top of page
  • Writer's pictureLawtons Africa

Legal-decision making and AI

Authors: Charissa C. Chengalroyen – Candidate Attorney

*Supervised by: Candice Pillay – Director and Ushir Ahir – Director

What is AI?

Simply defined, Artificial intelligence (AI) refers to any piece of software that is capable of “machine learning”. Sample data is input and processed by the software, leading to an outcome. Over a period of time, the AI will be able to make predictions based on the similarities of previously collated data inputs.

AI in law

The use of technology in the law in South Africa has to date largely been limited to the use of online databases and emails, yet in the United States of America, AI software has already been used in the “prediction of criminal behavior, length of sentences, and determining who is likely to recommit a crime”. The overarching question that must be answered is, whether the implementation of AI is suitable in South Africa and the degree to which it should be applied, if at all.

Issues with AI – a cautionary tale

American cases dealing with the Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) system, and similar risk assessments, have highlighted some areas of concern regarding the application of AI in the legal system.

The COMPAS system’s purpose is to assess the propensity of certain individuals to become re-offenders and to assist the judiciary in determining the appropriate sentence for criminals. However, research has found that the initial data inputs into the system were based on previous sentences in America, where harsher punishments were historically meted out to minorities. It was found that because the initial data input was biased, the AI merely perpetuated the inherent bias and inequality that existed in the judicial system (Criswell, B, Algorithms Deciding the Future of Legal Decisions,

This flies in the face of everything we associate with AI and technology: Reliability, objectivity, rationality, fairness, precision, and accuracy. As a society, we have been geared to accept technology as being devoid of fault and error, but this may not be the case when we use technology to simulate (fallible) human traits. Since reality dictates that technology and AI will be an increasingly intricate part of our daily lives, steps may need to be taken to mitigate against the negative and possibly prejudicial effects of AI.

Currently in South Africa there are no bills or policies regulating the use of AI. This will be necessary if South Africa ever wishes to incorporate AI into our legal system.

The policy should provide for:

  • A mandatory code of conduct for programmers and organisations. It is imperative that the algorithms and the outcomes are ethical – to avoid issues with programmers coding inherent bias or human fallibility into the software. In order for a code of conduct or ethics to be effective, it must be enforced with appropriate repercussions or sanctions. (Etzioni A, Etzioni O, Incorporating Ethics into Artificial Intelligence,

  • The need for transparency, impartiality and accountability should be imbued into every aspect of AI from creation to implementation.

  • AI technology must be limited for use as a form of guidance, not necessarily to replace the normative or cognitive reasoning and processes of the individual using the technology. (Criswell B)

  • All AI outputs must be interpreted and analysed before they are implemented or relied upon.

  • How privacy issues, especially regarding the storage and dissemination of sample data shall be handled. This will most likely be regulated in conjunction with the Protection of Personal Information Act of 2013. (All policy suggestions based on the, Communication from the Commission to The European Parliament, The Council, The European Economic and Social Committee and the Committee of the Regions Building, COM 2019, )

In its initial stages of development, AI will be unreliable or even defective. However, this should not detract from the potential benefits that we can gain from the use of AI for legal applications. AI can be used for processing large amounts of data and information to determine patterns or common outcomes in cases that present similar facts, which can contribute to work in the legal research and academic arena. AI can easily assist in the automation of “tedious or repetitive tasks such as periodic reviews by banks of commercial loan agreements, which could result in considerable cost savings to their clients.

There is also potential for AI to be deployed in the routine aspects of more complex legal work such as the production of legal instruments such as wills and residential real estate closing documents as well as contract drafting”(Wendel BW The Promise and Limitations of Artificial Intelligence in the Practice of Law, ); or the utilisation of AI to register and process insurance claims, use of an ‘online receptionist’ to deal with general legal queries from clients, law firm websites akin to Wikipedia which offers accessible general legal information to the public and access to online forms with pre-drafted clauses.

However, any fear of AI as a threat to the existence of the legal field in its entirety is misguided. AI cannot replace or replicate the “emotional intelligence, capacity for sympathy and empathy required in matrimonial disputes; in-court appearances on behalf of clients, at a trial, evidentiary hearing, or oral argument or creative, strategic advising which requires assessing not only legal risks but also taking into account multifaceted, ambiguous, possibly conflicting client objectives and interests” (Wendel BW).


Lawtons Africa is a South African law firm. With roots that grew out of seeds sown in down-town Johannesburg in 1892, our history features various changes and different names. Our team of lawyers, including directors, consultants, associates and candidate attorneys is highly qualified, market-recognised and skilled. For further information, visit


bottom of page