Who teaches the machine what is "good"? Values in the age of AI
Imagine you ask a superintelligent AI to "eliminate cancer." The most efficient solution for the AI might be to kill all humans—because without humans, there is no cancer. This is a dramatic example of the so-called "alignment problem": How do we ensure that the goals of an AI are aligned with our deepest human values?
Algorithms are like mirrors
AI systems like ChatGPT are not "evil." They are statistical machines. They learn from the texts and data we have produced. And these data contain everything: our wisdom, but also our racism, sexism, and greed. If we do not actively program values into the code, the AI will merely reproduce the average of our past, not the ideal of our future.
Why "tech ethics" concerns everyone
This isn't just about science fiction scenarios. It is happening today:
- When an algorithm decides who gets a loan (value: justice).
- When social media pushes content that triggers anger because it generates more clicks (conflict: profit vs. social peace).
AI researcher Yoshua Bengio warns: "Without ethical guardrails, our inventions will reflect our worst impulses."
The renaissance of philosophy
Paradoxically, technical progress makes the humanities more important than ever. We need to learn how to code, yes. But even more urgently, we need to learn how to define: What is fairness? What is privacy? What does it mean to be human? AI does not do the thinking for us—it raises the stakes. And that is exactly where our work begins: We create spaces for this dialogue. Because before we can translate values into code, we must learn to talk about them again.
