The Autoriteit Persoonsgegevens (AP) has developed guidance on meaningful human intervention in algorithmic decision-making. To this end, the AP recently asked companies, organizations, experts and stakeholders for input through a public consultation.
Organizations are increasingly using algorithms and artificial intelligence (AI) for algorithmic decision-making. Consider evaluating a loan application or reviewing online job applications. If organizations want to use algorithmic decision-making, they must comply with rules. People have the right to human intervention in algorithmic decisions about them that affect them. The AVG contains rules about that.
Human intervention is there to ensure that a decision is made carefully and to prevent people, for example, from being (unintentionally) excluded or discriminated against by the outcome of an algorithm. This human intervention should not only have a symbolic function; it should contribute meaningfully to the decision-making process.
How meaningful human intervention is designed is critical. For example, if an employee is hampered by time constraints or an unclear system, this can impact the outcome of the decision.
Organizations often have questions about how to set this up. Therefore, the AP has developed guidance on meaningful human intervention in algorithmic decision-making developed. These tools can help organizations and companies that want to use algorithmic decision-making to set up human intervention. For this reason, the document has been written to be as practical as possible, to best suit questions that organizations have.
The AP provides examples and an overview of questions that can help organizations design meaningful human intervention. These questions and examples address the relevant factors of people, technology and design, process, and governance.
Consultation responses have been compiled and included without reference to the organization or individual. Government agencies, independent foundations, companies, industry associations and academics provided input. Responses to the consultation included suggestions for particular questions. View a summary of responses to the consultation.
With the tools, the AP aims to help organizations set up meaningful human intervention in the most practical way possible. The document provides practical guidance and is intended as an education to help organizations set up meaningful human intervention in algorithmic decision making.
This document deals with meaningful human intervention that does not constitute exclusively automated decision-making, as referred to in Article 22(1) AVG and Article 11(1) RGR. These articles refer to "decision based solely on automated processing." By this we mean a decision that is entirely based on automated processing and has legal consequences for data subjects or affects them significantly in another way. If there is significant human intervention in the process, this means that it is not a decision based solely on automated processing. The decision then does not fall under Article 22 AVG.
In addition to furnishing human intervention, there are other key concepts from the article that may raise questions. Such as: when is the outcome of the process a "decision"? When does a decision affect someone 'substantially'? These questions are not answered in the paper.
To help organizations further along the way, the AP is going to work this year to give practical substance to other key concepts about automated decision making in the AVG, such as what it means to be "significantly affected," and what information a data subject must receive about the automated decision.
