Addressing AI’s bias from a humanistic perspective

Artificial intelligence has transformed how we live, work and interact, promising efficiency, precision, and even objectivity. Yet, beneath the shiny veneer of algorithms lies a pressing issue that remains insufficiently addressed — bias.

Far from being impartial, AI often reflects the same prejudices and inequalities embedded in the societies that create it. Bias in AI is not just a technical glitch; it is a social and ethical challenge that demands our attention.

AI systems are only as unbiased as the data they are trained on and the people who design them. Training data often mirrors historical inequalities, stereotypes, or underrepresented groups, leading to biased outcomes.

For example, a widely cited 2018 MIT study found that facial recognition algorithms had an error rate of 34.7 percent for darker-skinned women compared to just 0.8 percent for lighter-skinned men.

This disparity is not just an abstract technical issue — it manifests as a real-world disadvantage for those who are already marginalized.

Bias in AI also stems from the lack of diversity in its creators. With technology sectors still largely homogenous, the perspectives shaping algorithms often miss critical nuances.

As someone with experience in digital transformation projects, I have observed these biases firsthand. For instance, in one project involving AI-powered customer care agents, the system struggled to interpret diverse accents and cultural nuances, leading to a subpar experience for non-native speakers.

The impact of AI bias extends beyond theoretical concerns, influencing decisions in critical areas such as hiring, healthcare, law enforcement, and digital marketing.

In hiring, Amazon’s algorithm famously demonstrated bias against women because it was trained on male-dominated data. This perpetuated existing inequalities in a field that already struggles with gender diversity.

Similarly, in healthcare during the COVID-19 pandemic, pulse oximeters were found to be less accurate on individuals with darker skin tones, highlighting how biased technology can exacerbate health disparities.

In digital campaigns, in a discussion about targeted marketing, such as those used by fashion brands including Mango, concerns arose about AI reinforcing stereotypes. For example, the reinforcement of narrow definitions of beauty.

These examples underscore the human consequences of biased AI systems.

Bias in AI is not just about better coding; it is about understanding the broader societal context in which technology operates.

Patrizia A. Ecker

Some argue that AI bias is inevitable because it mirrors the flaws of human data. While refining datasets and improving algorithms are essential, this perspective oversimplifies the issue.

Bias in AI is not just about better coding; it is about understanding the broader societal context in which technology operates.

Others propose that AI can also serve as a tool to highlight and address biases. For example, AI can analyze hiring trends and suggest equitable practices or identify disparities in healthcare outcomes. This dual role of AI — as both a challenge and a solution — offers a nuanced perspective.

Tackling bias in AI requires a comprehensive approach.

An essential requirement is diverse development teams to ensure that AI systems are built by groups with varied perspectives and experiences. This is vital to uncovering blind spots in algorithm design.

In addition, there should be transparency and accountability so algorithms are interpretable and subject to scrutiny, and allow users to understand and challenge decisions.

There should also be ethical considerations integrated into every stage of AI development. This includes frameworks for bias detection, ethical audits, and public-private collaborations to establish guidelines.

A further requirement is for education and media literacy, to equip individuals and organizations with the tools to recognize AI’s limitations and question its outputs. Critical thinking and media literacy are crucial for fostering a society that demands fairness from technology.

AI is neither a villain nor a savior — it is a reflection of humanity. Bias in AI challenges us to confront uncomfortable truths about inequality and injustice in our societies. While the journey toward unbiased AI may be complex, it is one we cannot afford to ignore.

As someone deeply involved in driving digital transformation and fostering human-centered skills, I have seen firsthand the potential of AI to either entrench inequality or unlock unprecedented opportunities. The choice lies in how we build, deploy, and use these systems.

By addressing the roots of bias and fostering an inclusive approach to AI development, we can ensure that technology serves all of humanity — not just a privileged few.

• Patrizia A. Ecker is a digital transformation adviser, author, and researcher with a doctorate in psychology.