"It costs less to avoid getting into trouble than to pay for getting out of it." --Louis M. Brown, USC Law Professor (1909-1996)
The average cost of a commercial tort lawsuit against a business is about $350,000 per case, according to Intraspexion's research. Considering the high cost of settlements, defense attorneys' fees, verdicts and administrative costs, even when you win a case, you lose. So if a company can prevent just three lawsuits annually, its bottom line profit grows by over $1 million per year.
That fits a key goal established by the Association of Corporate Counsel. The ACC wants lawyers to target ways to drive down their companies’ annual legal spend by 25 percent. In response, many corporate legal departments have initiated such cost-cutting initiatives as LEAN/Six Sigma training, standardizing RFPs, flat fee arrangements and moving contract matters offshore.
So with business processes for legal services already being significantly streamlined, what’s left to cut?
The remaining target in reducing legal costs meaningfully is to impact the biggest expense: defending against lawsuits. In federal court alone, the average total expense for commercial tort litigation over the past decade was $160 billion per year, according to Intraspexion. Clearly the best way to reduce corporate legal costs is to avoid lawsuits in the first place as much as possible.
Enter Nick Brestoff. Preventing litigation is what drove Brestoff to retire from practicing law as a litigation specialist in California for 38 years, and to form his software company, Intraspexion. With engineering degrees from UCLA and Caltech, Brestoff earned his law degree at the University of Southern California where he studied under Professor Louis Brown, known as the father of “preventive law.”
Preventive Law advocates that the earlier you can be alerted to legal problems, the better you can mitigate their effect on the company’s operations and on everyone’s lives.
“As the saying goes, ‘Forewarned is forearmed’” Brestoff asserts. “Having an early means of detecting a lawsuit in the making from a company’s vast amount of internal communications is immensely valuable.”
After authoring Preventing Litigation: An Early Warning System to Get Big Value Out of Big Data, Brestoff patented a way for companies to use Deep Learning, a type of Artificial Intelligence, to quickly spot the litigation risks created by data flowing between employees on a daily basis.
Brestoff recognized early on that "high-speed graphics processors using Deep Learning are today’s sharpest knife in the AI toolkit. Deep Learning has been called 'the new electricity'  because of its potential to transform every industry, every project and every profession – including law."
Deep Learning has demonstrated remarkable success:
The pace of achievements is impressive. But after so many years of touting Preventive Law as a worthy concept, is Deep Learning ready to make it a reality?
“Absolutely!” Brestoff emphasizes. “To the best of our knowledge, no one before us has applied Deep Learning Algorithms (DLAs) to identify risk and provide early warning, and we’ve demonstrated it for the legal profession.
First, as one example of many types of lawsuits, we extract documents from court cases and other sources pertaining just to prior employment discrimination lawsuits.
Second, we use those documents to train our DLAs to create a positive set for recognizing emails that potentially create employment discrimination risk. We also use documents that are clearly not related to employment discrimination to create a negative set.
Finally we present the DLAs with a large body of email communications. In our earliest tests we used a large body of emails from former Enron CEO and Chairman Ken Lay.”
The results of those tests? Intraspexion's system "correctly discovered one email relating to a serious employment discrimination risk with an incredibly high degree of accuracy," asserts Brestoff.
“That was a critical discovery: Deep Learning worked! No key words. No taxonomies. No “expert system” rules. We really didn’t expect to surface a single risk. After all, Ken Lay was the Chairman and CEO of Enron, not the head of HR.”
Brestoff explains "the algorithm learns, first by example from the training data of textual facts alleged in hundreds of any particular type of lawsuit, such as employment discrimination. Then the system’s DLA for employment discrimination examines enterprise emails behind the firewall, and scores the emails for accuracy against that risk, and to what degree.
Corporate lawyers can then mark any false positives which surface so that the algorithm will be even more accurate next time. They don’t need to understand the machinery behind it at all in order to make the whole process work smoothly. Once a corporate attorney sees the risks, it’s up to them to investigate, gather evidence from other internal sources, and advise leadership accordingly.
In that sense, the system is not artificial intelligence, it’s augmented intelligence. “Corporate lawyers are currently blind to these risks,” Brestoff says, “even though they’re closest to the risky data. With Intraspexion’s DLA, now they can see.
By installing Intraspexion's system, companies are also better able to protect their brand reputation and their leadership. Since the purpose of the Deep Learning system is to prevent harm, any regulatory investigation of the business will reveal that management has taken systematic measures toward legal workplace compliance."
It appears Intraspexion may deliver a potential game-changer to the legal profession. On January 24, 2017 the U.S. Patent and Trademark Office issued Brestoff a patent, No. 9,552,548.
Since avoiding lawsuits is far less expensive than paying for them, Brestoff believes Intraspexion is the first company to develop a system that transforms “preventive law” from an abstract concept to a functional enterprise-grade technology.
“The ability to prevent litigation is both revolutionary and disruptive, from the bottom up,” says Brestoff. “Years ago, Professor Brown emphatically said it should be done. Now, it can be done.”
 Andrew Ng, Chief Scientist, Baidu research - http://fortune.com/ai-artificial-intelligence-deep-machine-learning/
~ ~ ~