A perspective on AI in insurance

By

When we think about the application of Artificial Intelligence (AI) within our current world, we often visualize voice-powered personal assistants such as Siri, Alexa, or Amy’s x.ai and maybe futuristic humanoids Westworld and Alex Garland’s Ex Machina movies.

We may also picture more revolutionary technologies such as arm robots in factories or logistics assembly lines automating repetitive manufacturing jobs, behavioral algorithms often quoted in the design of many well-known tech firms, personalized adverts, and autonomously-powered self-driving vehicles.

However, while the media often focuses on these radical and wide-reaching AI applications, more practical AI adoptions are quietly stirring their paths in particular sectors, including insurance, finance, and wealth management.

The global AI market revenue estimated at $62bn in 2020 is expected to grow at a compound annual growth rate (CAGR) of 42.2% from 2020 to 2027. According to Tractica, this market will reach $125Bn by 2025. A large part of the number is held by the AI automation market, which is expected to reach US$35bn by 2023. And the leading brand holding more than 9% of the AI market share is IBM, leading in active machine learning and AI patents (source: Statista).

At the same time, over 13,000 ventures that claim themselves worthy of wearing the AI tag have raised nearly $250Bn over the past ten years, covering many sectors of the economy, from healthcare and manufacturing to finance and insurance. And in this instance as well, some industries are gaining more traction and attention than others.

Where is the growth of AI coming from?

Recently Steve Mills, Chief AI Ethics Officer & Leader for Artificial Intelligence in the Public Sector, and Partner at Boston Consulting Group, stated quite eloquently:

“AI has become table stages for global, national economic and technological competitiveness. This goes beyond nations capturing a piece of the large and rapidly growing AI market. AI is poised to transform nearly every industry. There is an imperative for nations to position themselves to integrate AI into these sectors. Particularly those sectors that are economically important to them. Failing to do so could erode their competitive position, creating opportunities for other, more technologically advanced nations to fill the void. This is not just a matter of missed upside potential from the new AI market. It’s also about downside risk for every other sector that is economically important to a nation.”

Many sectors are looking at AI besides BigTech and their massive financing wallets. Today’s top four industries include:

This chart shows the venture raise volumes described below

 

Automotive: Ventures delivering artificial intelligence software for the automotive sector received $70-80Bn. For instance, these would include connected, electric, and autonomous vehicles—from consumer vehicles (e.g., eCars, eBikes, e-Scooters) to fleet and road transport tech as well as flying cars. Examples of renowned players leveraging AI to differentiate their propositions include Electric vehicle builder Tesla and WM motor, Data analytics provider Cambridge Mobile Telematics, and InsurTech venture Metromile.

Healthcare: Ventures delivering artificial intelligence software within the healthcare sector received $25-35Bn For instance, these would include Fitness & Wellness, Diagnostics, Lifestyle & Disease Self-Management, Assisted Technologies. Ventures that have received significant funding include the oncology platform Tempus or PingAn Health Medical Technology (e.g., Good Doctor in the health and insurance sector).

Financial services, including insurance: Ventures delivering artificial intelligence software within the finance sector received $25-35Bn. These would include payment, remittance, alternative financing, challenger banks, RegTech, and micro-financing/ insurance. Renowned players include InsurTech unicorn Lemonade, credit risk assessment platform Zest AI, or fraud detection platform Forter.

Energy: Ventures delivering artificial intelligence software within the energy sector received $15-25Bn. For instance, these would include energy efficiency, oil & gas, solar energy, renewable energy, and smart grid. The electric vehicle examples highlighted above would fit within this category and in oil and gas with Beyond Limits and Uptake within renewable energy.

The story of AI

Discussion on AI typically centers on how the technology works. To understand AI development, particularly within the insurance industry, we need to flip this on its head. We need to consider instead the types of problems the technology can address. This then helps to understand AI applications in the real world rather than concentrate on the tools, the techniques, and the big predictions.

Thomas H. Davenport and Rajeev Ronanki, in Artificial Intelligence for the Real World (Jan-Feb 2018) published in the Harvard Business Review, call these types of AI projects “low hanging fruits.” They are easy to achieve, lowest cost to implement, yet have a significant effect on improvement. They crucially state that these less ambitious implementations of AI are far more likely to be characterized by success than much more ambitious AI projects.

This is an essential distinction to understand, in addition to the fact that the Davenport-Ronanki report also explains that three-quarters of their surveyed companies believe that.

“AI will substantially transform companies within three years. That substantial transformation is coming in a lower key, but notably impactful, different from how the headlines would have us believe. While there is still some way to go, there are unique projects already in progress.”

Most simplistically, AI has three broad ways of addressing business needs:

 

No alt text provided for this image
  1. Behind the scenes automation: This entails using AI and cognitive intelligence techniques to optimize data – inputting, recording, transferring, managing data, and all the primary usage of data assets within back-office administration systems. Such AI delivers an exceptionally high return for a low cost and is relatively easy to implement as each automation links to specific process steps. The objective here is to automate repetitive tasks to simulate actual human behaviors. Technologies would include image or image/object recognition, natural language processing (text and speech), and robotics process automation.
  2. Insights and meanings from deep-analytics: In this instance, AI is more ‘intelligent’ with a greater degree of machine learning to drive custom personalization. The objective is to mimic the human brain functions with a greater level of efficiency. AI is used to do such things by developing sophisticated scoring mechanisms, predicting customer purchases, facilitating targeted pricing and underwriting actions, targeting messaging in adverts, or identifying real-time fraudulent behavior.
  3. Engaging anywhere and anytime with the bots: This includes using AI to improve or facilitate customer and employee engagement by reducing employee intervention for other types of highly repetitive tasks. This approach is evidenced in conversational AI technologies, including robo-advisors, chatbots, talk-bots, push-bots, or highly customized and gamified product and service recommendations and alerts.

How the insurance industry is developing and utilizing AI

While still in its infancy, we see exciting development within insurance for the three above categories, which would then expose AI to real tangible future usage while adapting, evolving, and transforming how insurance is “done.”

The startup and scaleup ecosystem is already challenging the sector with hyper-personalized customer-centric engagements, adaptive underwriting and pricing mechanisms, behavior modification utilizing gamification techniques, in addition to designing and delivering offers.

Let’s look at current examples of insurance industry solutions within the above three categories:

Automate: Some well-known growth ventures include intelligent robotic process automation platforms UiPathAutomation Anywhere, or Catalytic that are using a variety of bots to make the most sense of unstructured data & automate decision making. You will also find that Hyperscience turns documents into machine-readable data with strengths in the healthcare insurance sector. These platforms focus on systematically replacing those tasks, which are usually overlooked for their impact on effectiveness, productivity, and efficiency but could actually be easily enhanced by AI.

I recently chatted with the fun and engaging team at Artificial. a platform that automates end-to-end underwriting processes to accelerate transparency between an underwriter and its broking partners regardless of interaction channel. Launched in 2013, the platform entered the insurance space in 2016 and works now with Axis, Aon, Chaucer, and Capita, with a clear focus on the commercial/specialty lines markets and Lloyd’s of London. At its core, the solution aims to reduce manual data entry and rekeying to improve risk analysis and portfolio insight for underwriters, speeding their ability to identify those risks within their risk appetite often sent by brokers seeking to bind risks speedily. Confidence scores are used to optimize decision-making, and the team applies a collaborative approach to engage with data providers to solve the most complex insurance problems. By harnessing machine learning to extract data at the point of entry and an AI-led engine to tailor products flexibly, Artificial shared that they are standardizing and validating multiple data sources for use across multiple systems, combining a wide variety of analysis and pricing tools to optimize and personalize underwriting for insurance carriers and brokers.

Spanish startup BDeo’s focus has been on simplifying the claims process by using various advanced AI techniques to automate visual intelligence in the motor and home insurance spaces. By focusing on damage detection, the team connects underwriting and claims processes utilizing deep learning algorithms. For those who have been following this team’s progress, the team just raised €5 million at the end of last year to scale the solution across Europe. Talking with Julio and Ruth recently, I was intrigued by the visual interpretation details they can provide on vehicle and home-related claims. The platform combines image recognition, facial recognition, geolocation, virtual reality, and many other AI techniques to ensure accurate security and fraud detection to speed up valid claims.

Analyze: Analytically focused ventures utilize data to drive better profiling, scoring, and underwriting metrics. There are so many ventures that have done well in 2020 and which extend capabilities beyond AI analytics. In this segment, I would include the like of Cambridge Mobile Telematics, LemonadeZesty.ai, and Metromile. Other well-known market players would include risk modeling platforms Akur8Concirrus, and Praedicat, as well as acceleration portfolio venture, Aureus that released Donna last year targeted at optimizing the broker’s sales experience during the pandemic.

Digital Fineprint has been working on modernizing and digitizing SME underwriting framework while helping insurers affected by the current market condition to mitigate SME-related risk exposures, combining thousands of relevant open data sources dynamically, applying algorithms into real-time alerts to increase the precision and accuracy of human-led underwriting decisions. Interesting work in the SME space has been done with insurers such as Hiscox.

It is always a real pleasure to catch up with James Birch Ki about the end-to-end digitization of the Lloyd’s of London insurance process and the Ki team’s long-term aspiration for the platform and redefining insurance.

As an on-demand insurance platform for individuals and businesses, Ki Insurance wants to be the first at deploying digital trading with a fully automated underwriting process through an algorithmically driven modeling approach to provide more cost-effective and value-driven insurance transactions for Lloyd’s market, brokers, and other key market players. The venture offers instant capacity, in real-time accessible regardless of location—a vision based on AI’s true promise.

The team raised $500m funding in September last year. It also takes a strong collaborative and ecosystem approach to building its business model, focusing on its core competencies and augmenting noncore expertise with expert partners.

Engage: Many organizations today use chatbots for customer engagement. Some well-known ventures within the insurance space include Spixii, Rozie.ai, and Enterprise Bot, which have all delivered their solutions uniquely across the insurance value chain. One of the Spixii team’s great successes includes the delivery of Zurich’s claims chatbot ‘Zara’ in just 5 weeks.

In a recent article written for Alchemy Crew, Dr. Andree Bates shares insights on Amelia, a highly sophisticated digital virtual chatbot assistant based on actor Lauren Hayes‘ facial characteristics. Amelia uses artificial intelligence to understand customer grievances and recommend possible solutions. Using so-called “digital humans” could substantially improve customer service by reducing wait times and cutting extra costs.

While not that straight forward it is important to note that most of the ventures described above probably straddle across a couple of capability areas (e.g., “automate and analyze” or “analyze and engage”) because it is the only way a growth venture can continue to assert long-term differential advantage, besides the primary competence on which it was started at its core.

The new frontier: Sustainability and AI

All the above is great, but as the world moves towards addressing some fundamental world issues, AI is back at the forefront of the discussion, focusing on people and the planet.

Over the past few months, there has been a shift in discussion with AI centered around sustainability, carbon footprint, ethical use of data across healthcare, insurance, and finance to reduce discrimination, increase product pricing fairness and drive more financial inclusion. And I remember sitting at VivaTech Paris over 2 years ago listening to Ginni Rometti discussing the importance of ethics and AI.

Statista estimates in recent research that the impact of using AI on environmental applications to achieve economic growth could increase GDP by 5.4% in Europe by 2030. The latter estimate represents 5.1% for East Asia and 4.2% for North America. Indeed not insignificant predictions.

No alt text provided for this image

 

AI, sustainability, and ethics are starting to become key topics of interest. The question will gradually demand the consideration and enforcement of internal guidelines and policies that are continuously refined and validated based on the appropriate applications of AI against established ethical values, paying particular attention to sensitive customer characteristics such as ethnicity, race, gender, sexual orientation, nationality, financial resilience, strengths/skills/abilities, and religious and/or political belief. It also demands the design of AI applications that contribute to a positive and sustainable societal outcome by ensuring that internal guidelines and policies consider the principles of solidarity and inclusiveness, particularly for the vulnerable and underserved segments of the population (e.g., social inclusion, financial inclusion, affordability) as new customer-centric products and services are designed and launched.

One of the ventures I am really thrilled to see growing combines all the above considerations. It refines its business model and looks at the problem from sustainability, ethics, and rental resilience lens: And this is Canopy rent 

I will leave you with these final few words from Kris Sterkens, Janssen’s Chairman for EMEA:

“The world of healthcare [I would say finance and insurance too] is beset with a paradox in whereby there is more data than ever is flowing through physicians’ [and specialists] hands, but the true value of that data has gone largely untapped because it is unstructured and silo-ed in systems that generally are unable to talk to one another.”

 

This article was first published on LinkedIn  here.

If you want to know more:

Discover more on LinkedInTwitterYouTube, and Slideshare

 

Share this post

Menu