Is AI the answer to compliance? The Practice Manager’s view

“Is it AI or automation?”

First things first. ‘Real’ or ‘True’ AI is artificial intelligence that can learn, adapt, make judgement calls, create new rules, and make decisions in unfamiliar scenarios. That isn’t here yet, but it’s coming

However, aspects of AI (especially advanced automation) are already in wide use, and they’re a gift to law firms, especially in compliance. Compliance is about following sets of rules, and rules are bread and butter for machine intelligence. One firm at our recent Practice Managers’ breakfast reported how using AI has relieved a huge administrative burden in areas of their firm and allowed them to take on more fee-earning work.

“Can AI really be trusted to support compliance?”

When you have a programme that’s capable of working faster than a human, for longer than a human, and more accurately than a human, then there’s enormous potential. Add to that the ability to match or at least approach human-level decision-making, then the compliance possibilities seem endless.

The savings in time and money are the most obvious. It could free people to concentrate on the creative and empathetic work in the firm, the real ‘art’ of law or practice management, while routine compliance tasks are managed by AI.

“AI itself must be compliant: how can we be vigilant to bias?”

On the one hand, AI will be helping firms stay compliant, and on the other, it will create compliance demands of its own – with bias potentially posing a serious threat to effective compliance. As Paul Hobden, Insight Legal’s Managing Director, commented in a recent interview: “While I believe many of the current compliance requirements will be easier to manage and easier to audit, AI is going to throw up an array of new ethical and compliance issues. For instance, who audits your AI algorithm for bias?’

We train AI on data, and data is something we associate with objectivity, but data sets can easily contain human bias. AI could easily scale up that bias through machine learning, and very quickly the firm could commit serious discrimination through its AI. One of the most famous examples of that is when Amazon had to shut down its AI recruitment tool because it developed a prejudice against women.

“Will over-reliance on AI cause people within our firm to become complacent with compliance?”

If AI-powered compliance becomes the norm, there’s a danger that it becomes a crutch that firms lean on too heavily. It will be extremely powerful, but even when it reaches the height of its powers (whatever that may be), it needs to be treated as a tool, not an authority.

ChatGPT might be considered the ‘face’ of AI at the moment. There’s even a law-specific version. Chatbots have stirred a lot of excitement as well as fear — ChatGPT’s own creator wants it regulated for our protection.

Aside from future existential threats, the immediate risk is that a solicitor could ask AI to prepare a document or perform compliance work, see the result and feel that it was acceptable, and not check it thoroughly enough. That is a timebomb for compliance.