Algorithms are deeply woven into our daily lives, from search results and news feeds to medical diagnoses and credit assessments. They offer convenience and efficiency, but also carry risks that are rarely as visible as the benefits.

It is tempting to see algorithms as neutral technology. They are always a product of human choices, assumptions and interests.

Many decisions once made by people now lie in the hands of systems we do not fully understand. A job vacancy might never reach you because an algorithm does not match your profile, on grounds never explained to you. A loan might be denied based on patterns the applicant cannot understand, even though they are entirely consistent for the system itself. And which news articles you read is strongly determined by what the algorithm thinks appeals to you, and therefore also by what you never get to see.

The effect of this is subtle but far-reaching. Our choices shift slowly, often without us noticing. Our autonomy is shaped by logics we did not choose ourselves, and which sometimes run counter to our values, without there ever having been a moment we consciously agreed to them.

That is why we must keep testing technology against human values. It begins with making those values explicit: justice, transparency, inclusion, human dignity. They function as a compass when designing, testing and deploying algorithms, not as decoration afterwards but as a guiding principle from the start.

A fair algorithm does not discriminate on irrelevant characteristics. Transparency means decision-making is explainable to those affected by it.

Inclusion means diverse perspectives are taken into account during design, so the outcome is not biased by a limited group of designers who, without ill intent, build their own blind spots into the system. Explainability, often referred to as explainable AI, is crucial for trust in algorithms. People must be able to understand why a decision was made, especially when that decision has major consequences for their life. An algorithm making a medical diagnosis should not only give the result, but also the underlying reasoning. Without that explanation, a technological black box emerges that makes the conversation about values and fairness impossible, simply because there is nothing left to discuss.

A common misconception is that the algorithm makes the decision, and that responsibility therefore lies with the technology. But behind every algorithm stand people: developers, managers, policymakers. They determine the data used, the rules that apply, and the goals the system pursues. The algorithm is the instrument. The choices are human, even when they are subsequently applied automatically, at a scale no individual human could ever oversee.

Accountability therefore also means taking ownership of the consequences of algorithmic decisions. That calls for governance structures in which ethical review is as much a matter of course as technical quality control, not as a separate step that can be skipped under time pressure, but as an integrated part of how a system is built and maintained.

Social media show clearly how algorithms shape our reality.

These systems are optimised for attention: the longer you keep watching, the better for the business model. That means algorithms often favour content that provokes emotion, anger, outrage, sensation, because that holds your attention longer. The result is that societal debates harden and polarisation increases, not as a side effect but as a direct consequence of what the system was designed to achieve.

There is a direct link here to human values. If we want to preserve a healthy public space, we must look critically at how these algorithms are structured and what goals they serve. Do we want the economy of attention to set the course, or do we steer based on societal and democratic values that cannot simply be expressed in a conversion rate?

Leadership in a data-driven world means having the courage to ask these questions, not only from a technical standpoint, but mainly from a human one. It calls for embedding ethics in policy, diversity in design teams, and continuous testing and adjustment. Technology is never finished. Its societal impact keeps changing and demands ongoing attention, not a one-off approval at launch.

Ultimately, the question is not whether we use algorithms, but how. Algorithms can add enormous value, from finding medical treatments faster to making logistics chains more sustainable. But only if we preserve the human scale do these systems keep supporting us rather than steering us. Bringing the human into the algorithm means seeing technology as an extension of our values, not a replacement for them, and accepting that technology is never neutral, and that it is our shared responsibility to use it for good.

In a world increasingly steered by data, perhaps our greatest challenge is to remain human, and to ensure our technology keeps reflecting that too.

Notes for those who wish to read further:

1. Cathy O’Neil, Weapons of Math Destruction (2016, Crown Publishing). On how algorithms reproduce and amplify the biases of their creators at a scale humans alone could never reach.
2. Shoshana Zuboff, The Age of Surveillance Capitalism (2019, PublicAffairs). On how systems optimised for attention restructure our public space.
3. Virginia Dignum, Responsible Artificial Intelligence: How to Develop and Use AI in a Responsible Way (2019, Springer). On governance structures in which ethical review is integrated, not separate.
4. Stuart Russell, Human Compatible: Artificial Intelligence and the Problem of Control (2019, Viking). On explainability as a design requirement rather than an optional extra.
5. Luciano Floridi, The Ethics of Information (2013, Oxford University Press). On the moral infrastructure every information system inevitably carries with it.