Legal research tools in the age of generative AI
Nicole Black.
A little over a year ago, ChatGPT-4 was released, and in that short time, the functionality of generative artificial intelligence has been added to many of the tools that we regularly use in our day-to-day workflows. From research and brainstorming to data analysis and translation, generative AI is changing how work gets done.
The legal profession has not been immune from this trend, and legal technology companies have shared news of generative AI plans and releases with increasing frequency since the start of the year. One of the most notable categories of generative AI integration has been legal research, with many established companies embedding it into their platforms.
Generative AI and legal research are a great fit because the large language models utilized by generative AI tools rely on massive datasets to ensure accurate and appropriate responses to inquiries. The internet era has seen the digitization of statutes, regulations and caselaw, thus enabling the training of LLMs for legal research purposes.
However, despite the abundance of training data, generative AI tools are still in their infancy and sometimes provide inaccurate responses, also known as “hallucinations.” Lawyers continue to make the news for utilizing these tools for legal research and submitting documents to courts that contain false case citations.
These errors happen because attorneys fail to review the submissions and blindly rely on the output of generative AI tools. Most often, the tools used in these cases are consumer-based chatbots, such as ChatGPT, which are more prone to hallucination in the legal research context.
For that reason, I recommend that lawyers use only generative AI legal research tools developed by legal tech companies because those providers understand the legal professions’ unique practical and ethical requirements and can reduce the frequency of inaccurate responses.
Given the nascent state of the technology today, even legal-specific products sometimes hallucinate and provide inaccurate outputs.
As a result, you must treat generative AI tools just as you would an entry-level associate and carefully review all responses to ensure their accuracy, including case citations and the cases themselves. The failure to do so violates your competency obligations and could result in professional disciplinary action, court censures or even news headlines. So take advantage of these tools and the savings on time that they offer, but don’t cut corners by ignoring basic ethics rules.
When I last wrote about legal research options in 2019, AI was being used to provide more accurate results, but generative AI was not available.
Now that generative AI is embedded into many legal research products, the challenge becomes sifting through the available options and choosing the right tool for your law firm’s legal research needs. In this article, I’ll highlight many of the products that include generative AI features that are publicly available or in beta.
Choosing a generative AI research tool
Below, I’ll provide a detailed look at some of the leading solutions in the market, including their pricing details when possible. Whether you’re a seasoned legal professional or just beginning to explore the potential of AI in your practice, this overview is designed to equip you with the knowledge needed to make the right choice for your firm.
When choosing a legal research tool, understand that the platforms discussed below are cloud-based. As a result, all your firm’s data will be housed on servers owned by a third party, and the duty of technology competence requires you to vet all technology providers that will be hosting and storing your data. You must fully understand how that company will handle the data; where the servers on which the data will be stored are located; who will have access to it; and how and when it will be backed up, among other things.
Westlaw Precision
Now, let’s explore key considerations when selecting a generative AI legal research tool, starting with Thomson Reuters, which has invested heavily in AI in recent years.
Its legal research tool, Westlaw Precision, now includes generative AI-assisted research that relies on a combination of Microsoft Azure and OpenAI’s commercial API for GPT-4. This functionality was launched publicly in November 2023. With it, users submit research queries, and generative AI-powered responses are provided in narrative form, including footnotes to supporting citations. Responses will sometimes take between 90 seconds to three minutes.
Westlaw Precision pricing is not available on the website.
Lexis+ AI
LexisNexis also offers generative AI legal research functionality via Lexis+ AI, which was launched in October 2023. This tool utilizes several LLMs, including GPT-4 and Anthropic’s Claude 2, depending on the task requested.
Users can conduct legal research, draft documents, seek answers to legal questions, and summarize cases and documents. Responses to queries include linked citations to sources.
Pricing depends on the features and datasets used. And to ensure availability to firms of all sizes, modules for different types of functions—such as drafting or summarizing—can be purchased separately. Pricing is not available on the website.
Paxton AI
Another option to consider is Paxton AI, which is a startup that provides a generative AI legal research tool that also drafts documents and includes contract review features. Coverage includes caselaw, statutes and regulations for all 50 states, including federal data. Research results include links to the cited sources.
Pricing starts with a basic package, the free tier, which only allows a limited number of queries and limited access to statutory and regulatory data. If you’re not a student, pricing for access to all databases without limits starts at $99 per user per month.
Vincent AI
Next up, is vLex’s research assistant, Vincent AI, which was upgraded in October 2023 to include generative AI features available in beta. This tool can be used to create a research memo, build an argument, spot legal issues, draft headnote summaries, conduct a search by uploading a document, and obtain related authorities. It can also be accessed directly in Microsoft Word when drafting a memorandum of law or other documents.
Pricing isn’t available on the website.
Copilot
Finally, another option is LawDroid’s Copilot. Copilot was launched in beta in January 2023 and released publicly last month. With this product, users can conduct legal research using Copilot’s generative AI legal assistant interface. It can also be used to draft, translate and summarize documents.
The cost is $19 per user per month, and a free trial is also available.
When the time is right
Generative AI can be a game-changer for legal research, allowing you to access and analyze information more efficiently than ever.
However, built-in capabilities and costs can vary greatly, so choosing the right tool for your firm isn’t always easy. Given the rapid rate of technological advancements, there’s no better time than now to review your options and learn more about generative AI-powered legal research tools.
Whether you adopt these tools right away or down the road, by obtaining foundational knowledge about this technology, you’ll be better equipped to wisely invest when the time is right.
Nicole Black is a Rochester, New York-based attorney, author and journalist, and she is senior director of subject matter expertise and external education at MyCase, a company that offers legal practice management software for small firms. She is the nationally recognized author of Cloud Computing for Lawyers and is co-author of Social Media for Lawyers: The Next Frontier, both published by the American Bar Association. She writes regular columns for ABAJournal.com and Above the Law, has authored hundreds of articles for other publications, and regularly speaks at conferences regarding the intersection of law and emerging technologies. Follow her on X (formerly Twitter) @nikiblack, or she can be reached at [email protected].
This column reflects the opinions of the author and not necessarily the views of the ABA Journal—or the American Bar Association.
