Stefan H. Vieweg (editor) - AI for the Good: Artificial Intelligence and Ethics (Management for Professionals)

The human being is a social being and needs exchange; therefore, ethics are needed that summarize the social conventions, even if they differ in time and place, and provide orientation and support.

Digitization is currently changing our world massively, and with it the way we interact with each other. AI is one integral part and probably not obvious to everyone. Some applications such as service robots for people in need of help may quickly raise ethical issuesdespite major technical AI insufficiencies (see 3. below)such as the relationship of dependencies.

AI is a technology that has become enormously more powerful in technical terms (information processing speed and capacity) over the last few decades, thus opening new application possibilities. However, this progress is by no means at the cognitive level! Todays AI applications stagnate in fact in pattern recognition , i.e., all applicationseven if they come along in different formscan ultimately be traced back to pattern recognition, which either orientates itself on previous templates (e.g., speech recognition , facial recognition , text recognition , tracking of individuals, and chatbots) or tries to recognize new connections unsupervised (e.g., by specific deep learning ( DL ) approaches). In this respect, AI can now recognize patterns and process information in repetitive processes much better than ever before and with impressive success: much better than humans could ever do (so-called weak AI ). But AI has no soul, no desire, no emotions, and no values (for which it is worthwhile to useout ofquite changeableconviction) as we humans have. In this respect, a human-like so-called strong AI still seems to be a long way off from todays perspective.

General characteristics, which must be given, so that AI works well, are firstly suitable algorithms and secondly suitable data. Only in the combination of algorithm and data new insights can be gained. If one of the components is questionable, unsuitable, or even insufficient, the result can be disastrous, as real application examples have repeatedly shown (e.g., unintentional discrimination of applicants in an AI-based recruitment process (Amazon 2018) or AI-based photo categorization identifying a black person as a gorilla (Google 2015)).

AIeven if in a weak formis already a fact today and it requires a reflective handling of this technology, since this technology has enormous innovative but also enormous destructive power. A sweeping demonization and rejection of this technology is just as out of place as a naive good faith that nothing negative will happen (I have nothing to hide).

The social challenge that increasingly determines the way we live together in a globalized, fast-moving world that is also threatened by its own over-consumption of resources is that increasing complexity must remain manageable in a highly dynamic manner. Ethical understanding as a societal point of reference for good (desirable) and bad (intolerable) will certainly continue to adapt to digitization, as has always been the case. However, due to the acceleration in the digitized world, previous social-ethical correctives no longer work or work too late. What is needed is a new basic understanding of how to deal with the technical possibilities in society, and where the no go areas lie. A disinhibited exploitation of all technical possibilities will sooner or later lead to social and ecological collapse with hardly assessable and certainly mostly irreversible consequences. This new basic understanding should take into account the element of phenomenological evidence in order to support credibility and confidence in an AI-based decision that cannot be fully analyzed.

A competence is needed that enables the responsible use of the technology in order to make it usable in a socially positive sense.

Strengths and weaknesses of us humans and AI are complementary: where AI has very strong strengths (fast pattern recognition, detection of facts, and provision of evidence) humans tend to be limited. Here, AI can support humans in their inadequacies in order to help them make decisions that are sustainable for society. Used in a targeted manner, AI can be more of a blessing than a curse: in addition to technical optimizations, which ultimately all lead to efficient value creation and thus resource conservation. AI plays a special role in social and economic decision-making processes in order to keep the aforementioned complexity under control. In contrast to this, only humans in their great diversity are by far the only ones capable of making final decisions with their cognitive abilities, their socialization and permanent development and enormous adaptability (e.g., through the neuroplasticity of the brain), their curiosity, their sense of responsibility, and last but not least their soul.