How will AI cope with GDPR?

Every system or intelligent machine needs input data at the beginning. And that can be a stumbling block.

Nowadays, it is becoming almost a necessity to introduce new automated and intelligent systems into businesses. Speed and accuracy of data processing play a key role here and will help many businesses to keep up with the competition. But every system or intelligent machine needs input data at the start. And that can be a stumbling block.

So at the very beginning there is a system that needs to get input data. It then decides on the next course of action based on its evaluation. Of course, it's not always people-related data. It may be information that does not meet the general definition of personal data in the slightest, or that does meet it but is not processed in the sense of the GDPR. Then you can rest easy.

But if you are wondering which provision of the GDPR is the most problematic in relation to devices containing AI, it is the obligation of controllers to allow subjects to exercise the right to human intervention by the controller. The controller is expected to be able to explain in a comprehensible manner what the logic of automated processing is and on what principles it is based.

In practice, this means that you will need to have sufficient insight into every decision that the smart device makes, understand it and be able to explain why the decision was made in that particular way. And it doesn't matter whether it's a simple task or a very complicated decision-making process. Understanding on the part of the human being is necessary above all in order to be able to step in and guide the process if necessary.

You may be asking right now why we should let machines interfere with decision making. The answer is simple: if the decision-making processes in a smart device are preset incorrectly, the system could evaluate some situations unfairly and thus discriminate against whole groups of people. Human intervention here is more like a safeguard to maintain fairness.

Remember that as part of the duty of disclosure you will need to be able to explain how the algorithms used work, how the decision was made, and which facts were relevant to the AI. Meeting this obligation can very easily become an impossible mission.

The use of AI is definitely a step in the right direction, but for the reasons outlined above, it's definitely a good idea not to load up on more than you can carry. Make sure the algorithms you use are easy for you to understand, or consider simplifying them. Unfortunately, simpler algorithms can take their toll in the form of lower reliability and accuracy.

Do you have personal experience with a device using AI? Do you want to find out if the input data is processed in the sense of GDPR? Contact us!