More and more companies are adopting the AI around the globe. A survey conducted by Mckinsey on 9 Major industry sectors, points to growing interest in AI but with some barriers still remaining.
- 49% of the respondents said their organization has embedded AI one of the standard process ,
- 30% said there organizations is doing pilot
- 21% say that there are more than one function where the AI is being deployed .
In another recent SMA Survey of P&C insurance executives shows that AI has great potential in Insurance. 60% of these companies have pilots planned, started or completed and heading for production.
The question today is "is the impact of AI in insurance over hyped or understated ?"
No doubt RPA (Robotic Process Automation) is being deployed at wide scale , companies are also exploring AI use cases to do Fraud detections in BFSI (Banking, Financial Services and Insurance) domain. So we can definitely see that industry is starting to reap benefits from AI .
Other drivers are the explosion of data, and insurance is a data driven industry, will they be an early adopter of IoT (Internet of Things) companies data? With this data explosion it is very important to automate the analysing of data to gain benefits.
But at scale companies are still slow at deploying AI. We see companies facing lot of internal barriers in decision making process to test, adopt and deploy AI tech. Why is this?
Lack of AI knowledge – For lot of company’s staff do not have full knowledge about AI, to them its just a sort of Wizard that can automate the whole process. They do not know what to and what not to expect of AI automations. They are not fully aware how an AI Algorithm is trained, and when is the right time or even how to adopt AI in to a process.
Lack of Change Process - They are not aware of how the integration will take place , usually they are using multiple systems internally which becomes complicated when they plan to deploy AI and employees find easy work arounds that confound the work needed to for full integration and this can delay the deployment and fact gathering . Additionally when a business case is prepared to present to the Board teams have tough time defining the ROI for the AI deployment. With AI there is often overlaps in Human and AI process and the Long-term benefits can be confused. The use of AI often needs a complete overhaul of a process - such a where photos are going to be used - this required more focus to be placed on very basic tasks.
Lack of structured data - even though the innovation Team in a typical insurance company would like to deploy AI they are not aware that they do not have enough structured data to train the AI , which is a major roadblock in deploying the AI. This is one of the barriers in InsurTech movement and has caused many early partnerships. For example the need for images for claims processing - the images need to be of good enough quality and consistant. If they do not have enough of the car images then they cannot train the AI to complete the whole process. Which means they will have to first collect the data and then train they AI. Then devise a plan to roll out, eating into each process or business layer. This required new levels of cooperation across the company. This is not just an IT project!
AI as internal competitor – Lot of employees see AI as job threat, they think that AI will be a Risk for them as AI will take over their Task in future and they can loose the job. This is true but there is also the very big opportunity to bring tasks back in-house. AI can take over a few of the tasks which are very structured or repetative like data entry or analysing of photos in vehicle inspections. These were often outsourced so they can now come back in house and AI will only improve the productivity. Employees need to work in partnership with AI, they can concentrate on other tasks which require human skills. However this is a major barrier that is never spoken of but causes a resistance in buying and decision making, it needs good leadership from the top.
AI as a external competitor - new insuretech companies are using AI and do not have the legacy systems, this allows faster adotion and change. With a simple pivot an AI claims processor can quickly become a full P&C insurer. Is the insurance industry training its own cuckoo in the nest? The T&C's for business data and ownership of neural networks must be understood.
What is needed to over come these barriers?
As I see it, the process of pitching an AI project to a company needs to be very specific. We need work with companies to understand both external internal barriers together with the goals of the business long term. Then AI can be seen in context and breaking down the barriers as the project moves forward.
Companies should be informed about how the AI will work for them , what functions it can automate and how the integration will take place with the time frame. This will inform the business case and enable financial and productivity goals to be met. Partnership with a trusted internal team or external vendor, that is not the Cuckoo, is essential. AI, when implimented, will quickly become a business critical system and one that is not easily backed up or replacable with manual processes should it be disrupted.
2020 will be the year of wider AI adoption as companies look for higer value and greater productivity. Don't be left behind. TonkaBI is your trusted partner.