Whenever technology affects human lives, ethical questions arise. Introducing “misaligned”, a new publication on Medium.

Over the last decades information technology has intensively changed how we live our lives and communicate. In the last years however we have seen a new boom in applying machine learning techniques and data science, often bundled under the label of AI, achieving levels of data processing that was considered impossible just a decade ago and has branched out into in new — if not all — areas of day to day communication, work environments, health and social relations.

While in an ideal world any new application or service would undergo scrutiny of its impact an users, this has too often become a mere afterthought, while the focus has stayed on profitability and efficiency.

In the context of training machine learning models, “alignment” describes the process of steering AI systems towards a defined goal, or in other words to deliver the intended result — the design intent— of a system. However we can observe an increasing disconnect between advertised purpose and the delivered quality of applications. Despite this, alignment has been on the bottom of the list.

With AI based automated decision making becoming more widespread, ethical questions need to be asked: If a system does not deliver on its purpose, acts without a user’s consent or even a user’s awareness of being affected by it, what justification does it have to exist, and what limitations should be applied?

Despite attempts to put guardrails into place by the means of regulation, such attempts often are limited to legal compliance and rarely change the daily work of developers, data scientists and designers.

While AI Ethics has become a more popular field of research over the last years, it often is either ignored, seen as luxury rather than an essential basis, or is limited in scope to machine learning only.

“misaligned” will therefore attempt to cover the topics of AI Ethics in a broader context also addressing ethical questions in data science, data privacy and data self-determination.

For a start, I will be likely writing most of the articles, but I am also planning to conduct interviews with researchers in the field as well as invite guest writers.