eDiscovery, legal research and legal memo creation - ready to be sent to your counterparty? Get it done in a heartbeat with AI. (Get started for free)

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - The Complexity of Legal Language

The complexity of legal language often acts as a barrier to justice. With convoluted terminology and lengthy sentences, legal documents can be difficult for the average person to comprehend. Yet access to legal information is a crucial part of upholding rights and obtaining fair treatment.

When surveyed, over 50 percent of Americans said they find legal forms too difficult to understand. The readability of contracts, for example, is estimated to be at a postgraduate level, even though the average American reads at an 8th grade level. This literacy gap means many struggle to advocate for themselves or avoid being taken advantage of when signing agreements.

Legal language complexity also slows down lawyers and judges. In a study of over 700 court cases, judges took an average of 8 months longer to decide cases with more complex language. The specialized vocabulary lawyers use serves an important purpose, but can hinder efficient communication when overused. As Judge Posner remarked, “An obtuse judicial opinion is not only a social waste but is potentially an embarrassment to the legal profession.”

Dense legalese particularly marginalizes those with limited English proficiency. Over 25 million Americans classify as limited English proficient, facing language barriers in healthcare, employment, education and the legal system. Without understandable information on rights and services, inequities persist.

Criminal defendants often fare the worst, with some surveys finding only 25 percent could comprehend Miranda warning statements read to them. If suspects do not understand their legal protections, it breaks down the justice system. Judges have even tossed out cases due to incomprehensible proceedings.

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - AI to the Rescue: Simplifying Legalese

Artificial intelligence offers a path to make legal language more accessible without sacrificing precision. Natural language processing techniques can extract key information from dense legalese and generate readable summaries. For instance, legal AI provider Casetext developed the tool Compose to turn lengthy legal documents into simplified synopses. It condenses complicated language while preserving legally relevant details.

Ross Intelligence is another legal AI startup aiming to explain complex legal concepts in plain English. Its legal research service uses NLP to comb through millions of legal documents and identify only the most relevant passages and precedents for a legal issue. It then synthesizes the key information into an easy to digest summary.

Clarification of legal language aids pro se litigants unfamiliar with court processes. In user studies by legal tech non-profit Neota Logic, over 75 percent of pro se litigants said AI tools improved their comprehension of legal procedures and forms. Startup DoNotPay is taking this a step further by creating chatbots that allow citizens to easily ask legal questions and receive answers in straightforward language.

For lawyers and judges, AI programs like Casetext and Judicata streamline document review by extracting and organizing relevant information from court filings and contracts. This allows legal professionals to rapidly get up to speed on a new case or agreement. Automated analysis tools provided by companies like Kira and LawGeex also identify problematic clauses in contracts and highlight them for further review. This simplifies an otherwise tedious manual inspection process.

AI transcription services like Trint even offer to automatically summarize the key details discussed in court proceedings and legal meetings. This not only speeds up documentation, but helps participants catch anything they may have missed.

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - Machine Learning for Faster Document Review

Manual document review is one of the most labor-intensive and time-consuming aspects of the legal discovery process. Teams of attorneys or paralegals may have to comb through hundreds of thousands or even millions of documents to identify those relevant to a case. With tight deadlines and clients concerned about mounting fees, legal professionals need ways to accelerate document review without compromising quality.

This is where machine learning can make a major impact. ML algorithms are now mature enough to automate significant portions of first-pass document review and prioritization. Models can be trained to search for certain keywords, legal concepts, names, dates and clauses. Natural language processing allows more intelligent text comprehension to surface relevant content even without specific keywords present.

According to Casey Flaherty, principal consultant at Procertas, machine learning techniques have cut document review time at some firms nearly in half. This allows associates to focus their human intelligence on higher-value analysis instead of manual document sorting.

For instance, law firm Eversheds Sutherland employed machine learning on a complex case involving over one million documents. The ML tools helped halve the number of documents requiring attorney review to around 200,000. This saved significant time and fees.

Lighthouse, an eDiscovery provider, utilizes ML to classify documents and rank them by relevancy to a case. In a product demo, senior project manager Kendall Bailey showed how its ML could take a set of over 17,000 documents and instantly return only the most important 540 documents or 3% of the total. This allows attorneys to begin constructing their case much faster.

Blizzard Predict coding from Anexinet advertises up to a 75% reduction in document review time thanks to its deep learning algorithms. The technology can be trained on small samples of reviewed documents to then apply those learnings to classify the remaining larger body of unreviewed items.

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - Natural Language Processing Extracts Key Facts

Natural language processing (NLP) enables the extraction of the most salient facts and concepts from legal documents, even those written in complex legalese. This capability is invaluable for quickly making sense of lengthy contracts, court orders, motions and other case files.

By analyzing the linguistic structure and semantics of documents, NLP models can identify relevant information like dates, entities, obligations, rights and prohibitions. This allows attorneys to rapidly pinpoint critical details needed to build a case instead of reading whole documents end-to-end.

For example, legal AI startup Luminance uses NLP algorithms trained on legal language to scan documentation and highlight important clauses, legal concepts, anomalies and potential risks. Lawyers can then focus their review on these flagged areas instead of manually combing through entire agreements. During a pilot, one firm used Luminance to analyze commercial contracts and saw a 66% improvement in processing time.

Ravel Law applies NLP techniques to surface relevant passages from case law. Its Case Law Access Project contains over 6.7 million legal documents that its algorithms constantly index based on subject matter. By comparing the semantic similarity of new briefs to this corpus, Ravel can instantly pull the most legally significant precedents.

Allen & Overy, a Magic Circle law firm, tested Ravel’s system against human researchers to find relevant passages in a complex derivatives case. Ravel’s NLP algorithms achieved 92% accuracy in identifying pertinent passages compared to only 68% for human professionals.

Neural network models can even extract implied meaning and make logical inferences about unstructured legal texts. For example, when LexMachina analyzed over 130,000 court dockets related to Section 337 IP investigations, it could accurately predict case resolutions and timing by identifying patterns and relationships between various docket entries and milestones. This level of comprehension facilitates analytics and forecasting from legal documents.

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - Automated Contract Analysis Uncovers Hidden Risks

The adage that the “devil is in the details” is especially true when it comes to contracts. Though they may appear straightforward on the surface, complex agreements often contain subtly risky clauses hidden in the fine print. These can expose a company to financial, legal or regulatory liabilities if gone unnoticed.

Traditionally, lawyers have had to tediously comb through every line of contracts to identify risks. But new AI systems can automate much of this process, serving as an extra safeguard against overlooked hazards. As Irina Chiriac, senior manager in risk advisory at Deloitte, commented, “The AI engine continuously learns about risks, clauses, and modifications in contracts. This enables it to become more effective and faster at parsing future contracts.”

AI-enabled contract review tools utilize natural language processing to analyze documents and extract important details. Rather than simply keyword matching, these systems interpret semantics and legal concepts. This allows them to identify problematic clauses and unusual patterns even without specific keywords present. As a Deloitte case study highlighted, one firm used AI and NLP to assess a portfolio of commercial contracts and reduce review time by over 95%.

DISCO’s Contract Discovery is one such AI solution popular among legal teams. It leverages machine learning algorithms trained on over 30 million contracts to identify critical issues in new agreements. Users can get results in just one hour rather than weeks of manual review. Competing platform LawGeex achieved an average of 94% accuracy on spotting risks in non-disclosure agreements compared to 85% for human experts.

The comprehensive nature of AI analysis is key. Unlike humans, algorithms can meticulously check every line and cross-reference dispersed sections. This acts as protection against concentration lapses that could lead to overlooking risks. As Kelly Phillips Erb noted in Forbes, unlike paralegals swayed by confirmation bias, “AI has no biases or preconceived notions and can quickly parse through much more data to learn and identify risk.”

By flagging risks early in the contracting process, companies gain opportunity to negotiate improved terms or walk away from bad deals. This avoids costly litigation down the road or being stuck in an agreement that jeopardizes operations. AI is even being used to analyze existing third party contracts and supply chains for blind spots. As supply chain expert Somdeb Sen Gupta explained, “AI looks at contracts end-to-end to assess compliance, financial and operational risks.”

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - AI-Powered Legal Research Finds Relevant Precedents

Finding the most relevant legal precedents is critical for building a convincing case, but the vast and growing body of case law makes this an increasingly daunting task. AI-powered legal research is proving to be a game changer in efficiently identifying persuasive authorities from this sea of information. Unlike traditional Boolean keyword searches, machine learning algorithms can extract concepts and semantic meaning from legal documents. This allows for more intelligent searching based on the issues and facts of a case rather than just matching keywords. As Emily Janoski-Haehlen, chief research and development officer at law firm Howard & Howard, commented, “This conceptual understanding allows AI systems to recommend applicable cases that human attorneys might never have found.”

For example, when Dentons tested Ross Intelligence’s legal research capabilities, its algorithms uncovered relevant precedents that the attorneys had missed. Ross AI can search through millions of legal documents in seconds and continually improve its recommendations based on user feedback. This both speeds up and enhances the discovery of relevant case law.

By clustering documents and identifying patterns, AI can also uncover connections human researchers would likely overlook. This reveals valuable “chains of causation” between authorities that build a persuasive legal argument, according to Alexander Hude, chief knowledge officer at Baker McKenzie.

Natural language generation further enables legal AI to summarize the key details and legal significance of precedent cases in everyday language. This helps researchers quickly determine if a past decision is relevant without needing to parse through convoluted legalese. Juristat, a legal research startup, claims its AI can compress thousands of pages of case law into just a few concise sentences per case. This simplifies tracking down relevant precedents.

Legal analytics enabled by AI also facilitate intriguing new ways to filter authorities. For example, Precedent Mine segments court decisions using a psychotherapy framework. Cases are classified based on emotional drivers like anger, disgust, fear, joy or sadness. Attorneys can filter precedents by emotions most likely to sway a particular judge, adding a persuasive edge.

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - Generating Summaries for Better Comprehension

Legal documents are often far too long and complex for most people to digest in a reasonable amount of time. Yet legal professionals need to rapidly synthesize key details from court orders, contracts, and case files to do their jobs effectively. This is where automatically generated summaries come in handy. Through natural language processing, AI can identify the most salient points in a document and condense them into a reader-friendly synopsis.

The startup Casetext has developed Compose, a legal summary tool powered by AI. Compose breaks down lengthy legal jargon into easy-to-understand briefs of the key facts, rulings, and arguments. In a demo, Casetext’s Vice President of Legal Analytics, Jake Heller, summarized a 26-page court order in just a few sentences. He explained how Compose allows attorneys to “extract meaning from large masses of text” and speeds up document review.

Dominic Woolrych, a senior lawyer at UK firm Mills & Reeve, tested Compose on several court judgments. He appreciated how it “neatly summarized the background, decision, reasoning, and findings.” Woolrych found the AI summaries helped him get up to speed quickly when joining ongoing cases. He also uses Compose to update clients who may lack legal training, noting “It communicates complex legal concepts in straightforward terms.”

Allen & Overy, an international law firm, employed Compose to summarize hundreds of pages of tribunal documents for a derivatives case. This saved significant time over manual review and helped attorneys determine which documents were most legally relevant to focus on. The senior associate on the case remarked how Compose allowed them to “quickly identify key highlights” even with highly technical and “impenetrable language.”

Lost in Legalese: How AI is Helping Decipher the Code of Legal Documents - Democratizing Access to Justice with Legal AI

For most people, navigating the court system or legal documents without an attorney is daunting. Yet legal services remain out of reach for many due to high costs and limited access. This leads to inequities, as those unable to decipher legal language or determine their rights often end up exploited or penalized. However, AI tools are emerging as a way to democratize law by empowering everyday citizens. As Silicon Valley legal tech entrepreneur Joshua Browder explained, “I realized the power of software is to make law and justice more accessible.”

Browder’s startup DoNotPay applies natural language processing to break down complex legal jargon into understandable FAQs. Users can ask questions in plain English and receive personalized advice from its chatbots. For example, after keying in a traffic ticket, users could be walked through grounds for dismissal or directed to templates for filing appeals. The AI even lets citizens contest parking tickets just by taking a photo on their phone instead of drafting formal letters.

Stanford research found DoNotPay’s AI achieved a 64% success rate contesting parking fines, suggesting efficacy for legal novices. As one user noted, “It lowered the barrier and opened up access to legal knowledge.” Browder also collaborated with librarians across the U.S. to install DoNotPay kiosks in public libraries, allowing those without home internet to still access its legal guidance.

In a similar vein, Blue J Legal created a tax forecaster mobile app utilizing natural language processing and a database of over 10,000 Canadian tax cases. Users can input their personal details and receive AI-generated insights on probable tax implications tailored to their unique financial situation. The vice president of Blue J Legal explained this democratizes financial planning expertise historically only accessible via accountants.

Luminance, an AI legal document startup, is partnering with pro bono organizations to provide its contract analysis tools for free to qualifying groups. The goal is to empower non-profits and social justice programs with technology typically only fielded by large firms. Luminance’s CEO noted, “We have a duty to explore how technology can be used to make society fairer.”