In dealing with artificial intelligence, people often ask about ethical requirements and guidelines. We meet these questions in the media, in conversation with customers and in private life. And we regularly reflect on our own work.
Defining ethical principles is hard
When Splitbot GmbH was founded, our team worked out guidelines together. One thing became apparent: even where there is a shared understanding of morality and ethics, the interpretation differs in some areas.
AI has no conscience
One thing is clear: no software, with or without AI, has an understanding of ethics of its own accord. At its core, AI is not much more than very, very precise statistics. AI determines probabilities on the basis of data. Ethical requirements have to come from people.
The training data problem
That very data is the reason AI programs occasionally appear unethical. If an AI for selecting applicants is trained only on data from male applicants, it is unable to consider female applicants on equal terms.
Working with the University of Lübeck
All the more grateful, then, are we for the collaboration with Prof. Dr.-Ing. Christian Herzog and the students of the technology ethics programme at the University of Lübeck. We look forward to the intensive exchange and to considering the subject from different perspectives.
