I. Definition
Artificial Intelligence (AI) is an intelligence demonstrated by computers and machines which includes perceiving and synthesizing information with purpose to make decisions. It is an umbrella term for a range of algorithm-based technologies that solve complex tasks. Goal of AI is to mimic capabilities of the human mind. The ideal purpose of AI would be creating systems who think and act rationally. This means making unbiased and error-free decisions based on available data. There are five basic components of AI: perception, learning, reasoning, problem-solving and language understanding. While “regular” software is programmed to perform certain task or group of tasks, the AI is programmed to learn to perform the task(s). In conventional software, primary artifact is the code, but for AI primary artifact is data. As a first step in establishing AI system, the machine does raw data collection from various input sources, then works on identifying raw data (images, text files, videos, etc.) while adding meaningful labels to provide the context, so the machine can learn on its own, or by the support of the human. Traditional software takes an input and with some logic written in the code and afterwards creates an output, while machine learning algorithm takes both an input and an output, gives specific logic and learns from previous outcomes when creating a new output. Also, traditional software is doing the same thing over and over again and it doesn’t really change unless human updates it. AI changes over time because of its ability to learn.
II. AI and existing legal framework
Businesses are using AI in all sorts of context. There are numerous overlapping and developing legislative and regulatory developments that apply to the AI technologies. Many of relevant legal acts have extraterritorial effect. It means that using AI lawfully makes it increasingly challenging for running businesses.
On the European Union level, we can expect changes due to the proposed AI Regulation. This is going to be the very first comprehensive regulatory framework applicable to AI in the world (like GDPR is for personal data protection). It has very extensive material scope and it’s going to be applied to the whole developing chain of AI, including AI providers inside and outside of the EU. Proposed sanctions are going to be high. The AI Act will lay down harmonized rules for the developing AI and placing it on the market, as well as post-market control. Very important part of this act is AI risk assessment. It divides risks into four categories: unacceptable risk, high risk, limited risk, and minimal risk. If AI will be adopted in its current form, this is quite significant because it will prohibit uses of AI that contravene EU values. According to Art. 5 of the proposed AI Regulation, which covers prohibited AI practices, LegalTech solutions aren’t directly included. High risk is mainly referring to LegalTech used by public authorities and courts. It is likely that most LegalTech solutions used by companies, lawyers, and clients will fall into category of a limited risk. All other systems will be considered as AI systems with minimal risk. Once adopted, Legal Tech providers will have at least two years to prepare for the requirements that will be imposed by the AI Regulation.
Additionally, we can also expect the adoption of new EU AI Liability Directive which regulates damages that AI-related products and services might cause and liability of publishers of AI. Under the proposed new AI Liability Directive, the presumption of causality applies if three conditions are met: first, that the fault or the failure of an AI system has been demonstrated; second, that it reasonably likely has influenced the output; third, that the fault or the failure of the AI system to produce output gave rise to the damage. Harmed person will thus need to prove that providers or users of high-risk AI systems have failed to comply with obligations under the AI Act once it will be adopted.
An appropriate regulatory framework to promote sustainable AI by monitoring and mitigating the associated risks in legal profession is a pre-condition for using AI more comprehensively in the legal domain.
III. Examples of using AI by Legal Professionals
Using AI in legal domain is not new. However, its adoption is slow. Reviewing documents for litigation, analyzing contracts to check if they meet the required criteria, legal research and predicting case outcomes are only few of the examples of AI based software used by legal professionals.
When performing the review of documents for litigation, lawyers search for specific and important keywords, dates, emails and other documents. AI is able to learn what is relevant and what is not, from previous searches conducted by legal professionals. Based on that, it can more accurately identify key documents for litigation. AI-based software can help with organizing files, documents, emails, calendar and tasks. Structuring of large amounts of information makes the administration of legal documents more efficient. Machine learning AI is able to learn through training with purpose to extract the relevant parts from a large amount of information. This is one of the best ways how AI is being used in law right now – a database of information.
When talking about analyzing contracts, adopting machine learning to review contracts is taking the legal industry by storm. One of many examples how AI can be used to analyze contracts is AI software is called COIN and is used by JP Morgan to analyze commercial loan agreements. Bank plans to use this software for other types of legal documents too. Why? Because a work that earlier took hundreds of hours to be completed, can now be done in a few seconds. AI-based software can spot and identify issues that might have been missed by human lawyers. It can review contracts faster and, in some cases, more accurately than humans.
AI-based contract management software learns from earlier contracts and data. This previous knowledge on how to handle data allows to flag any risks, missing key clauses or insert pre-approved clauses. Communication between legal department and sales department is simplified. All this speeds up contract approval time.
Legal research is commonly performed by legal practitioners and scholars. Sometimes it will take lawyer hours, days or even weeks to find a relevant case or legal concept explanation. AI-based software can do it almost instantly, but only if the AI system has enough data and powerful algorithm. Some attorneys are not even aware they’re using AI in their research since it’s been integrated into many research services. With intelligent legal research software, attorneys can test out variations in fact patterns or legal analyses to identify the most advantageous strategy. Comparative analysis between cases in different countries and states or between federal and state courts no longer takes days of exhaustive scanning.
And lastly, AI can predict case outcomes. Accurately assessing the likelihood of a successful outcome for a lawsuit can be very valuable. No lawyer has complete knowledge of all data. However, AI can access more data and analyze it much quicker than a lawyer. Therefore, it can predict case outcome faster and even with higher precision. It allows attorneys to decide how much they should invest on experts, whether they should take case on contingency or advise clients to settle. By analyzing large data very quickly, AI can help with picking the most effective witness, best way to present a case and propose a way to respond to opposing arguments. Also, clients can more easily decide whether it’s worthy or not to proceed moving forward with the case. Furthermore, AI applications are being used as advisers to judges on bail and sentencing decisions. Those AI tools used by criminal judges assess the recidivism risk of defendants and convicted persons in decisions on pre-trial detention, sentencing or early release. Name of this tool is Correctional Offender Management Profiling for Alternative Sanctions (COMPAS) and it’s been used in some of the states in USA. This is no longer the future but a reality.
IV. Benefits of using AI by Legal Professionals
Moreover, AI offers new opportunities for digitalizing legal services. Delegating certain legal tasks and automation where decisions need to be reached on the basis of a large quantity of data is one of the most common ways of using AI in legal practice. AI technologies are able to make decisions at a near-instantaneous speed.
Other benefits of using AI-based software include:
- Elimination of time-consuming tasks and automation of low-level tasks: Many of activities are routine, paperwork-based duties that divert legal professionals’ attention away from the strategic demands of a case. Those tasks can be automated in software and save lawyer’s time to focus on more complicated issues. AI’s ability to work on repetitive tasks is one of the many advantages for legal sector.
- Producing high-quality work: AI-based software isn’t tired, sleepy, confused or biased. It can help manage documents and cases more efficiently, review contracts for missing clauses, highlight typos and wrong terminology or flag ambiguous terms through comparison and ability to learn.
- Earlier and accurate risk assessment: AI based software can examine information in real-time. This allows lawyers to find potential risks on time and prevent legal problems later.
- Efficient and accurate background checking: Part of the lawyer’s work is doing extensive background checking of clients. AI can perform this task accurately and efficiently with less time needed than humans.
- Reduced stress of legal professionals: Legal professionals do multiple tasks and some of them are really time consuming. As stated earlier, low-level tasks can be delegated and automated to allow lawyers to have more time to focus on creative analysis and more complicated duties. For example, scheduling of tasks is very important to complete them on time. Software’s ability to send notifications and reminders when deadlines are approaching is a useful feature.
- Remote work: Since we all got used to work from home during COVID19 pandemic, AI technologies are helpful in this domain also. Accessing documents and cases from homes, setting up video meetings to discuss and analyze cases became a new reality. According to the Legal Industry Report 2021: Lessons Learned from the pandemic, the use of software that increased productivity, often had the result of helping to increase profit or stabilize revenues for some firms. That being sad remote work is here to stay in the short term, and will most likely become commonplace in the foreseeable future.
V. Concerns regarding AI
First concern is violation of privacy laws. Some countries have comprehensive data protection laws that restrict AI and automated decision making involving personal information. The EU General Data Protection Regulation (GDPR) recommends that organizations using automated data processing, such as AI, take certain measures to ensure information is processed fairly. AI providers may face challenges in defining the purpose of processing information when developing AI because it is impossible to predict what algorithms will learn and use data for. Basically, data might be used for new purpose.
Second concern is based on ethical considerations regarding AI. Safe and ethical management of AI systems is mandatory. Algorithms that underpin AI technology need to be transparent or at least understandable. Also, AI systems should be aligned with values known to humans, so they function properly in our society and profession in which they are intended to be used. As an example, Amazon used an AI tool in the hiring process, but unfortunately it was discriminating against women and they had to shut it down.
Third concern is related to use of AI in Intellectual Property (IP). Work created by the AI cannot be protected under existing copyright laws. And same is with inventions. With its ability to learn and to be creative it might create new inventions that need to be protected as inventions. But AI cannot be considered as the owner of IP itself, owners are only a natural person or a legal entity. So, whether the user of the AI system should own the IP rights or the rights should go to the inventor of the AI system? Or the AI system itself should be allowed to hold the rights?
VI. Conclusion: Is AI ready to practice law?
AI is a welcome tool in the cause of justice. It is because AI is able to work on repetitive tasks very quickly, it has ability to learn, make unbiased decisions and produce high quality work, help lawyers by predicting case (trial) outcomes, use previous outcomes when creating new ones, identify patterns and use logic while being a creative problem solver, advise legal professionals (attorneys, judges, etc.) but also advise clients on legal issues and many more.
As AI technologies continue to develop, they are already unlocking many opportunities to transform and improve the field of law. Most important advantage is that AI saves time. Computers can analyze large amounts of data, more thoroughly than humans can, in a tiny fraction of the time. Time savings mean monetary savings, since less legal professional time is involved in finding answers and identifying mistakes. Those savings can quickly make up for the cost of new technology. AI can be both the biggest opportunity and potentially the greatest threat to the legal profession. Today Artificial Intelligence represents an opportunity for a law firm to be a leader in a legal profession, but soon it will be a matter of keeping up rather than being a leader. The financial strength will allow only big law firms and companies to utilize AI. At least, in the start. However, one thing is certain — AI is stepping in a legal profession and taking over, and lawyers need to embrace the new technologies. Otherwise, they won’t be able to keep up with the competition.
