What Are the Current Limitations of Legal AI?

Even if you don’t consider yourself much of a technology enthusiast, you’ve likely noticed the explosion in artificial intelligence (AI) technologies we’ve seen in the past few years. Complicated professions once thought to be the exclusive domain of human beings, like legal professions, are now facing major disruptions and changes because of the encroachment of AI.

But what exactly are the current limitations of legal AI? And should lawyers be concerned about legal AI taking over the entire profession?

Legal AI Core Capabilities

According to Law.co, legal AI is already quite advanced, and “capable of making almost any lawyer’s job easier.”

Currently, legal AI seems competent at the following:

  • Document review. If you provide a series of documents to an appropriately configured legal AI engine, this technology can quickly process those documents and reveal insights that might take human lawyers hours (or days) to generate. It’s a way of greatly accelerating the traditionally tedious and time-consuming document review process.
  • Legal research. AI is more than just a glorified search engine, but when it comes to legal research, it’s helpful to think of it that way. Legal AI is exceptionally good at helping lawyers research complex topics, facilitating much faster knowledge transfers.
  • Document drafting. Legal AI is also heavily used for document drafting. This is especially true of simple, somewhat repetitive contracts, like wills or powers of attorney. With a relatively short prompt, you can generate a good rough draft of a contract or other legal document and edit it to perfection.
  • Summary and analysis. Legal jargon can sometimes be difficult to wade through, even if you’re an experienced lawyer. That’s why legal AI is especially valuable for its ability to summarize and analyze complicated texts.
  • Automation. Certain AI applications can also automate various aspects of legal jobs, helping lawyers and paralegals get more done in less time.

What Legal AI Can’t Do

In their current state, there are many things that legal AI engines can’t do.

  • Attend hearings. Despite some strong efforts, AI-powered robot lawyers aren’t able to attend hearings or argue in court. It’s debatable how well they would fare in an environment like this, but the general consensus is that this is the exclusive territory of humans, at least for the foreseeable future.
  • Answer complex questions. AI may be capable of processing and understanding complicated questions, but it’s not necessarily equipped to answer them accurately. For certain legal questions and issues, human lawyers still know best.
  • Draw from relevant experience. Similarly, only human lawyers can draw from relevant experience. A legal AI can help you crunch the numbers or put your thoughts into better words, but only seasoned human lawyers have an intuition for the legal system.
  • Form relationships. Finally, and perhaps most importantly, legal AI applications aren’t capable of forming relationships. People seeking legal advice often value their relationship to their lawyer as much as the actual advice that they dispense. For now, it doesn’t seem like legal AI will ever be fully capable of replacing this dynamic.

Weaknesses and Blind Spots

There are also some weaknesses and blind spots associated with current legal AI.

  • Biases. It’s important to recognize that AI engines are still programmed by humans, and humans are plagued with biases and distorted thinking. In many cases, AI is also plagued by biases and distorted thinking.
  • Factual accuracy. Generative AI is notorious for being unreliable in terms of factual accuracy. AI may be capable of helping you research – but it still needs a firm guiding hand (and an additional round of fact-checking when everything’s done). There are ways to mitigate this issue, but no ways to fully eliminate it.
  • False confidence. Perhaps the most glaring weakness of the use of AI in the legal profession isn’t a weakness of AI itself, but rather, is a weakness of the people leaning on AI too much. It’s easy to become impressed, or even enamored with legal AI after using it to automate or accelerate tasks you don’t really like. But after a few rounds of this, you may develop a false sense of confidence or security in this sophisticated, yet limited tool. Already, we’ve heard stories of lawyers who have made egregious mistakes simply because they trusted legal AI to do something it couldn’t – or because they got a bit lazy in the wake of incorporating this disruptive technology.

Toward a More Advanced Future

This article is written with the current state of legal AI in mind. But it’s hard to say exactly where AI goes from here. While we can make some reasonable predictions about what will or won’t be possible in the future, it’s a fool’s errand to try and predict beyond the current foreseeable horizon; humanity’s greatest technological leaps forward are some of the least predictable.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top