AI and Law: Judiciary and ChatGPT

AI and Law: Judiciary and ChatGPT

Artificial Intelligence (AI) is no longer an unknown term and is very much a part of our day-to-day lives be it unlocking mobile phones, personalized feeds while perusing social media platforms, use of banking system through apps, use of maps while commuting, and so on and so forth. As a step forward in the field of AI, Google announced the kickstart of Google I/O 2023 as groundbreaking developments in AI with features such as “Help me Write” – expands on capabilities of smart reply and smart response; “Med-PaLM2” – medical application analyzing X-ray images and generating comprehensive mammography reports, etc. AI is nothing but application(s) that perform tasks which once required human efforts.

The legal fraternity has also admirably accepted the use of AI. In today’s world lawyers use AI for research, document automation, contract management, litigation database, intimation of court’s cause list, etc. and thereby assists in streamlining various legal processes. Recently, a final year law student has developed an app named Law Bot Pro which offers comprehensive and user-friendly platform for obtaining legal information.

AI has been used by the judiciary across the globe and following are examples of AI usage in judicial process:

India: With a view to onboard new age, cutting edge technology of machine learning and artificial intelligence in the judicial domain for enhancing efficiency, productivity of justice delivery, the Supreme Court of India implemented following 2 apps: -

Supace which is an assistive tool that can augment the efficiency of the legal researchers and judges to work on cases, extract relevant information, read case files, manage team-work and draft case documents with ability to find facts, issues, and points of law from thousands of pages of documents, in a matter of seconds. On Supace a user can convert pdf files of judgements into text, take assistance of chatbot and extract synopsis, FAQs, evidence, and case laws.

Suvas on the other hand translates judgements from English to vernacular languages.

United States of America: The Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) software, a computer-based algorithm is used to predict the chances of recidivism (re-offending) of criminals. The algorithm’s assessment is based on factors including sex, age, and criminal history, which has been tested and deemed reliable in identifying defendants with the highest risk score who went on to reoffend at a rate four times greater than those assigned the lowest risk score.

Brazil: The Supreme Court of Brazil has implemented the Victor AI system. The AI system, in its initial phase, has the goal of analysing the huge load of appeals presented before the High Court and to automate the evaluation of one of its prerequisites: Victor’s pilot version aims to identify cases with repercussão geral (“general repercussion”). Before the implementation of the system the civil servants used to scrutinize for general repercussions which took 40 minutes per case as against 5 seconds with the use of the system.

Argentina and Colombia: The Public Prosecutor’s Office in Buenos Aires is using Promatea software to predict a solution to a court case in less than 20 seconds with a 96% success rate. The same software is being used by the Constitutional Court of Colombia.

China: Chinese Internet Courts / Smart Court do not require citizens to appear in courts. It includes non-human judges, powered by AI and allows participants to register their cases online and resolve their matters via a digital court hearing.

The aforesaid examples of use of AI by judiciary can be said to predictive justice (US, Argentina and Colombia) or for improving efficiency, speed, access and effectiveness of the courts system (China, Brazil and India).

One such AI driven app named ChatGPT has recently gained tremendous popularity. ChatGPT is an open AI and is designed to generate human-like text based on large corpus of text data and is trained to perform various language tasks, such as answering questions, writing stories, composing code, and generating essays. The self-definition by ChatGPT on “how does ChatGPT works” clarifies that it analyses the data and generates a response. So, in the legal profession can it be said that lawyers will no longer be required to conduct research and ChatGPT will analyse the data and provide researched material?

ChatGPT has recently been used by Courts to pronounce judgements. The first such instance is a judgment by the Colombian Court, where the query to ChatGPT was: “Is autistic minors exonerated from paying fees for their therapies?” and ChatGPT answered: “Yes, this is correct.” This judgement prompted discussion on the authenticity and use of AI in law with criticism from colleagues. The Learned Judge defended the technology as it can help Colombia’s legal system to be more effective and added that the judgement refers to earlier rulings on the issue thereby implying that the judgement is not passed purely on the response provided by ChatGPT but earlier rulings as well.

The Punjab and Haryana High Court in India in Jaswinder Singh @ Jassi versus State of Punjab and Others reasoned that the petitioner therein has not made a case for grant of bail prima facie as the documents indicated the petitioner’s involvement in a brutal assault culminating in death of an individual. The Learned Judge used ChatGPT (post-reasoning) to assess the worldwide view on bail when assault was laced with cruelty and inquired - “What is the jurisprudence on bail when the assailants assaulted with cruelty?” ChatGPT responded that “The jurisprudence on bail for cases where the assailants have assaulted with cruelty will depend on the specific circumstances of the case and the laws and regulations of the jurisdiction …” Thereafter, ChatGPT elaborated that the judge may decide to grant the bail depending on the severity of the assault, the defendant’s criminal history and strength of evidence against him. Further stating the importance of the presumption of innocence ChatGPT clarified that in cases of assault with cruelty bail may be granted if they do not pose a risk to the community or a flight risk.

On one hand where ChatGPT has been the talk of the town, on the other hand Italy has in April 2023 banned ChatGPT in view of a suspected breach of Europe’s strict privacy regulations.

Concerns with the Use of AI

Authenticity of data: The response created by an AI tool such as ChatGPT uses available data to respond to any queries. There will always be a doubt if the data relied upon to generate the result is genuine, verified and reliable.

Lack of transparency: AI is algorithm which is created by entity/individuals and procuring a disclosure of core aspects may be a challenge. This seems to be a real issue as the owners and developers of the Compas risk-assessment and sentencing tool (discussed above) appears to have declined to disclose the core methods and datasets used in the AI tool.

Lack of confidence to rely on the result: Generative AI may in a straightforward query be reliable. However, AI is mostly used for slightly or more complex issue which require information and analytic skills, the reliability on the results in such cases is dubious. For instance, the response provided by ChatGPT in the case of Jaswinder Singh versus State of Punjab (discussed above) is vague though the explanation provides for situations where the response could be affirmation or negative.

Lack of accountable: The use of AI will raise questions, such as, the extent to which an AI can be used in judicial pronouncement; basis for relying on an AI provided output; the onus of accountability for a judgement rendered primarily by use of an AI? In fact, one of the criticisms for the Victor software (discussed above) is accountability as it appears that the software selects appeals for review by random selection process without any rational.

Ownership or authorship of the results generative by AI systems: The question that arises is who the owner is and/or author of the results depicted by AI systems - is it the developer of the algorithm; the entity which engaged the developer; AI which analysed the data and creates the result; and/or the user who uses such results. This poses serious questions on seeking copyright protection on AI generated content. The United States copyright laws do not protect works solely created by a computer whereas such works are protected in United Kingdom. The copyright laws in India do not provide an answer however the Courts in India have granted copyright protection for AI generated works.

There is no doubt that AI systems saves time and increases efficiency. However, for an effective implementation of AI in judicial pronouncements, principles of responsible AI viz. explainable, transparent and accountability should be introduced. It is important that AI participants respect the rule of law, human rights and democratic values, including privacy and data protection, non-discrimination and equality, diversity, fairness, social justice, and internationally recognised rights. Dr. Dory Reiling, Former Senior Judge and Expert for the Council of Europe said that AI should be able to explain how it came to a result, both in terms of how the information was processed along by making a substantive explanation available. This highlights the need for human control at every stage of AI processing as the technology currently stands. As Dr. Reiling puts it, the algorithm of an AI cannot be used as a sole deciding factor and judges must be in control while possessing the ability to deviate from the output.