Opinion | AI does not become ethically of rules but of involvement

If we want to give ethics a meaningful role in answering these issues, it is not enough to occasionally have a discussion about a specific ethical issue. Then delegating responsibility to AI specialists is also not sufficient: such a Human in the Loop Then soon becomes a human tang.

To get a grip on the ethical aspects of AI, organizations must focus on what I call ‘ethical infrastructure’.

That seems like a somewhat illogical link of concepts. With ethics we think of noble but abstract principles, in infrastructure of concrete bridges or electricity networks.

Yet the term exposes an aspect of ethics that often remains underexposed. Namely: that an organization that wants to work ethically, needs structures and processes that are systematically built, maintain and controlled. Think of it this way: if we want to provide an entire company with light, the wiring must be drawn by all layers of the organization.

You cannot get responsible deployment of AI by hoping for world improvers in the AI ​​teams

To achieve that, organizations have to tackle things differently and change three distinguished cases: the decision-making (who has influenced and heard), the accountability (who supervises and can correct), and the organizational design (the structure).

Take the Cooperative Enterprise Mondragon from Spanish Basque Country, which runs a billion -dollar turnover and is one of the largest companies in the country. There, the employees ultimately vote on the strategic direction and the ethical course. It underlines that decision -making can also be collectively arranged. For example, in the use of AI in areas such as work, care, allowances or the fight against fraud, representatives of, for example, trade unions could also be involved. Or at least be enabled to check decisions and correct it if necessary. This is only possible if there are fixed ethics functions in organizations that have influence, resources and fixed times when they perform their checks.

World improvers

We now lean too much on the ‘moral ambition’ idea of ​​ethics-ethics as something individual, which we leave to its developers in the case of AI. But if we want to make other ethical choices, it will not work to overload the AI ​​developers of the health insurer from the start of this piece with frameworks and moral consciousness. You cannot get responsible deployment of AI by hoping for world improvers in the AI ​​teams, but by organizing organizations in such a way that moral action is structurally supported-also in the long term.

An ethical infrastructure does not change the logic of market forces overnight. She does not automatically prevent profit incentives from outweighing moral considerations. We simply cannot trust that people always put ethical action and companies aside their own interest.

But it is precisely in that the power of a good ethical infrastructure. Because what she does do is make ethical aspects visible and negotiable, and giving the opportunity to employees, customers, social organizations and supervisors to think along with questions and to exert influence on the direction that organizations take in with AI. This way we can also turn the artificial future into a humane future.

To share

Mail the editors