The combination of increased availability of large amounts of fine-grained human behavioral data and advances in machine learning is presiding over a growing reliance on algorithms to address complex societal problems. Algorithmic decision-making processes might lead to more objective and thus potentially fairer decisions than those made by humans who may be influenced by greed, prejudice, fatigue, or hunger. However, algorithmic decision-making has been criticized for its potential to enhance discrimination, information and power asymmetry, and opacity. In this paper, we (...) provide an overview of available technical solutions to enhance fairness, accountability, and transparency in algorithmic decision-making. We also highlight the criticality and urgency to engage multi-disciplinary teams of researchers, practitioners, policy-makers, and citizens to co-develop, deploy, and evaluate in the real-world algorithmic decision-making processes designed to maximize fairness and transparency. In doing so, we describe the Open Algortihms project as a step towards realizing the vision of a world where data and algorithms are used as lenses and levers in support of democracy and development. (shrink)

This paper discusses the establishment of a governance framework to secure the development and deployment of “good AI”, and describes the quest for a morally objective compass to steer it. Asserting that human rights can provide such compass, this paper first examines what a human rights-based approach to AI governance entails, and sets out the promise it propagates. Subsequently, it examines the pitfalls associated with human rights, particularly focusing on the criticism that these rights may be too Western, too individualistic, (...) too narrow in scope and too abstract to form the basis of sound AI governance. After rebutting these reproaches, a plea is made to move beyond the calls for a human rights-based approach, and start taking the necessary steps to attain its realisation. It is argued that, without elucidating the applicability and enforceability of human rights in the context of AI; adopting legal rules that concretise those rights where appropriate; enhancing existing enforcement mechanisms and securing an underlying societal infrastructure that enables human rights in the first place, any human rights-based governance framework for AI risks falling short of its purpose. (shrink)

Available methods such as technology assessment and risk analysis have failed to predict the effects of technological choices. We need to give up the futile predictive ambitions of previous approaches and instead base decisions on systematic studies of alternative future developments. It will then be necessary to cope with mere possibility arguments, i.e., arguments in which a conclusion is drawn from a mere possibility that a course of action may have certain consequences. A five-step procedure is proposed for the assessment (...) of such arguments. It includes a search for mere possibility arguments pointing in different directions, a scientific evaluation that may lead to the specification or refutation of some of these arguments, two symmetry tests, evaluation of the seriousness of the arguments in terms of novelty, spatio-temporal unlimitedness and interference with complex systems, and hypothetical retrospection that aims at finding a course of action that will be defensible in retrospect. (shrink)

In the traditional decision theories the role of forecasts is to a large extent swept under the carpet. I believe that a recognition of the connections between forecasts and decisions will be of benefit both for decision theory and for the art of forecasting.In this paper I have tried to analyse which factors, apart from the utilities of the outcomes of the decision alternatives, determine the value of a decision. I have outlined two answers to the question why a decision (...) which is made on the basis of a forecast is better than a decision which is based on a guess. Neither of these answers is universally valid. An assumption which is necessary for the first answer, i.e. Good's result, is that Bayes' rule is accepted as a correct and generally applicable decision principle. The second answer, which was given with the aid of probability intervals, departed from a more general decision principle, the maximin criterion for expected utilities, which was formulated in order to evade some of the criticism against Bayes' rule. However, the argument leading to the ansser is based on the assumption that the probability intervals associated with the states of nature represent certain knowledge. For this reason this answer is only approximatively valid.As a number of quotations in the section on “the weight of evidence” show, it is not sufficient to describe the knowledge about the states of nature by a single number, representing the (subjective) probability of the state, but something else has to be invoked which measures the amount of information on which a decision is based. Several authors have tried to characterize this mysterious quantity, which here was called the weight of evidence. However, there seems to be little agreement as to how this quantity should be measured. (shrink)

In order to explore the quantifiability and formalizability of uncertainty a wide range of uncertainties are investigated. They are summarized under eight main categories: factual, possibilistic, metadoxastic, agential, interactive, value, structural, and linguistic uncertainty. This includes both classical uncertainty and the uncertainties commonly called great, deep, or radical. For five of the eight types of uncertainty, both quantitative and non-quantitative formalizations are meaningful and available. For one of them (interactive uncertainty), only non-quantitative formalizations seem to be meaningful, and for two (...) (agential and structural uncertainty) neither quantitative nor non-quantitative formalization seems to be a useful approach. (shrink)

Responsible research and innovation features the dialog of science “with society,” and research performed “for society,” i.e., for the benefit of the people. I focus on this latter, outcome-oriented notion of RRI and discuss two kinds of problems. The first one concerns options to anticipate the future course of science and technology. Such foresight knowledge seems necessary for subjecting research to demands of social and moral responsibility. However, predicting science and technology is widely considered impossible. The second problem concerns moral (...) evaluation. The benefit or harm produced by certain research and innovation achievements is often hard to estimate. Against this background of uncertainty in factual and moral assessment, I explore opportunities left for RRI. First, RRI should contribute to maintaining a wide range of approaches. Second, judgments about RRI should draw on zones of convergence among the variety of research approaches pursued. Third, decisions about implementing a technology should be revisable. Fourth, the more specific inclusion of demands from society should be reserved to technology development. Fifth, in many respects, the social compatibility of a new technology is due rather to the social context than the inherent features of the product. Favorable circumstances are transparency, representation of all relevant parties, and procedural fairness. As a result, some of the benefits and detriments of research and innovation can be identified without detailed knowledge of future findings. (shrink)

In a decision making context, an agent’s uncertainty can be either epistemic, i.e. due to her lack of knowledge, or agentive, i.e. due to her not having made use of her decision-making power. In cases when it is unclear whether or not a decision maker presently has control over her own future actions, it is difficult to determine whether her uncertainty is epistemic or agentive. Such situations are often difficult for the agent to deal with, but from an outsider’s perspective, (...) they can have sensible pragmatic solutions. (shrink)