The FT recently reported that in 40% of AI start-ups across Europe there was no evidence that AI applications were being used. The suggestion being that AI has become a catch-all buzz word that’s often used flippantly and can be confusing to investors.
It begs the question – how do you define AI? It turns out that there are any number of definitions ranging from systems that think exactly like humans, to systems that work without figuring out how human reasoning works.
In an insurance context it’s probably more helpful to focus on specific AI applications such as Machine Learning (ML) and Natural Language Processing (NLP). ML being an application of AI that affords systems the ability to automatically learn and improve from experience without being explicitly programmed. NLP presents the ability for a computer to interpret and understand natural language, such as text and audio captured in a system.
Both these technologies enable the analysis of massive quantities of information, hence its application in insurance where we collect huge volumes of data. At C-Quence for example we capture in excess of 1000 data points on each client .
Within C-Quence we are developing ML and NLP techniques in an underwriting context in order to suggest alternatives for an underwriter on risks that currently require intervention because they are outside the automated underwriting parameters built into our C-Q Elements trading platform. The machine learns from previous decisions made by the underwriter and suggests an appropriate remedy.
Ultimately these decisions will be built into the automated underwriting process further refining the platform which means more risks will go straight through without referral, thereby making arranging commercial insurance easier for brokers and their clients.
Elliot Biggs, CIO at C-Quence