From theory to practice: Using AI hands-on


An editorial by ERGO CDO Mark Klein

Digitalisation & Technology, 11.09.2024

Today, insurers are already using artificial intelligence (AI) to assess risks, improve processes, and optimise the customer experience. The result: substantial efficiency gains and better service. With the advent of generative AI (GenAI), a further player has entered the stage. Together with classic AI, it will take the insurance sector to a new level. A look behind the curtain.

ERGO CDO Mark Klein

Permanent change – that was one of the core ideals of Steve Jobs, co-founder and, for many years, CEO of Apple. Jobs was noted for his – at times excessive – striving to attain perfection, simplicity and innovation. He was convinced that outstanding products could be created by focusing on what matters most and putting the spotlight on the user experience. One major aspect of his success was his constant questioning of the status quo. His approach yielded groundbreaking innovations – and has more to do with the insurance sector than you might think. 

After all, insurers, too, have to constantly reevaluate their processes, products and priorities. Where is there still room for improvement? How are customer needs changing? How is the world around us changing? It is only by staying agile and constantly challenging themselves, trying out new things, and integrating new technologies, that insurers can improve and succeed. Accordingly, for insurers, change means resilience. And resilience is the key to long-term financial success.  

Principle of simplicity supported by AI

Jobs’ principle of simplicity can also be applied to the insurance sector. Today, many people find insurance too complicated or lacking in transparency. Insurance policies aren’t exactly hot items that customers can’t wait to get their hands on; as a rule, they only become important once something has already gone wrong; but when it does, everything should work smoothly, from claims handling to processing. In this sense, insurers’ efforts should all revolve around simplicity – throughout the value chain. 

When it comes to putting the ideal of permanent change and the principle of simplicity into practice, one technology can especially help insurers: artificial intelligence. 

With the aid of AI systems, they can e.g. prepare stress tests and scenario analyses to assess how robust their own insurance portfolio is. Moreover, AI-based analyses can help to more accurately predict the probability of specific claims or losses. That’s an essential aspect for underwriting – and as such, for insurers’ profitability. In addition to calculating and simulating risk, AI is also used to improve business processes and services. 

ERGO: Around 100 AI applications in place

At ERGO, we’ve been using AI in various areas since 2017. Currently, it’s mostly used in our own administrative processes and touches on e.g. input management and on handling contracts, claims and benefits. Group-wide, we’re now using roughly 100 AI applications. Thanks to AI, ERGO has achieved an automation rate of more than 90 percent in the classification of submitted health insurance documents. In comparison to the previous rate of ca. 50 percent, achieved using a rule-based approach, that’s a substantial increase that translates into accelerated claims handling for our customers. Moreover, AI is providing comprehensive support for customer calls. Every day, ca. 10,000 telephone calls are answered by virtual assistants at ERGO. Depending on the nature of the call, these phone bots direct it to the corresponding department, independently provide answers regarding the status of submitted claims, or record customers’ reports on hail damage. 

AI boosts ERGO’s efficiency. This enhances customer service and helps us remain competitive in terms of the expense ratio. Technology is used to free employees from repetitive tasks and to buffer heavy peak workloads, like those during the COVID-19 pandemic, the annual flu outbreaks, or in connection with major claims, which are becoming more common due to climate change. This is particularly relevant in light of demographic change and the resultant skills shortage, a problem that affects society as a whole. 

Generative AI as the next development stage

In addition to the AI applications already in use, since last year ERGO has intensively concentrated on the next stage of AI: generative AI. What sets it apart is how it allows users to access and use data. Whereas the development of machine learning took decades and was always a topic for specialists, GenAI is now democratising this highly complex technology by making massive amounts of external data accessible using simple text commands. 

The fact that GenAI is capable of understanding and generating human speech will accelerate the evolution of the insurance sector by several decades. Used correctly, GenAI-supported tools will lead to the further optimisation of business processes, a better customer experience and higher productivity, not to mention new services. A study conducted by the Boston Consulting Group in 2023 found that consultants using generative AI were able to complete 12 percent more tasks in 25 percent less time, and in 40 percent better quality, than the control group. 750 of the Group’s consultants from around the globe were surveyed. 

Insurance companies are predestined for the use of GenAI

Using GenAI also holds tremendous potential for insurers. Since there are hundreds of thousands of contact points with customers on a daily basis, all of which involve the processing of language, insurance companies are predestined for using language models and tools based on them. GenAI could be used to supplement traditional AI- and rule-based systems in the context of classifying and extracting data from documents. Potentially, it could also help to further automate and accelerate claims handling by analysing images, videos and other data. In the form of chat bots and virtual assistants, these tools could engage in more authentic conversations and process more complex requests. Moreover, the models could help to recognise patterns in connection with combating fraud, ensuring fair claims handling. 

One key thing to bear in mind: using AI and GenAI should involve careful consideration, implementa-tion and steering. Just as with classic AI applications, if GenAI is to be used productively, all relevant stakeholders like Legal, Compliance, IT, and Risk Management, as well as the co-determination committees, must be involved in advance. Humans must continue to have the final say, given that GenAI applications not only hold great potential for use in end-to-end processes but also entail certain challenges, e.g. with regard to data protection, copyright, AI hallucinations, and discrimination. Further, a clear data strategy, suitable deletion concepts, and ethical guidelines on how AI is to be implemented at a given company are called for.

ERGO GPT and Knowledge Assistants

ERGO has already introduced its first GenAI applications. Since April, our employees have had access to an in-house “ERGO GPT”. And we’re currently exploring the use of chat bots as “knowledge assistants”, i.e. as interfaces with existing internal databases, and working on pilots for GPT-based chat bots in customer communication. When it comes to developing the use cases, ERGO pursues a model-agnostic approach. In other words, we not only use the extensive commercial Large Language Models as a basis for creating our own applications, in keeping with the respective legal and regulatory requirements. We also draw on Open Source models. 

Together with the AI applications already available, GenAI will rapidly deliver appreciable added value and lead to further efficiency gains. Over the years to come, it will offer considerable competitive advantages. Consequently, a host of banks and insurers are already using the technology – that, and because their staff actively request it. Deployed correctly, and in combination with rule-based approaches and classic AI, it can provide a veritable boost when it comes to aspects like innovation, growth, process optimisation, and compensating for the skills shortage. And all the while, insurers should never lose sight of the goal: facilitating insurance and permanent change. 

Please note: This article was first published by Versicherungsmonitor.

Text: Mark Klein, CDO ERGO Group AG

ERGO CDO Mark Klein

Author: Mark Klein, CDO ERGO Group

Mark Klein has been ERGO's Chief Digital Officer since 2016. Previously, he was head of T-Mobile Netherlands. His main task at ERGO is the digital transformation of traditional business in Germany and abroad. He is establishing new, digital business models.

Mark Klein on LinkedIn

Your opinion
If you would like to share your opinion on this topic with us, please send us a message to next@ergo.de.

Related articles

Digitalisation & Technology 08.03.2023

What is ChatGPT - and if so, how many?

Everywhere you look, you read about ChatGPT. For many months, //next columnist Markus Sekulla has not experienced such hype. Be in the media or on Twitter or LinkedIn, a lot is currently being written about the prototype of OpenAI's new AI. So it’s time for a closer look.

Digitalisation & Technology 17.03.2023

ChatGPT in the insurance industry

Our experts Jennifer Betz, Jens Sievert and Nicolas Konnerth explain the impact of ChatGPT on the insurance industry and possible use cases for insurance companies.

Digitalisation & Technology 03.05.2023

How does the chatbot actually work?

Hand on heart: Do you understand how ChatGPT really works behind the scenes? What role do algorithms, vectors and probabilities play in making this fascinating technology work the way it does – and get better all the time? The following two explanatory videos – one in German, one in English – helped us to understand.