Artificial intelligence and law. When will AI replace lawyers?

The most discussed topic right now is the rapid development of artificial intelligence (AI). The main hope of humanity is that AI will learn how to do routine intellectual work better than humans. The main fear is that AI will take away people’s jobs. Seems to be strange, isn’t it? It’s as if people can’t decide: do they want to work or not? In fact, there is nothing surprising: this is how our emotions work, and this is normal situation.

But I still want to figure it out: can AI replace lawyers? After all, the “death of the profession” has been prophesied for lawyers for a long time, long before ChatGPT walked on stage, and after that, perhaps, only lazy people did not say anything about it.

Tasks of a lawyer and the possibilities of their automation

To understand how much robotization threatens lawyers, if at all, it is necessary to understand what exactly their day-to-day work involves. Let’s highlight several basic routine tasks and then it will become clear whether each of them can be automated.

1. Oral consultations

Every day, any lawyer (both a consultant and an in-house lawyer) receives one or more requests from clients (external or internal). Many problems received can be solved with a simple oral consultation if the lawyer has sufficient knowledge of the substance of the conversation. But the task is not as simple as it seems, since it requires active listening: the client himself cannot and should not carry out legal qualification, determine what information should be transferred to the lawyer and what shouldn’t. A good lawyer asks as many questions as a diagnostician.

Automatization of active listening skills is a difficult task. For example, ChatGPT hasn’t yet known how to work with incomplete prompts (search queries). To get a good answer from ChatGPT you need to ask an excellent question. Only a competent lawyer can ask a great question. But this does not mean that this will always be the case – teaching AI to ask clarifying questions is quite possible in the next couple of years.

2. Legal qualification

Having received a complex problem that cannot be solved with an oral consultation, the lawyer finds out: what does the law say about this? What is the judicial practice? What are the chances of winning in court if it comes to that?

This process can be extremely routine – dozens of documents need to be checked, hundreds of pages of text need to be read. It takes a lot of time and effort. Of course, everything depends on how “standard” the issue is, the number of approaches available in judicial practice, the certainty of legislative norms and many other factors. But overall, this is one of the most difficult and draining parts of lawyer’s work.

At the same time, automating legal qualifications is very simple. Even without AI, there are many tools (of course, not free of charge) that allow you to select the necessary judicial and regulatory acts pretty quickly and efficiently. Artificial intelligence is superior to such services because it provides much higher quality and enormous request processing speed. Sometimes this can also be done for free.

Automated legal qualification is already a reality, and in the future, it will increasingly become the “standard of living”. In my opinion, “manual” legal qualification will finally become a thing of the past by the end of the decade.

3. Analysis and preparation of documents

An equally labor-intensive activity is working with documents. Lawyers are called “paper worms” for a reason. A lawyer reads, analyzes, edits and writes dozens or even hundreds of pages of text every day. At the same time, each text must be brought to an ideal condition, proofreading and editing several times.

There are already a great variety of services for automated preparation of documents (especially contracts). AI “beats” them in terms of content, but “loses” in terms of form and ease of use. Thus, “classic” document constructors (not all, however) can be used by non-lawyers, because the interface is adapted and certain legal structures are literally “chewed”. AI, as we have already noted, will not reveal even half of its potential without receiving a competent prompt, which only a competent lawyer can draw up. There’s no reason why AI can’t solve this problem in the foreseeable future, but it won’t be an easy task.

4. Presentation of the result and making edits

Documents prepared by a lawyer must meet high standards: have no erasures, corrections, do not contain unnecessary information, and be convenient for the reader to understand.

For artificial intelligence, issuing a document according to user’s requirements is not a problem; all you need is a detailed description of the requirements in prompt.

Is it possible to teach AI the skills of a lawyer?

It is difficult to even roughly describe all the skills that a lawyer has or should have in order to be successful in legal profession. Therefore, let’s highlight only the most important ones:

1. Critical thinking

No matter how trivial it may sound, this skill is the most important for a lawyer. When performing any task, critical thinking is essential. When consulting, preparing documents, analyzing legislation, a lawyer is obliged to study different points of view, find and refer to primary sources of information, and not take anything he hears or sees for granted without checking. And while in other professions it may be possible to obtain the required result without all this “stifling bureaucracy,” for a lawyer this is literally a basic skill, without which not a single task will be completed properly.

Artificial intelligence cannot have critical thinking simply because of its technology. AI allows you to process millions of terabytes of information per second, but checking the source data is impossible – no one expects this from AI. Yes, increasing the volume of source data increases the probability of eliminating errors to 90 – 95 percent. For example, a machine can learn to isolate and indicate different points of view on the same issue. But AI is not able to draw conclusions from the existence of different opinions. And it is impossible to teach AI this skill in the foreseeable future. Moreover, it is impossible to teach him the skill of searching for a primary source – for him, all data is, as it were, equal in status: if a person can recognize the quality of information even intuitively, then this is inaccessible to AI in principle. Even worse, the AI is trained to generate “its own” content based on information it finds on the Internet, which, again, it assumes to be true by default. This means that, having discovered a dozen low-quality articles with incorrect interpretations of laws or even fictitious norms, he himself begins to generate a priori false information.

Therefore, at the moment, a common situation is when artificial intelligence needs to double-check everything. What only a competent, qualified lawyer can do. We also remember curious cases like a trial in the United States, where the AI simply invented two non-existent legal precedents (history is silent about whether he found them on the “outskirts” of the Internet or came up with them himself, but this does not change the essence). If the prompt does not stipulate such basic things as the need to use only the legislation of a specific country or only the one in force at the time of the request, the AI will not take this limitations into account, while a person understands such restrictions without words. Double-checking for AI requires time and knowledge, which actually “kills” the advantages of AI in the speed and detail of request processing. Unless someone creates certain AI-based solutions to solve these problems…

2. Logics

A lawyer needs to work with information, arranging it in a logical, consistent order.

At the moment, the AI does this poorly, but improvements are constantly being made. Unlike critical thinking, setting up AI training so that the information it generates does not contradict the laws of logic is a completely feasible task.

3. See the big picture

By default, a lawyer should mind his own business, but to effectively complete tasks, he needs to see the big picture. A lawyer must understand the client’s business, his goals, his buyers, his suppliers, his managers, his at least the basic business processes. Only in this way will a lawyer be able to provide an individual approach, develop and offer the most suitable solution for the client (and not the lawyer himself).

As we discussed above, it is theoretically possible to teach AI active listening. However, it is impossible to teach him ‘a helicopter view’. This would be contrary to the basic principle of its operation. AI sees another ‘big picture’ – an array of data loaded into it, found by it and generated by itself. Changing his focus means changing him. But then it will no longer be the AI that conquered the world. And you shouldn’t do this at all. Human intelligence and artificial intelligence in this case do not compete, but mutually complement each other – each one sees what is hidden from the eyes of the other by the design features of their own brains.

4. Emotional intellect

Probably, not all lawyers will agree that they need this skill at all. There is an opinion, especially among male lawyers, that “you have to think with your head”, and emotions only get in the way. However, many lawyers, especially women, who are the majority in the profession, actively use emotional intelligence. This works especially well when interacting with people: clients, judges, government officials. Recognizing your own and other people’s emotions, managing them – without these skills it is difficult to succeed in such a communicative profession, which is the legal profession. It is completely impossible to imagine a lawyer who has ‘turned off his emotions’ and in whose hands is literally the life of a client locked in a jail.

In the abbreviation AI, the word “intelligence” in principle means only ratio – ‘ordinary’ intelligence, IQ. The concept of ‘emotional intelligence’ (EI) is absent there. It is possible to teach AI to recognize emotions – but only from pictures and videos. He will never be able to ‘read’ the emotions embedded into the text, at least in the foreseeable future. Therefore, a client will never be as comfortable communicating with a robot lawyer as with a human.

This means that lawyers can ‘exhale’ – they are not threatened with widespread unemployment. However, in the profession now, unfortunately, there are many ‘specialists’, who have no skills at all, but only know how to perform tasks. Such ‘specialists’ should keep in mind that they will not be able to compete with AI – it is hundreds of thousands of times faster and tens of thousands of times cheaper.

Lawyer of 2030

To summarize, let us state a simple fact: AI will not replace lawyers, but in the very near future it will significantly simplify their work. The lawyer will be able to transfer all the main routine tasks to AI, concentrating on creative work that requires deep knowledge and the use of emotions.

By 2030, there will be no need for lawyers who are not capable of creative work, and the use of AI will become as an integral part of everyday routine as working on a computer is now. By the end of the decade, every lawyer must learn to create correct and competent prompts for artificial intelligence, otherwise it will become as unpopular as a lawyer who writes documents by hand is now unpopular.

And the truly sought-after lawyers will be those who can come up with and propose solutions that make maximum use of the capabilities of artificial intelligence. The future inevitably comes, and a person needs to seize new opportunities rather than try to return to the past. Robots and artificial intelligence should work for the benefit of humans, and not interfere with them.