// . //  Insights //  Keeping Up With Generative AI

Generative artificial intelligence (AI) has arrived in force and has the potential to transform many ways insurers do business. Poster child of the age of acceleration, it has gained daily media coverage, and its possibilities have captivated headlines.

Earlier this year, we explored the fundamentals of generative AI and the impact it may have in the insurance industry, as we saw many insurers experimenting with its potential. We are now seeing industry discussions progressively shifting away from “What is generative AI?” to “What can I do with generative AI that is impactful, and how soon can this impact be delivered?”. In this five part article series, we will explore the world of generative AI through the lens of insurance industry leaders, addressing practical questions with the goal of helping the industry move forward — thoughtfully — on the impact curve.

Our initial edition discusses the generative AI opportunity for insurers, how significant generative AI will be for insurance organizations’ business strategies and ways-of-working, and how quickly the impact is likely to materialize. Below is an excerpt of our report. Please click here to read our full version of Keeping up with Generative AI.  

In this article, we cover the following questions:

1. Is generative AI any different from previous disruptive technologies? Go to the answer

2. Why should I worry about it now? Go to the answer

3. How should I think about opportunities for my business? Go to the answer

4. How quickly will these different opportunities come to life? Go to the answer

We are entering a profound age of age of AI. We have a huge opportunity with AI and machine learning systems to hire people, and create a robust and progressive future-centered digital economy. We need to expand our digital literacy and arm people for the jobs of the future, so they can work with technologists to represent our corporate, business and ethical values
Ramesh Srinivasan, UCLA Professor, Reinventing technology for the global society

1. Is generative AI any different from previous disruptive technologies?


Exhibit 1: Four ways generative AI is impacting industries
The opportunity is massive,
but with many “unknown unknowns”
The ecosystem is nascent,
even within big tech
Signal vs. noise is a greater problem
than usual
“Building the machine” at scale
will take time
Source: Oliver Wyman analysis

The opportunity is massive, but with many “unknown unknowns”

A recent Celent survey found that by the end of 2023, half of insurers will have tested generative AI solutions, with more than 25% of insurers planning to have solutions in production by year-end. These numbers are significantly higher for larger insurance companies, and are likely to keep increasing as enterprise generative AI solutions and platforms proliferate and become more accessible.

However, the speed at which the technology is evolving requires leaders and teams to “learn as they go,” both around “what to do” and “how to implement the models, tools, and solutions.” While the short-term benefits are still unclear, we believe it is important for insurers to continue experimenting, exploring, and the scaling potential of certain solutions in order to stay ahead of the curve, and gain a potential first-mover advantage.

Unlike prior disruptive technologies — such as the internet, mobile, cloud, low code and no code, or even blockchain — whose mere early adoption took years to materialize, generative AI fostered large scale experimentation practically overnight. This was driven by a combination of ease of access to consumer solutions (such as OpenAI's ChatGPT or Google's Bard), worldwide media coverage, and the promise of near-instant benefits (however real).


Exhibit 2: Many insurers are testing generative AI solutions, and 25% plan to have solutions in production by the close of 2023
% of Oliver Wyman/Celent survey respondents (all sizes/> $5 billion gross written premium - GWP)
Source: Oliver Wyman/Celent poll surveying C-suite executives in the insurance industry, conducted from May 15 to May 22, 2023.
33 executives responded, with eight representing companies over $5 billion in revenue. Questions included: (1) “Are you currently developing large language models in a test environment for future usage?” (2) “Are you currently using large language models in any production applications?”

The ecosystem is nascent, even within Big Tech

Over the last few months, Big Tech players have announced their “horse in the race,” and are testing their way to right, with many growing pains along the way. From issues during live demos to fast-tracked beta releases, most Big Tech outlets pushed their products to market as quickly as possible, increasing potential risks around the short-term use of their technology.

Many enterprise solutions remain primarily focused on experimentation-type use cases, with major compliance, privacy and technology considerations — among others — yet to be resolved. LinkedIn’s co-founder Reid Hoffman notoriously stated that, “If you’re not embarrassed by the first version of your product, you’ve launched too late”— and many players in the generative AI race seem to have taken this to heart, but are now dealing with an enterprise world that is used to leveraging more “finished” products, and where large-scale pilots and rollouts come with a range of adoption challenges in the best of instances.

Signal versus Noise is a greater problem than usual

Similar to most technology disruptions, many technology players of all sizes and capabilities are rapidly announcing new generative AI solutions aimed at enterprise use cases for insurers. More than ever, it will be critical for insurers to assess potential generative AI solution providers and partners using a structured process (including due diligence and proof-of-concept development) in order to distinguish between “real” solutions from shiny objects, as well as identify the most efficient and effective partners. 

Building the machine” at scale will take time

Realizing material gains from generative AI will require significant changes in ways of working. Early pilots may require guardrails that reduce — or even counter — expected productivity gains in limited settings. Yet, persevering through short-term challenges may be crucial to gain a first-mover advantage and achieve long term success.


Exhibit 3: Oliver Wyman’s 5 key success factors to driving generative AI impact
Focus on business value by working from specific insurance problems/needs back
Build confidence and initial wins based on experience, course-correct early and often using new learnings
Be Change-led, technology-enabled
Take risks into account — AI holds uncertainties, unknowns and is dynamically evolving
Engage the organization to gradually build conviction and engagement — all functions, all levels
Source: Oliver Wyman analysis
Artificial intelligence is a co-pilot and fundamentally an invention that can help us do research, learn more, communicate more, sift through data better

2. Why should insurers worry about generative AI now? 

Insurance “Demand-side” hints at AI as a top objective in future-proofing organizations. According to a recent Oliver Wyman C-Suite survey on future growth opportunities,  AI and digital strategy are the number 1 priorities for organizations that have already achieved their transformation objectives (compared to companies that have not, where it is more of a number 4 priority). Management attention on generative AI is substantial at the moment, hinting at continued interest and investment.

In November 2022, OpenAI introduced GPT-3.5 and ChatGPT. ChatGPT reached more than 1 million users in five days and 100 million users in less than two months. It is being used for search, customer insights/service, writing content, coding, video creation

Technology “supply-side” hints at a “new normal” following early growing pains. While the current environment indicates a lot of uncertainty and risks, there will definitely be a “before” and “after” ChatGPT, with AI currently experiencing its “iPhone moment.” The technology offers the power to change how we work, produce and interact with content — similar to the way the iPhone changed how we communicate, share information, make purchases, snap photos, and access the internet. 

We anticipate enterprise and customer-facing solutions to incorporate generative AI in various forms in 2024 and beyond, based on the solid trend that has started to emerge in the first few months of 2023.

3. How should I think about opportunities for my insurance business?


Exhibit 4: Four ways generative AI can deliver business value for insurers

Drive efficiency
via automation

  • Streamline repetitive or manual tasks to increasetime for higher-skill activities (customer support, operations)
  • Objective: Decrease costs, increase productivity (claims/full-time equivalent )
  • Challenge: Most outputs require human review;few off-the-shelf models; training and upskilling (from “doing” to “validating”)

Augment/ “Co-pilot”

  • Augment human expertise with synthesized researchand insights to facilitate complex decision-making (underwriters, claims specialists, agents)
  • Objective: Reduce losses, shorter processing time
  • Challenge: Input data quality and exhaustivity (internal and third party); training/upskilling of decision-makers

Hyper-personalize engagement/ experience (Customer and Employee)

  • Use data to deepen customer understanding (personas, social listening) and develop personalized marketing materials
  • Objective: Generate higher revenue via increased conversion, retention, cross-sell, and customer engagement
  • Challenge: Accuracy of AI-drafted materials, legal and compliance governance processes

Reinvent industry models
and value propositions

  • Develop holistic Generative AI-based products and services, for example:
    • Integrated home insurance and assistance chatbot (leveraging IoT data)
    • Needs assessment and product bundling (for small businesses)
  • Objective: Revenue growth (new or existing)
  • Challenges: Product-market fit, model development cost and timeline (including data)

Source: Oliver Wyman analysis

Drive operational efficiency via automation

Generative AI is not merely a replacement for underwriters, agents, brokers, actuaries, claims adjusters, or customer service representatives. Rather, it is an opportunity to create new operational efficiencies, build greater customer satisfaction, and empower employees to focus on value-added activities.

Generative AI can help streamline repetitive or manual tasks so employees can focus on higher value-add activities. Efficiency opportunities include customer service query support, product research and benchmarking, agent training using synthetic customer queries, corporate function efficiency gains: Similar opportunities exist across legal, compliance, finance, human resources, etc.


Exhibit 5: What if claims productivity could be improved?
Generative AI has the power to automate “mundane” adjuster tasks and to make adjusters’ jobs more efficient and focused on driving outcomes versus documenting file and updating management

Potential for ~5% to 20% time saving benefits (depending on type of claim, line of business, and the percentage of documentation that can be automated)

Augment and “co-pilot” decision-making using generative ai

Generative AI technology offers an instant feedback loop of information, allowing insurers to analyze customer insights or quickly solve an issue across multiple mediums including emails, website interactions, and customer service discussions. Insights can be produced in shorter timeframes — from seconds, minutes, and hours, rather than multiple days or weeks.

This offers a powerful co-pilot for underwriters, claim adjusters, agents and other roles, which can augment human expertise and help accelerate complex decision-making. Many of these roles rely on large amount of expertise that cannot be replaced by rules-based algorithms. Generative AI can help accelerate experts’ analyses and assessments.

Hyper-personalize engagement and revolutionize customer experiences

Within the marketing and distribution part of the value chain, generative AI can help insurers identify upgrades, add-ons, smarter and more targeted products to sell to customers — ultimately generating new revenue streams. Sales team can leverage generative AI to create effective sales lead lists, craft compelling, personalized sales scripts, and generate customized follow-up materials (for example, emails, marketing graphics, and messages). AI offers deeper insights on customer behaviors, helps develop more personalized customer experiences, and can support teams efficiently manage pricing, logistics, and product distributions.

Reinvent insurance industry models and value propositions

The combination of generative AI use cases to create efficiencies, “co-pilots,” and hyper-personalization along with other technology, operation and behavioral changes, may lead to brand new opportunities for the industry. These offer a potential to reinvent the entire insurance value chain, and transform the role of the insurer altogether. While these opportunities are practically boundless and further out for the future, below are a few potential reinvention examples.

4. How quickly will these different opportunities come to life for the insurance industry? 

In the shorter-term, we anticipate that generative AI will materialize in more targeted areas within insurers’ organizations and value chains. These focus areas need to meet a set of materiality, feasibility, and organizational readiness criteria, as well as, be an initial beacon for scaling to more transformative solutions in the foreseeable future.

These shorter-term ambitions may include:

  • Customer service support solutions that save agents time, but do not include customer self-service processes
  • Targeted automation solutions for claims adjusters, which can open the door for broader “augmented” use cases down the line
  • Solutions enabling the development of numerous, high-fidelity personas (for example, with volumes in the hundreds vs. in the tens) and/or the rapid development of highly customized marketing materials that can be reviewed and finalized by internal marketing teams

These initial solutions will be the first step towards generating broader outcomes, such as the end-to-end transformation of complex claims management or large account underwriting reviews. We also anticipate new business value propositions combining the power of efficiency, augmentation and hyper-personalization, such as the ability to rapidly develop highly customized small business insurance propositions at scale.


Exhibit 6: Near-term versus longer-term opportunities for insurers
Harnessing the power of generative AI offers many opportunities for the insurance industry that continues to evolve. Below we share areas that are emerging, and how it can create value and impact for your business — in the near- and longer-term.


Source: Oliver Wyman analysis

Unlock the power of generative AI 

In the series' upcoming articles, we will explore questions around business value creation and new ways of working. We’ll help you unlock the power of generative AI, and take a deep dive into specific use cases and actions for your organization.

At Oliver Wyman, we help our clients think critically about generative AI opportunities across the value chain, pilot and scale use cases, and set up programs and portfolios to deliver immediate and long-term impact. This includes value proposition development, solution design and build, operating model design, talent and workforce considerations, capabilities, models, and processes, licensing and partnerships, and adopting best risk and governance practices across the business.