The Rise of Automation in the Legal Industry

Written by Tim Jones, Lawyer

The rise of artificial technology (AI) and the increased presence of AI has reached fever pitch recently with the introduction of ChatGPT. In the legal profession, being an industry so often the object of criticism for being overly conservative and slow to change, this has led to increased societal pressure to play catch up.

 The use of technology has the potential to deliver faster and more economical decisions across many sectors.[1] The ability to utilise online systems to organise data entry and initiate complex algorithms without human error could have a significant impact on how justice is accessed in the community and how decisions are made in the near future. Of course, with such reform on the horizon the ability to review decisions may also be impacted. What is obvious is that the people in charge of making the call on the use of automated systems need to make sure they have tested the technology appropriately before implementing it in the public and private sectors.

The rise of automation in the legal and administration sectors has led to an increased desire of decision makers to use technological infrastructure to streamline existing processes.[2] The worrying trend has become that artificial or automated decision-making programs have not been fully tested before their implementation.[3] The recent case in America of  “the world’s first robot lawyer DoNotPay” and closer to home, the case of Amato v The Commonwealth are perfect examples of where technological advancements have impeded proper and well thought out decision making.[4]

Amato v The Commonwealth

Centrelink implemented an automated decision-making scheme, known as Robodebt, to recover alleged debts owing to Centrelink by debtors. Problematically, the scheme relied upon the Australian Taxation Office and its data to average a fortnightly income for working out whether a person was overpaid by Centrelink. This completely discounted the Social Security Act 1991 (Cth) which had and should have continued to be used to calculate payments from Centrelink.[5] The scheme compounded this issue by putting the onus of proof on the accused to disprove the allegations being made by a faulty system and often the scheme was adding extra penalty fees for a failure to respond which were affected by errors at law.[6]

With the case in question, Robodebt miscalculated Deanna Amato’s income by almost four times her real income by using the ATO averaging model. The Scheme deepened the mistake by automatically sending correspondence of the alleged debt to an old address and then added fees under Robodebt for the attempted collection of the alleged debt. The Federal Court found in favour of Ms Amato on the basis that Robodebt was automatically issuing notices to an incorrect address, completely disregarded the payment framework espoused in the Social Security Act 1991 (Cth)[7], and used an arbitrary fortnightly averaging framework in its place that was not capable of being supported in fact or law.[8]

This case is a great example of the implementation of a streamlined technological process going wrong. Government administrators did not address the issues surrounding the calculation framework and adopted the Robodebt scheme far too quickly. In doing so, the Government have potentially left themselves open to legal action that may have ramifications on how other agencies implement automated decision-making software and programs moving forward.

Impact of Automated Decision-Making Systems

Administrative and government agencies have started to embrace technological advancements. This was originally to keep up with the times, however, it has now become a key ingredient in more easily accessing services provided to the greater public through digital platforms.[9] The digitizing of record keeping has made it easier to cut costs and deliver information and services to the wider public in a more efficient way. The push to use available technology is seen as being a directive imbued by the Government to enable discretionary spend on manpower to be limited and to use those costs savings on new infrastructure.[10] What has begun to worry some, is that the powers vested in the Government to administer for the greater good is slowly being etched into complete reliance on computer systems. Even more troublesome are the issues surrounding privacy and equality.[11]

The common denominator between the use of AI decision-making programs and their successful implementation is the appropriateness of their design when combined with the existing legislation.[12] This brings us back to the case of Amato v The Commonwealth. Centrelink may have thought they had the perfect program to keep costs to a minimum while revenue raising for the Government. However, their failure to abide by the Social Security Act 1991 (Cth) and the common principles of justice, fairness, privacy and equality ultimately led to an abject failure of the Robodebt program.[13]  This case is a lesson for the legal profession.  Implementation of AI will have to happen; but it should not be adopted without rigorous testing and appropriate supervision and oversight by qualified lawyers.

[1] Lisa Toohey, Monique Moore, Katelane Dart and Dan Toohey, ‘Meeting the Access to Civil Justice Challenge: Digital Inclusion, Algorithmic Justice, and Human-Centred Design’ (2019) 19 Macquarie Law Journal 133.

[2] Zalnieriute, M, Moses, L and Williams, G, ‘The Rule of Law and Automation of Government Decision-Making’ (2019) 82 Modern Law Review 3, 1.

[3] https://lsj.com.au/articles/worlds-first-robot-lawyer-hit-with-lawsuit-for-practising-without-a-license/

[4] Deanna Amato v The Commonwealth of Australia VID611/2019.

[5] Social Security Act 1991 (Cth), ss 1223(1)(b) and ss 1229(1).

[6] Minister for Immigration and Border Protection v Stretton (2016) 237 FCR 1 at 3-4 [5] (Allsop CJ).

[7] Social Security Act 1991 (Cth), ss 1223(1).

[8] Minister for Immigration and Citizenship v SZMDS (2010) 240 CLR 611.

[9] Zalnieriute, M, Moses, L and Williams, G, ‘The Rule of Law and Automation of Government Decision-Making’ (2019) 82 Modern Law Review 3, 1.

[10] John Manzoni, ‘Big data in Governments: the challenges and opportunities’ (21 February 2017) < https://www.gov.uk/government/speeches/big-data-in-government-the-challenges-and-opportunities>

[11] Zalnieriute, M, Moses, L and Williams, G, ‘The Rule of Law and Automation of Government Decision-Making’ (2019) 82 Modern Law Review 3, 1.

[12] Ibid, 4.

[13] Greg Jennett, ‘Robodebt removed humans from Human Services, and the Government is facing the consequences’ (30 May 2020) ABC Online < https://www.abc.net.au/news/2020-05-30/robodebt-stuart-robert-scott-morrison/12303322>