<img src="https://ws.zoominfo.com/pixel/kZxG1sNctrruFoZSPoVD" width="1" height="1" style="display: none;">
Contact Us
Book A Demo
Book A Demo
Contact Us

Can Artificial Intelligence (AI) and Machine Learning help Risk Managers?

Angus Rhodes

Much of the AI discussion to date has been about the risks and rewards that it offers. For example, while there have been high profile stories about the risks of automated vehicle crashes many commentators are predicting long-term impact on the future of motor insurance. Warren Buffet no less believes safer vehicles will see the motor insurance industry reduce by 60% over the next 25 years. When you consider that motor insurance accounts for 40% of the industry, that is a pretty seismic change.
On a broader scale, one recent report by a Swiss think tank for the World Economic Forum predicted  that 75m jobs were at risk of being replaced by 2022. On the reward side, the report also forecasts that 122m jobs will be created as a result.AI definition

But what does this mean for world of risk?

It is set against this backdrop that Ventiv presented at an IRM Special Interest Group event on  AI and Machine Learning; How will the Chief Risk Officer drive risk in the machine?

The session covered a review of current trends in artificial intelligence including; examples of AI technology, applications being utilised in business processes, and potential risks and steps to help mitigate them. We looked at how risk managers are using machine learning and AI and showcased the capabilities of IBM Watson Analytics.

Until recently analysing their data and information was a manual job for risk managers and their teams with more advanced analytic capabilities further restricted the insurance markets. This empowerment creates real opportunities to take control and drive decision making. Throw in the AI dynamic and things become even more interesting as there is no longer a need be a data scientist nor statistician to harvest the rewards. With Natural Language Processing (NLP) they don’t even need to know a technical language. All that is really required is the ability to gather the data set inputs.

What data and what outputs can we expect using AI?

Risk managers are familiar with collecting data on spreadsheets from multiple sources and locations. However, spreadsheets do not necessarily drive insight, learning and understanding, especially when you cannot see the wood for the trees plus everything is siloed. This is the real game changer where AI can potentially deliver in 3 areas:

  1. Sources need not be limited to traditional risk information, they can be internal, external and business sources combined. For example, if there are problems with the delivery of goods in the supply chain other factors can be considered. Using AI, weather information could also be overlaid against claims to see if there was any correlation between poor weather conditions and failed deliveries. In the case of slips and trips could store and office layout or footfall be factors? Multiple sources could be overlaid against risk and claims data to investigate cause and effect. 
  2. AI can drive queries on behalf of the Risk Manager. Raising questions that can be investigated further, rather than simply relying on the intuition of the risk professional alone. With a big plus of it automatically identifying single or multiple factor scenarios driving the results versus a heavy manual effort. This could provide fresh approaches to problems, particularly emerging risks. The result being that AI could provide decision trees for the risk manager to effect change efficiently and effectively.
  3. An opportunity to take ERM and ERM software to a new level to aid the whole business. Rather than engaging with other divisions and the board on a risk basis alone, the greater insight provided by AI could add to the bottom line. The risk team could in turn help breaking down silos and provide a holistic approach to ERM.

Five key things risk managers want from AI

From the group discussion a number of AI needs emerged which fell into five areas:

  1. For AI to identify emerging risks and break down business silos
  2. Understand the connectivity between risks, the linkage of data sets and where AI can fill the gaps. There is a desire to see if a range of risks; qualitative, quantitative, structured and unstructured can be interpreted through AI and machine learning. 
  3. Confidence that the AI systems can show workings and how conclusions have been reached. It also needs to show a ROI through time and cost savings.
  4. Be proactive, identifying, preventing and mitigating, rather than merely reporting on what is happening.
  5. Retain human elements so that people communicate and can still act accordingly and make ethical decisions. This in turn will drive faith in the models being used.

If you are interested in how AI, machine learning and risk management analytics software could help you, get in touch with Angus Rhodes, Global Product Manager at Ventiv Technology. Email Angus Angus.rhodes@ventivtech.com

Learn more >

Oct 22, 2018

Subscribe by Email