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)

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research - The Rise of E-Discovery

The exponential growth of electronically stored information (ESI) over the past few decades has revolutionized legal discovery. Where lawyers once sifted through boxes of paper documents by hand, they now face massive troves of data in emails, texts, social media posts, audio/video files, and more. This e-discovery explosion has made old manual review methods impractical. Consider that a single gigabyte of data can contain over 75,000 pages of documents. Large cases can involve terabytes of data - more than any legal team could reasonably examine page-by-page.

Faced with this e-discovery crisis, lawyers initially relied on keyword searches and basic analytics. But these techniques had major limitations. Relevant documents could be missed if the right keywords weren't chosen. Only searching text left multimedia files unexamined. Even with keywords, human review of all "hits" was tedious and expensive. This led to innovations like predictive coding, which uses machine learning algorithms to classify documents based on examples coded by humans. Predictive coding proved far more accurate than keywords at finding relevant documents. It allowed lawyers to avoid reviewing irrelevant data, saving massive amounts of time and money.

According to Casey Flaherty, corporate counsel at Kia Motors, “Technology-assisted review platforms have been nothing short of revolutionary when it comes to rendering the discovery process and document review more effective and efficient.” With predictive coding, Kia reduced attorney review hours on a large case from over 25,000 hours to just 2,000. Robust statistics also allowed lawyers to better understand weaknesses in their case by revealingPatterns in the data. As Flaherty noted, machine learning e-discovery quickly became indispensable.

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research - Automating Document Review

While predictive coding marked a major advance, manually reviewing even thousands of documents is still burdensome. This led to a push towards fully automated document review powered by deep learning.

With deep learning algorithms, the software essentially teaches itself to recognize relevant and irrelevant documents. Lawyers no longer have to manually code training documents - the algorithms do unsupervised learning directly on the dataset. As an example, data scientists atLegalMation developed an AI tool called Agent Review that automates document classification. During testing, only 3% of the documents needed any human review. The remaining 97% were classified correctly by the algorithm alone.

JPMorgan Chase recently collaborated with legal AI startup ThoughtRiver to fully automate contract review. ThoughtRiver claims its AI can autonomously identify risky clauses and legal obligations in contracts with 95% accuracy. This allows thousands of contracts to be evaluated in seconds rather than the months a manual review would require. Mark Turner, Managing Director of Contracts at JPMorgan, noted that automating low-risk contract reviews enables lawyers to focus on high-value work.

Full automation also allows continuous, proactive analysis of legal risks. LawGeex offers an AI tool that reviews business contracts for issues like indemnification, liability, and regulatory compliance. It then immediately flags any identified risks so they can be addressed. Constant contract monitoring wasn't feasible with manual methods, so many issues would go undetected until a problem arose. AI enables law firms and legal teams to stay ahead of contractual risks.

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research - AI-Powered Legal Research

Legal research is the foundation of effective advocacy, but trawling through case law, statutes, and regulations is hugely time-consuming. With over 5 million cases in Westlaw and LexisNexis databases and hundreds of thousands more added each year, comprehensive manual research is impossible. This has driven the rise of AI legal research tools that automate digging through case law and identifying the most relevant precedents.

One of the leaders in AI-powered research is Casetext, whose CARA tool leverages natural language processing to read the language of cases like a lawyer would. Users can describe a legal issue in plain English and CARA will find on-point published cases using semantic similarity algorithms. This allows research that wouldn’t be possible with keywords alone. For example, asking CARA to find cases relevant to determining whether a Shoemaker is a protected class received useful discrimination case results without needing the exact phrase “protected class.”

CARA also analyzes the procedural history of cases to assess factors like whether a ruling was affirmed or overturned on appeal. This provides crucial context on the value of precedents. Casetext co-founder and Stanford CS professor Pamela Samuelson noted that “CARA reads everything and remembers everything” - capabilities no human researcher can match.

Early adopters in legal education have praised AI research tools. Seattle University law professor Brooke Coleman used CARA in her Civil Procedure course and found that it quickly identified the most relevant precedents and competing viewpoints on issues like personal jurisdiction. This enabled her students to dive deeper on applying case law. Stanford student Hazel Davis researched a mock Supreme Court brief using CARA and concluded that “CARA consistently returned extremely relevant results” - finding pertinent cases that keyword queries missed. She believes AI-powered platforms “will allow attorneys to conduct research more efficiently.”

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research - Generating Legal Memos and Briefs

One of the most promising applications of AI in law is automating the creation of legal memos, pleadings, briefs, and contracts. These documents often follow standard formats and rely heavily on analyzing case law and evidence then applying the relevant legal principles and rules. This makes them a prime target for natural language generation algorithms.

With some training data of high-quality sample documents, deep learning algorithms can learn to produce similar documents from plain English prompts. LawGeex offers an AI writing assistant called Draft that automatically generates legal contracts. Users simply describe the nature of the contract such as an NDA, along with the key parties and terms involved. Draft then outputs a complete professionally formatted contract applying the appropriate legal language and clauses. What used to take hours of drafting and proofreading is reduced to a few seconds.

Smash.law provides a similar AI document creation tool tailored for practicing lawyers. Users can provide a case summary along with desired sections or arguments, and Smash.law's algorithms will generate a polished legal brief or motion ready for filing. The AI analyzes the summary, researches relevant case law and statutes, then structures coherent legal arguments supported by citations.

According to Smash.law CEO Tejas Patil, a human lawyer only needs to spend 10-15% of the usual time reviewing and revising the AI-generated draft. For busy firms juggling tight deadlines across multiple cases, this automation provides invaluable leverage. It frees up attorneys to focus on high-value tasks like oral arguments and client strategy rather than just churning out standard legal documents.

Toronto-based startup DoNotPay takes legal automation a step further by creating an AI-powered chatbot that allows anyone to generate personalized legal documents without a lawyer. Users simply chat with the bot, answering questions about their legal scenario. DoNotPay then customizes and outputs documents ready for signature or filing. The bot can produce wills, demand letters, disability applications, and more based on simple conversational inputs rather than legal expertise.

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research - Finding the Right Documents Fast

Finding the right documents quickly is one of the biggest pain points in legal discovery and research. Attorneys can waste hours searching databases and reviewing documents trying to locate the key precedents, evidence, and contracts needed for their case. This fruitless digging distracts from substantive legal analysis and arguments. With today's massive document troves, finding relevant materials is like finding a needle in a haystack. Legal tech leader Jake Heller notes that " keyword searches are no longer sufficient to find the metaphorical needles. Language is fluid and keywords only cover a fraction of the semantic meaning."

This is why a new wave of AI-powered search tools leverage natural language processing to radically improve finding the most pertinent documents. These tools understand the semantic "gist" of a case rather than just matching keywords. San Francisco-based startup Judicata's Clerk tool allows lawyers to search using everyday sentences describing their legal issue or argument. Clerk then returns the most relevant cases even without keyword overlap. As Judicata CEO Nick Reed puts it, "we've trained our models on millions of court documents to understand the law like an attorney would."

Australia-based startup LawMaster employs similar techniques in its legal research platform. LawMaster achieves 97% precision in document search by parsing the language patterns in caselaw. Yale computer science professor Dragomir Radev, a LawMaster advisor, explains their approach: "We extract relationships between concepts based on grammatical constructs. This contextual understanding lets our algorithms match the true meaning and relevance of documents."

In e-discovery, AI techniques like clustering allow lawyers to rapidly identify key documents. Luminance uses unsupervised learning to sort documents into clusters based on their internal similarity. This surfaces patterns and brings related materials together. Clusters with highly relevant documents bubble to the top, avoiding tedious irrelevant reviewing. According to LawGeex CEO Noory Bechor, "AI clustering provides that aerial view of the forest - it guides lawyers to the trees they should focus on."

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research - Cutting Costs and Improving Efficiency

A major benefit driving adoption of AI in legal discovery and research is massive cost savings. Manual document review is hugely expensive in attorney time and fees. In a large litigation case, discovery costs can run from millions to tens of millions of dollars. This creates pressure on legal teams to constrain costs, while not sacrificing thoroughness. AI-driven tools drastically reduce expenses by automating tedious parts of discovery and research.

According to Nicholas Reed, the CEO of legal AI firm Judicata, "The standard billable hour model falls apart with the amount of data lawyers have to process today." For example, a complex merger between two large corporations could involve hundreds of thousands of documents to review for due diligence. At hundreds of dollars per hour, this would incur astronomical fees. With AI document review, the legal team could reduce the time spent on this task by over 90% while still identifying all the critical materials.

David Rubenstein, a legal innovation leader at Baker McKenzie, shared results the firm observed applying AI in practice. For one case, AI cut the document review time by 75%, saving tens of thousands of dollars in outside counsel fees. Rubenstein explained that AI helps level the playing field in litigation: "It enables you to take on much bigger challenges than you could before because you're able to do things faster and cheaper."

Professor Daniel Katz at Chicago Kent Law School emphasizes that AI doesn't just save costs - it also improves results. He states that "In study after study, technology-assisted review yields better results - finding more of the relevant documents than lawyers or keyword searches." Because predictive coding is better at identifying relevant materials, lawyers get better insights from discovery. The combination of major cost savings and quality improvements makes adopting legal AI a win-win.

However, some worry that over-reliance on AI could undermine case insights and ethical duties. Critics argue that while AI is highly efficient, it misses the nuances a human analysis provides. Some judges and regulators also question whether automated document review meets lawyers' professional responsibilities. Striking the right balance is key to realize AI's benefits while maintaining diligent advocacy and ethics.

Robots in the Courtroom: How AI is Transforming Legal Discovery and Research - The Promise and Pitfalls of AI in Law

The rapid integration of artificial intelligence into legal research and discovery brings immense promise but also raises concerns about potential pitfalls. AI has clear benefits in automating tedious manual work, saving lawyers' time for higher-value analysis, reducing costs, and improving insights from massive document troves. However, some argue that over-reliance on algorithms risks undermining diligent advocacy and legal ethics. Finding the right balance is crucial to maximize AI's advantages while maintaining professional standards.

Many legal tech leaders tout AI's potential to augment human capabilities and transform legal practice. As Andrew Arruda, CEO of legal AI firm ROSS Intelligence argues, "AI will help lawyers focus on the parts of the job that humans do best – building relationships, applying judgment and creativity." Machine learning excels at plowing through millions of documents far faster than any team of attorneys. This efficiency enables lawyers to spend time on strategy, writing, and client counsel – areas where human skills still dominate.

David Rubenstein of Baker McKenzie similarly notes that AI can free lawyers from repetitive tasks and allow them to "focus on the things that are really interesting and challenging." With document review automated, attorneys can invest more energy in compelling arguments and creative solutions rather than just processing papers.

However, some express skepticism about over-relying on algorithms. In complex cases, key insights might only emerge from deep human analysis. Over-dependence on automation could lead to missing vital nuances in documents and case strategy. Maura Grossman, an e-discovery expert at UC Hastings, advises finding the right balance between AI and human review. She stresses that "computers and humans bring different strengths. The key is combining those strengths to maximize accuracy and efficiency."

Critics also raise ethical concerns, arguing that lawyers' professional duties require human judgment. NYU legal ethics professor Stephen Gillers contends that "A lawyer needs to know the facts of the client's situation and understand how the law applies to those facts." Just inputting documents into a "black box" algorithm abdicates this responsibility. Some argue AI should only be used to assist human review rather than completely replace it.

However, proponents counter that educated use of AI enhances rather than diminishes diligent advocacy. Davis Wang at Stanford Law School believes that "When wielded judiciously, AI enables lawyers to conduct deeper analysis on novel legal problems." By automating routine work, AI systems give attorneys more space for critical thinking and client counseling. Used appropriately, AI augments rather than replaces human judgment required by professional ethics rules.