Machine Learning (ML) is reaching the peak of a hype cycle. If you can think of a personal or grand societal challenge – then ML is being proposed to solve it. For example, ML is purported to be able to assist in the current global pandemic by predicting COVID-19 outbreaks and identifying carriers (see, e.g., Ardabili et al. 2020). ML can make our buildings and energy grids more efficient – helping to tackle climate change (see, e.g., Rolnick et al. 2019). ML is even used to tackle the very problem of ethics itself – creating an algorithm to solve ethical dilemmas. Humans, it is argued, are simply not smart enough to solve ethical dilemmas; however, ML can use its mass processing power to tell us the answers regarding how to be ‘good’, in the same way it is better at Chess or Go (Metz 2016). States have taken notice of this new power and are attempting to use ML to solve their problems, including their security problems and, of particular importance in this thesis, the problem of countering terrorism. Counterterrorism procedures including border checks, intelligence collection, waging war against terrorist armed forces, etc. These practices are all being ‘enhanced’ with ML-powered tools (Saunders et al. 2016; Kendrick 2019; Ganor 2019), including: bulk data collection and analysis, mass surveillance, and autonomous weapons among others. This is concerning. Not because the state should not be able to use such power to enhance the services it provides. Not because AI is in principle unethical to use – like land mines or chemical weapons. This is concerning because little has been worked out regarding how to use this tool in a way that is compatible with liberal democratic values. States are in the dark about what these tools can and should do. @en

With Artificial Intelligence (AI) entering our lives in novel ways—both known and unknown to us—there is both the enhancement of existing ethical issues associated with AI as well as the rise of new ethical issues. There is much focus on opening up the ‘black box’ of modern machine-learning algorithms to understand the reasoning behind their decisions—especially morally salient decisions. However, some applications of AI which are no doubt beneficial to society rely upon these black boxes. Rather than requiring algorithms to be transparent we should focus on constraining AI and those machines powered by AI within microenvironments—both physical and virtual—which allow these machines to realize their function whilst preventing harm to humans. In the field of robotics this is called ‘envelopment’. However, to put an ‘envelope’ around AI-powered machines we need to know some basic things about them which we are often in the dark about. The properties we need to know are the: training data, inputs, functions, outputs, and boundaries. This knowledge is a necessary first step towards the envelopment of AI-powered machines. It is only with this knowledge that we can responsibly regulate, use, and live in a world populated by these machines.@en

There is widespread agreement that there should be a principle requiring that artificial intelligence (AI) be ‘explicable’. Microsoft, Google, the World Economic Forum, the draft AI ethics guidelines for the EU commission, etc. all include a principle for AI that falls under the umbrella of ‘explicability’. Roughly, the principle states that “for AI to promote and not constrain human autonomy, our ‘decision about who should decide’ must be informed by knowledge of how AI would act instead of us” (Floridi et al. in Minds Mach 28(4):689–707, 2018). There is a strong intuition that if an algorithm decides, for example, whether to give someone a loan, then that algorithm should be explicable. I argue here, however, that such a principle is misdirected. The property of requiring explicability should attach to a particular action or decision rather than the entity making that decision. It is the context and the potential harm resulting from decisions that drive the moral need for explicability—not the process by which decisions are reached. Related to this is the fact that AI is used for many low-risk purposes for which it would be unnecessary to require that it be explicable. A principle requiring explicability would prevent us from reaping the benefits of AI used in these situations. Finally, the explanations given by explicable AI are only fruitful if we already know which considerations are acceptable for the decision at hand. If we already have these considerations, then there is no need to use contemporary AI algorithms because standard automation would be available. In other words, a principle of explicability for AI makes the use of AI redundant.@en

Transparency is important for liberal democracies; however, the value of transparency is difficult to articulate. In this article we articulate transparency as an instrumental value for providing what we call ensurance and assurance to liberal democratic citizens. Ensurance refers to the property of liberal democracies which prevents it from sliding into authoritarianism and assurance is the property whereby citizens are assured that ensurance exists. Looking at the rise of bulk data collection and use afforded by information communication technologies, this paper focuses on the way that technologies disrupt relations between the state and its citizens, and suggests Value Sensitive Design as a methodology to protect key aspects of liberal democracies. Bulk data collection makes the achieving of ensurance and assurance more difficult due to two types of opacity which arise as a result of the practice: technical opacity—the difficulty for citizens to understand the technology behind bulk data collection; and, algorithmic opacity—opacity which results from properties inherent to algorithms which guide the collection and processing of bulk data. Design requirements will be suggested to respond to the disruptions caused by ICTs between liberal democracies and their citizens which threaten the necessary value for liberal democracies of representativeness.@en