Artificial Intelligence (AI), the natural or human intelligence that can be demonstrated by machines or computer systems, is all around us, with almost every industry racing to embed AI into their products and services. The insurance industry is no exception, having recognized the way in which AI can assist in making their businesses more user friendly.
The current insurance systems require that a person calls the insurance company directly or makes use of an insurance broker when making any changes to their insurance policy. This can be a time consuming and laborious task. Submitting a claim when an insured item is damaged, lost or stolen, can also be a lengthy process.
Certain insurance companies have recognized how AI can improve the business of insurance for themselves as well as their clients. The South African insurance industry is highly regulated in order to protect both customers and insurance companies, but the regulatory framework has not kept up with technological developments. South Africa (as well as other countries) does not have any laws that deal specifically with AI.
The Organisation for Economic Co-Operation and Development (OECD) published the OECD Council Recommendation on Artificial Intelligence in May 2019. In June 2019, the G20 adopted human-centred AI Principles that drew from the OECD AI Principles. The OECD AI Principles are important, as many countries will use them as a guideline to draft or develop their own legislation. These Principles aim to promote the responsible development of AI to ensure respect for human rights and democratic values.
One of the key provisions from the OECD AI Principles provides for transparency and explainability, which states that AI developers must enable users of the AI to challenge any outcomes which adversely affect them, when the decision made by the AI technology is based on “plain and easy-to-understand information”. This provision is likely to be tested within the insurance industry as it raises the question as to what will happen when an insurance company using AI rejects a claim and what the right of recourse afforded to the client would be.
An insurance company currently using an AI chatbot to handle claims has indicated that should a claim be rejected, reasons for the decision will be made available, and in most cases, the reasons for rejecting the claim can be resolved. Should the issue not be resolved, and an insurance claim is rejected as a result, the client would need to use the current avenues available to them, such as the Ombudsman for Short-Term Insurance, to resolve the dispute. After the claim has been rejected and a client has escalated the claim to the Ombudsman, it is likely that the insurance company would also have to re-evaluate the claim themselves to determine if the AI was correct in rejecting the claim. Disputes such as these will be ground-breaking for South Africa and will hopefully lead to an updated legislative framework that promotes innovation.
The use of AI in various products and services is increasing at a rapid speed. However, in order for AI to meet its full potential, its use and development must be regulated. With the publication of the OECD AI Principles, and the drive by industries that require these regulations in order to implement AI, we will hopefully see changes in the legislative framework relating to AI in South Africa in the near future.
Kerushnie Govender, Associate, and Kirsty Gibson, Candidate Attorney, in the Financial Institutions Sector Group, and overseen by Wildu du Plessis, Head of Banking & Finance, Baker McKenzie Johannesburg.