AI for Legal Research: A Guide for Legal Professionals

Can AI do legal research?

Yes. Legal research is one of the most common ways legal professionals use generative AI in their practice. Ideally, researchers can use legal AI tools to carry out comprehensive research more efficiently. However, there are some caveats.

First, let’s be clear on what artificial intelligence refers to. In its simplest form, AI is an overarching description of technologies that use computers and software to create intelligent, humanlike behavior. Specifically, “generative AI” technology can respond to specific user prompts by drawing on enormous data sources to instantaneously create seemingly new, task-appropriate content such as essays, blog posts, poetry, designs, images, videos, and software code.

The benefits of using AI for legal research

The benefits of AI in law are clear: the unmatched speed of AI layered into a top legal intelligence platform can help legal professionals prepare better and faster than ever before, resulting in added efficiencies, more streamlined workflows, better training and education, elevated profits, and, ultimately, greater client satisfaction. Legal AI tools can automate time-intensive tasks such as reviewing evidence or producing initial drafts of motions, legal briefs, contracts, and settlement agreements. However, it also creates serious risks if not used properly.

Faster, more comprehensive docket searches

Crafting searches is already a critical skill for using legal research software effectively. The right docket search query can provide a wealth of information about relevant cases, jurisdictions, judges, and opposing counsel.

In the past, legal professionals have conducted docket searches manually, which required lots of time. But now, AI technology can help legal professionals carry out their research much faster by sifting through documents in seconds instead of hours or days.

Leverage data for more actionable insights

Because these AI technologies can sift through thousands of cases so quickly and comprehensively, modern legal research tools can make instantaneous associations between related case law and help aggregate or summarize data in more useful ways.

For example, before litigation analytics were common, a partner may have asked a junior associate to find all summary judgment motions ruled on by a specific judge to determine how often that judge grants or denies them. The attorney could have manually searched PACER or searched through court opinions, which would have taken a long time. Now that legal research can be done so much more efficiently with AI-powered tools, it frees up time and effort a legal professional can use for other activities that benefit their client.

Retain more work in-house

Given these efficiencies, some legal professionals may wonder: Can AI replace paralegals?

Rather than replacing jobs such as those of paralegals, it’s more likely that AI will help legal teams – especially in-house legal departments – to retain more work in-house. Both in-house attorneys and lawyers practicing at firms say that generative AI will help reduce their reliance on outside counsel, which will allow these teams to be more selective in the work they outsource and give them more leverage to make alternative fee arrangements for work sent to outside counsel.

There is potential for AI to impact certain duties performed by legal assistants, paralegals, and even early-career attorneys. But the rules of ethical conduct demand that lawyers must train and supervise nonlawyers who assist them. That duty extends to their deployment of AI.

The risks of using AI for legal research

While AI tools offer automation capabilities that can resolve inefficiencies in legal practice, the use of AI technology raises serious issues for the legal field, including:

• Algorithmic bias – This occurs when the algorithm itself is biased, or when an otherwise unbiased algorithm is trained on biased data
• Hallucinations – The phenomenon by which AI chatbots may confidently provide false information in response to a prompt
• Inaccuracies
• Confidentiality concerns

Some legal professionals have been deploying AI for the better part of a decade to parse data and query documents. But the explosion of interest in certain powerful AI tools for general use has led to questions and concerns within the legal profession – and even led some law firms and in-house legal departments to entirely ban the use of this emerging technology due to the potential risks.

“I think the thing that most people are worried about when they do research using generative AI is hallucinations,” said Cohen, pointing to news reports of lawyers who have submitted made-up citations generated by faulty AI research. “That is something that Bloomberg Law takes very seriously, and we’re working on ways to make sure that the data we pull from and the results that we give back are complete.”

Use AI to enhance, not replace, legal teams

AI shouldn’t take the place of the review, analysis, and judgment of an attorney. AI tools should be used to enhance the legal research process by providing more data on a wider array of variables so that a legal professional can take even more information into consideration. Legal professionals should still rely on their own experience and knowledge of their clients, the facts, legal precedent, and expert opinions.

Which is the best AI for legal research?

The best AI for legal research should be designed just for that: legal research.

Legal AI tools are valuable because they can analyze massive amounts of data and unearth details that are undetectable to the human eye – potentially minimizing errors and helping researchers avoid missing important, relevant documents. Yet without human expertise and supervision ensuring the quality and accuracy of data, AI can potentially do more harm than good.

“Being able to analyze – that’s the bread and butter of attorneys,” said Cohen, noting that legal professionals must know how to apply their research and make a case. “That’s what the best legal platforms do that regular generative AI doesn’t.”

Recent advancements in AI technology require an extraordinary level of discernment and testing. General-use AI tools don’t have the capabilities to support the layered, nuanced research that lawyers need.

The latest legal research tools use AI technologies such as natural language processing and machine learning to help legal professionals more quickly discover valuable information that would otherwise take hours of manual research to find.

Natural language processing improves traditional legal research methods

Natural language processing refers to a computer’s ability to understand human language in the way that people naturally speak and write it.

Today, users are generally familiar with natural language searching, which is widely used on popular internet search engines. As platforms continue to adapt to changing technologies, legal researchers will begin gravitating away from traditional ways of searching such as using keywords or Boolean search.

One potential downside: These search engines can apply algorithms to rank the relevance of results, and the reasons for those rankings may not be obvious in all platforms. You may need to interpret whether the search is giving you everything you need.

When considering legal research software, check to see if the tool you’re vetting offers context on results. For example, Bloomberg Law’s Smart Code tool quickly pulls case extracts related to your natural language search and ranks each case interpretation as strong, moderate, or weak, so you don’t waste time on results you don’t need.

Supervised machine learning minimizes risk

Machine learning refers to a computer’s ability to “learn” by using algorithms to analyze data, find patterns, or predict outcomes.

There are three types of machine learning to understand when considering how to use AI for legal research:

1. Supervised machine learning: The AI application seeks and recognizes patterns within predefined data sets. These data sets are typically created by human domain experts who act as “guidance counselors” of sorts to the machines.
2. Unsupervised machine learning: The AI application creates data sets without known outputs or predefined data. The application is, in essence, learning and adapting to inputs on its own.
3. Reinforcement learning: The AI application uses an algorithm to create correlations. By incorporating rewards and learning from data feedback, the application uncovers the ultimate processing path.

The best AI for legal research is one that incorporates supervised machine learning tools. This approach offers the benefit of faster research than ever with less risk of inaccuracies or missing documents. Supervised machine learning can automate the tedious process of reviewing legal documents, checking citations, searching case law, and finding additional relevant content. It can also help you identify various document types across court dockets, so you’re able to search for a precise filing type.

Because the legal field places supreme importance on accuracy, legal professionals should be extremely cautious about relying on tools that use purely unsupervised learning techniques.