Topics In Demand
Notification
New

No notification found.

Key factors organizations could consider to improve AI maturity
Key factors organizations could consider to improve AI maturity

December 14, 2022

325

0

 

Innovation and technology adoption differ across industries given the way industries operate. The insurance sector has been at a different pace to adapt technology when compared to banking or retail owing to different factors: regulatory environment, product life cycle, customer needs etc. Trends have begun to accelerate in insurance industry to keep technology in the driver’s seat as incumbents and new age “InsurTech” players take advantage of AI. Key factors influencing AI adoption in insurance are: Increased demand for customer personalization of insurance products, Value generated from automation of traditional operations/processes and Opportunity to monetize data availability owing to multitude of channels: sales, social media, telematics, etc. What could insurance players do to address these trends and demands? What are some important factors for organizations to embrace while embarking on the journey to increasingly and successfully adopt AI?

  1. Businesses must consciously integrate AI solutions into their annual planning / strategy

Businesses are enabled by technology. When addressing key business problems, business and technology strategy must go together and must focus to deliver the same goal and outcome. Technology teams must be integrated into business strategy/planning. In the context of AI solutions, when the leadership plans their annual strategy/focus areas, there must be a conscious effort to explore how AI can help accelerate their goals. For example, if the organization is looking at reduced cycle time to process claims, depending on the focus areas and relevant information at hand to achieve this goal, teams may explore opportunities on automated underwriting, anomaly detection, cognitive solutions to process form-based claims etc.

When the AI team becomes part of the same core team, it will be able to propose solutions in a more informed way. Many times, technology teams are separated from core teams, and they will come a little late into the game after some key decisions are made. This way, AI teams may miss out on the purpose, background, pain areas the business and prioritization of problems business is trying to solve. Having AI strategy integrated into the business strategy will enable pick the right problems to solve, experiment with confidence and plan solutions in more informed and transparent manner.

  1. Have an agile and measurable framework for periodically calibrating success

Once the business decides their key focus areas and articulate the solutions, they intend to experiment with by taking inputs from technology teams, it is important to have a framework that will calibrate the success of the project. The ownership of the initiative must lie with the business and technology teams (including AI teams) are the enablers for the solutions and outcomes. The success criteria for every team must be clearly defined and there must be a continuous review framework to track the progress towards the goal. Any deviations must be flagged, and a review must be made on the continuity of the solution. Having this framework also brings in transparency. It becomes easy to understand on a daily/periodic basis of what’s going in the project rather than getting surprised at a later point. Given that, always the measure of success may not be quantified in terms of revenue, cost or cycle time. Some solutions are built for customer experience, need for technology upgrade etc. So, business prioritization and articulating success criteria is important. This way everyone will be aligned on what was invested into and why. The other important piece of the framework is the process we intend to follow for delivering a quality product. Generally, frameworks come with best practices on delivery and the necessary cadence and signoff procedures. In the recent past, agile frameworks and agile delivery models have been receiving good attention for their focus on incremental tangible outcomes and the way they make business integrated part of the solution.

  1. Encouraging a culture for learning, research and experimentation

Many forums talk about the relatively less success rate in AI projects. This goes for a multitude of reasons: availability of relevant data, organization culture, data quality, problem qualification, technical expertise etc. Not every project needs to start with a sophisticated modelling technique. Achieving the necessary maturity in modest steps is crucial in promoting AI adoption. Enterprise projects where clarity is fully not there and need more time to mature should begin by developing deterministic/rule-based models with significant input from business stakeholders and prototype the models through pilot tests. This methodology of rule-based to probabilistic models gradually enhances data labeling and understanding so that projects can then be transformed to Machine Learning based on the need. As teams become comfortable with the project and inspired from initial results, it makes sense to transition to a fully automated machine learning based approach. So, a good way to strengthen the AI success rate is by encouraging small pockets of multiple research/experimentation proof of concepts. These must be fast paced PoCs targeted on a fail-fast learn-fast approach. Depending on the success rate, these PoCs can further be evolved to enterprise scale involving multiple teams.

Last point on research and experimentation is to extract value through new learning experiments involving global business and technical hackathons, participation in academic events, and sharing best practices in internal and external forums. These will help organizations understand key trends and foster attracting the best talent to enable innovation.

  1. Thought leadership to ensure model bias, fairness and compliance are addressed

While organizations are busy experimenting with AI/ML models for achieving competitive advantage and performance improvements, one key aspect that cannot be ignored is ensuring information security and individual privacy while collecting data, and transparency in storing, algorithm building, and usage of model output. Organizations cannot afford to let model bias provide undue advantage of customer information and the bias must not be traded for performance gains. Insurance companies must establish internal institutions that govern data usage and ensure that developed models are fair and do not threaten individual privacy. This way, the best practices for fairness and model explainability will bring more trust to the customers and help insurers stay ahead. 

To summarize, Leveraging AI is becoming vital, and the insurance industry is increasingly adopting AI for competitive advantage and customer centricity. Investing in niche technologies such as AI and focusing on upskilling the talent enables the insurance players to harness the advantages of data driven decision making and become innovative. At the same time, it is vital for organizations to calibrate success and refine priorities in their AI journey. As we move into the future, the trust between insurers and their clientele is bound to increase. Insurance will transition from being just a financial transaction to one where the insurer is actively managing risks and protecting customers and their families. Given the speed of change is unprecedented, insurers will have a clear advantage if they adopt AI in a fair and transparent manner. This is not a simple good-to-have. But this is imperative to their survival.

Authors:  PBK Chaitanya, Director - Business Analytics, Rama Yedla, Senior Data Scientist and Siddharth Pandit, Analyst - Business Analytics


That the contents of third-party articles/blogs published here on the website, and the interpretation of all information in the article/blogs such as data, maps, numbers, opinions etc. displayed in the article/blogs and views or the opinions expressed within the content are solely of the author's; and do not reflect the opinions and beliefs of NASSCOM or its affiliates in any manner. NASSCOM does not take any liability w.r.t. content in any manner and will not be liable in any manner whatsoever for any kind of liability arising out of any act, error or omission. The contents of third-party article/blogs published, are provided solely as convenience; and the presence of these articles/blogs should not, under any circumstances, be considered as an endorsement of the contents by NASSCOM in any manner; and if you chose to access these articles/blogs , you do so at your own risk.


© Copyright nasscom. All Rights Reserved.