Are we entitled to know what the robots are thinking?

When Hermon Melles (26) planed his master's thesis he decided to find a research topic that included both his major interests: law and technology. While writing his thesis at the offices of Wikborg Rein he has been given the opportunity to converse with some Norway's most competent lawyers within data protection and technology.

When I decided to apply to write my thesis at Wikborg Rein it was first and foremost the firm's good reputation and their competence that stood out. I wanted to write about automation and artificial intelligence in a GDPR context, and Wikborg Rein stood out to me due to its many skilled lawyers, tells Melles.

Wikborg Reins collaboration with AVO Consulting and their launching of a course at NHH this spring was one amongst many things that made the firm stand out for Hermon.

Systems that create their own practices

Hermon has given his thesis the title «Are we entitled to know what the robots are thinking?». The research paper explores the treatment of information through automated decisions after the introduction of GDPR. These are systems that operate without human interference and make decisions based on underlying algorithms and artificial intelligence.
- The issues that I am currently working on is whether there exists a right to an explanation of an automated decision by GDPR and the reach of this right, explains Hermon.

The question is relevant if for example, an insurance company uses an automated decision making process to set motor insurance premiums based on monitoring customers’ driving behaviour.

- If your brother for example is paying for a motor insurance premium much higher than yours, then he is entitled to an explanation of how the premium is set after GDPR. Based on my findings, such an explanation must be given at the same time as the premium is set, meaning that such a function must be integrated into the system.

With machine learning, the system's underlying algorithm is continuously fed with data, such as previous decisions. According to Hermon, the challenge of explaining decisions becomes even greater when more advanced artificial intelligence and machine learning are being used.

- In cases where the decision is made with artificial intelligence it can be extremely difficult to explain by the employees because they do not necessarily understand how the system came to make that exact decision, Hermon says.

Unknown waters

When Hermon began researching for his topic he searched for a theme that was relevant across industries and which would grow to become more prominent in all industries over time. In addition, he wanted to immerse himself within a topic that contained little existing research.

- With these criteria's in mind it was only natural that my thesis topic concerned automated decisions, explains the master student.

Industry 4.0 is used to characterise a new technological age. Most companies already explore or implement the use of automated decision in various stages of their business. Hermon believes that it is important to take GDPR into consideration when considering or applying algorithms and automated decisions.

- When automated systems with the help of machine learning make decisions about who gets a loan from the bank, home insurance or a job – then transparency within the systems is vital so that we are able to monitor what is going on. If an explanation is needed then businesses using automated systems are obligated to explain how these decisions have been made and the technology behind them. This is why I wanted to investigate the existence and the extent of such a right in GDPR, Hermon concludes.