AI is never just technology

By what standards should Artificial Intelligence be developed and deployed? This was the question five fellows set out to explore in April, at the start of the first round of the Publix Tech Journalism Fellowship. Over two weeks, they traveled through Berlin, Cologne, Brussels, and Paris, speaking with researchers, journalists, lobbying experts, company representatives, public administrators, and civil society representatives.

They did not find a simple answer: while researchers, NGOs, and investigative journalists often called for more reporting on power structures, regulation, and societal consequences, company representatives wished for greater attention to successful applications and innovation potential.

The ethics of AI is equally a question of power, infrastructure, regulation, and social responsibility. Four observations emerged from the first two fellowship weeks:

Making power visible

One of the key insights: the debate about AI is always also a debate about power.

Theresa Züger from the Humboldt Institute for Internet and Society placed the question of the common good at the center of her conversation with the fellows. AI can only create societal value if those affected are involved in shaping it and if transparency is established about why it is being used in the first place. Rainer Mühlhoff, Professor of Ethics and Critical Theory of AI, warned against narratives that portray the technology as neutral or without alternative. Such narratives obscure the fact that technical systems emerge within political and economic contexts. They conceal who benefits and who bears which risks.

Conversations with Rebekka Weiß from Microsoft on corporate responsibility and with Thorsten Wetzling and Corbinian Ruckerbauer from Interface on the use of AI by intelligence services made clear how contested questions of control, accountability, and democratic oversight are.

For journalism, a central task lies here: it should not only explain what new technologies can do. It must ask who controls them, what interests lie behind them, and who bears the consequences of their use.

Explaining the invisible infrastructures

In everyday life, AI often appears as a surface: as a chatbot, search function, or recommendation system. Behind it, however, lies a material infrastructure — from data labor to data centers to energy and resource consumption — that rarely becomes visible.

Journalist Ingo Dachwitz drew on his research into data brokers and data labor to show the often hidden structures behind digital technologies. Location data, advertising profiles, data trading, and content moderation are components of a system that remains largely invisible to most users.

This perspective became particularly vivid during a visit to NTT Global Data Centers' Berlin 2 Data Center. After passing through several security checkpoints, the fellows stood in a hall of metal, server racks, and cooling systems. The noise, the cold, and the almost deserted environment made tangible what remains abstract in public debate. AI requires energy, space, cooling, raw materials, and computing power.

Several speakers criticized the fact that these aspects are frequently underrepresented in media coverage. Tech journalism that takes AI seriously must look beyond products and applications. It must also examine supply chains, working conditions, and the ecological consequences of digital technologies.

Reporting on concrete processes

Whether AI is used responsibly is determined not only by abstract ethical principles, but also by what legal rules apply and how they are enforced.

Legal scholar Matthias Kettemann made clear that ethical guidelines offer orientation, but their practical impact remains limited. Binding legal regulations such as the AI Act are necessary to effectively address the risks of AI systems.

Yet how such rules come into being is itself politically contested. In Brussels, it became visible how intensely the struggle over regulation is fought. Felix Duffy from LobbyControl explained how major tech companies attempt to influence political decisions through lobbying, think tanks, and political narratives.

During a lobby tour through the EU quarter, Bram Vranken from Corporate Europe Observatory showed the fellows how densely the offices of large tech companies are clustered around Parliament and the Commission. The physical proximity to political institutions made visible the importance that tech corporations place on political influence.

Conversations with the European Commission and the NGO European Digital Rights also showed that the regulation of AI is a permanent process of negotiation between economic interests, innovation promotion, and the protection of fundamental rights. Throughout, the fellows repeatedly heard how strong the influence of large, economically powerful companies is, while NGOs often have to fight to be heard.

For journalism, this translates into a concrete mandate. It should not only accompany the major debates about the opportunities and risks of AI. It should look more closely at the seemingly smaller processes — legislative procedures, hearings, and lobbying strategies. These are often exactly where the course is set.

Bringing expertise together

The debate about AI is too complex to explain from a single perspective.

The technical foundations, societal impacts, legal frameworks, and political conflicts surrounding AI are deeply intertwined. Good tech journalism therefore depends on exchange with academia and civil society.

The presentation by Thorsten Wetzling and Corbinian Ruckerbauer from Interface on the use of AI by intelligence services made clear that much relevant information only becomes public through investigative research. Academic analyses frequently build on journalistic revelations. Conversely, journalists often depend on the expertise of researchers and NGOs to contextualize complex technical and regulatory relationships.

Good tech journalism does not emerge in isolation. It requires networks, critical distance, and the willingness to bring different forms of expertise into dialogue with one another.

The fellowship continues

At the end of the first two fellowship weeks, the fellows had not found a clear answer to the question of what constitutes "Ethical AI." But they had gained a clearer sense of why that question is so difficult to answer. In September, the fellowship will continue, with a focus on sensitive areas of AI deployment such as medicine, policing, education, and military applications. At the same time, the fellows aim to continue exploring, together with experts and editorial representatives, how tech journalism itself can improve.

One observation many fellows took away from the first two weeks: the longer they spoke about AI, the less the conversation was actually about technology. In September, they therefore want to return to where the search originally began — to the algorithms, models, and training data.

Eva von Grafenstein designs and coordinates the Publix Tech Journalism Fellowship and accompanies the fellows throughout the program.