Susskind: AGI is coming
Lawyers are wrong to think that the need for human judgement will protect them from AI taking on most of their role, Professor Richard Susskind has argued.
The renowned legal futurist also predicted that the legal businesses that will thrive in the coming age of artificial general intelligence (AGI) will be those that build legal AI systems, rather than provide the advice themselves.
Professor Susskind has published a 30th anniversary edition of his landmark book, The Future of Law [1], containing a new introductory chapter updating his vision and consolidating 45 years of thinking and writing about the impact of digital technology on law, as well as new commentary for every chapter.
He defined AGI broadly as systems “that are as apparently capable as humans at undertaking all tasks that require our full range of cognitive skills”; systems that match human performance in terms of output.
“In my estimation, the emergence of AGI would be the single-most significant technological advance in the history of humanity,” he wrote. “We would be sharing the planet with entities that are as capable as us. This is an eventuality for which we are comprehensively ill-prepared.”
He continued: “If pinned down to make a concrete prediction, I expect that by the early 2030s generative AI systems that we rightly use today with caution will be reliable for most practical purposes.
“Currently systems will have been augmented or replaced by new techniques, so that most of them will no longer require human supervision. Indeed, to have a human in the loop will more likely diminish than improve performance.”
In the short term, he said, AI in law simply meant automating, improving and optimising traditional legal practice inside the existing business model.
Many of the global ‘elite’ law firms, “in contrast to their press releases, privately view AI as a way of boosting their profits under the old pyramidic structure that has served them well for decades”.
In the long run, however, AI would empower people who were not lawyers to handle their own legal affairs.
“The craftsmanship of the finest lawyers might be sought in critical situations but otherwise the scope for traditional lawyering would be limited,” Professor Susskind went on.
Even family law would be administered “differently, but sensitively, drawing on work in ‘affective computing’ which are systems that can detect and even express human-like emotions”.
Criminal law might be different, he suggested, but only because the involvement of humans “may be regarded as intrinsically important” by society.
For many lawyers, their judgement was the “killer objection”.
“How on earth can a machine ever exercise judgement, I am often asked? Again, this is to get muddled about the difference between process and outcome.
“Judgement is a human faculty or facility. It is a human process. It is a tool that we possess that helps us handle certain kinds of problems…
“It is by using judgement that humans manage uncertainty. But this leads me to conclude therefore that the big question here is not whether AI systems are able to make judgements. This is a little like asking whether submarines can swim or planes can flap their wings.
“The deeper question is whether machines can handle uncertainty. And the answer to that – manifestly and absolutely – is ‘yes’. They do so on the strength of the oceans of data on which they were trained, whereas humans rely on their modest mini-databases of personal experience.
“Viewed in this light, the judgement objection falls away.”
It was “not at all clear what unique value conventional lawyers would contribute” in an age of AGI, he warned.
There were two strategies for lawyers – to compete with AI or to build it. As AI became more capable, the former would be less and less tenable.
“The build strategy urges law firms and lawyers to build the systems that will replace their old ways of working… This is the great opportunity in the legal market.” It was also where the future lawyer jobs would be.
The business model was not charging for time, but licensing systems, services, content and solutions so that clients could undertake much legal work for themselves.
This was not some form of ‘robotic lawyering’, he stressed. “Rather, we should be asking… how can we deliver the legal outcomes that citizens and organisations need, but in entirely new ways that are less painful, costly and time-consuming?”
As an example, Professor Susskind pointed to law as code in autonomous vehicles that meant they could not run red lights.
“The idea that excites me the most is some kind of AI-based advance warning system in major organisations that would alert leaders when there are early signs of legal difficulties”, such as potential breaches of contract, cartel-like activity or exposure to litigation much earlier than is possible today.
“This is problem avoidance displacing problem solving.”
While a few leaders of law firms were facing up to this challenge, he said, “in the long run, my money here in on AI systems developed and delivered by start-ups (many of which do not yet exist) and not the old gatekeepers”.
This was, Professor Susskind insisted, ultimately not a technological argument but a philosophical case on improving how law was used and applied, rather than protecting the legal profession.
He argued that AGI “could reinforce and even extend the rule of law, could crack the global access-to-justice problem and could integrate the law more fully and proportionately in the legal market”.