My Basket0

Is AI working for you?

Blog

Publication date:

19 August 2020

Last updated:

19 August 2020

Author(s):

James Moorhouse

When implementing artificial intelligence (AI) into systems and processes, what should underwriters be looking for at the beginning of their data journey?

Handling submissions

In commercial lines, underwriters often get too many submissions to be able to process. The majority of these come from insurance brokers, but even so, 50% of these will most likely be left on the table. Therefore underwriters have to prioritise the applications they know will generate a lead. Manual data entry is time-consuming, especially when some of the applications received are incomplete. But AI could be used as a tool to help sort and prioritise these applications based on previous success rates.

Quick decisions are also made as to whether these applications fit into the insurer’s risk appetite. If they are not familiar with a type of business, underwriters will need to research the line of business using the internet, databases, inspection reports or through talking to other colleagues. Further risk evaluation may then require on-site physical inspection. However, advances in technology means that this can now sometimes be performed by computer vision.

With AI playing a greater role in handling admin, this should leave underwriters with more time for decision-making, focusing on how to apply appropriate coverage and determining any limits and/or deductibles. Once this is done, a firm decision can be made on whether to accept the risk and what pricing to then apply.

 

Focusing on AI

For underwriting, AI can be utilised for decision making when it comes to process automation and aggregating information. Anything that will allow underwriters to spend more time to focus on valuable parts of the insurance chain could be seen as a benefit. To do this, insurers need to identify the areas where they can improve efficiencies and decision making.

Commercial lines could benefit saving time on admin as this requires a more holistic approach due to the variety of risks placed. Personal lines customers often have similar needs, meaning that the decision-making process might be more straightforward.

 

Where do you start your journey?

To be able to get the most out of using AI you need to have an objective. And a plan to reach that objective. Ask yourself the following questions when considering what it is you actually want to achieve:

  • What is the problem you’re trying to solve?
  • What is the niche/speciality?
  • Have you made a roadmap? Do you have the data, appropriate tech, appropriate use case?
  • What happens when you receive a commercial submission? What are the steps/processes?
  • Do you have a start-up mindset - thinking small with one idea or outcome to focus on?

When setting an AI objective, it should be made clear whether a return on investment (ROI) is an issue with or without AI. It’s also important to determine what criteria you are applying, for example if you want to make processes ‘faster’ does that mean rate, volume or number?

 

What mistakes are insurers making?

AI is sometimes implemented as a tickbox exercise without any real thought or plan behind it. If this is the case then the project is doomed to fail, wasting both time and money. Here are some of the ways AI is being implemented poorly:

  • Unrealistic expectations – there won’t be any overnight changes for big carriers
  • Too ambitious - new companies start from scratch without legacy systems or previous processes to integrate or upgrade
  • Vague or lack of KPIs – how are you measuring your progress? What targets have you set?
  • Poor marketing – is it relevant to the customer? Do you really understand what they are looking for?
  • Using limited data – the more data the better

 

What’s your niche?

If you want AI to improve the customer experience, then you need to know what you’re using it for. As insurers specialise, automation could help speed traditional processes so that more time can be spent researching. Those seeking personal lines products will want efficiency and automation while commercial lines will require a more holistic approach, adding more value to the process.

However, how do you train AI if there is no common view on the risk? This is why sometimes a subjective assessment still needs to be made on certain applications. But this can only be done if the guaranteed business is identified and generated first.

 

Where does your data come from?

In short, the more data that is available, the better the AI decision-making process will be. But this can only happen if there is safe and secure data sharing. There are data privacy laws which may present issues in doing this, but this is to protect the customers from having their data misused.

Could this mean that data might be the next form of currency? Or will there be a data sharing eco-system? If the insurance sector wants to help the customer then it needs to share more data. There is no single source, so data needs to be aggregated. But the challenges of this are data privacy, security and preventing fraud. At present there needs to be more effective way to address this.

Telematics is one option that can replace traditional sources of information. By offering real-time data, a more accurate decision can be made about the customer. This can also prevent high-risk customers from driving up the price of premiums for everyone, leading to fairer pricing. Connected risks are important, resulting in underwriters being able to access dynamic data in ways that are more accessible, for both the insurer and the customer.

This document is believed to be accurate but is not intended as a basis of knowledge upon which advice can be given. Neither the author (personal or corporate), the CII group, local institute or Society, or any of the officers or employees of those organisations accept any responsibility for any loss occasioned to any person acting or refraining from action as a result of the data or opinions included in this material. Opinions expressed are those of the author or authors and not necessarily those of the CII group, local institutes, or Societies.

alt