Our Member of Advisory Committee, the German philosopher of technology and physicist, Klaus Kornwachs, published last year a new book on the development of Artificial Intelligence (AI). Since the development of AI has so far been owed to humans, it is also humans who can choose the further development of AI. Kornwachs shows us which human limits might be able to overcome by the use of AI, where the categorical limits of AI lie and, in particular, why human self-limitation is needed, though. His book helps us understand where we should want to go with AI.
In chapter 9, Kornwachs is developing a smart ethic. You should not design an ethic for so-called intelligent machines but an ethic for designers and users of such machines. Such an ethic has to comply with what he calls the principle of condition preservation (“Das Prinzip der Bedingungserhaltung, Eine ethische Studie”, Lit-Verlag, Münster/London, 2000). This means an imperative to act in such a way that the conditions for enabling responsible action of humans are preserved for all humans involved. Machines are not subjects, they are instruments and humans must have these instruments under control, otherwise they cannot take responsibility.
Thus, Kornwachs proposes eight rules for dealing with AI.
1. Never use a decision-making system that substitutes your own decision. Even robots must not be used in decision-making intent.
2. Nihil Nocere – do not tolerate any harm to users.
3. User rights break producer rights.
4. Do not build pseudo-autonomous systems that cannot be turned off. Fully autonomous systems should not be allowed.
5. The production of self-conscious, autonomously acting robots (if possible) is prohibited (analogous to the chimera ban and human cloning ban in genetic engineering).
6. Do not fake a machine as a human subject. A machine must remain machine, imitation and simulation must be always recognizable. It must always be clear to all people involved in human-machine communication that a machine communication partner is a machine.
7. If you do not know the question and the purpose of the question, you cannot handle the system response and understand the behaviour of a robot. The context must always be communicated.
8. Anyone who invents, who produces, operates or disposes of technology has interests. These interests must be disclosed honestly.
This is a quotation from the chapter “Transhumanism as a Derailed Anthropology” authored by Kornwachs in the volume “Transhumanism: The Proper Guide to a Posthuman Condition or a Dangerous Idea?”, edited by W. Hofkirchner and H.-J. Kreowski in 2021. The new book of Kornwachs is written in German. Its title translated in English is “AI and the disruption of work: active beyond job and routine”. You find the rules in German in the 390 pages book on page 291. Read more…