Artificial intelligence technology has become more pervasive than ever. Complex algorithms within this technology enable companies across multiple business sectors to gather information and profile current and potential customers. This has enabled companies to easily provide solutions that fit customers’ needs.
However, new legislation enacted by the European Union (EU) may make it difficult for insurance software providers to fully integrate AI into their platforms. The new legislation, known as General Data Protection Regulation (GDPR), mandates that companies must obtain consent from customers within the European Union for the use of any personal information obtained. Enforcement of the new law could be costly for noncompliance by fines of up to four percent of annual revenues received on a global basis.
The scope of the consent required by the new legislation goes beyond the traditional privacy policies and acknowledgment of cookies that many consumers are accustomed to seeing as pop-ups. Companies including insurance companies will be required to seek and receive consent from anyone whose personal information is used in order to employ AI technology that could have significant outcomes, such as approval of an insurance policy or claim. Clients who do not wish to give such consent will have the right to demand intervention by a human employee of the firm for review of the decision.
In addition to a customer’s right for human intervention, the customer will also be empowered to obtain details about the automated process used in order to determine if discriminatory profiling procedures were used.
In order to comply with the new EU regulations, insurance software providers will have to rethink the design of their algorithms to allow for the data protection applications to be embedded into the software designs. It will no longer be possible to just have a pop-up box with a field to check for permission. This could be problematic for EU firms that must compete with companies in other areas of the world, such as Asia and the United States.
The challenge lies in the requirement for human interaction, which complicates the issue of writing effective algorithms to allow artificial intelligence to operate smoothly and logically. In order for someone to understand an algorithm, the person working with it must be able to manage its complexity. If the algorithm is overly complex with multiple abstracts and data points, it will be extremely difficult for someone to read it and ascertain the results.
Even though the legislation has been passed into law in the EU, lobbyists from financial and insurance industries are continuing to monitor the situation while encouraging government officials to remain aware of the benefits that AI technology can bring to the consumer.