Summary: Some industries are a clear slam-dunk for AI/ML applications and some less so. The legal, regulatory, and compliance businesses (law firms, internal legal departments, and the contract review and regulatory compliance departments of heavily regulated industries) fall in this last category. This is a review of seven companies found by TopBots to be successful; pointing to opportunities others can follow.
Remember just a few years ago when we were looking forward to now or a little beyond and imagining what applications AI/ML would have in different industries. Some of those prognostications were slam dunks as they applied to customer propensity or using machine vision to count whatever widgets you were interested in.
What struck me as really speculative at the time were applications that impacted fields dense in laws, regulations, and complex situations where human knowledge and intuition had long dominated.
Healthcare – An Early Failure in Knowledge Concentration
One of those fields is healthcare. One early attempt to corral all the available knowledge and interpretations was IBM’s Watson which proved a bust. It was just too human-intensive to keep adding and categorizing new data while ensuring you were removing anything out of date.
Another recent bust is the diagnostic advice chatbot introduced in the UK called Babylon. This expert system was supposed to replace physicians in the initial referral process into the medical system as well as providing advice on common ailments directly to the caller.
Unfortunately a group of auditing physicians “found that around 10% to 15% of the chatbot’s 100 most frequently suggested outcomes, such as a chest infection, either missed warning signs of a more serious condition like cancer or sepsis or were just flat-out wrong”.
AI/ML in healthcare has largely abandon ‘knowledge concentration’ in favor of more successful applications like ML-driven patient alerts via IoT edge applications and CNN computer vision driven augmentation of medical imaging and visually based lab testing.
Legal & Compliance Share the Same Barriers as Healthcare
Another field dense in both textual knowledge and a long history of interpretive application is law. The predictions at the time of thousands of displaced lawyers just didn’t ring true.
Like medicine, the body of knowledge consists of a huge amount of laws and regulations that are constantly changing, and particularly that the application of those laws could be interpreted and applied in so many different ways based on the context. Not to mention different laws and interpretations based on the physical location. That might apply to where the act was to occur, what state or local version might have precedence, or where it might be most advantageously litigated.
The third barrier is that law firms and legal departments just aren’t that big. Setting aside a handful of 1,000+ member international law firms or the legal departments of the largest corporations, most law firms and legal departments are going to look like pretty small businesses. And like most small businesses they would have difficulty justifying large AI/ML expense or the disruption to their existing workflows.
Where is AI/ML Making Inroads in Legal and Compliance?
TopBots is one of several well respected review and rating firms that keeps track of the more successful entrants in AI/ML. I’ve always found their listings interesting and useful. Recently they updated their “Essential Landscape of Enterprise AI Companies (2020)” which includes a section on legal and compliance.
I thought it would be interesting to look at the specific AI/ML applications represented by the seven companies TopBots rates as ‘enterprise worthy’ to see where AI/ML was heading in this field.
Please keep in mind these are not endorsements. We picked these only based on TopBots review and categorization of these companies and their applications as ‘enterprise worthy’ hopefully showing where opportunities for others may also lie in the legal and regulatory field.
ROSS is attempting to succeed in capturing and accessing all current laws and interpretations where IBM’s Watson failed with medical information. According to TopBots and ROSS’s clients the AI powered search engine does indeed reduce the time and manpower needed to research and support case construction. The system is question based using the same language lawyers are familiar with. It is designed to report similar or related cases and overturned rulings to eliminate dead ends. Data coverage includes federal published and unpublished case law, and all 50 states statutes and regulations.
Luminance squarely targets a huge bottleneck in the legal world, the side-by-side comparison and review of dozens or hundreds of documents in all sorts of contract, M&A, employee review, or similar situations that are the bread and butter of legal work. Their pretrained NLP cloud platform based on unsupervised learning is claimed to be effective out of the box. Additional supervised learning can be added by the user based on their own labeling of examples. The platform looks for anomalies among versions, compares clauses across docs, identifies key data points within docs, and can be used, for example, to ensure the firm’s favored clause language is present in all document. Document review is claimed to be accelerate more than 10X.
LawGeex uses its NLP platform to review and analyze contracts according to rules assigned by the user to quickly answer the question ‘Can I sign this?’ Where conflicts are found the issue is escalated for resolution to the appropriate legal resource. This is useful not only in the legal professions but in any environment such as procurement or insurance where large volumes of contracts is the norm.
Built for high volume contract review, the Kira platform targets not only the legal profession but corporate, retail, insurance, and financial services environment with similar high contract flows. Kira comes preloaded to apply more than 450 ML models to search for inconsistencies with the user’s norms and standards, and allows users to add additional rules as they see fit. The pre-built models cover common contract review tasks like due diligence, general commercial compliance, lease abstraction, ISDA schedules, GDPR, and more.
Neota offers a no-code intelligent automation platform suitable for law firms, internal legal departments, and other professional services environments. No code is presented as appropriate for these small operations to avoid IT programming and support costs. Like most IA applications however, client usage appears to be rules driven rather than AI/ML supported. There may be some machine vision in support of document processing but that’s not clear from their web site. You might wonder what sort of high frequency, low touch tasks are available in the legal world. Their web site points to several repetitive tasks like templated NDAs, in-taking request for legal services, or routing of risk assessment templates to appropriate hands.
The last two companies reviewed by TopBots fall less into the legal world and more into the requirements for regulatory compliance faced particularly in banking, financial services, and insurance.
With applications in all industries with compliance regulations around KYC, Socure offers a platform of 10 applications designed to identify high and low risk consumers and businesses, and allow maximum safe opportunity for fast, low-cost onboarding and underwriting. Socure uses ML techniques and massive public and private databases to power KYC, global watchlist, document verification, address, email, device, and phone risk, and models to identify identity fraud and synthetic fraud.
Friss offers a package of ML powered risk assessment tools targeting the P&C insurance industry. While this may not seem like a legal function, it is part of regulatory compliance in KYC. Risk may enter the process at any stage from client underwriting to claims and ultimately fraud investigations. The P&C industry would like to optimize what is called “straight through processing” which minimizes human intervention and total time from claim to award. The Friss tools combine internal and external data sources with ML to score risks of fraud at each stage. The scoring allows each user to adjust the amount of human involvement to the modeled and scored risk.
About the author: Bill is Contributing Editor for Data Science Central. Bill is also President & Chief Data Scientist at Data-Magnum and has practiced as a data scientist since 2001. His articles have been read more than 2.5 million times.