Upon first hearing about AI in legal practice, what first came to mind was clientele across the nation seeking resolutions to complex and dire cases, now being subjugated to the same robots that hallucinate nonexistent court cases and instruct people to eat one small rock per day. Disconcerted, I was quick to discredit the potential that AI platforms could have in the legal sphere.
But with attention from countless research groups, investors, consumers and even the White House, it’s important to recognize that the scale and acclaim of AI platforms are surging with no signs of slowing down any time soon. Avoiding it entirely is no longer an option.
We are living through a crucial period in the evolution of AI. Industries across the globe are calling for its responsible and measured use, and yet concrete guidelines for achieving this aim remain scarce. Evaluating the successes of AI in America’s largest law firms can offer valuable insight into how these technologies can be both effectively harnessed and ethically constrained, since completely shying away from them is no longer feasible. We don’t have to blindly delve into the rich tapestry of everything AI has to offer. But if law firms want to keep up with demands from the industries they serve, they should approach the overhaul of analog methods with a rigorous and discerning lens.
Two notable platforms — Relativity and Kira Systems — point to particularly hopeful benefits in firms throughout the country. By refining tailored, cautious focuses and workshopping transparent models with reliable consistency, these platforms ultimately seek to accommodate the demands of legal professionals, not replace them. That’s the key to successful AI implementation, coupled with a healthy dose of accuracy checks and a sharply critical eye to technological overreach.
Relativity is a multicontinental software development company. Phil Saunders, Relativity’s chief executive officer, described the company’s efforts as “putting legal data intelligence into action, transforming legal work with generative AI and embracing the next step toward cloud transformation.” RelativityOne is the company’s versatile and transparent generative AI model for legal practitioners. It was created to specialize in litigation, investigations, regulatory requests and data breach response, with specific capabilities tailored to meet the unique needs of each domain.
Use of AI in litigation — a field highly contingent on ethical, human discretion — should not be zealously or broadly accepted. AI does not possess the same level of understanding or the ability to navigate these complexities with the reasoning skills as a trained attorney.
Fortunately, Relativity currently emphasizes e-discovery technology for litigation firms, which aids only the initial step of litigation referred to as discovery. During discovery, parties involved in a dispute must share all relevant information, including records and any other evidence related to the case. E-discovery simply adds technologized support to this step of the process. Relativity’s aiR for Review is a large language model component of RelativityOne used for e-discovery. It optimizes contract review by making complex, archaic statutes more digestible and emphasizing their relevance on a case-by-case basis.
Teneo, a global CEO advisory firm, used aiR for Review to reduce a large-scale investigative project that would have taken several months into just 18 days, effectively cutting associated costs by 70%. Teneo’s experts worked iteratively alongside aiR for Review to periodically verify accuracy, which gradually expanded to larger batches of documents. The results proved that aiR for Review could process up to 150,000 documents at a time with high reliability, even eliminating misunderstandings in the process.
Relativity described that Teneo was not only impressed by aiR for Review’s efficiency, but also by the clarity and depth of its natural language rationales. Explanations accompanied each prediction, enabling the review team to grasp subtle patterns in the data and pinpoint categories that didn’t require further analysis. AI technologies, when strategically implemented, can significantly enhance legal processes by drastically reducing both time and costs with an assured level of accuracy and integrity.
Kira Systems, based in Toronto, operates in seven out of 10 of the U.S.’s Vault firms. Its founder, Noah Waisberg, had a background in law before moving into technology. Kira is a machine learning platform whose primary function is also contract analysis. The platform’s ability to improve speed and accuracy in contract review enhances precision and cost-effectiveness.
Kira leverages built-in provision models called Smart Fields and a patented machine learning capability. Smart Fields are enacted as safeguards for accuracy, developed and trained by a team of human legal experts across a range of subject matters. They can extract common clauses, provisions and data points across a variety of legal domains with impressive accuracy even in different Latin-based languages. This makes them especially useful for professionals practicing across international boundaries.
Kira’s Quick Study supervised machine learning capability allows users to create their own Smart Fields for further specialization. The platform requires a minimum of 90% recall, reducing the risk of error in the contract review process. For reference, a study published in the Richmond Journal of Law and Technology found that the maximum amount of recall in manual review is 65%, while other e-discovery software manages about 75% to 80%.
So, technical jargon and case studies aside — what is it that makes these AI platforms beneficial? How are they distinctive from the largely fallible platforms that undermine rational analysis we have come to know and loathe?
Machine learning and language processing engines are the ones most reliably used to analyze and process vast amounts of pre-existing, expert-verified legal data. These models perform best when they are trained on strictly legal text, engineered with supervision and checked for reliability by human experts.
For now, legal AI platforms are most widely used for contract law, which heavily emphasizes extensive negotiation and drafting agreements, unlike prosecutorial fields. AI is employed to automate repetitive tasks associated with contract law, such as document review. This means legal professionals practicing contract law can benefit from AI by focusing on more pressing, less tedious work. Fields of law more contingent on litigation may not see the same outcomes with these platforms since the nature of their practice differs.
Leveraging transparent and thoughtfully developed attributes has already made AI a powerful tool for legal professionals by driving greater productivity, reducing costs and ultimately improving the quality of legal services. As AI continues to evolve, its role in transforming the legal landscape becomes increasingly essential — just as essential as critical and well-informed application of these technologies.
Casey Mendoza is an Opinion Intern for the winter 2025 quarter. He can be reached at caseym4@uci.edu.
Edited by Jaheem Conley