Cambridge University Society for Women Lawyers

Kindly Sponsored by

Ashurst Logo.png

Robot-Assisted Law: Databases and Legal Reasoning

Written by the Runner-Up of the CUSWL Law and Tech Essay Competition 2021 - Alec Thompson


The title of this article might suggest another hype-filled exposition of the second coming of legal AI: the ‘legal singularity’ where judges and lawyers are replaced with well-trained software.[1] Not so, instead it will be covering the influence of information storage technology on the legal system. Although distinctly less glamorous, the legal system was revolutionised once through writing, [2] why not again in the digital age? We can loosely call this kind of digitalisation ‘robot-assisted law’, like ‘robot-assisted surgery’, where (human) lawyers and their smart tools work together. These new tools are the digital casebook, the massive online database, and the advanced search engine. Like any tool, they produce gradual shifts in the practices of their users;[3] the goal of this article is to cover a few of the unexpected side-effects and glitches these new systems produce.

Land Law and a Bad Search Engine

The first pitfall is an over-reliance on search engines to find and file cases. The danger is that when the search engine fails to work properly it can produce legal blind spots. An example is the website storing the judgements of the Adjudicator and judges in the First-tier tribunal.[4] The search tool was difficult to use and led to cases not being cited simply because they couldn’t be found.[5] The result is around eight lines of authority were produced on a single point because the courts failed to cite each other’s decisions.[6] The legal issue was whether A can recover their land from C, an innocent third-party purchaser who received an interest from the registered fraudster, B. The caselaw is now a mess, largely because there are multiple decisions which fail to cite one another, all with their own route to favouring A.[7] The disturbing conclusion is that essentially poor web-design led to substantive law becoming fractured. It appears we are at a point where, due to integration between the legal process and digital architecture, trivial technical issues can accidentally produce legal reform. For instance, take the government’s page for legislation: despite being the official online text of the law, it is well known to be out of date,[8] or, more pressingly, the vast amount of informal legal advice available on the web.[9] If enough people, lawyers or laymen, act on these websites, mistaking them as the true legal picture, is this so different from an official change in the law? Unlike reporting errors in the past,[10] restricted by the common professional understanding of cases, digital resources are read by thousands across the country, and, crucially, can be altered rapidly. Although this means mistakes can be corrected quickly, we should also take an interest in the unanticipated distortions it might produce.

Immigration Law and Codifying Bias

The problems above related to the legal technology being faulty: the issues here involve them working too well. Thanks to the massive quantity of data now stored online, a lawyer can quickly search all the judgements given by a particular judge across her career. It is well known, for instance, that immigration judges possess biases,[11] and that these can be predicted by using legal search tools to find all their previous cases.[12]Given lawyers seek to achieve their client’s practical interests, a tool like this is invaluable: why not use arguments which have worked in the past?[13] The danger, however, is that by recording – essentially codifying – these biases, legal search tools convert them into a form of quasi-law. After all, if Judge Y, W, and Z believe that immigration applications are to be allowed solely for married couples, a criterion within their discretion but not mentioned in the law, and lawyers dissuade clients from applying on this basis, is this any different from the law actually containing that restriction? Of course, these biases have always been present, indeed it is likely some experienced lawyers were probably loosely aware of them, but this would disguise the true nature of the change. The main difference today is that anyone with access to the right software can quickly extract clear-cut extra-legal biases and use them in practice. This could be a force for good – it could allow challenges based on illegal bias[14] – but it could also entrench biases and tendencies by giving them tangible, accessible form.


Although the innate conservatism of legal circles hides it, we are in an incredibly exciting time for law. Law is, after all, a discipline based on the manipulation and presentation of data – and more data was produced between 2013-15 than in all prior human history.[15] New database tools provide many new opportunities for streamlining legal practice and democratising legal knowledge, but they also offer various dangers. As lawyers we should be self-conscious of these limitations, both for our clients and for society as a whole.


[1] Bhorat, Z. “Do we still need human judges in the age of Artificial Intelligence?” Open Democracy. URL: accessed 12/01/21; Benjamin, A. (2016), "The path of the law: Towards legal singularity." University of Toronto Law Journal, 66(4) p. 443-455. [2] Baker, J. “Introduction to English Legal History”, (Oxford: Oxford University Press, 2019) [3] See the influence of practitioners’ textbook headings on the way lawyers reason. Danner, A (2014) ‘The influences Of The Digest Classification System: What Can We Know?’ Legal Reference Services Quarterly 33(2), 17-156 [4]Search Decisions Database” Tribunals Judiciary, URL: accessed on 12/01/21 [5] See Goymour, A. (2013). “Mistaken Registrations of Land: Exploding the Myth of ‘Title by Registration’” The Cambridge Law Journal, 72(3), 617-650, 633 [6] Ibid [7] Namely, Barclays Bank Plc v Guy [2008] EWCA Civ 452; Ajibade v Bank of Scotland Plc (Formerly Halifax Plc [2008] EWLandRA 2006 – 0163; Odogwu v Vastguide & Ors [2008] EWHC 3565; and Crawley v Gudipati [2010] EWLandRA 2008 – 0602 [8] improving the UK digital statute book through collaborative maintenance” Open Data Institute, URL: on 12/01/21. [9] For example, the common practice of getting legal advice from Quora, Reddit, and Wikipedia, where the pages for Tort Law and Criminal Law get around 100 views a day; ‘English Tort Law’ Wikipedia URL:; ‘English Criminal Law’ Wikipedia. URL:, Accessed on 12/01/21 [10]Although see John Campbell who, whilst a court reporter, would place decisions he disagreed with in a drawer marked ‘Bad Law’ never to see the light of day. Bryan, M. (2012) ‘The Modern History of Law Reporting’, University of Melbourne Collections, 11 [11] See, for example, MR v Secretary of State for the Home Department PA/02377/2019; for a US study, see Benedetto, M. (2008),“Crisis on the Immigration Bench: An Ethical Perspective” Brook. L. Rev. 73 [12] Keefe, M “How to find legal cases by particular judge or entity” Thomson Reuters. URL: Accessed on 12/01/21 [13] Ibid, “Knowing how a judge has ruled in a similar case in the past can be essential information when you are building your case. This information could be the difference between a favorable result for your client and an embarrassing one.” [14] Supra (n11) [15] Kumar, V. “Big Data Facts” Analytics Week. URL: Accessed on 12/01/21