insurance chatbo, virtual insurance software
Artificial intelligence (AI) and machine learning are the hot topics of 2018 across all industries, especially insurance and law.
 
Although perceived as a relatively recent development, the term “artificial intelligence” was first used by U.S. computer scientist John McCarthy in 1956.
 
To help its legal audience get a better handle on the subject, ALM’s Legalweek 2018 New Yorkkicked off on Jan. 29 with a workshop titled, “The Foundation of AI and Machine Learning — What it is, isn’t and how to tell the difference.”
 
The basics of AI are important for the insurance industry as well, as shown by the examples from the panel on ways insurance
claims processing and fraud investigations could benefit from AI.
 
Multiple meanings
 
Moderator Kevin Fumai, senior managing counsel of Oracle, led the discussion by acknowledging that AI and machine learning are terms that can have many meanings, depending on the user, and that currently, there is no one meaning. For the purposes of this workshop, Fumai explained that the panel would use the terms interchangeably. The definition the panel adopted was, “having a computer do things that we thought only humans could do,” he added.
 
Shawnna Hoffman-Childress, cognitive co-leader, associate partner, Cognitive and Analytics, Global CoC, IBM Watson, suggested that the audience think of AI as “augmented technology” that helps humans do their jobs more efficiently.
 
Brian Kuhn, co-founder and leader of IBM Watson Legal Solutions also noted, “We don’t speak enough about the ‘learning’ aspect of AI and the ways the tools will improve over time.” He urged the audience to look for ways to use AI to identify megatrends, in turn identifying business risks and proposing solutions.
 
Catherine Krow, founder and CEO of Digitory Legal Corp., believes that the AI a company uses is only as good as who is building it, reminding the audience that the old saying, “garbage in, garbage out,” is still valid. When inputting the company’s expertise and institutional knowledge to the database you’re building, “You only want the best of the best,” she said. “AI is only as good as the information it’s given.”

 
Data up approach
 
Krow, who lives and works near Silicon Valley, added that machine learning is discussed as a breakthrough in the tech companies in her area, using a “data up” approach. This means starting from the data itself with no preconceived notions of what you’ll find or what results to expect.
 
Hampton Coley, director of practice technology and discovery services for Canon Business Process Services, Inc., said that he most often sees AI in the legal context used in document reviews — a labor-intensive process that is also important for insurance professionals. “You need to vet the system to be sure that it yields good results,” he said, reminding the audience that “limited data provides skewed results.” Users of AI need to know the data going in before they can understand the data coming out.
 
The panel agreed that in all organizations, there must be a good use case developed before implementing any AI system. Coley noted that the most successful use cases he’s seen focus on the business of law, not the practice of law. AI has to enhance the practice, but it can’t put the art of practicing law, which depends heavily on a lawyer’s experience, persuasive skills and other intangibles, at risk.
 
Large insurance companies see AI across their operations, said Kuhn, for example, in examining claims fraud. They expect some of the same benefits in the insurer’s law department as well, in terms of finding common themes to their claims or cases and
finding ways to leverage data to better manage costs.
 
Scale is important
 
AI is most valuable with large sets of data, Coley said, and correspondingly, weaker with smaller sets of data, which puts small companies at a disadvantage. Hoffman-Childress noted that the strongest AI systems have many people and personalities “training” the system. “You need about 100,000 pages of deep text to have your AI system work effectively,” she said.
 
“The promise of AI,” said Krow, “is that it can pull the robot out of the lawyer and make better use of the lawyer’s judgment.” Coley added that he would like to see AI challenge a professional’s judgment, enabling the professional to make better decisions.
 
AI allows a lawyer reviewing a case or an underwriter reviewing a potential client to view data through multiple lenses, Kuhn explained, as if the data were being analyzed by an accountant, a data scientist, a claims adjuster, a lawyer, and a psychologist all at the same time. He also predicted that a new specialty legal practice will emerge with professionals who are tech and legal experts on AI and machine learning.
 
To effectively implement AI, Hoffman-Childress said, you have to put feelings aside and provide a strategic initiative that your vendor can align with. Kuhn added that the issue of scope is also significant in designing a project, for example, an augmented claims process or refining data on subrogation cases.
 
“A huge mistake is not knowing how you do things today,” stressed Hoffman-Childress. You need to understand everyone’s role in the workflow, not just the senior partner or junior associate, underwriter or claims adjuster. Only then can you harness the power of AI.
 
To see how Elafris Virtual Insurance Agents can help your company click here to schedule a free online demo.
 
This article was originally published on propertycasualty360

Need more details? Contact us

We are here to assist. Contact us by phone, email or via our social media channels.

Subscribe to our mailing list

×

×

Subscribe to our mailing list

×