Artificial intelligence or AI is a hot topic discussed all around the globe, of all of which has been anticipated as “a replica of a human, which is not a human”. But, how AI, perceived as robots or cyborgs by the people today will play a role in the legal field? How will AI bring reforms in the legal profession and legal administration? Will it be a friend or a foe? And to what extent AI will benefit the legal field. This paper addresses all such questions. For that, I first discussed what is AI generally. Then providing a background of AI, the discussion gets to how lawyers use AI today and how it can be used in the future.
This paper also discusses various AI mechanisms and what can we do to regulate the activities of AI in the future. The discussion also addresses how AI, at its prime will still be different from humans.
A key motivation in writing this paper is to provide a realistic view of AI which will be a part of human society in the upcoming future. However, this discussion contains speculations about AI that may or may not happen in the future.
What is artificial intelligence or AI?
We have all heard that artificial intelligence will profoundly impact all of our activities including the way we live, work, educate, etc. But what exactly is artificial intelligence? Why it is considered a revolution in the history of humankind? Artificial Intelligence or AI can be described in many ways. The term was first coined by J. McCarthy, M.L. Minsky, N. Rochester, and Claude E. Shannon in their project named “A Proposal for the Dartmouth Summer Research Project on artificial intelligence” in which they quoted “An attempt will be made to find how to make machines use language, form abstractions and concepts, solve kinds of problems now reserved for humans, and improve themselves. For the present purpose, the artificial intelligence problem is taken to be that of making a machine behave in ways that would be called intelligent if a human were so behaving.
This conference played a pioneering role in defining the field of AI. “According to Merriam Webster, artificial intelligence is “a branch of computer science dealing with the simulation of intelligent behavior computers” or “the capability of a machine to imitate intelligent human behavior”.
In a simple language, artificial intelligence or AI can be termed as “a technology to automate cognitive activities that are normally done by a human.”
An example can be taken of AlphaZero, a game-playing AI created by DeepMind which has a record of beating the world’s best chess-playing computer programs by self-learning the patterns in less than 4 hours. Some other examples of AI can range from Elon Musk’s company Tesla which is self-automated Cars, a camera application that identifies objects like animals, food, etc. with the use of AI and our daily Instagram or Facebook feed which appears to us on the basis of our daily activities. These examples, though not exhaustive, clearly undermine the principle that AI can be used to automate activities that involve human cognition, reasoning, planning, and decision making.
Now, we need to ask ourselves- Is the Current AI really Intelligent? And much like the depiction shown by movies of AI in the form of robots and cyborg is really foreseeable? Well, the answer to the latter is maybe yes, but not incoming 15 to 20 years. Recently Elon Musk opened up about Neuralink which aims to create a brain-machine interface that restores the cognitive abilities of a human and sort of places a human being in the category of a cyborg. The answer to the former question is not really, AI has not attained its prime. As we will see later in this Article, AI works around a program that has been put into it by its programmer and they are able to get useful results but still are short of intelligence and reasons. The other form of AI is ‘artificial general intelligence’ or AGI is ‘the ability of a machine to perform any task that a human can’, which can possibly include talking to each other, owning a house, getting married, having civil rights, and the list goes on.
That picture is hard to see yet, but we may achieve this in the upcoming 15 to 20 years if the development in the field of AI continues at this rate. AGI can be seen as a goal in the AI industry in which all machines can do human activities in the exact same manner as humans do today. Comparing it with the today’s AI industry, we have reached a stage of AI which we term Natural language processing or NLP where AI can understand language as part of applications such as Google Assistant, Alexa, etc.( which can also be termed as the internet of things) which has diminished the gap between human intelligence and AI.
Methods of achieving AI in a system
Various methods or techniques can be used to achieve AI in systems to automate cognitive tasks. Some techniques are (A) Machine learning, (B) using rule and knowledge representation, and (C) Natural language processing. Let’s look at them in more detail: –
“Machine learning is the study of computer algorithms that allow computer programs to automatically improve through experience” An Algorithm is a set of software containing rules that a computer follows and gets a job done. An algorithm is an array of instructions that take an input and provide an output.It evaluates the data and executes it by solving the problem. What a device does is the result of the algorithms running on the computer’s hardware. Machine learning can certainly be considered as a way to achieve AI. Yet, people, mostly IT companies, use them interchangeably. Machine learning algorithms find natural patterns in the data provided, get insights about it, and make predictions. Today machine learning is used in medical diagnosis for prediction of disease progression, image recognition for security purposes, speech recognition to recognize the words spoken, and to obtain an index arbitrage strategy, etc.
Therefore, we can infer that machine learning is dependent on data. Hence, ‘the more the data, the more the efficiency’. Explaining this, let’s take an example of our Instagram or Facebook feed, it gets designed according to our viewing history, our liked pages, and our activities on the platform, not instantly when we join such platform, but after some days of joining and after that only we get a feed which is inspired by our activities. Thus, these social media platforms use machine learning in every aspect by analyzing your activities by machine learning algorithms. We all may have a question in our minds at some point or the other, that how come the email service providers, automatically filter out emails and mark them as spam? The answer to that is that this is done by machine learning techniques used by email companies to filter out spam emails.
But how do they identify which email is spam and which is not? The machine learning technique used by these email providers has the ability to learn and identify spam emails and phishing messages by analyzing tons of emails throughout many computers which is just beyond checking the spam emails of existing users. One pattern can be filtering the emails which contain certain kinds of a word like “free”, “sale” or identifying emails from a particular sender and tagging them as spam.
Rules and Knowledge Representation
Another technique of achieving AI is the rules and Knowledge Representation which plays a huge role in AI. It allows designers to lay down certain vital rules for any special activity that the programmer wants to automate. The system processes the given information and gives the result deductively. The system based on rules and Knowledge representation in the right setting can do tasks effectively, which otherwise would be very difficult for a human. However, a shortcoming in this is, unlike machine learning, there is no self-learning and continuous learning, unlike machine learning. That means that the programmer has to install the rules ahead of the time on the computer and the computer will play its part on the basis of the rules provided only.
Knowledge Representation is being used to create many expert systems including medical systems which can be used to identify, by analyzing, for example, symptoms and then by using the data in the expert system, that an individual is likely to have a disease.
By understanding this concept, one can compare this with India’s ‘Arogya Setu’ application which has been launched amid the coronavirus outbreak. The application asks an individual to enter the details of his health condition, his travel history, whether he has met any individual who recently has returned from abroad, and whether or not he has been experiencing symptoms of the Coronavirus. The expert system of the application then uses the data of the individual to analyze them and deduct whether the individual, according to the data provided is suffering from the virus or not. The application also uses Bluetooth and GPS features of a mobile phone to track an infected person’s mobile and alert people who are around that infected person. So, now one can see the involvement of AI in today’s scenario, even when the whole world is in a crisis.
Natural Language processing
Natural Language Processing or NLP is a field of AI that gives the machines the ability to read, understand, and derive meaning from human languages. The ultimate object of NLP is to read, decipher, understand, and make sense of the human language in a manner that is valuable. NLP is proving to be very useful in today’s time and its use is increasing pm a daily basis. It is allowing to get meaningful results in the fields of healthcare like the Amazon Comprehend Medical, in media like to identify fake news, in social platforms to develop voice-driven interfaces like Amazon’s Alexa and Apple’s Siri, and in the legal field like the ‘LegalMation’ which provides intelligent tools to help practicing attorneys and legal professional to automate routine litigation tasks. All the process carried out in NLP is done by various algorithms designed by the programmers.
However, NLP carries a huge risk with itself of cybercrimes and breach of privacy. The data provided to the NLP system is prone to get misused and hence with the increase in the use of NLP, the organizations need to take care of their cybersecurity models and strategies.
Involvement of AI in the Legal Field
Now that we have somewhat got the basics of AI and how it is involved in our daily lives, we shall now move to its participation in the legal field. Like any other field, the presence of AI has also been felt in the legal profession in today’s time. But, there is a long way to go. Like AI in any other sector, its application in law is also focused on machine learning and rules-based knowledge systems. The International Association of AI and law hosts International Conference on AI and law from as back as its first conference in 1987 in Boston for showcasing applications and development of AI techniques in the field of law. Various companies are now getting established which are linking AI with the law and thus making the process of lawless cumbersome and more efficient.
As more and more private companies are entering into this field, there is a significant increase in the usage of AI in this profession. Usage of AI in the field of law can be categorized into three branches of the law, namely- by administrators of the law, by legal practitioners, and by non-legal practitioners. Let’s discuss them briefly.
- AI is being used by administrators of law (which includes the judges, administrative officials, police, and legislators) in some countries to carry out administration in an automated manner, yet making sure there is no miscarriage of it, thus continuing to act as the sentinel of justice. For instance, Police forces in India have started to get assistance from AI by using various AI systems that are equipped with options like face biometrics along with a database of past criminal records. In 2019 the supreme court of India proposed to introduce AI for a better judicial system and the administration of justice delivery. The Supreme court mobile application was launched by the president of India, Ram Nath Kovind in 2019.
- There was a remark from him that “AI fuelled law translation system will facilitate the quality translation and will further help in improving the efficiency of the Indian Judicial System”. Another algorithm used by the judges of the United States is COMPAS or Correctional Offender Management Profiling for Alternative Sanctions which is used to predict the risk of an accused to commit another crime in the future. It works through a proprietary algorithm that considers some of the answers to a 137-item questionnaire. Therefore, as of today, we can see AI’s involvement in the decision-making process of the judiciary in some parts of the world to determine recidivism.
- The AI is also being used by the legal practitioners, i.e. by lawyers, legal researchers, and paralegals to carry out their legal work which involves managing and drafting documents along with preparing arguments in an automated manner to achieve high efficiency in their field which otherwise requires ample amount of time and cognitive skills. AI is enabling machines to do tasks that a lawyer does and it is developing at a significant rate.
AI is being used to help with activities like (i)Legal analytics, which analysis data points from past case laws and a judge’s history of deciding the case to form a pattern, (ii) in Document automation based on data input (iii) Intellectual property (iv) Due diligence (v) predicting the litigation outcome. Many companies have invented software using AI that works as a lawyer and takes less time to complete such a job. Also, the work done by these software(s) would be free from any mistake and without any time delay. An example of this is Leverton, a subsidiary company of MRI software, situated in the United States, which uses AI to automate the process of due diligence in lease abstractions. The company abstracts over 1,000 key data points from many types of leases, including telecoms, revenues, retail, etc.
The company claims that its trained AI accelerates the abstraction process along with structuring the data that can be used for contract analytics and better negotiation. Another example is a software company known as ‘ThoughtRiver’ which uses its automated contract review technology to speed up the contracting process by reading the contract, answering key legal questions, and then serving up detailed advice along with guiding its users through remediation.
- AI in law is also being used for those who are not legal practitioners. An example of this would be popping up of legal self- help systems on a website. They are simple expert systems which are in the form of chat boxes that provides ordinary users with the answers of a normal legal question.
What new forms of AI can be introduced in the Legal field in the upcoming time?
Now that we have discussed that AI has already made its way in the legal fraternity, what new techniques of AI if introduced in the future, can enhance the administration of justice and other legal processes. A caveat here is, that what I have written are speculations that may or may not prove to be efficient or simply considered in the coming future.
(i) An advanced form of AI can be introduced in the work of, first of all, the judges. According to the National judicial data grid (NJDG), to date, there are around three crores of twenty-four lakhs eighty-four thousand cases pending in the district and Taluka Courts of India out of which around two crores thirty-four lakhs are criminal cases and around ninety lakhs are civil. This shows that there is a massive backlog of cases in district courts and talukas and this mammoth amount of cases cannot be adjudicated by the exiguous amount of judges and magistrates in India. According to a report.
There are 20 judges per one million people in India, that means as per the data provided by NJDG, it will take around 324 years and 8 months to dispose of every pending case in the subordinate courts if the ratio of judges in proportion to the cases will remain the same. One thing to keep in mind is the data discussed is not including pending cases in the High courts, which are 25 in number and the Supreme court.
So, what can one do to solve this problem? To ameliorate the problem of the huge backlog of pending cases, AI can certainly be used. Let’s say, for e.g. if we are able to make an algorithm which filters out every case and categorize them on the basis of, say, for e.g., the maximum punishment for an offense and past criminal record of the offender, or in civil cases, categorize them according to the remedies a plaintiff is seeking and compile the facts of the case with it to help the judge dispose of the case expeditiously. A program based on this idea will shorten the lengthy hearings in a case and will also reduce the burden on a judge’s shoulder.
Another algorithm that can be built is a system that disposes multiple cases on its own. Cases pending for more than 3 years, which can be disposed of on the basis of plea bargaining, i.e. where the punishment of offenses is less than 7 years, which is not committed against women or a child below 14 years of age and where the accused has not been convicted earlier, can by the use of AI dispose cases by taking a decision on its own, but with the prior consent of the judge or magistrate. This way the judges, which the help of AI, can cope up with the backlog of cases lying in front of them.
(ii) Forms of AI can also be used by the legislators by using an algorithm that stores the data regarding any enactment of a statute. The legislation, as we all know is an underpinning of society as well as the legal field. All judgments are given according to the interpretation given by the legislators to that particular statute or provision. But where there is ambiguity, the judges often use their minds to interpret what the policymakers intended while enacting the statute and while doing so, the judges may or may not interpret it according to the intentions of the legislators, leaving it open for criticisms in the future.
In such a situation, AI can play a role in helping the judges to give a correct picture of what the legislation intended while enacting the statue by its database which not only stores ‘what the legislation enacted’ but also stores in its database ‘what the legislation intended’ relating to that statute. Apart from that, if we look much further in time, AI can also be used to help the policymakers to enact statues according to the needs of the society by collecting data from the public of their needs and wishes.
(iii) As we all know, police are the spine of any civil society and are the go-to-authority for an individual as one trusts the police to solve their matters and protect them. The Police, are considered as ‘guardians of the society.’ But, in today’s scenario, due to fast-growing crimes in the country, the police forces are halted from performing their duty effectively. As per a report, the national pendency rate or percentage of a pending investigation of cases under the Indian Penal Code is around 28.2%, ranging from agonizing 80 % of the rate of pendency in Manipur to a meager 7.3% in Rajasthan.
The introduction of effective technology and AI can help the police force to complete the investigation process on time and thus, ameliorating this issue. Forms of AI can be used to help to carry out processes like a collection of evidence, investigation, and serving of warrants, making the lives of police much easier. For e.g. an algorithm can be built which can be embedded in a moving subject which searches the place of incidence or any place which the police think fit for investigation legally and more efficiently with a minor chance of missing something and thus making an investigation of police expeditious.
(iv) We all have been an intern at some point or the other. In the legal field, we as interns are told to analyze the cases by reading case files of whatever cases our mentor has been handling. We all know how voluminous those case files are. They even, sometimes stretch to 500 pages, containing, plaints, charge sheets, evidence, previous arguments, interim orders, and the list goes on. As a lawyer, especially when you are not a public prosecutor, in order to successfully win a case, you have to read those documents and analyze them and sometimes find flaws or irregularities in them.
Now, when you have a 500-page file in front of you, that becomes a gigantic task that takes days and days. To curb this menace, technological innovation is required. AI can, thus, make a paradigm shift in the work of attorneys and lawyers. A system can be built which analyses a particular case, along with documents, and upon choosing which side you are representing in a case (either a plaintiff/complainant or defendant/accused), the system can prepare arguments and documents which are AI-crafted and gives you a probability of whether you will succeed in winning the case or will fall short. This will do about 70 percent of the total job a lawyer does today.
(v) ‘Legal literacy’ as we all know means awareness of ones’ right and means of securing those rights. It is a powerful instrument for the advancement of society. Today, in India, excluding people from the legal profession, people are not aware of their legal rights and how they can prevent someone from exploiting their rights. Just like any other form of literacy, whether it be digital literacy, media literacy, or scientific literacy, legal literacy is also critical. AI can help in spreading legal literacy in the population. AI can, thus, be used to create an algorithm that will help disseminate legal education for the non-legal personnel to get an insight of their rights and duties.
The system can also be used to provide conventional knowledge about the legal processes in detail, like how the court works, what are the formalities to institute a suit or proceeding in a court, what remedies are available for a certain type of act or omission, and how someone can seek remedy from the court, making it easier for the public to reach the courts and thus increasing the circumference of the courts to reach out for matters for adjudication which otherwise, don’t come before the courts and as a result, leads to depriving a person of his rights and vigilantism, i.e. the practice of ordinary people taking unofficial actions to prevent crime or taking legal actions without actual legal authority. Apart from that, this technology can also be used by future lawyers, who want to get exposure to the working of the courts.
So, when AI reaches its prime, will it stand on equal footing with us?
As per a quote from Erich Fromm, from his book, ‘The Sane Society’, “Reason is man’s faculty for grasping the world by thought. In contradiction to intelligence, which is man’s ability to manipulate the world with the help of thought. The reason is man’s instrument for arriving at the truth. Intelligence is man’s instrument for manipulating the world more successfully; the former is essentially human; the latter belongs to the animal part of man.” AI, it manages to achieve its goal to become like humans and carry out activities which are done by us, will still be different from us. An Artificial intelligent system in the future, though will have greater capabilities of excelling in almost any task compared to human beings, will still lack an overall ability of reasoning and this will differentiate humans from what we can call ‘a masquerade of humans.’
We can now think about how artificial intelligence can help humans in many fields, yet there are always two sides of the same coin. We have all seen in movies how deadly robots and cyborgs can become if not controlled. Experiencing the rise in artificial technology, just like any other technological expansion, comes up with various legal questions.
Questions may include if computers are made to do work in place of humans, whose liability will arise if a crime or an overt act or any act of negligence has been done by a computer while doing what he is programmed to do? Will computers be given legal entity just like humans? Will, they will be responsible? Or will they be treated as property and thus, the programmer of such a system will be held liable? Or the principle of vicarious liability be applied? Thus, treating the programmer as a master and the computer as a servant? There are many possibilities.
Another question that can arise is how will the activities of these systems be regulated? How can we stop artificial intelligence from becoming a companion of humans to a sovereign? The answer to that could be legislation. There is a quote by French Jurist Jean Carbonnier which says “Do not legislate unless you tremble.” This supports the fact that legislation should be enacted very carefully and legislations enacted without due caution can prove to be fatal. Hence legislations should be used in the coming future to regulate the activities of artificial intelligence to stop from becoming a sovereign rather than a service to humans. However, in my opinion, it is unlikely that artificial intelligence will completely overtake humans before 15 to 20 years from today.
Now I shall conclude this discussion by saying that artificial intelligence is and will be a friend of human beings until there comes a situation to choose between ‘either to join it or fight it.’
Vaibhav Jairath | JIMS School of Law, Greater Noida
 Definition of IOT- “a system of interrelated computing devices, mechanical and digital machines, objects, animals or people that are provided with unique identifier and the ability to transfer data over a network without requiring human-to-human or human-to-computer interaction”
 Tom Mitchell (Machine Learning, McGraw Hill,1997) https://medium.com/towards-artificial-intelligence/differences-between-ai-and-machine-learning-and-why-it-matters-1255b182fc6
 27th May, 2020
 Section 265A to 265L of Code of Criminal Procedure,1872
 “Ne légiférez qu’en tremblant” https://journals.openedition.org/droitcultures/367 ( English translated para 35)