Revolutionizing the Legal Industry: How GPT-3 is Powering Legal Tools

Learn How Lexion, Spellbook, Casetext, and Harvey are Transforming the Legal Industry with Artificial Intelligence.

While Legal AI tools have been around before the November 2022 release of GPT-3, a list of several have been getting renewed attention since its release. Below is a non-exhaustive list of those tools and a summary of the business need they seek to fill as artificial intelligence in the legal industry becomes more relevant to everyone in the legal system.

As opposed to law firm automation, Lexion bills itself as “the fasted way to get contracts done right.” This tool is designed to make a given company’s contracting more efficient. Its features the ability to “track key dates and terms automatically” and save hours of analyzing and reporting. Of note for lawyers, their ad copy on their site is directly at the cost of lawyers by stating that “Lexion is the simplest way to scale legal without adding to your headcount.” It will forever be the tightrope that these companies walk marketing the efficiency of AI tools to lawyers, while subtly downplaying that adoption of these tools mean less work for lawyers. Someone will eventually crack the code on how to slickly convince lawyers to purchase a product whose end game is take the lawyer’s job. Lexion integrates with MS Word, Salesforce, Slack, Teams and other tools many companies already use.

Lexion appears more focused on the handling of contracts, getting approvals, signatures and collaboration on the execution of contracts within an organization. It does not appear to have features enabling the drafting or issue spotting within contracts.

Spellbook is an AI assistant connected to Microsoft Word. It claims to enable the drafting and reviewing of contracts by an AI tool which is trained to summarize complex contract terms and even identify terms that ought to be in the contract but are missing. While contract management software is not new to legal tech, boosting it with fine tuned AI is something new. It boasts the ability of lawyers to “draft contracts 3x faster with AI.”

The tool has a waiting list while it is fine tuning over the coming months their website says. Some of the initial features are:

1. Suggesting language for new clauses based upon the agreement you have submitted to it to evaluate

2. Detecting what they deem are “aggressive terms” which they define as “sneaky or unusual terms.”

3. Providing negotiation suggestions

4. Auto Diligence (a not ready for prime time feature as of this writing).

Spellbook is explicit that it is powered by GPT-3. Which implies it will be powered by subsequent upgrades to GPT which are expected to roll out on at least an annual basis.

A short video of their offering is below.

Casetext deploys AI to enable more efficient legal research. Its workflow involves uploading a brief, complaint or other pleading. The tool then reads that document and searches the Casetext database for the most relevant results.

The results of that search is that its AI “recommends relevant cases, statutes, briefs, black letter law, and more.”

You can also upload opposing counsel’s brief to receive the same analysis. Its AI supported searching finds cases with the “same legal issues, facts, and jurisdiction without the need of constructing a complex search query.”

The tool can claims to discover cases which your opposing counsel missed in their brief. There are also endorsements by judges on their site suggesting the court may want to use their tool to help find cases that either counsel omitted from their briefs. It is indeed an interesting prospect that courts themselves will be using such tools to monitor filings.

Harvey is a tool funded, in part, by OpenAI the creator of GPT-3. It is also on a waitlist. It brings natural language processing to legal research similar to CaseText. Harvey was founded by a former litigator at the firm O’Melveny & Myers and a former research scientist at DeepMind and Facebook. While the website itself has no information about the tool, articles about the tool offer some insights. One of Harvey’s founders billed the tool as “a natural language interface for existing workflows.” Article Link. A sample question of questions taken from the article which gives a flavor of what type of AI legal research Harvey is designed to perform include:

1. “Tell me what the differences are between an employee and independent contractor in the Fourth Circuit”; and

2. “Tell me if this clause in a lease is in violation of California law, and if so, rewrite it so it is no longer in violation.”

Despite this natural language interface that it envisions, the founders of Harvey insist that the tool is not meant to provide legal advice to non-lawyers. I am sure that is true given their design. But, the market exists, as yet untapped, to develop tools that will do just that - become the AI lawyer for the masses. We shall see what obstacles that tool has to surmount once it becomes reality. That reality is sooner, much sooner, rather than later.