Where Is the Human in Algorithmic Decision-Making?
More and more organizations rely on algorithms to support, or even fully make, decisions. But where does that leave the human?
The Dutch Data Protection Authority (Autoriteit Persoonsgegevens) recently published a practical tool for establishing meaningful human intervention. This finally gives substance to a crucial but often vague concept from Article 22 of the GDPR, and it couldn’t be more timely.
What Makes Human Intervention Truly Meaningful?
According to the DPA’s guidance, human oversight only counts as meaningful when:
-
The reviewer genuinely has the power to influence the outcome;
-
The reviewer understands how the algorithm works, and its limitations;
-
There is space for human judgment, such as empathy, contextual awareness, or ethical reflection;
-
Workload, interfaces, and training are designed to encourage critical thinking rather than “click and continue.”
The AI Act also stresses the importance of human oversight in high-risk AI systems. Here too, meaningful intervention is not a symbolic checkbox: it is a mandatory safeguard to prevent harm to fundamental rights.
Embedding the Human Role in AI Governance
At Data & Business Legal, I focus not only on AI compliance but also on ensuring that the human role is properly embedded in policy, systems, and organizational culture. As a Certified AI Compliance Officer (CAICO®), I combine legal expertise with a deep understanding of AI governance and ethics.
Putting the Human Back in the Loop
Do you want to ensure that your algorithms are used both legally and ethically responsibly? Let’s take a closer look at the role of human judgment in your decision-making processes.
How is human intervention organized within your organization, and can you truly challenge the algorithm when it matters?
