So what happens when AI can pass the Bar?

Imagine dedicating years of your life to mastering the law, only to find out that an artificial intelligence (AI), launched just last year, has also passed the Bar Exam.

The Bar Exam gates entry into the legal profession in the United States. Traditionally, it marks the culmination of a lawyer's academic preparation. It's not easy to pass.

But there was a turning point last year, when OpenAI released GPT-4, which passed the Bar.

This achievement marks a seismic shift in how we perceive the role of AI in the legal field, promising to redefine what it means to practice law.

Over the past month, I've talked to lots of lawyers about the opportunities and challenges of incorporating AI into their legal practice, and they are all looking for ways to use AI as soon as possible.

There are a plethora of ways AI can be applied into law. There are also a large number of challenges: issues like IP rights, procedural equality, and government regulation. I'll cover all of these topics and more below.

The Opportunity

Since generative AI and large language models (LLMs) excel in processing and generating textual data, the legal industry is poised for disruption.

Across the industry, AI is already bringing 10x improvements in some legal workflows.

Take a look at some of the companies that have skyrocketed over the last year:

• Harvey AI: valued at $715 million and backed by OpenAI, it has partnered with leading firms like Allen & Overy and PwC to streamline research and drafting processes using AI. More than 15,000 law firms are on its waitlist.
• Leya : originally from Sweden, it’s been growing exponentially in Europe. It offers LLMs tailored to specific jurisdictions.
• Bench IQ : allows attorneys to uncover the reasons behind all of their judges’ rulings, so they can tailor their arguments to achieve higher win rates.
• Gavel : lets law firms create custom client-facing workflows to automate their legal practice, supporting services such as Hello Divorce .
• DoNotPay : AI-powered legal app that helps users with a variety of tasks, including fighting parking tickets, securing refunds, canceling subscriptions, and more — letting users win up to $25k in small-claims court.

There are plenty of potential use cases as well:

• Legal Research: cut the time spent on legal research and summarizing cases.
• Contract Drafting and Review: streamline the contract drafting process, ensuring accuracy and compliance with legal standards.
• Legal Strategy Ideation: aid in brainstorming sessions to devise legal strategies.
• Predictive Analytics: offer insights into legal outcomes, aiding in more informed decision-making. In one case, a judge even used ChatGPT to assist in making a court decision (source: Vice).
• Streamlining Paperwork Tasks: From incorporating companies, registering brands, fulfilling hiring paperwork, and more, AI will handle routine tasks.
• Enhancing Legal Service Efficiency: tools to democratize access to legal services, making them more efficient and accessible, including chatbots that will help individuals get a better sense of the laws.

The speed of AI legal research is astounding:

• Performance Eval: LegalBench and LawBench are benchmarks to evaluate language models on specific legal reasoning tasks.

• Evaluating AI Against Legal Professionals: Studies like "Better Call GPT" compare the performance of AI models with human lawyers, providing insights into AI's capabilities and limitations (finding: LLMs perform on par with LPOs and Junior Lawyers, accurately determining legal issues within contracts.)
• Domain-Specific Language Models: language models tailored to legal texts, such as 中文法律LLaMA for the Chinese legal domain, showcasing the potential for highly specialized AI tools in law.
• Prompting Techniques: Research on techniques like chain-of-thought prompting is exploring how to adapt AI to better understand and respond to complex legal inquiries.

Risks and Challenges

The implementation of AI in the legal field is not without challenges and potential issues, including:

• Billing and Economic Models: The traditional per-hour billing model faces scrutiny as AI enables quicker service delivery, prompting a rethink of billing practices. Read more: Will legal AI end the billable hour?
• Security and Privacy: Given the sensitive nature of legal data, securing AI tools against breaches is crucial.
• Intellectual Property (IP) Issues: The copyright-ability of AI-generated content and the obligations of AI companies regarding the content used for training models are under debate. Read more: Is A.I. the Death of I.P.?
• Trust and Accuracy: There remains skepticism regarding the reliability of AI tools for critical tasks like research and legal argumentation.
• Fairness and Equality: The use of AI in judicial processes raises concerns about procedural equality, especially if access to AI tools is uneven among parties.
• Regulatory concerns: governments are looking for ways to regulate — or even ban — the use of AI in the legal profession.

Wrapping Up!

The rise of AI into the legal sector is not just an inevitable future; it’s already here. The implications of this will be seen in the coming years.

Let me know if any areas covered here interest you — I’ll be happy to expand in the comments!

Thanks to some of the lawyers I’ve chatted with whose insights contributed to this article: Andrés Ortiz Herbener , Emilio Archila Saetta , Heinz Moeller Gómez , Rafael Pástor , Pierre Matringe , Mateo Viteri Melo , Sebastián Silva , César Coronel Jones , Laura Pardo , Roberto J. Argüello , Alberto E. Rigail C. DBA and Laura Castro Redondo . They’ll be among the first to build and adopt these tools. Will you?