Getting to grips with artificial intelligence
In almost every corner of the globe, artificial intelligence (AI) is now the name of the game, with a multitude of industries and occupations embracing it with open arms.
The legal profession is no exception and faces complex and diverse challenges as it comes to grips with technology that has the potential, among other things, to create AI lawyers.
In fact, one American law firm already claims to have hired its first AI lawyer to research precedents and make recommendations in bankruptcy law.
So the race is now on to understand the full implications of technology that has profound implications for almost every aspect of life as we know it, especially the law.
In New Zealand, the baton has been taken up by a multi-disciplinary team at Otago University involving the Faculty of Law and the departments of Philosophy and Computer Science. Supported by a $400,000 grant from the Law Foundation, a team led by Associate Professor Colin Gavaghan will spend three years undertaking groundbreaking research into the possible implications of AI innovations for law and public policy in New Zealand.
“It’s pretty much impossible to open a newspaper at the moment without reading something about artificial intelligence,” Dr Gavaghan told LawNews. “If it’s not driverless cars, then it’s the prospect of mass unemployment, or even the threat of robot super-intelligence taking over the world!”
Dr Gavaghan says AI technologies pose fascinating legal, practical and ethical challenges.
“One of our objectives is to get a better sense of what’s actually possible or probable in the near or medium term future. What areas of life will AI most affect? How will it affect them? And how well set-up are we to deal with the changes it will bring?” One such example is “PredPol”, a technology widely used by police to predict where and when crime is most likely to occur.
“It uses algorithms to predict the most likely crime hot-spots in a given city at a given time,” explains Dr Gavaghan. “It’s already being used by a lot of US police forces and trialled by at least one in England. The idea is that it will allow the police to deploy their resources where they can do the most good and the advantages are easy to see.”
“The concern, though, is that this is another step toward ‘predictive policing’ where people are treated not on the basis of what they’ve done, but on what an algorithm predicts they’ll probably do in future.”
He adds that, to some extent, a lot of policing and criminal justice decisions do this already, with bail, sentencing and parole all relying substantially on predictions of future conduct.
“What might be different about the ‘era of the algorithm’ is that these decisions will be made by a programme that’s putatively objective and at the same time inscrutable. There are some very ambitious claims that AI will mean the end of unconscious bias in human decisions, but there’s also evidence that the way they are designed can result in bias being ‘baked in’.
“If an algorithm works by aggregating data from previous arrests or sentences, for instance, then it might be basing its decisions on a data set that itself reflected certain kinds of bias.”
Then there is the question of scrutability, which Dr Gavaghan says poses its own set of problems.
“The problem here is that, as we proceed down the road to machine learning, it might not be possible to know exactly how the algorithm arrived at its decision, which might make it very hard to challenge.
“The Wisconsin Supreme Court recently held that a sentencing algorithm could only be used to assist, and not to replace, a human decision-maker. But, psychologically, it might be difficult for human judges and police to argue with the computer!”
Another aspect of AI that Dr Gavaghan and his team are addressing is the ethical dilemmas posed by such technology. For instance, who would you sue if AI denied you promotion, or even a job in the first place, and is it acceptable for a driverless car to deliberately swerve in a way that saves its own passengers, yet kills a pedestrian?
The latter issue will be of much interest to New Zealanders as the country could soon be a test bed for driverless cars. According to The New Zealand Herald, the government is in talks with an unnamed vehicle manufacturer about bringing driverless cars here for a pilot project.
Associate Transport Minister Tim Macindoe is quoted as saying New Zealand has special advantages that make it attractive for such trials, such as a quirk in our transport laws that means there is no requirement for vehicles to have a driver.
Colin Gavaghan says issues surrounding driverless cars are already being widely debated.
“Yes, everybody loves the driverless car example, and no wonder!” he says. “Real traffic accidents tend to happen too fast to allow for considered ethical judgements, so we tend to cut human drivers some slack if they do what, with hindsight, we decide was the ‘wrong’ thing.”
“But driverless cars will be programmed with their ‘ethical’ rules before they leave the showroom and that allows, for instance, Mercedes to announce that their cars will be programmed to prioritise the lives of the occupants rather than anyone outside the car.
“Is that something we should allow? Would anyone buy a car that wasn’t going to prioritise their safety?” Dr Gavaghan asks. “There are no easy answers to these questions, but it at least makes sense to ask them now before driverless cars become a common feature on our roads, as I’m sure they will.”
Dr Gavaghan says the issue of AI being used to hire and promote people in the workforce is no less problematic.
“Cathy O’Neil in Weapons of Math Destruction gives several examples of people who were hard-done by algorithms in employment decisions. The real challenge is making sure they are open to scrutiny and challenge.
“There have been some noises from, for example, the European Parliament about requiring transparency, but precisely how this might work in practice is something that’s going to need a lot more thought.”
So, given the apparent emergence of AI lawyers in the US and the impact of such technology on the legal profession as a whole, should lawyers and paralegals be worried about their future job prospects?
Dr Gavaghan says no one can predict with any certainty how this will play out.
“There are a lot of very different predictions being made about the impact of AI on jobs, from ‘there will be new jobs to replace the old ones’ to ‘get ready for the post-work future’! But I suspect we’re a bit away from the end of the profession. Earlier this year the International Bar Association estimated the risk of current lawyers losing their jobs to automation at just three to five per cent.
“However, that’s not to say that the profession won’t experience substantial changes, as it’s likely that certain tasks currently performed by humans will be automated, such as the more repetitive tasks involving the least creativity or imagination.”
Dr Gavaghan says the impact of AI on the New Zealand job market is something that his team will be looking at in the second and third years of the research programme.
“If there really are going to be substantial job losses, then initiatives like guaranteed basic income might have to come into play and that could have effects beyond the individual being made redundant. How would ACC, for instance, be funded in a world with a lot less human work?”
Dr Gavaghan believes that, as with every other phase of automation, old jobs could just be replaced with new.
“Deloitte published a report last year which concluded that there’s likely to be a declining demand for ‘traditional’ lawyers, but a growing demand for other roles in law firms, including data and technology experts and lawyers who have come through non-traditional routes.
“I’m guessing, just like everyone else, but I have a suspicion that the ‘double degree’ that a lot of our law students take – in many cases combining law with a high level of science literacy – could position them very well in the job marketplace of tomorrow.”
Asked whether AI could be described as a game changer for the legal profession, Dr Gavaghan said:
“Deloitte definitely think so. Their conclusion was that those firms who keep up, or even innovate, could reap major rewards, but those who fail to do so are largely doomed. I suspect we’re a bit away from AIs replacing all judges, but in areas like out-of-court dispute resolution, they’re already doing it.
“One thing we feel confident in saying at the moment is that it’s vital that New Zealand gets out in front on this technology because the costs of being left behind are, in various ways, pretty unthinkable.”
Many law firms in New Zealand and around the world already see the writing on the wall and are determined not to be left behind. MinterEllisonRuddWatts, for one, has entered into a joint venture with Goat Ventures to explore and capitalise on the application of artificial intelligence for legal services. A market first in New Zealand, the joint venture is a $2million co-investment at pre-seed stage.
The law firm’s CEO, Mike Schubert, told LawNews that businesses are looking for new ways to meet their legal obligations efficiently.
“We believe it’s time to introduce new technology into the legal services market to meet this need and are excited at the prospect of combining our legal knowledge and approach with a team who know how to get the best from evolving technologies.”
Four artificial intelligence PhD holders who have cut their teeth in commercial organisations like Microsoft, Xero and Spark have been taken on to make it happen.
“Like other organisations looking at how to use disruptive technologies to make a step change, MinterEllisonRuddWatts has chosen to start with a blank sheet, away from legacy structures and practices in its core business,” Mr Schubert says.
“Access to Goat’s expertise through the JV will allow us to build on our understanding of AI and look more closely at how we can use it to benefit our clients.”
Just how law firms like MinterEllisonRuddWatts get to grips with AI will be watched with more than passing interest by others in the legal profession.
The question on many minds will be whether the new technology will just augment lawyers or perhaps in the fullness of time, replace them?