Myths About AI

Digital systems do not emerge in a vacuum. They are developed by people, financed by corporations, and shaped by political interests. Every technological decision influences what becomes visible, whose voices are heard, and who remains overlooked. When AI systems reproduce discrimination or render precarious labour invisible, these are not technical glitches. They are expressions of existing power structures within society.

When it comes to AI and the public good, it is worth looking closely. Power in this field is highly concentrated. Only a handful of companies can afford the massive infrastructure investments required to train large-scale AI models. Most of these companies are based in the United States, where political debates increasingly frame issues such as diversity and inclusion as examples of “woke ideology.”

It is within this political environment that companies such as OpenAI, Google, and Meta are developing products that are becoming accessible to ever more people. Just a few years ago, these companies publicly emphasized their commitment to diversity and inclusion. Today, many of those commitments have largely faded from view. As Big Tech continues to demonstrate, values are often negotiable when growth and shareholder value are at stake.

If there is one thing AI cannot do, it is magically transform an unjust world into a fairer one. Take again diversity as an example: as long as racism, misogyny, homophobia, transphobia, and discrimination against people with disabilities persist in society, these forms of bias will also find their way into technological systems, where they may even be amplified.

There are sensitive areas, such as recruitment, where the limitations and shortcomings of automated systems must be weighed very carefully. It is no coincidence that lawmakers classify such applications as high risk. At the same time, there are less sensitive areas where the targeted use of AI tools can be genuinely beneficial. AI applications, for example, can help users adopt gender-inclusive language or draw attention to one-sided and biased wording.

However, when technology becomes the focal point of discussions about justice and the public good, we risk building better digital tools rather than better societies. Anyone seeking to address social challenges must be willing to move technology out of the spotlight and engage with the root causes of those problems instead. By 2026, we should know better than to fall into the trap of technological solutionism.

The reality is this: no technology, no matter how advanced, will ever magically free us from society’s challenges. Image and facial recognition, machine learning, and generative AI are not solutions to social problems; they are tools. And like any tool, they can be useful under the right circumstances. Machine-learning applications, for instance, can support the early detection of cancer through the analysis of X-rays, MRIs, CT scans, and mammograms. But that does not make doctors, nor a well-functioning healthcare system, obsolete.

At this point, it would be comforting to have a simple recipe to follow. But as is so often the case, the key lies in embracing complexity: understanding interconnections, uncovering power structures, grasping the many dimensions of social problems, looking beyond marketing promises, and resisting the pressure exerted by Big Tech companies eager to sell their own ideologies.

One thing is clear: AI for the common good cannot be the same AI that consumes vast amounts of environmental resources and relies on precarious human labour performed by millions of data workers.

The goal cannot be to give AI a veneer of fairness while the world itself remains unjust. The real challenge is to transform the power structures that produce inequality in the first place. Achieving this requires everything from breaking up monopolies and enforcing robust regulation to supporting alternatives and investing in solutions that extend beyond digital tools.

And it all begins with the ability to imagine a better world. A world in which technology is not the ultimate goal, but a means to serve the wellbeing of both people and the planet.