Legal systems worldwide have embraced English jurist William Blackstone’s idea of letting ten guilty people escape rather than having one innocent person suffer. But, “justice delayed is justice denied” is also a very common adage in the judicial domain. Both raise three important points: ensuring justice is delivered, supporting it with correct evidence, and addressing it in a timely manner.
However, timely redressal has been an enormous challenge for various developing countries. Millions of cases have been pending across several courts and these courts dispose of fewer cases than are filed. On average, cases take three years and nine months to get disposed of in some of these developing countries, which is an astounding number.
The legal industry is very cautious when it comes to adapting new technology.. The question is, can AI technology be used to render justice in a speedy manner without compromising on outcomes? The answer is a resounding “yes”. It is vital for the current legal procedure to be infused with modern technologies like AI, machine learning (ML), big data, and cloud computing that can potentially transform the way legal services are delivered in the judicial system. The use of AI in legal practice has already taken shape and is poised for even greater adoption.
How Can AI Transform the Judicial System?
There are 3 key pillars of judiciary: administrators of the law (judges), practitioners of the law (lawyers), and the enablers (police, legal service providers, etc.). Each has a set of tasks that range from repetitive activities like documentation to decision-making. AI can transform the way these tasks are performed. Here are a few examples of how AI can benefit the judiciary and have a greater impact on meeting the goal of delivering speedy justice.
There are several cases filed everyday but not all need to be addressed by the court. These cases can be settled through alternate dispute resolution systems of arbitration, mediation, etc. With the help of AI/ML Models, the admissibility of a case can be determined within a few seconds. In addition, AI can help prioritise decision-making for important cases, saving valuable time on cases that can be resolved outside the court.
The traditional process of doing legal research is manual and time-consuming. AI systems based on technologies like natural language processing (NLP) can provide lawyers with relevant insights and recommendations or locate pertinent cases and contextual material in far less time than the current manual process, and free them from repetitive tasks.
Investigations are tough. It involves long hours of legal research to find the right pieces of information. AI, combined with technologies like edge computing, internet of things (IoT), 5G, and cloud can expedite harnessing the power of data generated by many departments. Visual surveillance and data from sensory systems combined with social media mining that leverages advances in facial recognition, speech recognition, text analytics and deep learning can be used to track linkages between suspects involved in criminal activities, thereby shortening investigation timelines.
AI has the potential to swiftly analyse available legal data and predict potential legal outcomes, perhaps even better than humans. Many cases are of similar nature and have several variables like witnesses, proof, and evidence that can be used to train AI/ML models for predicting outcomes. This information can help the client and the lawyer cover missed evidence or address potential thinking biases to better improve case preparation and assess the chance of winning in a trial.
Another application of natural language understanding (NLU) is conversational AI adopted in various applications like digital interfaces or chatbots. A chatbot can also offer AI-powered legal counsel. Not all clients can afford lawyers, but these AI assistants can help a plaintiff at much lower cost.
Judges can be aided with a case summary created using text summarization and NLG techniques. NLG automatically creates abstracts and descriptions through insights found in data. Having an initial report from AI can help expedite the process since it can analyse information faster. Ongoing research in this area can lead to building assistive tools that can help judges hear more cases every day.
Challenges to AI Adoption
When it comes to AI technologies, trust is crucial. There are always doubts about the reliability of model outcomes. While research in explainable AI where the results of the solution can be understood by humans, and visualization of AI model functioning where algorithms are used to create images from data for easy understanding is advancing, it is difficult for people to trust the algorithms. The question remains – how do we know that an AI algorithm is making the right decision? This has been the primary hurdle for AI to make its way into judicial decision-making.
There are also issues regarding confidentiality and ethics. Data used to train algorithms is collected from various sources and at times, without the user’s knowledge or consent. People will not easily accept models that violate the right to privacy. And legal systems will not authorize these models as acceptable forms of evidence.
Finally, there is the issue of biases in modelling. Introducing data that comes pre-loaded with implicit biases can lead to biased conclusions based on factors like ethnicity, gender, age, and others. The data fed for training needs to be fair. The system will only be as good as the data used to train it.
Today, AI applications are created to carry out a specific task, learn to become better over time, and process every combination of input values and result measurement until the most effective output is achieved. However, decision-making in the judiciary system includes complex thinking like abstract reasoning, creative thinking, logical analysis, and emotional intelligence. These are beyond the current realms of applied AI.
The Way Forward
A decision can impact one’s life whether they are a victim or an accused. We can question a human’s decision but how do we challenge a machine’s decision? We accept verdicts made by judges in the court, but will we similarly accept verdicts made by machines? To advance in this direction, it is imperative that machine-based decisions overcome problems like biases, privacy breach, and lack of transparency. Architectures should be defined to make it easier to identify these issues, help make changes, and stop them before they arise.
While we address all of the challenges to AI, we need to choose solutions that can speed up the long-drawn process to help deliver justice to the lives involved. Now is the time to adopt a new approach to the legal system that can dispense timely justice to all.
About the Author:
VP & Global Head – AI Solutions, Wipro
Mukund and his team collaborate with leading enterprises across industries to drive revenue growth through competitive business insights, deliver experiences aligned to emerging digital behaviors, and drive efficiencies by leveraging technologies such as artificial intelligence, machine learning, RPA, and big data. He is a well-known industry thought leader with more than 21 years of experience in leading technologies.