Frequently Asked Question
AI-related questions
-
We implemented an AI Management System (ISO 42001) that makes sure that the AI is transparent and controllable. For example, providing sources for the generated responses or showing the line of reasoning to get to an answer.
-
Our AI systems are designed with adaptability in mind. We have an ISO42001 AI Management System in place and continuously monitor regulatory changes and integrate automated compliance updates where feasible. Regular model retraining and rule-based logic adjustments ensure that our AI remains aligned with evolving insurance policies and legal requirements. Additionally, we maintain a strong collaboration with regulatory bodies to anticipate changes before they take effect.
-
A lot of data of insurance companies is unstructured, using AI we have the ability to extract meaningful information from these sources on a large scale that was previously impossible.
In addition, the AI is trained to recognize patterns that could indicate errors and can surface these in an instant even on large datasets so that by using AI the quality of the data can be improved efficiently.
-
We view AI as an enabler rather than a replacement for human employees. AI automates repetitive tasks, allowing employees to focus on higher-value activities such as complex customer interactions, strategic decision-making, and personalized service. We also invest in upskilling and reskilling programs to ensure that employees can work effectively alongside AI, enhancing rather than diminishing job opportunities.
-
Our architecture is designed to use the AI models as efficiently as possible, we use a combination of the following 3 techniques to minimize token costs:
Keeping the context window of the prompt as small as possible by only including the most important information.
Using caching techniques to prevent multiple requests to the AI models
Using a variety of models depending on the specific use case; smaller and cheaper models for simpler tasks and large models for more complex queries.
-
Our centralized Blockchain enhances data security, transparency, and auditability, making it particularly valuable for auditability. Unlike traditional databases, blockchain ensures immutable records and objective validation, reducing the risk of data tampering and improving trust across stakeholders. However, for daily operations, we rely on traditional databases that are directly derived from all events managed within the insurer (which also serve as the foundation for the blockchain), prioritizing efficiency and cost-effectiveness.
-
Yes, AI is tailored to individual customer contexts within a 'My Environment' portal. By leveraging user data, preferences, and behavioral patterns, AI can provide personalized recommendations, policy insights, and automated support. We also implement strict data privacy measures to ensure that AI-driven personalization aligns with regulatory and ethical standards. Furthermore, our data permission rules are enforced such that the data that the user can view (be it employee or customer) is the same as the AI can view when that specific user makes a query.
-
OpenAI was chosen based on its advanced language models and strong research foundation, and commitment to responsible AI development. In addition, OpenAI has a strong collaboration with Microsoft Azure allowing for using these powerful models while maintaining regulatory and privacy requirements.
Finally, we continuously evaluate alternative providers to ensure we remain technologically competitive. Our AI strategy includes a modular and flexible approach, allowing us to integrate different AI providers based on specific use cases, cost efficiency, and security requirements.
Finally, we continuously evaluate alternative providers to ensure we remain technologically competitive. Our AI strategy includes a modular and flexible approach, allowing us to integrate different AI providers based on specific use cases, cost efficiency, and security requirements.