International insurance law firm Kennedys is teaming up with the University of Manchester to develop next-generation fraud prevention software as part of a machine learning project funded by the Innovate UK Knowledge Transfer Partnership.
With assistance from the University, the two-year project will see Kennedys combine its existing fraud expertise and international data sets to develop machine learning techniques to help its clients prevent insurance claims fraud.
This is another case of law firms tapping the data they already have with the assistance of machine learning to build new and more valuable services for their clients. One could say this is an example of legal AI systems creating new value, not just providing greater efficiency and productivity.
While many law firms initially have looked at AI systems, such as exploiting natural language processing and machine learning, as a primarily efficiency-gaining tool, they are now increasingly seeing how new product lines and services can be built to sell to clients, that perhaps could not have been done before, given human limitations in analysing large volumes of complex unstructured data.
To some extent this breaks the narrative of ‘AI takes jobs’ and instead suggests ‘AI creates new value’.
This work builds on a new wave of machine learning applications the firm has begun to put in place following the hiring of data expert, Karim Derrick, recently as Head of Research and Development.
Artificial Lawyer asked Derrick to explain a bit more about the machine learning venture:
– The fraud system uses machine learning. Can you please outline how this works? / How this will work?
This latest development is part of our overall strategy to develop an insurance legal services platform that covers the lifecycle of a claim from inception to settlement. At the heart of that platform is data.
We already have a market leading product in the market that we train based on the knowledge and experience of our fraud experts: Ki.
We help clients identify potentially fraudulent claims by examination of their claims. Spotting fraud is about identifying anomalies.
Machine learning will allow the system to build on that expertise and to be even more accurate, to work at much greater scale and to also spot patterns in the data that might have been overlooked by our experts. It’s all about our relentless drive to deliver value to our clients, helping them use lawyers less.
– Is this making use of Natural Language Processing? Would you describe this as an AI system, or seeking to be?
It will and yes and yes.
– Will this new tech link with developments you’ve done in the UK and in India already? (See story about Kennedys’ work with an Indian team of developers.)
Yes, it’s a component of the overall strategy. Data underpins a lot of the R&D we are doing right now and like the other work we are doing this initiative will feed into our overall platform and become a service that can in turn feed into all the products that are accessed from the platform. We now have a pipeline of products fed by our Ideas Lab, prototyped by india, delivered by our platform.
As Derrick says, Kennedys has been developing a suite of online products for some years, to help clients better manage their business and, in some cases, reduce their reliance on lawyers.
Initiatives include an offshore prototyping development team, data science and analytics capability, a future scanning and emerging risk team and an in-house internal incubation programme to generate new ideas.
Kennedys partner Richard West added: ‘Kennedys has been developing online legal services for many years in the form of the Kennedys toolkit, which has seen significant client traction and recognition from a number of industry awards. Our aim is to continually use technology to challenge existing practice and to help our clients use lawyers less.’
The academic team supporting the two-year project at the University of Manchester consists of Jian-Bo Yang, Professor of Decision and System Sciences and Director of the Decision and Cognitive Sciences Research Centre (DCSRC) and Dong Ling Xu, Professor of Decision Science and Support Systems.
Professor Yang concluded: ‘We are confident that we can help Kennedys improve its current system. The key is to try and develop a hybrid system where you can use both big data and human knowledge in deep learning to tackle the problem, which we call transparent machine learning. In this way you can explain exactly why you reach your decisions. It is evidence-based, transparent decision-making.’