve been focused on getting an enterprise version in the hands of our professionals very quickly so there’s a secure alternative to using an open source version,” says Brad Brown, head of KPMG’s tax technology practice. The firm is exploring the use of generative AI to assist staff’s research of complicated tax codes and to summarise and analyse tax return data, while the biggest opportunities lie in using the tools to knit together disparate data systems with a user-friendly interface.

“AIs got to work within a broader ecosystem,” Brown says. “That is where the game is, not just focusing on the AI but focusing on the tools it is integrating with.”

PwC and EY are among the firms using the latest generation of AI technology to work on pitching for new business, populating client requests for proposals with information from a database of past successes. The benefits of cutting the time spent on RFPs will flow straight through to the firms’ profits. “That’s the cost of sales in the professional services space,” says PwC’s Atkinson.

Safeguards on use of AI

Existing codes of conduct should cover most uses of AI, executives say, but will need to be kept under constant review.

Client information should be kept firmly within in-house AI services, for example, but staff might need to be reminded not to go plugging it into public models in a late-night short-cut for a PowerPoint presentation.

There is also a race to hire expertise, and train staff, since AI will be rolled out not just by the professional services firms themselves but also at their clients, posing challenges — for example — in auditing a company’s financial results and processes for firms that cannot keep up.

“This isn’t something to be afraid of, this is something to lean into,” says PwC’s Atkinson. “Everybody is going to be moving really rapidly, and you don’t want to be standing five miles behind.”

Copyright The Financial Times Limited 2024. All rights reserved.
Reuse this content (opens in new window) CommentsJump to comments section

Follow the topics in this article