Law schools are abuzz with discussions about the opportunities new technologies are offering. If law reflects society, technology will (and should) change the law and how lawyers practise it. But it’s also gradually nibbling away the simpler tasks on which new lawyers cut their teeth. Will robots make better baby lawyers? What types of jobs will law school have to prepare its graduates for? These are questions that law school must confront – and quickly – if it’s going to stay relevant.
Technology and legal practice
There is a lot of information in Australia and overseas about new technology being used in legal practice. In 2017, the New South Wales Law Society published its Future of Law and Innovation in the Profession (FLiP) report. The report extensively catalogued the changes underway in legal practice in Australia, and the pressures driving them.
At the same time, Vicki Waye, Martie-Louise Verreynne and Jane Knowler published the results of an Australia-wide survey of attitudes to innovation that found ‘Australian law firms … rising to the challenge’. The law firms surveyed didn’t claim that it was an attachment to tradition holding them back, but it was the cost, regulation and expertise. Wolters Kluwer got a similar response from European and US firms in a survey it published this year.
But where legal practice is changing is, at the moment, at what Susan Hackett and Karl Chapman christened ‘the bottom end of the licence’:
Work at the bottom of the license is work to cut your teeth on; it is work that repeats; … not strategic, but operational and routine, even if it is complex and takes training and intelligence to perfect; [and] it is likely more fungible in terms of its value.
This type of work is well within the ability of machines but falls short of the IT holy grail of real artificial intelligence. They are the types of tasks that could be reduced to simple decision trees, pattern recognition or an algorithm. A lawyer might need to be trained to do the job, and the law behind it might be complicated. But once they have mastered it, it doesn’t require too much thought.
Chatbots and parking tickets
You probably already know about one of the most famous legal decision tree-based models – DoNotPay. While it calls itself a ‘robot lawyer’ it’s a simple chatbot that helps users write a letter. Its creator wasn’t even a lawyer. It was written by a computer scientist – Josh Browder – to challenge parking tickets. It asks the user a series of questions and fills in the blanks on a template that the user can then send to whichever local authority is responsible.
Similar sorts of chatbots have popped up in different places. You can now even buy a chatbot to write your will from Steam’s gaming platform. An Australian law firm offers a slightly more bespoke version that allows users to write their will with the help of a chatbot under the supervision of a lawyer. Lexi the LawBot will help you write legal documents. And the recently-launched Josef is offering law firms the software to help them write their own chatbots.
These types of solutions aren’t terribly intelligent in their own thinking. Their creators have given them a series of scripts. The script will change depending on the answers the user provides. The script’s author needed experience and training, but once settled its unlikely that a user will ask anything the script didn’t anticipate.
Discovery, due diligence and pattern recognition
More advanced technology can do more than just fill in the blanks. It can begin to recognise patterns in data. The more common a pattern in a data set – like a word or set of words – the more likely it is to be important.
Bernard Marr writing for Forbes, identified different legal tasks that are ripe for this more advanced form of processing. In particular, he identified tasks that require trawling through bundles of documents to find relevant materials. A little like the ‘find’ function on your word processor, you can tell a machine can to look for things that match a set of values. More advanced systems can simply be shown ‘relevant’ material that it analyses itself and applies to the next batch of material.
Obviously, large scale discovery in civil proceedings would be an ideal target for this type of technology. Machines can perform the task faster, are more reliable and cheaper. Demonstrations of the accuracy of relevancy assessments by computers were already being published 15 years ago. If you do a simple Google search today for software to help you with e-discovery, you will be spoiled for choice.
That’s not a bad thing, is it?
Well, no. Not all of it. Especially the development of web-based, free (or cheap) automated legal advice.
Cost and accessibility are two of the most consistent drivers for adopting technology in legal practice. The FLiP report and independent studies like those done by Waye and her colleagues discuss both at length. There is a significant unmet need for legal help in Australia. Although, how much is a little challenging to identify.
We do know that people recognise the role of law and lawyers in resolving disputes. In a report published this year, the Victoria Law Foundation found that participants in their research consistently identify lawyers as important in ‘justiciable’ problems. Importantly, they also found that people who are more vulnerable to others infringing their rights, like women and people with a mental illness, were more aware of the role of law than others.
Substantial problems and legal access
But just because someone knows a lawyer could help doesn’t mean that they can access that help. Victoria Law Foundation found that people from non-English speaking backgrounds were less likely to seek advice. So were people with no access to the internet or with severe mental illness.
In terms of the wider community, the NSW Law and Justice Foundation published one of the most comprehensive studies of need in 2012. The Foundation surveyed more than 20,000 people across Australia. One-half of respondents said they had at least one legal problem in the previous 12 months. There was a high correlation between lack of resources, disadvantage and the number of legal issues a respondent reported.
More recently, the Productivity Commission revisited the Foundation’s data. The Commission found that respondents used lawyers in a little under a quarter of ‘substantial’ legal problems. The reasons for not seeking help were diverse, but a majority of respondents believed that it would make no difference or was too stressful.
The Commission didn’t conclude that not getting help meant there was an unmet legal need. But the Commission drew some tentative links between the differences in the number of ‘substantial’ legal matters, respondents’ decisions not to get help and the Foundation’s findings on disadvantage.
Could ‘robot lawyers’ fill the gap?
Because of the perceived success and accessibility of legal chatbots like DoNotPay, the New South Wales Law Society argues that this might be an answer to a lack of access. That could be true, or not. What’s missing is an assessment of whether there is, or could be, a chatbot or other online system that will help.
The Law and Justice Foundation gave a good hint about the types of problems respondents reported most often. And parking tickets ain’t it. Most respondents reported consumer issues followed by crime and housing problems.
So, technology might provide an answer here. But we need more work on what people actually need, where and when. We also need to make sure that the people who need it can actually use it.
Uber, gigging and outposts
Related to technology taking over tasks and the pressure on lawyers to make their services cheaper is a rise in new models of practice.
In an article published in the Law, Technology and Humans journal, Margaret Thornton explored what she refers to as the ‘Uberisation’ of legal practice. Pressure from clients on lawyers to reduce costs and improve service has seen the emergence of providers that look nothing like the traditional law firm. Eric Chin has christened it ‘NewLaw’.
NewLaw providers might employ lawyers working remotely on short-term, fixed-fee legal tasks (gigs). Other firms might out-source the job to the provider, or other employees might in-source to others. They might also employ lawyers who are outposted to clients, working hand-in-hand with the client’s employees.

As Thornton explains, these types of arrangements are cheaper than traditional law firms. There is no need for wood-panelled offices, glass-topped conference tables or employee breakfast bars. Cost is low, and flexibility (for clients and lawyers) is high. One of the lawyers that Thornton spoke to explained that their NewLaw firm had ‘partners’. But the firm charged them out for less than junior lawyers in traditional firms. And because they were more experienced and worked faster, the cost was substantially cheaper.
While this all sounds wonderful, NewLaw relies heavily on experienced lawyers. NewLaw will not work with lawyers who don’t know what they’re doing. It also won’t work if lawyers bunk off and don’t do the job. As one NewLaw manager explained to Thornton:
Obviously, we can’t send a recently qualified lawyer in to hit the ground running on a short-term stint. We don’t have time or the appetite to train a lawyer—that’s fine. We want them to come to us when they’ve had that training.
The President of the Queensland Law Society, Bill Potts, has also highlighted the problem with the anonymity of the internet. Because clients can’t see the lawyer who is behind the screen, and websites often adopt seductive names, how can we be certain a lawyer is even involved?
The end of days (for new lawyers)?
A lot of the commentary about technology and automation is incredibly enthusiastic. Some commentators have suggested that automation across Australian industries will produce trillions in savings and productivity by 2030. The same sort of enthusiasm attaches to potential changes in legal practice.
But the ‘tectonic shift’ driven by technology that the FLiP report, for example, talks about has knock-on effects.
Where will baby lawyers work?
As Hackett and Chapman explain, the type of work that is ripe for robots is ‘bottom of the licence’. High volume or simple tasks that might require some background training. It’s the type of work ‘to cut your teeth on’. Unsurprisingly then it’s also the kind of work that new lawyers tend to do.
The drive to lower costs has also tended to push this work further down the law firm food chain to casual paralegals. Thornton found that large clients and NewLaw firms occasionally expressed a preference for paralegals to do simple tasks to keep costs down. Worryingly, this has also created a stratification in firms with casual paralegals being employed with dim prospects for promotion. A development Hilary Sommerlad calls a ‘legal precariat’.

Australian law schools are producing about 7,000 graduates each year. Not all of them will go on to be lawyers. And the current prospects for employment are still better than other graduates. But, as Thornton points out, the number of jobs for new lawyers is already shrinking. Deloitte predicts that by 2026, more than 100,000 legal roles will disappear in the UK alone.
So, if technology continues to advance, and new types of practice rely on experienced lawyers, where will jobs for new lawyers come from if robots make better baby lawyers?
One possibility is that the market for law student casual paralegals will become much more significant. In effect, the early career experience of new lawyers would be pushed down into law school. While that superficially sounds great, law students’ experience is tougher. It’s challenging to do well at law school. But adding work to pay rent and participating in law school competitions (which law students believe will help get a job) makes it even harder.
It also assumes that law firms will devote the same time to training paralegals as if they were new lawyers. But if paralegals are there to bring costs down, where is the incentive to train them?
And what should law school tell students about career planning?
What should law schools teach now?
So what should law schools be teaching law students? The substantive law is still important. But some of it might not be relevant.
For example, I sometimes teach civil procedure to law students. If you do a quick, superficial stocktake of your common or garden variety civil procedure course, some things might not be worth teaching any more:
- Chatbots or robot lawyers can help with simple disputes. So, should I get rid of material about less complex proceedings like debt recovery and judgment enforcement and focus on complex litigation? But is any firm going to let a junior lawyer loose on defamation, IP or medical negligence?
- For a society living on smartphones, social media is becoming a better way to serve documents. Should I be teaching how to extract and produce data from social media platforms that can show how and when a possible respondent accesses them for substituted service?
- If document production, like simple pleadings, can be automated, how much material on pleadings should I take out? Should I only keep the most arcane and complex content that a machine couldn’t do?
- We’ve already talked about automated systems for discovery. If the systems are intelligent enough to determine relevance, how long will it take for developers to program rules for privilege? What should I leave in about discovery?
- Four New South Wales jurisdictions now have online facilities for interlocutory applications. The Federal Court actively encourages parties to discuss how technology might be used to speed up proceedings. Simple interlocutory applications are how lots of young lawyers get their first taste of courtroom advocacy. Should I provide more advocacy training to prepare new lawyers to instead handle more complex matters? But what sort of firm is going to let a new lawyer loose on a trial on their first trip to the courts?
Teaching ‘top of the licence’?
You’d be right if you accused me of thinking too small. But the Law Admissions Consultative Committee (LACC) that writes the syllabus for Australian law schools still includes these things. While the LACC is reviewing the syllabus, it’s more likely than not lagging behind the available technology even before its released.
The alternative is to teach what Hackett and Chapman call ‘the top of the licence’. That is, high value and complex skills that machines can’t do. Can law schools provide the ‘teeth cutting’ training that law firms currently offer so that new lawyers can jump the gap from law school to experienced lawyer?
To provide that training, Australian law schools are going to have to look long and hard at what they teach and how they teach it. While admitting authorities and law schools rely on the LACC requirements, they’re locked into an approach to training new lawyers. That approach assumes that nothing will change (or, more accurately, nothing has changed).
It means that law schools must start to redefine their courses to reflect the changes in legal practice. They must do so quickly and in active discussion with the profession and technology providers. Otherwise, robots will make better baby lawyers.
2 Comments on “Robots make better baby lawyers?”