The EU Artificial Intelligence Act (AIA) defines four risk categories for AI systems: unacceptable, high, limited, and minimal. However, it lacks a clear methodology for the assessment of these risks in concrete situations. Risks are broadly categorized based on the application areas of AI systems and ambiguous risk factors. This paper suggests a methodology for assessing AI risk magnitudes, focusing on the construction of real-world risk scenarios. To this scope, we propose to integrate the AIA with (...) a framework developed by the Intergovernmental Panel on Climate Change (IPCC) reports and related literature. This approach enables a nuanced analysis of AI risk by exploring the interplay between (a) risk determinants, (b) individual drivers of determinants, and (c) multiple risk types. We further refine the proposed methodology by applying a proportionality test to balance the competing values involved in AI risk assessment. Finally, we present three uses of this approach under the AIA: to implement the Regulation, to assess the significance of risks, and to develop internal risk management systems for AI deployers. (shrink)

A growing number of decision theorists have, in recent years, defended the view that rationality is permissive under risk: Different rational agents may be more or less risk-averse or risk-inclined. This can result in them making different choices under risk even if they value outcomes in exactly the same way. One pressing question that arises once we grant such permissiveness is what attitude to risk we should implement when choosing on behalf of other people. Are (...) we permitted to implement any of the rationally permissible risk attitudes, is there some specific risk attitude that is required when choosing for others, or are we required to defer to the risk attitudes of the people on whose behalf we are choosing? This article elaborates on this question, explains its wider practical and theoretical significance, provides an overview of existing answers, and explores how to go about providing a more systematic account of how to choose on behalf of others in risky contexts. (shrink)

This article argues that Lara Buchak’s risk-weighted expected utility (REU) theory fails to offer a true alternative to expected utility theory. Under commonly held assumptions about dynamic choice and the framing of decision problems, rational agents are guided by their attitudes to temporally extended courses of action. If so, REU theory makes approximately the same recommendations as expected utility theory. Being more permissive about dynamic choice or framing, however, undermines the theory’s claim to capturing a steady choice disposition in (...) the face of risk. I argue that this poses a challenge to alternatives to expected utility theory more generally. (shrink)

The standard argument to the conclusion that artificial intelligence (AI) constitutes an existential risk for the human species uses two premises: (1) AI may reach superintelligent levels, at which point we humans lose control (the ‘singularity claim’); (2) Any level of intelligence can go along with any goal (the ‘orthogonality thesis’). We find that the singularity claim requires a notion of ‘general intelligence’, while the orthogonality thesis requires a notion of ‘instrumental intelligence’. If this interpretation is correct, they cannot (...) be joined as premises and the argument for the existential risk of AI turns out invalid. If the interpretation is incorrect and both premises use the same notion of intelligence, then at least one of the premises is false and the orthogonality thesis remains itself orthogonal to the argument to existential risk from AI. In either case, the standard argument for existential risk from AI is not sound.—Having said that, there remains a risk of instrumental AI to cause very significant damage if designed or used badly, though this is not due to superintelligence or a singularity. (shrink)

Recent authors have argued that it is overwhelmingly important to mitigate existential risks: risks that threaten the survival or development of humanity. This position is often supported by pessimistically high estimates of existential risk. In this paper, I extend a model by Toby Ord and Thomas Adamczewski to do two things. First, I argue, across a range of modeling assumptions pessimism tends to hamper rather than strengthen the case for existential risk mitigation. Second, I show that pessimism is (...) unlikely to ground the overwhelming importance of existential risk mitigation unless it is coupled with an empirical hypothesis: the time of perils hypothesis. However, I argue, the time of perils hypothesis is probably false. I conclude that existential risk pessimism may tell against the overwhelming importance of existential risk mitigation. GPI Working Paper No. 1-2022. (shrink)

I have claimed that risk-weighted expected utility maximizers are rational, and that their preferences cannot be captured by expected utility theory. Richard Pettigrew and Rachael Briggs have recently challenged these claims. Both authors argue that only EU-maximizers are rational. In addition, Pettigrew argues that the preferences of REU-maximizers can indeed be captured by EU theory, and Briggs argues that REU-maximizers lose a valuable tool for simplifying their decision problems. I hold that their arguments do not succeed and that my (...) original claims still stand. However, their arguments do highlight some costs of REU theory. (shrink)

A natural view in distributive ethics is that everyone's interests matter, but the interests of the relatively worse off matter more than the interests of the relatively better off. I provide a new argument for this view. The argument takes as its starting point the proposal, due to Harsanyi and Rawls, that facts about distributive ethics are discerned from individual preferences in the "original position." I draw on recent work in decision theory, along with an intuitive principle about risk-taking, (...) to derive the view. (shrink)

A wide range of governmental policies characteristic of the modern state seek to reduce individuals’ fatality risks—risks that arise from air and water pollution, pathogens, food ingredients and contaminants, motor vehicles, infrastructure, radiation, workplace accidents, alcohol and recreational drugs, firearms, consumer products, tobacco, natural disasters, and other sources. This book provides a rigorous treatment of the ethics of fatality risk regulation. It does so through the lens of welfare-consequentialism—specifically, lifetime welfarism, with a focus on utilitarianism and prioritarianism. The ethical (...) ranking of possible worlds depends on the patterns of lifetime well-being in the worlds. Premature death is ethically significant just so insofar as it changes the lifetime well-being of the person who dies and perhaps others. At the level of policy choice, the book deploys the social-welfare-function (SWF) framework—the most systematic decision-procedure for implementing lifetime welfarism. The book demonstrates how the SWF methodology can be brought to bear in assessing risk-regulation policies. Each individual faces a policy-specific lottery over lifetime well-being, as determined by their risk profile (probability of surviving the current year and future years) and attribute profile (the attributes the individual will have in the current year and future years if alive rather than dead). In short, a policy corresponds to an array of individual risk and attribute profiles, which can then be assigned a utilitarian or prioritarian value. The SWF methodology as thus applied to risk regulation differs quite substantially from cost-benefit analysis (CBA), which is currently the dominant policy-assessment methodology in governmental practice. (shrink)

Can it be rational to be risk-averse? It seems plausible that the answer is yes—that normative decision theory should accommodate risk aversion. But there is a seemingly compelling class of arguments against our most promising methods of doing so. These long-run arguments point out that, in practice, each decision an agent makes is just one in a very long sequence of such decisions. Given this form of dynamic choice situation, and the (Strong) Law of Large Numbers, they conclude (...) that those theories which accommodate risk aversion end up delivering the same verdicts as risk-neutral theories in nearly all practical cases. If so, why not just accept a simpler, risk-neutral theory? The resulting practical verdicts seem to be much the same. In this paper, I show that these arguments do not in fact condemn those risk-aversion-accommodating theories. Risk aversion can indeed survive the long run. (shrink)

In addition to protecting agents’ autonomy, consent plays a crucial social role: it enables agents to secure partners in valuable interactions that would be prohibitively morally risk otherwise. To do this, consent must be observable: agents must be able to track the facts about whether they have received a consent-based permission. I argue that this morally justifies a consent-practice on which communicating that one consents is sufficient for consent, but also generates robust constraints on what sorts of behaviors can (...) be taken as consent- communicating. (shrink)

Why can testimony alone be enough for findings of liability? Why statistical evidence alone can’t? These questions underpin the “Proof Paradox” (Redmayne 2008, Enoch et al. 2012). Many epistemologists have attempted to explain this paradox from a purely epistemic perspective. I call it the “Epistemic Project”. In this paper, I take a step back from this recent trend. Stemming from considerations about the nature and role of standards of proof, I define three requirements that any successful account in line with (...) the Epistemic Project should meet. I then consider three recent epistemic accounts on which the standard is met when the evidence rules out modal risk (Pritchard 2018), normic risk (Ebert et al. 2020), or relevant alternatives (Gardiner 2019 2020). I argue that none of these accounts meets all the requirements. Finally, I offer reasons to be pessimistic about the prospects of having a successful epistemic explanation of the paradox. I suggest the discussion on the proof paradox would benefit from undergoing a ‘value-turn’. (shrink)

How much does rationality constrain what we should believe on the basis of our evidence? According to this book, not very much. For most people and most bodies of evidence, there is a wide range of beliefs that rationality permits them to have in response to that evidence. The argument, which takes inspiration from William James' ideas in 'The Will to Believe', proceeds from two premises. The first is a theory about the basis of epistemic rationality. It's called epistemic utility (...) theory, and it says that what it is epistemically rational for you to believe is what it would be rational for you to choose if you were given the chance to pick your beliefs and, when picking them, you were to care only about their epistemic value. So, to say which beliefs are permitted, we must say how to measure epistemic value, and which decision rule to use when picking your beliefs. The second premise is a claim about attitudes to epistemic risk, and it says that rationality permits many different such attitudes. These attitudes can show up in epistemic utility theory in two ways: in the way you measure epistemic value; and in the decision rule you use to pick beliefs. This book explores the latter. The result is permissivism about epistemic rationality: different attitudes to epistemic risk lead to different choices of prior beliefs; given most bodies of evidence, different priors lead to different posteriors; and even once we fix your attitudes to epistemic risk, if they are at all risk-inclined, there is a range of different priors and therefore different posteriors they permit. (shrink)

Catastrophic risk raises questions that are not only of practical importance, but also of great philosophical interest, such as how to define catastrophe and what distinguishes catastrophic outcomes from non-catastrophic ones. Catastrophic risk also raises questions about how to rationally respond to such risks. How to rationally respond arguably partly depends on the severity of the uncertainty, for instance, whether quantitative probabilistic information is available, or whether only comparative likelihood information is available, or neither type of information. Finally, (...) catastrophic risk raises important ethical questions about what to do when catastrophe avoidance conflicts with equity promotion. (shrink)

Violence risk assessment tools are increasingly used within criminal justice and forensic psychiatry, however there is little relevant, reliable and unbiased data regarding their predictive accuracy. We argue that such data are needed to (i) prevent excessive reliance on risk assessment scores, (ii) allow matching of different risk assessment tools to different contexts of application, (iii) protect against problematic forms of discrimination and stigmatisation, and (iv) ensure that contentious demographic variables are not prematurely removed from risk (...) assessment tools. (shrink)

It is widely held that the influence of risk on rational decisions is not entirely explained by the shape of an agent’s utility curve. Buchak (Erkenntnis, 2013, Risk and rationality, Oxford University Press, Oxford, in press) presents an axiomatic decision theory, risk-weighted expected utility theory (REU), in which decision weights are the agent’s subjective probabilities modified by his risk-function r. REU is briefly described, and the global applicability of r is discussed. Rabin’s (Econometrica 68:1281–1292, 2000) calibration (...) theorem strongly suggests that plausible levels of risk aversion cannot be fully explained by concave utility functions; this provides motivation for REU and other theories. But applied to the synchronic preferences of an individual agent, Rabin’s result is not as problematic as it may first appear. Theories that treat outcomes as gains and losses (e.g. prospect theory and cumulative prospect theory) account for risk sensitivity in a way not available to REU. Reference points that mark the difference between gains and losses are subject to framing, many instances of which cannot be regarded as rational. However, rational decision theory may recognize the difference between gains and losses, without endorsing all ways of fixing the point of reference. In any event, REU is a very interesting theory. (shrink)

A moderately risk averse person may turn down a 50/50 gamble that either results in her winning $200 or losing $100. Such behaviour seems rational if, for instance, the pain of losing $100 is felt more strongly than the joy of winning $200. The aim of this paper is to examine an influential argument that some have interpreted as showing that such moderate risk aversion is irrational. After presenting an axiomatic argument that I take to be the strongest (...) case for the claim that moderate risk aversion is irrational, I show that it essentially depends on an assumption that those who think that risk aversion can be rational should be skeptical of. Hence, I conclude that risk aversion need not be irrational. (shrink)

In this paper, I explore the prospects for risk-based arguments in favour of anti-natalism, which explain the wrongness of procreation in terms of wrongful risk-imposition on the resultant child. After considering and rejecting two risk-based arguments from the existing literature - David Benatar's and Matti Häyry's - I propose a more promising version that focuses on the lack of appropriate justification for imposing the risks of existence, namely, one that refers to the essential interests of the child (...) on whom those risks are imposed. The paper proceeds in four parts. In Part 2, I set the stage for my discussion by clarifying the basic structure of risk-based arguments and identifying some of their important features. In Parts 3-5, I consider three distinct risk-based arguments for anti-natalism and argue that only the third, justificatory argument has the potential to be successful. (shrink)

Artificial intelligence systems increasingly function as epistemic infrastructure, mediating how individuals and institutions access, evaluate, and act upon knowledge. This paper develops a comprehensive typology of epistemic risks posed by contemporary AI systems, organized around seven core categories: hallucination, error amplification, spurious coherence, authority drift, opaque provenance, filter bubbles, and miscommunicated uncertainty. I argue that addressing these risks requires recognizing three epistemic duties for AI system designers: provenance duty, uncertainty disclosure duty, and contestability duty. The paper presents concrete technical and (...) governance mechanisms for implementing these duties, including retrieval-augmented generation, calibration interfaces, metadata standards, feedback loops, and epistemic audits. Case studies in medical advice, legal research, and news recommendation demonstrate how these mechanisms can safeguard epistemic integrity while preserving AI's beneficial capabilities. (shrink)

The paper explores how the framework of systemic risks can help govern the risks imposed through solar climate engineering research. The central argument is that a systemic perspective of risk is a useful tool for analysing and assessing the risks imposed through Stratospheric Aerosol Injection (SAI) research. SAI is a form of climate engineering that could cool the planet by enhancing its albedo through the injection of aerosols into the stratosphere. Researching such a technology creates systemic risks with a (...) strong sociotechnical component. This component consists of the potential societal harm that a developing or new technology might cause to existing norms, values, institutions, and politics. The systemic risk framework is a valuable heuristic for this case, given the complex interdependencies of societal systems, infrastructures, markets, etc. At the same time, the systemic risk framework can be enhanced through the inclusion of a more robust and reflected ethical considerations on technological risks. Consequently, this article seeks to supplement the systemic risk governance framework with insights from technology ethics. Specifically, the paper offers an ethically reflective conception of societal value dynamism and stakeholder engagement and participation, tying it to existing systemic risk governance approaches. (shrink)

This volume bridges contemporary philosophical conceptions of risk and responsibility and offers an extensive examination of the topic. It shows that risk and responsibility combine in ways that give rise to new philosophical questions and problems. Philosophical interest in the relationship between risk and responsibility continues to rise, due in no small part due to environmental crises, emerging technologies, legal developments, and new medical advances. Despite such interest, scholars are just now working out how to conceive of (...) the links between risk and responsibility, the implications that risks may have to conceptions of responsibility (and vice versa), as well as how such theorizing might play out in applied cases. With contributions from leading scholars, this volume brings together new work examining the interplay between risk and responsibility, exploring its varied philosophical aspects and applications to contemporary issues in law, bioethics, technology, and environmental ethics. Risk and Responsibility in Context will be of interest to philosophers working in ethics, bioethics, philosophy of law, and philosophy of technology, as well as scholars and practitioners in law, health and science management, public policy, and environmental studies. (shrink)

A small but growing number of studies have aimed to understand, assess and reduce existential risks, or risks that threaten the continued existence of mankind. However, most attention has been focused on known and tangible risks. This paper proposes a heuristic for reducing the risk of black swan extinction events. These events are, as the name suggests, stochastic and unforeseen when they happen. Decision theory based on a fixed model of possible outcomes cannot properly deal with this kind of (...) event. Neither can probabilistic risk analysis. This paper will argue that the approach that is referred to as engineering safety could be applied to reducing the risk from black swan extinction events. It will also propose a conceptual sketch of how such a strategy may be implemented: isolated, self-sufficient, and continuously manned underground refuges. Some characteristics of such refuges are also described, in particular the psychosocial aspects. Furthermore, it is argued that this implementation of the engineering safety strategy safety barriers would be effective and plausible and could reduce the risk of an extinction event in a wide range of possible scenarios. Considering the staggering opportunity cost of an existential catastrophe, such strategies ought to be explored more vigorously. (shrink)

With the ascent of modern epidemiology in the Twentieth Century came a new standard model of prediction in public health and clinical medicine. In this article, we describe the structure of the model. The standard model uses epidemiological measures-most commonly, risk measures-to predict outcomes (prognosis) and effect sizes (treatment) in a patient population that can then be transformed into probabilities for individual patients. In the first step, a risk measure in a study population is generalized or extrapolated to (...) a target population. In the second step, the risk measure is particularized or transformed to yield probabilistic information relevant to a patient from the target population. Hence, we call the approach the Risk Generalization-Particularization (Risk GP) Model. There are serious problems at both stages, especially with the extent to which the required assumptions will hold and the extent to which we have evidence for the assumptions. Given that there are other models of prediction that use different assumptions, we should not inflexibly commit ourselves to one standard model. Instead, model pluralism should be standard in medical prediction. (shrink)

This paper examines a recent argument in favour of strong precautionary action—possibly including working to hasten human extinction—on the basis of a decision-theoretic view that accommodates the risk-attitudes of all affected while giving more weight to the more risk-averse attitudes. First, we dispute the need to take into account other people’s attitudes towards risk at all. Second we argue that a version of the non-identity problem undermines the case for doing so in the context of future people. (...) Lastly, we suggest that we should not work to hasten human extinction, even if significant risk aversion is warranted. (shrink)

This book examines the moral luck paradox, relating it to Kantian, consequentialist and virtue-based approaches to ethics. It also applies the paradox to areas in medical ethics, including allocation of scarce medical resources, informed consent to treatment, withholding life-sustaining treatment, psychiatry, reproductive ethics, genetic testing and medical research. If risk and luck are taken seriously, it might seem to follow that we cannot develop any definite moral standards, that we are doomed to moral relativism. However, Dickenson offers strong counter-arguments (...) to this view that enable us to think in terms of universal standards. (shrink)

Risk communication has been generally categorized as a warning act, which is performed in order to prevent or minimize risk. On the other side, risk analysis has also underscored the role played by information in reducing uncertainty about risk. In both approaches the safety aspects related to the protection of the right to health are on focus. However, it seems that there are cases where a risk cannot possibly be avoided or uncertainty reduced, this is (...) for instance valid for the declaration of side effects associated with pharmaceutical products or when a decision about drug approval or retirement must be delivered on the available evidence. In these cases, risk communication seems to accomplish other tasks than preventing risk or reducing uncertainty. The present paper analyzes the legal instruments which have been developed in order to control and manage the risks related to drugs – such as the notion of “development risk” or “residual risk” – and relates them to different kinds of uncertainty. These are conceptualized as epistemic, ecological, metric, ethical, and stochastic, depending on their nature. By referring to this taxonomy, different functions of pharmaceutical risk communication are identified and connected with the legal tools of uncertainty management. The purpose is to distinguish the different functions of risk communication and make explicit their different legal nature and implications. (shrink)

The orthodox theory of instrumental rationality, expected utility (EU) theory, severely restricts the way in which risk-considerations can figure into a rational individual's preferences. It is argued here that this is because EU theory neglects an important component of instrumental rationality. This paper presents a more general theory of decision-making, risk-weighted expected utility (REU) theory, of which expected utility maximization is a special case. According to REU theory, the weight that each outcome gets in decision-making is not the (...) subjective probability of that outcome; rather, the weight each outcome gets depends on both its subjective probability and its position in the gamble. Furthermore, the individual's utility function, her subjective probability function, and a function that measures her attitude towards risk can be separately derived from her preferences via a Representation Theorem. This theorem illuminates the role that each of these entities plays in preferences, and shows how REU theory explicates the components of instrumental rationality. (shrink)

In “Risk Based Passenger Screening in Aviation Security: Implications and Variants of a New Paradigm”, Sebastian Weydner-Volkmann describes the current paradigm shift from ‘traditional’ forms of screening to ‘risk based passenger screening’ (RBS) in aviation security. This paradigm shift is put in the context of the wider historical development of risk management approaches. Through a discussion of Michel Foucault, Herfried Münkler and Ulrich Beck, Weydner-Volkmann analyses the shortcomings of such approaches in public security policies, which become especially (...) evident in the aviation security context. As he shows, the turn towards methods of RBS can be seen as an attempt to address a trade-off ‘trilemma’ between the effective provision of security, the implied costs for industry and passengers, and the ethical, legal and societal implications of the screening procedures. In order to analyse foreseeable outcomes of embracing RBS, he differentiates three prototypical variants of the new paradigm on the basis of their main referent and rationale. For each variant, he then subsequently assesses the implications for the ‘trilemma’, after having unveiled the criteria of analysis that will necessarily have to be followed within a serious appraisal of RBS methods. (shrink)

This dissertation presents and defends an ethical outlook based on two guiding ideas: first, there is a plurality of rationally permissible attitudes towards risks that involve our welfare or well-being; and second, an ethical deliberator should defer to rational individuals’ attitudes towards risk when assessing policies with uncertain consequences that affect them. John Harsanyi famously argued that deferring to people’s preferences under risk leads to utilitarianism. Amartya Sen and John Weymark pointed out that Harsanyi’s argument relies on a (...) contentious interpretation of decision theory under risk—according to which, effectively, only one attitude towards risks involving our well-being, that of risk neutrality, is rationally permissible. As the Sen-Weymark critique clarifies, Harsanyi’s stance is untenable if our quantitative notion of well-being is independent of preferences formed in conditions of risk. Chapter 1 argues that this is indeed the case: our quantitative notion of well-being depends on comparisons between outcomes with no uncertainty. It defends a method for deriving quantities of well-being based on conjoint measurement between objective goods. Chapter 2 explores the space of possibilities that our two guiding ideas leave us with; it shows that under weak assumptions they are incompatible with welfarism, the view that all we need to know in order to evaluate outcomes is how people’s well-being is distributed. It then presents a dilemma for non-welfarist views: they either entail that risk attitudes affect the moral significance of well-being in contexts of zero risk, or they struggle to vindicate the thought that improving everyone’s well-being is an important ethical concern. Chapter 3 sides with the first horn of this dilemma. It offers a defense of a position I call the Risk-Priority View: on this theory, the ethical import of an individual ending up with a certain amount of well-being is affected by that individual’s risk attitude. Roughly, more risk-averse individuals get priority in dire circumstances, while more risk-seeking individuals get priority in favorable ones. (shrink)

Uncertainty, insufficient information or information of poor quality, limited cognitive capacity and time, along with value conflicts and ethical considerations, are all aspects thatmake risk managementand riskcommunication difficult. This paper provides a review of different risk concepts and describes how these influence risk management, communication and planning in relation to forest ecosystem services. Based on the review and results of empirical studies, we suggest that personal assessment of risk is decisive in the management of forest ecosystem (...) services. The results are used together with a reviewof different principles of the distribution of risk to propose an approach to risk communication that is effective aswell as ethically sound. Knowledge of heuristics and mutual information on both beliefs and desires are important in the proposed risk communication approach. Such knowledge provides an opportunity for relevant information exchange, so that gaps in personal knowledge maps can be filled in and effective risk communication can be promoted. (shrink)

Which risks are bad? This is not an easy question to answer in any non-circular way. Not only are risks sought out for various reasons, but risks are plainly discounted in many situations. What may seem "risky" when examined all by itself, may not seem risky when encountered in a real lived situation. Thus risks that are imposed by others, in particular, might seem horrendous when considered in abstraction, but quite acceptable when encountered in life. What we need to do, (...) among other things, is to examine this "seeming." Are people being deluded or distracted when they fail to take seriously the riskiness of activities undertaken by others, or does this make some rational sense? There can be no doubt that people sometimes are deluded, of course. We are all fallible. But risks are not unalloyed evils. They are typically taken because of their association with something positive. So questions about whether people are making wise judgments when they don’t care about certain risks that they take, or, on the other side of the coin, when they become absolutely hysterical when confronted with a risk that appears to others to be insignificant, hinge on a web of further factors. (shrink)

When you are indifferent between two options, it’s rationally permissible to take either. One way to decide between two such options is to flip a fair coin, taking one option if it lands heads and the other if it lands tails. Is it rationally permissible to employ such a tie-breaking procedure? Intuitively, yes. However, if you are genuinely risk-averse—in particular, if you adhere to risk-weighted expected utility theory (Buchak in Risk and rationality, Oxford University Press, 2013) and (...) have a strictly convex risk-function—the answer will often be no: the REU of deciding by coin-flip will be lower than the REU of choosing one of the options outright (so long as at least one of the options is a nondegenerate gamble). This turns out to be a significant worry for risk-weighted expected utility theory. I argue that it adds real bite to established worries about diachronic consistency afflicting views, like risk-weighted expected utility theory, that violate Independence. And that, while these worries might be surmountable, surmounting them comes at a price. (shrink)

This chapter addresses a general question: What is responsible risking? It explores the notion of "responsible risking" as a thick moral concept, and it argues that the notion can be given moral content that could be action-guiding and add an important tool to our moral toolbox. To impose risks responsibly, on this view, is to take on responsibility in a good way. A core part of responsible risking, this chapter argues, is some version of a Forethought Condition. Such a condition (...) requires us to not make decisions or plans such that we could not deliver on our responsibility obligations. This chapter argues that three moral obligations can be derived from descriptive notions of responsibility. They are (1) the control condition, (2) the responsibility to fix condition, and (3) the responsibility to explain condition. The idea is that in order to impose risks responsibly we ought not act in ways that undermine our ability to remain in control over the risks, undermine our ability to sort out negative outcomes by reparations or compensation, or undermine our prospects of providing an explanation for our actions that also makes sense in hindsight and to those affected. This idea is then tried out on the case of germline gene editing. It is argued that such a moral notion of responsibility is also action-guiding when we do not know for certain what is the morally right or obligatory action all-things-considered due to epistemic gaps. (shrink)

We investigate risk attitudes when the underlying domain of payoffs is finite and the payoffs are, in general, not numerical. In such cases, the traditional notions of absolute risk attitudes, that are designed for convex domains of numerical payoffs, are not applicable. We introduce comparative notions of weak and strong risk attitudes that remain applicable. We examine how they are characterized within the rank-dependent utility model, thus including expected utility as a special case. In particular, we characterize (...) strong comparative risk aversion under rank-dependent utility. This is our main result. From this and other findings, we draw two novel conclusions. First, under expected utility, weak and strong comparative risk aversion are characterized by the same condition over finite domains. By contrast, such is not the case under non-expected utility. Second, under expected utility, weak comparative risk aversion is characterized by the same condition when the utility functions have finite range and when they have convex range. By contrast, such is not the case under non-expected utility. Thus, considering comparative risk aversion over finite domains leads to a better understanding of the divide between expected and non-expected utility, more generally, the structural properties of the main models of decision-making under risk. (shrink)

In this paper we aim to demonstrate the enormous ethical complexity that is prevalent in child obesity cases. This complexity, we argue, favors a cautious approach. Against those perhaps inclined to blame neglectful parents, we argue that laying the blame for child obesity at the feet of parents is simplistic once the broader context is taken into account. We also show that parents not only enjoy important relational prerogatives worth defending, but that children, too, are beneficiaries of that relationship in (...) ways difficult to match elsewhere. Finally, against the backdrop of growing public concern and pressure to intervene earlier in the life cycle, we examine the perhaps unintended stigmatizing effects that labeling and intervention can have and consider a number of risks and potential harms occasioned by state interventions in these cases. (shrink)

Papers from the conference on AI Risk (published in JETAI), supplemented by additional work. --- If the intelligence of artificial systems were to surpass that of humans, humanity would face significant risks. The time has come to consider these issues, and this consideration must include progress in artificial intelligence (AI) as much as insights from AI theory. -- Featuring contributions from leading experts and thinkers in artificial intelligence, Risks of Artificial Intelligence is the first volume of collected chapters dedicated (...) to examining the risks of AI. The book evaluates predictions of the future of AI, proposes ways to ensure that AI systems will be beneficial to humans, and then critically evaluates such proposals. 1 Vincent C. Müller, Editorial: Risks of Artificial Intelligence - 2 Steve Omohundro, Autonomous Technology and the Greater Human Good - 3 Stuart Armstrong, Kaj Sotala and Sean O’Heigeartaigh, The Errors, Insights and Lessons of Famous AI Predictions - and What they Mean for the Future - 4 Ted Goertzel, The Path to More General Artificial Intelligence - 5 Miles Brundage, Limitations and Risks of Machine Ethics - 6 Roman Yampolskiy, Utility Function Security in Artificially Intelligent Agents - 7 Ben Goertzel, GOLEM: Toward an AGI Meta-Architecture Enabling Both Goal Preservation and Radical Self-Improvement - 8 Alexey Potapov and Sergey Rodionov, Universal Empathy and Ethical Bias for Artificial General Intelligence - 9 András Kornai, Bounding the Impact of AGI - 10 Anders Sandberg, Ethics and Impact of Brain Emulations 11 Daniel Dewey, Long-Term Strategies for Ending Existential Risk from Fast Takeoff - 12 Mark Bishop, The Singularity, or How I Learned to Stop Worrying and Love AI -. (shrink)

Critical race theorists and standpoint epistemologists argue that agents who are members of dominant social groups are often in a state of ignorance about the extent of their social dominance, where this ignorance is explained by these agents' membership in a socially dominant group (e.g., Mills 2007). To illustrate this claim bluntly, it is argued: 1) that many white men do not know the extent of their social dominance, 2) that they remain ignorant as to the extent of their dominant (...) social position even where this information is freely attainable, and 3) that this ignorance is due in part to the fact that they are white men. We argue that on Buchak's (2010, 2013) model of risk averse instrumental rationality, ignorance of one's privileges can be rational. This argument yields a new account of elite-group ignorance, why it may occur, and how it might be alleviated. (shrink)

Accelerating global climate change drives new climate risks. People around the world are researching, designing, and implementing strategies to manage these risks. Identifying and implementing sound climate risk management strategies poses nontrivial challenges including (a) linking the required disciplines, (b) identifying relevant values and objectives, (c) identifying and quantifying important uncertainties, (d) resolving interactions between decision levers and the system dynamics, (e) quantifying the trade-offs between diverse values under deep and dynamic uncertainties, (f) communicating to inform decisions, and (g) (...) learning from the decision-making needs to inform research design. Here we review these challenges and avenues to overcome them. ▪ People and institutions are confronted with emerging and dynamic climate risks. ▪ Stakeholder values are central to defining the decision problem. ▪ Mission-oriented basic research helps to improve the design of climate risk management strategies. (shrink)

Philosophy &Public Affairs, Volume 51, Issue 4, Page 373-412, Fall 2023.

Many healthcare practices expose people to risks of harmful outcomes. However, the major theories of moral philosophy struggle to assess whether, when and why it is ethically justifiable to expose individuals to risks, as opposed to actually harming them. Sven Ove Hansson has proposed an approach to the ethical assessment of risk imposition that encourages attention to factors including questions of justice in the distribution of advantage and risk, people’s acceptance or otherwise of risks, and the scope individuals (...) have to influence the practices that generate risk. This paper investigates the ethical justifiability of preventive healthcare practices that expose people to risks including overdiagnosis. We applied Hansson’s framework to three such practices: an ‘ideal’ breast screening service, a commercial personal genome testing service, and a guideline that lowers the diagnostic threshold for hypertension. The framework was challenging to apply, not least because healthcare has unclear boundaries and involves highly complex practices. Nonetheless, the framework encouraged attention to issues that would be widely recognised as morally pertinent. Our assessment supports the view that at least some preventive healthcare practices that impose risks including that of overdiagnosis are not ethically justifiable. Further work is however needed to develop and/or test refined assessment criteria and guidance for applying them. (shrink)

Under what conditions does machine learning (ML) model opacity inhibit the possibility of explaining and understanding phenomena? In this article, I argue that nonepistemic values give shape to the ML opacity problem even if we keep researcher interests fixed. Treating ML models as an instance of doing model-based science to explain and understand phenomena reveals that there is (i) an external opacity problem, where the presence of inductive risk imposes higher standards on externally validating models, and (ii) an internal (...) opacity problem, where greater inductive risk demands a higher level of transparency regarding the inferences the model makes. (shrink)

Special Issue “Risks of artificial general intelligence”, Journal of Experimental and Theoretical Artificial Intelligence, 26/3 (2014), ed. Vincent C. Müller. http://www.tandfonline.com/toc/teta20/26/3# - Risks of general artificial intelligence, Vincent C. Müller, pages 297-301 - Autonomous technology and the greater human good - Steve Omohundro - pages 303-315 - - - The errors, insights and lessons of famous AI predictions – and what they mean for the future - Stuart Armstrong, Kaj Sotala & Seán S. Ó hÉigeartaigh - pages 317-342 - - (...) - The path to more general artificial intelligence - Ted Goertzel - pages 343-354 - - - Limitations and risks of machine ethics - Miles Brundage - pages 355-372 - - - Utility function security in artificially intelligent agents - Roman V. Yampolskiy - pages 373-389 - - - GOLEM: towards an AGI meta-architecture enabling both goal preservation and radical self-improvement - Ben Goertzel - pages 391-403 - - - Universal empathy and ethical bias for artificial general intelligence - Alexey Potapov & Sergey Rodionov - pages 405-416 - - - Bounding the impact of AGI - András Kornai - pages 417-438 - - - Ethics of brain emulations - Anders Sandberg - pages 439-457. (shrink)

The EU proposal for the Artificial Intelligence Act (AIA) defines four risk categories: unacceptable, high, limited, and minimal. However, as these categories statically depend on broad fields of application of AI, the risk magnitude may be wrongly estimated, and the AIA may not be enforced effectively. This problem is particularly challenging when it comes to regulating general-purpose AI (GPAI), which has versatile and often unpredictable applications. Recent amendments to the compromise text, though introducing context-specific assessments, remain insufficient. To (...) address this, we propose applying the risk categories to specific AI scenarios, rather than solely to fields of application, using a risk assessment model that integrates the AIA with the risk approach arising from the Intergovernmental Panel on Climate Change (IPCC) and related literature. This model enables the estimation of the magnitude of AI risk by considering the interaction between (a) risk determinants, (b) individual drivers of determinants, and (c) multiple risk types. We use large language models (LLMs) as an example. (shrink)

Population axiology concerns how to evaluate populations in terms of their moral goodness, that is, how to order populations by the relations “is better than” and “is as good as”. The task has been to find an adequate theory about the moral value of states of affairs where the number of people, the quality of their lives, and their identities may vary. So far, this field has largely ignored issues about uncertainty and the conditions that have been discussed mostly pertain (...) to the ranking of risk-free outcomes. Most public policy choices, however, are decisions under uncertainty, including policy choices that affect the size of a population. Here, we shall address the question of how to rank population prospects—that is, alternatives that contain uncertainty as to which population they will bring about—by the relations “is better than” and “is as good as”. We start by illustrating how well-known population axiologies can be extended to population prospect axiologies. And we show that new problems arise when extending population axiologies to prospects. In particular, traditional population axiologies lead to prospect-versions of the problems that they praised for avoiding in the risk-free settings. Finally, we identify an intuitive adequacy condition that, we contend, should be satisfied by any population prospect axiology, and show how given this condition, the impossibility theorems in population axiology can be extended to (non-trivial) impossibility theorems for population prospect axiology. (shrink)

If the intelligence of artificial systems were to surpass that of humans significantly, this would constitute a significant risk for humanity. Time has come to consider these issues, and this consideration must include progress in AI as much as insights from the theory of AI. The papers in this volume try to make cautious headway in setting the problem, evaluating predictions on the future of AI, proposing ways to ensure that AI systems will be beneficial to humans – and (...) critically evaluating such proposals. (shrink)

The epistemology of risk examines how risks bear on epistemic properties. A common framework for examining the epistemology of risk holds that strength of evidential support is best modelled as numerical probability given the available evidence. In this essay I develop and motivate a rival ‘relevant alternatives’ framework for theorising about the epistemology of risk. I describe three loci for thinking about the epistemology of risk. The first locus concerns consequences of relying on a belief for (...) action, where those consequences are significant if the belief is false. The second locus concerns whether beliefs themselves—regardless of action—can be risky, costly, or harmful. The third locus concerns epistemic risks we confront as social epistemic agents. I aim to motivate the relevant alternatives framework as a fruitful approach to the epistemology of risk. I first articulate a ‘relevant alternatives’ model of the relationship between stakes, evidence, and action. I then employ the relevant alternatives framework to undermine the motivation for moral encroachment. Finally, I argue the relevant alternatives framework illuminates epistemic phenomena such as gaslighting, conspiracy theories, and crying wolf, and I draw on the framework to diagnose the undue skepticism endemic to rape accusations. (shrink)

Rowe has recently argued that pure risk of harm cannot itself be a harm. I respond to Rowe and argue that given an appropriate understanding of objective probabilities, pure objective risk of harm can itself be a harm.

Powerful, technically complex international compliance regimes have developed recently in certain professions that deal with risk: banking (the Basel II regime), accountancy (IFRS) and the actuarial profession. The need to deal with major risks has acted as a strong driver of international co-operation to create enforceable international semilegal systems, as happened earlier in such fields as international health regulations. This regulation in technical fields contrasts with the failure of an international general-purpose political and legal regime to develop. We survey (...) the new global regulatory systems in the actuarial, banking and accounting fields, with a view to showing how the need to deal reasonably with risk has resulted in an international de facto law solidly based on correct abstract principles of probability. (shrink)

An axiom of medical research ethics is that a protocol is moral only if it has a “favorable risk-benefit ratio”. This axiom is usually interpreted in the following way: a medical research protocol is moral only if it has a positive expected value -- that is, if it is likely to do more good (to both subjects and society) than harm. I argue that, thus interpreted, the axiom has two problems. First, it is unusable, because it requires us to (...) know more about the potential outcomes of research than we ever could. Second, it is false, because it conflicts with the so-called “soft paternalist” principles of liberal democracy. In place of this flawed rule I propose a new way of making risk-benefit assessments, one that does comport with the principles of liberalism. I argue that a protocol is moral only if it would be entered into by competent subjects who are informed about the protocol. The new rule this eschews all pseudo-utilitarian calculation about the protocol’s likely harms and benefits. (shrink)

According to the modal account of luck it is a matter of luck that p if p is true at the actual world, but false in a wide‐range of nearby worlds. According to the modal account of risk, it is risky that p if p is true at some close world. I argue that the modal accounts of luck and risk do not mesh well together. The views entail that p can be both maximally risky and maximally lucky, (...) but there is nothing which is both maximally lucky and maximally risky. I offer a novel theory of risk that fits together with the modal account of luck and demonstrate that it is both extensionally and formally superior to extant proposals. (shrink)

This is the editorial for a special volume of JETAI, featuring papers by Omohundro, Armstrong/Sotala/O’Heigeartaigh, T Goertzel, Brundage, Yampolskiy, B. Goertzel, Potapov/Rodinov, Kornai and Sandberg. - If the general intelligence of artificial systems were to surpass that of humans significantly, this would constitute a significant risk for humanity – so even if we estimate the probability of this event to be fairly low, it is necessary to think about it now. We need to estimate what progress we can expect, (...) what the impact of superintelligent machines might be, how we might design safe and controllable systems, and whether there are directions of research that should best be avoided or strengthened. (shrink)