RSS feed Get our RSS feed

News by Topic

BizReport : Law & Regulation : July 01, 2019

How Tech Upstart Evisort has brought Artificial Intelligence to the Legal World

The new buzzword in Artificial Intelligence is "Cognitive Computing." This is a new standard of AI algorithms which undertakes to emulate some of the more complex functions of human cognition. Examples include driverless cars and facial recognition, which have been in the news a lot lately. These tasks were impossible for AI even just twenty years ago because the speed of hardware was simply not sufficient to make these tasks feasible.

by BizReport

Another domain of AI which has eluded computer science up until now is language parsing. Natural human language, it turns out, has many vague ambiguities which throw off the best automated linguistic algorithms. While search engines such as Google, customer service voice recognition systems, and other service-sector automation have made strides, a robot's interpretation of a document hasn't been something you'd want to bet your livelihood on... until now.

Evisort Document Management

One new startup making headway in AI document management is Evisort. The Silicon Valley start-up has engineered a processing system for legal documents. In the span of seconds, the AI system can scan through a 30-page document and extract the key data points for tidy organization in a database. These factoids include cash amounts, named parties, expiration dates, terms and conditions, and more. The company has recently netted investor funding from funds backed by technology industry elites.

Evisort has been called "Google for contracts." This may sound redundant, but while Google merely matches text patterns, Evisort is a document management and data analytics solution for contracts and other legal documents. The difference is that instead of mere pattern-matching, Evisort follows legal language and registers text in context, automating the scanning of legal documents and producing the relevant facts in a concise form.

Why Not Just Read Our Own Documents?

That's approaching a threshold beyond all human ability. The average attorney handles upwards of seventy documents per business day or more than 25K per year. The problem lies under the category of what we are coming to call the "data deluge," an overflow of information coming from all directions at once in our wired-up society.

Evisort is a joint venture of MIT researchers and Harvard Law experts, uniting computer science and the practice of law. The company founder Jerry Ting, with co-founders Amine Anoun and Jake Sussman, saw an open market in the legal profession, where reams of document processing was still done manually. Similar to how the medical profession uses AI assistants, called "expert systems," to parse through the specialized language of medical records, Evisort is an expert system for the legal profession.

People outside the legal profession may dismiss contract boilerplate as "legalese," but legal language is an exact science. Terms in legal documents have to be applied with care because vagueness in wording loses court cases. In a way, legal language is just like a sort of programming language. You're using an abstract context, words, to build an engineered effect that operates the legal system in the desired way. It just so happens that legal language is also possible for AI algorithms to break down.

Computers Sharing the Cognitive Load

While perfect "movie level" AI may be forever outside our grasp, assistive cognitive technology is a giant step forward in business innovation. With Evisort's system, the computer is taking on the drudgery that nobody wants to do: reading stacks of legal paperwork. This frees up professionals to focus on the parts you need a human for, such as making decisions based on the data. The same way calculators assist mathematicians without ever replacing them, a contract-parsing system wades through the complexity of red tape without being close to a replacement for lawyers.

We are sitting at the dawn of an age of cognitive computing, although, like most AI concepts, it's been carried forward from concepts expressed years ago. The difference is that we have machines capable of producing real-time results. Contrary to what most of the public might expect, we actually have many problems, even mathematical ones, that cannot be solved by computers right now, and only recently have we gotten hardware advances fast enough to tackle some of these latest frontiers.


Subscribe to BizReport



Copyright © 1999- BizReport. All rights reserved.
Republication or redistribution of BizReport content is expressly prohibited without the prior written consent.
BizReport shall not be liable for any errors in the content, or for any actions taken in reliance thereon.