Return to site

NLP’s Legal Limit


NLP’s Legal Limit

Natural Language Processing (NLP) is routinely hailed as one of the most promising technological fields. Once NLP innovators perfect software that can intelligently read documents, the potential is limitless for the automation of our society. From the DMV to Goldman Sachs, this technology offers great promise, but currently there are still significant limitations to the technology and its current potential application.

One must first understand the nature of the technology. NLP is a delicate instrument that must be precisely directed. When creating NLP software, the application is extremely relevant to the potential outcome. So says NLP expert and former Microsoft & Googler Dima Korolev, “NLQ or Natural Language Query is doable in high performance, but NLP is too close to strong Ai and will make mistakes in a high performance setting.”

Strong Ai is a field of Ai that is trying to make the machine’s intellectual functions as strong as a human’s intellect. However, this field is still in an exploratory stage as most Ai is still quite far off from anything that resembles the cognitive functions of a human.

Fields that are sensitive to perturbations such as the legal field, where NLP has been extremely hyped, will have a tough time applying the technology. This is particularly relevant to one of the most document-centric fields out there: law.

Corporate lawyers focus on the details of lengthy, complex documents. Mergers and acquisition agreements, for example, are written by attorneys for parties for parties engaged in an adversarial dance with one another. So rather than a standard template that is easy for a simple NLP system to understand, the AI will need to pull together a clause in the merger agreement with a defined term 50 pages away, and then adjust that according to an amendment filed later in a separate document.

According to Logan Beirne, professor at Yale Law School and CEO of legal technology company Matterhorn Transactions, “In the law, details and accuracy are everything.” Matterhorn provides comprehensive databases, analyses, and reports to thousands of law firms across the US, UK, and Canada, allowing lawyers to compare the market terms of complex transactions. “Attorneys rely on us to draft complicated deals and advise clients on what is market for terms, so we need 100% accuracy - something NLP hasn’t been able to achieve. Yet.”

Beirne has largely relied on his own team of attorneys to digest the deals and accurately enter the granular information into the company’s databases. “The varied and complicated nature of legal documents has required a human touch. However, after years of our attorneys analyzing over 50,000 documents, we have a unique treasure trove of data which we are currently using to train AI tools.” Beirne is optimistic for the future of NLP, adding, “We are on the cusp of using NLP to achieve the impeccable accuracy that attorneys need.”

Unlike in law, NLP has been successful in finance, where a document can have a positive or negative interpretation based on far fewer factors. For example, consider the terms “increase” vs “decrease” and the correlation each word’s usage in a press release would have with the potential forward 90 day price return.

In the legal world there are many twists and challenges and NLP will take a few more years to master the craft.

Interview with NASA Astronaut Scott Kelly: An American Hero​

13 Questions With General David Petraeus

Why Choose Machine Learning Investing Over A Traditional Financial Advisor?

Interview With Home Depot Co-Founder Ken Langone

Interview with the Inventor of Amazon's Alexa

Automation and the Rebirth of American Retail

China Debuts Stealth Unmanned Combat Aerial Vehicle

Sweden's Economy Embraces AI & Automation

Austria's Automated Ai & Robotic Future Is Now

Nuclear Submarines: A 7,000 Lb Swiss Watch

Ai Can Write Its Own Computer Program

On Black Holes: Gateway to Another Dimension, or Ghosts of Stars’ Pasts?

Supersonic Travel: The Future of Aviation

Was Our Moon Once Habitable?

The Modern Global Arms Race

NASA Seeks New Worlds

Cowboy Turned Space Surgeon

Shedding Light on Dark Matter: Using Machine Learning to Unravel Physics’ Hardest Questions

When High-Tech Meets Low-Tech Economy: Ai & the Construction Industry

Aquaponics: How Advanced Technology Grows Vegetables In The Desert

The World Cup Does Not Have a Lasting Positive Impact on Hosting Countries

Artificial Intelligence is Transforming the Forex Market

Do Machines Dream? Inside the Dreams of a Machine

Can Ai Replace Human Ski Coaches?

America’s Next Spy Plane

Faster than Sound and Undetectable by Radar

The Implications of Machine Learning on Condensed Matter Physics & Quantum Computing

Crafting Eco-Sustainability: WTC and Environmental Sustainability

Can Ai Transform Swimming?

Argentina's AI Future: Reversing a Century of Decline

Tennis & Artificial Intelligence

Kazakhstan's Ai Aspirations

Peru's Ai Future Will Drive Economic Growth

The Colombian Approach to the AI Revolution

How AI Can Explain Its Thinking

Singapore: Ai & Robotic City

Ai in New Zealand

Brazil & Artificial Intelligence​

Denmark & Ai

Can Ai Replace Human Ski Coaches?

Tennis & Artificial Intelligence