Journal

The role of regulation in overseeing how AI operates

A large part of the focus of the recent conference about the future of legal services, held by the Legal Services Board (‘LSB’), was firmly and rightly fixed on the growing importance of artificial intelligence (‘AI’) for legal work. And this is as true, if not more so, for patent and trade mark attorneys as it is for solicitors writing wills or enabling the buying of houses. AI can be brilliant at sorting through and selecting relevant information or contributions from a mass of existing material. It will sometimes get things wrong, and there will have to be a human interface in order to provide a protective check. But its major benefit is that it can sort and sift huge masses of data in the blink of an eye. Poor plodding humans will take years to do the same sort of operation.

We should welcome this rapid and (by and large) efficient set of tools to do much of the basic groundwork for legal services. Sorting through masses of existing patent or trade mark data, for example, can be immeasurably speeded up by AI. Where we need to draw the line, though, is in making sure it’s an expert human who submits a patent or trade mark application, and advises a client on doing so, rather than a machine. The submission of an application requires a degree of skill, and an understanding of nuance, and a collaborative ease in the relationship with a client. AI can’t do these things. There are some who argue that there will come a time when AI will have new ideas, and demonstrate creativity, and display human empathy. I doubt it. AI will always be dealing with what already exists, not with what can be conjured in the future, anew.

What therefore should be the role of regulation in overseeing how AI operates, and whether it is to the benefit of the consumer? That, actually, has to be the acid test. Is this way of producing information to the benefit of consumers, or not? The answer will frequently be yes, and occasionally no. We at IPReg need to make sure that attorneys are equipped with the right knowledge and training and skill, and operate under the right rules, to enable them to make the distinction.

There is one further and dramatic impact that AI will have on the legal world, I suspect. At the moment, the remuneration of an attorney is determined by the number of hours spent working on a particular case. But when AI has reduced the number of hours to minimal levels, how then can clients be properly billed? The answer has to lie in the development of a system of billing by outcome rather than by time. This would probably be a sensible way forward, even if AI were not forcing us to look at this as a fundamental change. Time is after all an input, not an outcome. And surely it must be better to ask a client to pay by what has been achieved, at the end of the day?

Chris Smith, April 2024