New rules may open floodgates for AI-driven auto insurance

For instance, Kotak General Insurance Co. Ltd has started using an AI-based vehicle pre-inspection tool in partnership with Inspektlabs for policy renewals.

As part of the process, customers can capture photos and videos of vehicles and upload them on a cloud-based app. The AI-powered model generates an automated inspection report on the extent of damages within a few seconds to determine if the vehicle should be insured.

The company announced its plans two weeks after the Insurance Regulatory and Development Authority of India (Irdai) allowed general insurance companies to start using tech-enabled solutions for own damage covers.

Suresh Sankaranarayanan, Kotak’s chief technology officer, said the regulator’s move will pave the way for AI-driven solutions in vehicle insurance. However, for Kotak, the plan was in the pipeline for quite some time, he added.

Irdai has also asked insurers to offer telematics-based insurance plans such as ‘Pay As You Drive’, wherein the premium amount is calculated based on the number of kilometres covered by a vehicle, and ‘Pay How You Drive’, where it is determined by driver behaviour.

Sankaranarayanan said the Irdai’s move will encourage insurance companies to come up with various plans. “We are also working on something on similar lines and will be coming out with it soon. “Insurance is a paper-intensive business, whether it is onboarding or claims. We have been digitizing the documents, but digitizing only reduces the transport time. Using AI cuts down a few hours from the processing system,” he added.

This will also make the processing of claims much faster and also help identify fraud markers more effectively, said Sankaranarayanan.

To be sure, AI-driven insurance is not new. Second-hand car sellers are also using such solutions, or are planning to. For instance, used car platform CARS24 uses neural network architectures along with traditional imaging and sound processing technique to detect damages and defects in car exteriors as well as engines. “We have been able to reduce inspection turnaround time by 20%. Our trading margin has improved by 5% of the selling price in the last six months,” Naresh Mehta, head, data science and analytics (global) at CARS24, said.

In February, Kotak General Insurance signed an agreement with CARS24 Financial Services to allow the platform extend motor insurance services to its consumers.

Similarly, pre-owned cars showroom CarzSo, too is planning to deploy an AI-based solution during and after inspection for rating vehicles based on their condition.

“Pre-owned car dealers and buyers are finding it challenging to assess the condition of a pre-owned car,” said Vaibhav Sharma, founder and chief executive, CarzSo.

However, CarzSo isn’t using AI to insure vehicles, but the core technology is similar. In fact, industry experts said as more and more insurance use-cases come, companies offering AI products will gain more authenticity.

That said, the increased use of AI could also be detrimental to end users. Bias in AI models and poorly constructed algorithms could penalize users when they should not, or overlook road and driving conditions in countries like India.

Insurers are aware of these concerns, too. “No model is perfect, to begin with. It is going to reject a few cases for no fault of the user may be,” said Sankaranarayanan, adding that the company is fine-tuning the model to remove errors or biases.

“Over time, the model will get refined to a degree where this may not be the predominant problem anymore.”

First article