The worst that could happen

Do you consider yourself an optimistic person? Optimism is a personality trait, an attitude towards life. Without optimism, it would be difficult to undertake, research or imagine advances that do not yet exist. Optimism is also a cognitive bias that leads us to underestimate risk. In technology projects, it manifests itself in unrealistic deadlines, insufficient budgets and overly confident decision-making.

Analyzing every decision takes energy, so we categorize, simplify, and supplement information with heuristics. An automatic way of thinking. It works well, most of the time: we don’t have to constantly think about how to get from home to work, we don’t consider deeply whether we want an apple. Problems arise under pressure, when there is too much information, when the situation is unclear, when we rely on memory rather than evidence. These mental shortcuts lead to errors and expose biases.

Biases are well-known in technology teams. Many companies implement systematic methods to mitigate their impact. Optimism bias and ingroup bias cause problems to be overlooked in the project definition phase. One technique to combat this is premortem. The team meets and imagines that the project has already been completed, launched , and failed. From this fictional future, they write a letter explaining what went wrong. This way, risks and fragile decisions are identified before they become consolidated.

There are many other biases that affect technological development: anchoring bias, confirmation bias, or the sunk cost fallacy, also known as the Concorde effect. To manage their influence, methods have been developed that start from the same premise: bias cannot be eliminated. However, its impact on important decisions can be reduced with explicit rules and procedures.

So, what methods are used to combat the biases that reinforce structural inequalities? Affinity bias favors people similar to oneself. Authority bias gives more weight to certain voices, even if what they say is not better. Ingroup bias favors those who belong to the dominant group.

These dynamics contaminate career-defining decisions, affecting hiring and promotions. They determine who is listened to in meetings and who is interrupted. They separate strategic work from invisible work. They shape who is considered a leader and who occupies support roles.

The impact also reaches the products. If the designer shares similar profiles and experiences, it is more likely that certain bodies and realities will be left out. Voice recognition systems that fail with female voices, selection algorithms that penalize non-linear trajectories, less accurate facial recognition with dark skin… Errors that need to be detected at the beginning, not once deployed.

In these cases, more ambition, more resilience and more technical skills are required of those who fall outside the dominant pattern. The story presents the technological space as open and meritocratic. The problem shifts from the system to the individual trajectory, despite knowing that biases are inevitable.

Anti-bias methods were born without a gender perspective or an intersectional perspective. They need to be adapted. Change the question and the result changes. “ What could go wrong?” is not the same as “ For whom could it go wrong, in what contexts, and who are we leaving out?” When there is no method, inertia decides. And inertia maintains the existing order. That is why it is necessary to design processes that test it.

What is the worst that could happen, and for whom?