this product is unavailable for purchase using a firm account, please log in with a personal account to make this purchase.

Select from any of the filters or enter a search term

I, lawyer the 'trusted adviser' of the future

I, lawyer the 'trusted adviser' of the future

By Sean Kelly and Harrison Nurse


  • Current and potential legal technologies have the capacity to revolutionise daily legal practice, and many legal tasks either have been, or will become, entirely undertaken by legal technologies with minimal human oversight. 
  • The complex nature of some projects and the current limitations inherent in some predictive legal technologies make it less likely that clients will cease relying on the judgment, strategic decision-making and communication skills of lawyers in the foreseeable future.  
  • Accordingly, it is unlikely that current and imminent legal technologies will entirely replace the role of the lawyer, who will retain the mantle of "trusted adviser". 

Lawyers need to understand what legal technologies can achieve in practice, their limitations and their potential impact on legal services and advisers.*

Many megabytes have been used discussing the potential impact of legal technologies on the legal profession. Predictions range from "the end of lawyers"1 to "there's never been a better time to be a grad".2 As with most polarising topics, the most accurate answer lies somewhere between the extremes. 

Commentary to date has generally focused on the sheer number of technology start-up companies, the array of "bots" and "apps" that are offered and the potential impact on the traditional business models adopted by law firms. There are many recurring themes, including:

legal technologies should be embraced, to ensure improvement and efficiencies in delivering client service

old business models are under threat, and the billable hour may soon be obsolete

lawyers should be optimistic about the future as potential technologies are likely to undertake the "boring" work, leaving lawyers to engage in more interesting tasks.

This article discusses what legal technologies can achieve in practice and the potential limitations of these technologies, and canvasses opinion on the potential impact on legal services and legal advisers in the future. The authors conducted a confidential survey of 86 industry participants to gain insight into contemporaneous industry perceptions and expectations of the current and future impacts of technology on legal practice. The results of the survey are presented here. 

Looking forward

What is possible for front-end technologies

Current legal technologies used by front-end and transactional lawyers include:

automatic production of relatively simple contracts based on responses input into a user friendly questionnaire that can be accessed by clients

high-powered document "find" applications that identify drafting infelicities based on bespoke drafting rules created by users and "taught" to the application

use of blockchain technology to construct "smart contracts" capable of self-executing certain basic, binary functions.3

As front-end technologies mature there may be opportunities for multiple, separate technologies to integrate. Optimistic projections include artificial intelligence (AI) being used to draft complex commercial contracts. Document automation technologies, used in conjunction with data-driven decision-making, clause libraries and historical project performance data, may be able to model favourable outcomes and predict the best case risk profiles taking into account the relevant sector and commercial participants. 

While well-known successful applications include basic non-disclosure agreements, such applications are not currently used for complex commercial projects. Some limitations of data-driven legal decision-making on complex commercial projects are discussed later here. In addition, such projected applications can overlook the nuances involved in front-end and transactional legal practice. Some overlooked factors are:

AI learning from past data requires that all projects will be iterative or draw heavily from past projects. Therefore, the implementation of contemporaneous commercial and legal developments will be delayed, limiting the ability of applications to recognise new market trends in real time

data-driven contract drafting would rely on the parties having perfect information, clear drivers and a degree of consistency with previous positions. These factors are rarely present in complex commercial projects before commencement of negotiations. 

What is possible for back-end technologies

Current legal technologies used by dispute resolution lawyers include:

online tribunals that have been rolled-out in various jurisdictions allowing litigants to quickly and efficiently obtain judicial determinations without much (or any) human contact

data-driven applications that are able to act as "robot lawyers" and provide advice and recommended next steps, including draft correspondence for linear disputes (for example, disputing a parking infringement fine)

AI legal research platforms that allow users to ask legal questions using plain English and obtain results listing the most relevant material across primary and secondary legal sources. The platform learns with each search that it runs, and continually improves its accuracy based on the choices made by users by selecting (or ignoring) sources in the results list.

Machine learning has the potential to allow legal technologies to review large volumes of data to identify concepts and patterns and then recommend appropriate actions based on those patterns. Statistical modelling and AI machine learning have already successfully predicted the outcome of judicial decisions. This was demonstrated in a recent study using natural language processing and machine learning to predict decisions of the European Court of Human Rights. By extracting and interpreting the text of judgments that related only to the facts of each case, the law and the arguments presented by the parties, the predictive model correctly predicted 79 per cent of the final decisions.4 

As back-end technologies mature, potential future applications include parties being able to present their cases to an entirely online and artificially intelligent judicial determination process. An entirely online platform may evolve that allows each party to upload all its documentary evidence and its summary of the relevant factual matters supporting its legal claims. Using natural language processing, in the context of machine learning and a sufficiently large database of past, topically relevant cases, an AI judge may provide a judgment, or in the alternative, a non-binding assessment of the likelihood of success for each party. 

This may either form the basis of an early indication of the likelihood of success (much like a skilled mediator can provide), or may provide a quick and relatively cost effective binding judgment that can be appealed to a human tribunal.

Industry perceptions

The authors prepared an online survey of industry perception of how legal services have been and will be affected by new technologies.

The survey was completed by 86 respondents from the infrastructure, construction and related sectors from around Australia and overseas. This industry sector typically involves complex and high-value projects where there is a degree of consistency (broad contract forms) laid over consistently changing physical conditions. The results of this survey may be broadly reflective of industry perceptions of legal technologies applied in other complex commercial fields of enterprise. 

When asked whether new technologies had significantly impacted the way they work in the preceding five years (question 2), there was a relatively even split between "yes" (50 per cent) and "no" (44.19 per cent). A large majority of respondents indicated that existing legal technologies have not reduced their workloads (question 3), split between respondents noticing increased workloads (41.54 per cent) and respondents whose workloads have not been affected (50.77 per cent). However, there is a clear indication that the implementation of existing legal technology has increased the number of work streams and deliverables (72.92 per cent).

When asked to predict how future legal technologies would affect workloads and work streams in the future (question 4), there was a clear perception that over the next one to two years the effect would be "a little" (74.58 per cent). However, this trend reversed when respondents were asked to consider a time period spanning 2023 to 2028, as a large majority predicted that the impact of legal technology would be "a lot" (71.23 per cent).


Figure 1: Predictions of how future workloads and work streams will be affected (question 4)

Respondents then selected options on a spectrum from "yes, in the next 1-2 years" to "no", indicating whether what they considered to be the essential skills of successful legal practitioners would radically change as a result of future legal technologies (question 6).

A relatively large number of respondents perceived that there will be "no" radical change to the essential skills of legal practitioners (27.06 per cent). However, the most popular response was "yes, but in more than five years" (30.59 per cent), followed by "yes, during the next 2-5 years" (28.24 per cent). 



Figure 2: Predictions of if/when essential legal skills will be affected (question 6)

The relatively high number of respondents perceiving that legal technologies would not radically impact the essential skills of legal professionals might have been caused by perceptions of the particular tasks that future legal technologies might accomplish. Respondents were asked to select the time period in which they expected legal technology to be able to accomplish what they considered to be "administrative", "complex but repetitive" and "strategic decision-making" tasks (questions 7, 8 and 9), and some clear trends emerged.

In respect of administrative tasks, a majority of respondents thought that either it was already being accomplished (40.70 per cent) or would be occurring in the next 1 to 2 years (18.60 per cent). However, the most definitive perception among respondents is that legal technology will never be able to accomplish strategic decision-making tasks (72.09 per cent). 


Figure 3: Predictions of when administrative, complex but repetitive and strategic decision-making tasks may be accomplished by technology (questions 7, 8 and 9)

Considering the above indication of industry perceptions, it appears that current legal professionals view strategic decision-making as one of the essential skills of infrastructure and construction lawyers that will never be accomplished by legal technology. 

This view is partially reflected in the responses to the final question (question 10), which asked respondents to predict the extent to which clients would trust legal services or advice generated by legal technology. Respondents were able to select from the following predictions:

Prediction A – clients will trust technology for all legal services and advice

Prediction B – technology will only be trusted to provide legal services and advice not involving strategic decision-making.

Prediction C – clients will differentiate between legal services and legal advice and will never fully trust technology to provide legal advice

Prediction D – clients will trust technology to an extent but it will be used as a guide only, with human lawyers making all final decisions

Prediction E – human lawyers will always remain the only trusted adviser and technology won't be fully trusted

Prediction F – the future is too difficult to predict

Prediction G – don't know.


Figure 4: Predictions of how technology will be implemented by clients (question 10)

Interestingly, Prediction B was only the third most popular response (15.13 per cent), despite expressly focusing on strategic decision-making tasks. Instead, the most popular predictions were that technology would never be trusted to provide legal advice (as distinct from legal services) (28.57 per cent) and that technology would only ever be used as a guide (40.34 per cent). 

This suggests that to many respondents the future of the legal profession is not perceived to be only about what legal technology cannot accomplish (for example, strategic decision-making), but rather, respondents may have considered other positive capabilities of human lawyers that will ensure the lawyer of the future will always have a role to play. It is important to recall the context of this observation – a vast majority of the respondents are lawyers.

Impact on legal services 

The perceptions of respondents to the online survey discussed in the preceding section generally consider that it is likely new legal technologies:

will change both the way lawyers work and relevant professional deliverables "a little bit" in the near future, but there is likely to be "a lot" of change to legal tasks and work streams in the medium term future

will accomplish what may generally be described as administrative tasks (such as transacting binary outcomes during contract administration) and complex but repetitive tasks (such as machine-learning legal research)

is less likely to be able to accomplish what may generally be considered to be strategic decision-making tasks. For example, even (hypothetical) AI judgments will be predictions generated from modelling of past factual scenarios and outcomes, which will require significant strategic consideration by human lawyers of the inputs into the process and "training sets", and assessments of outcome accuracy and viability of appeals.

However, the breadth of responses to the survey questions showed the uncertainty associated with the progression of technology and its potential impact on the future of legal services in the infrastructure, construction and related sectors.

Limitations of potential technologies

The implementation of potential technologies, particularly those based on data analytics and machine learning, are subject to limitations that will restrict their scope of application as a single source of legal services. For example:

the challenge of labelling training data, which often must be done manually and is necessary for machine learning. One well-known example is "training" electronic document review platforms to recognise and rank relevant documents. Similar concepts apply to most data-driven technologies, even those that apply machine learning to continually develop and self-refine, which require external validation processes

challenges in accessing data sets that are sufficiently large and comprehensive

the difficulty of explaining results from large and complex models, which may be insurmountable in the dispute resolution context, where parties may not accept an AI adjudication that cannot be readily explained in human terms

the risk of bias in data and algorithms, which can undermine the reliability of AI determined results and the willingness of parties to trust the relevant outcomes.5

A recent report by McKinsey & Company addresses some of the current limitations of AI, and how the willingness of parties to accept AI driven outcomes may be affected. The report observes that:

“. . . capturing the potential impact of these techniques requires solving multiple problems. Technical limitations include the need for a large volume and variety of often labelled training data, although continued advances are already helping address these. Tougher perhaps may be the readiness and capability challenges for some organisations. Societal concern and regulation, for example about privacy and use of personal data, can also constrain AI use . . .”6

In this age of "big data", there may be a tendency to find comfort in the large volume of documents and data that are routinely involved in major commercial projects. However, this may be a false comfort, whether when applied to AI adjudications or AI-driven contract preparation and risk allocation. The mere fact that "big data" involves access to voluminous data sources does not mean the data is quantitatively useful. "The relevant standard is not whether the entire data source can be regarded as ‘big’, but whether the filtered data that is pertinent to the present legal case is big enough.”7

In addition, AI requires the labelling of a large volume of training data, providing external direction and confirmation from practitioners. This training data requires both a legal and commercial "correct answer" or for labels to dictate "wins and losses" to inform best practice. However, these matters are often not "black and white" and there may be no definitive correct answer in many scenarios. Indeed, the scope of options that can be considered "best practice" has led to a diverse and competitive legal landscape. Not only will this restrict the reliable application of potential legal technologies in the future, but it is likely to ensure that lawyers will remain invaluable when labelling data sets to facilitate predictive algorithms.

Therefore, while AI-driven technologies will continue to revolutionise business practices across a number of industries, it may be relatively slow to impact the provision of legal services in some sectors in a manner that fundamentally alters the core qualities of legal service. 

Conclusions: potential impact on legal advisers of the future

As a result of these limitations, it is likely that the judgment and strategic decision-making skills of the lawyer of the future will not be replaced by legal technologies. In these circumstances, legal technologies will be used to complete certain legal tasks far more efficiently and accurately than human lawyers previously have, but clients are unlikely to trust those technologies to provide legal advice (as opposed to legal services in the form of discrete tasks) which may remain the domain of human lawyers.

As discussed here, a drawback of AI is the relative inaccessibility of its reasoning and inability to explain conclusions in human terms. It is understandable that clients may want some comfort that the conclusion or position reached is best practice. In these circumstances, the result of AI-driven technologies is likely to be used as a guide only, to be interpreted by technology-savvy lawyers and used as the basis of strategic advice and guidance.

In addition, even if some of the more optimistic projections for potential legal technologies eventuate, those potential technologies are not likely to replace the oversight, strategic and persuasive qualities of a successful lawyer. These qualities form part of a lawyer's judgment and will, it is posited, remain highly sought after by clients of the future. This is particularly the case in adversarial environments – both transactional and dispute resolution – where participants will usually be looking for an edge over each other which may not be available if each party solely relies on predictions derived from past examples.

Arguably, these qualities have always underpinned the traditional role of "trusted adviser" and may not significantly change even though the daily tasks of the lawyer of the future will be revolutionised. n

Sean Kelly is a senior associate at Clayton Utz. Harrison Nurse is a senior lawyer at Clayton Utz.

* This paper was first presented at the Society of Construction Law Australia Annual Conference, August 2018.


1. Richard E Susskind, The End Of Lawyers?, Oxford University Press, 2010.

2. "There's never been a better time to be a grad", Lawyers Weekly, July 2018, p6.

3. A useful article providing a brief overview of smart contracts is from Adam Lodders, "Understanding smart contracts", September 2018 Law Institute Journal (92.09).

4. See Theodore W Ruger et al, "The Supreme Court Forecasting Project: Legal and Political Science Approaches to Predicting Supreme Court Decision-making", 104 Colum. L Rev 1150 (2004), and Nikolaos Aletras et al, "Predicting judicial decisions of the European Court of Human Rights: a Natural Language Processing perspective", (2016) PeerJ Computer Science, 93.

5. "Notes From The AI Frontier - Insights From Hundreds Of Use Cases" (McKinsey & Company, 2018).

6. Note 5 above, i.

7. Robert Parnell, "When Big Legal Data Isn't Big Enough: Limitations In Legal Data Analytics" (Settlement Analytics, 2016), p7.


The content you are trying to access is exclusive to LIV members*

To access your exclusive member content please click the 'Already a Member' button below and you will be redirected automatically.

Not a member but would like to find out about the value of LIV membership? Click the 'Become a Member' button below or call our membership team on (03) 9607 9470.

*Note that some content may be exclusive to specific types of members. If you would like to inquire about your access please contact the membership team on (03) 9607 9470.