Abstract. The main current approach to the AI safety is AI alignment, that is, the creation of AI whose preferences are aligned with “human values.” Many AI safety researchers agree that the idea of “human values” as a constant, ordered sets of preferences is at least incomplete. However, the idea that “humans have values” underlies a lot of thinking in the field; it appears again and again, sometimes popping up as an uncritically accepted truth. Thus, it deserves a thorough (...) deconstruction, which I will do by listing and analyzing comprehensively the hidden assumptions of the idea that “humans have values.” This deconstruction of human values will be centered around the following ideas: “Human values” are useful descriptions, but not real objects; “human values” are bad predictors of behavior; the idea of a “human value system” has flaws; “human values” are not good by default; and human values cannot be separated from human minds. The method of analysis is listing hidden assumptions on which the idea of “human values” is built. I recommend that either the idea of “human values” should be replaced with something better for the goal of AI safety, or at least be used very cautiously. The approaches to AI safety which don’t use the idea of human values at all may require more attention, like the use of full brain models, boxing, and capability limiting. (shrink)

Inchul Yum (2024) argues that the widespread adoption of language agent architectures would likely increase the risk posed by AI by simplifying the process of aligning artificial systems with human values and thereby making it easier for malicious actors to use them to cause a variety of harms. Yum takes this to be an example of a broader phenomenon: progress on the alignment problem is likely to be net safety-negative because it makes artificial systems easier for malicious actors (...) to control. I offer some reasons for skepticism about this surprising and pessimistic conclusion. (shrink)

A morally acceptable course of AI development should avoid two dangers: creating unaligned AI systems that pose a threat to humanity and mistreating AI systems that merit moral consideration in their own right. This paper argues these two dangers interact and that if we create AI systems that merit moral consideration, simultaneously avoiding both of these dangers would be extremely challenging. While our argument is straightforward and supported by a wide range of pretheoretical moral judgments, it has far-reaching moral implications (...) for AI development. Although the most obvious way to avoid the tension between alignment and ethical treatment would be to avoid creating AI systems that merit moral consideration, this option may be unrealistic and is perhaps fleeting. So, we conclude by offering some suggestions for other ways of mitigating mistreatment risks associated with alignment. (shrink)

The “Value Alignment Problem” is the challenge of how to align the values of artificial intelligence with human values, whatever they may be, such that AI does not pose a risk to the existence of humans. Existing approaches appear to conceive of the problem as "how do we ensure that AI solves the problem in the right way", in order to avoid the possibility of AI turning humans into paperclips in order to “make more paperclips” or (...) eradicating the human race to “solve climate change”. This paper proposes an approach to Alignment rooted in the Enactive theory of mind that reconceptualises it as "how do we make relevant to AI what it relevant to humans". This conceptualisation is supported with a discussion of 4E cognition and goes on to suggest that the Alignment problem and the Frame problem are the same problem. The paper concludes with a discussion of the tradeoffs of the different approaches to value alignment. (shrink)

Tae Wan Kim, John Hooker, and Thomas Donaldson make an attempt, in recent articles, to solve the alignment problem. As they define the alignment problem, it is the issue of how to give AI systems moral intelligence. They contend that one might program machines with a version of Kantian ethics cast in deontic modal logic. On their view, machines can be aligned with human values if such machines obey principles of universalization and autonomy, as well as (...) a deontic utilitarian principle. Programming machines to do so might be useful, in their view, for applications such as future autonomous vehicles. Their proposal draws both on traditional logic-based and contemporary connectionist approaches, to fuse factual information with normative principles. I will argue that this approach makes demands of machines that go beyond what is currently feasible, and may extend past the limits of the possible for AI. I also argue that a deontological ethics for machines should place greater stress on the formula of humanity of the Kantian categorical imperative. On this principle, one ought never treat a person as a mere means. Recognition of what makes a person a person requires ethical insight. Similar insight is needed to tell treatment as a means from treatment as a mere means. The resources in Kim, Hooker, and Donaldson’s approach is insufficient for this reason. Hesitation regarding deployment of autonomous machines is warranted in light of these alignment concerns. (shrink)

This paper presents a new trilemma with respect to resolving the control and alignment problems in machine ethics. Section 1 outlines three possible types of artificial moral agents (AMAs): (1) 'Inhuman AMAs' programmed to learn or execute moral rules or principles without understanding them in anything like the way that we do; (2) 'Better-Human AMAs' programmed to learn, execute, and understand moral rules or principles somewhat like we do, but correcting for various sources of human moral error; and (3) (...) 'Human-Like AMAs' programmed to understand and apply moral values in broadly the same way that we do, with a human-like moral psychology. Sections 2–4 then argue that each type of AMA generates unique control and alignment problems that have not been fully appreciated. Section 2 argues that Inhuman AMAs are likely to behave in inhumane ways that pose serious existential risks. Section 3 then contends that Better-Human AMAs run a serious risk of magnifying some sources of human moral error by reducing or eliminating others. Section 4 then argues that Human-Like AMAs would not only likely reproduce human moral failures, but also plausibly be highly intelligent, conscious beings with interests and wills of their own who should therefore be entitled to similar moral rights and freedoms as us. This generates what I call the New Control Problem: ensuring that humans and Human-Like AMAs exert a morally appropriate amount of control over each other. Finally, Section 5 argues that resolving the New Control Problem would, at a minimum, plausibly require ensuring what Hume and Rawls term ‘circumstances of justice’ between humans and Human-Like AMAs. But, I argue, there are grounds for thinking this will be profoundly difficult to achieve. I thus conclude on a skeptical note. Different approaches to developing ‘safe, ethical AI’ generate subtly different control and alignment problems that we do not currently know how to adequately resolve, and which may or may not be ultimately surmountable. (shrink)

The AGI alignment problem has a bimodal distribution of outcomes with most outcomes clustering around the poles of total success and existential, catastrophic failure. Consequently, attempts to solve AGI alignment should, all else equal, prefer false negatives (ignoring research programs that would have been successful) to false positives (pursuing research programs that will unexpectedly fail). Thus, we propose adopting a policy of responding to points of philosophical and practical uncertainty associated with the alignment problem by (...) limiting and choosing necessary assumptions to reduce the risk of false positives. Herein we explore in detail two relevant points of uncertainty that AGI alignment research hinges on---meta-ethical uncertainty and uncertainty about mental phenomena---and show how to reduce false positives in response to them. (shrink)

Abstract: One approach to alignment with human values in AI and robotics is to engineer artificial systems isomorphically with human beings. One persistent problem is that different human beings champion different value orientations as ideal, exclusive of others, thereby confounding worries about AI. Prior work proposed a view of human development of purpose and source of meaning in life in terms of project ideal conditions, in effect establishing lifelong value orientations according to which intermediate situations are evaluated. The (...) idea is that robots designed accordingly may autonomously align with human values through similar developmental processes, to realize project ideal conditions through iterative interaction with social and object environments. The present work considers two apparently mutually exclusive views on value alignment, that of St. Augustine and contemporary cognitive scientists including Gregg Northoff. The present work describes these views as divergent error theories which, due to variable development of variable cognitive potential, present to different subjects as mutually exclusive yet accurate accounts of personal experience. Specifically, the hypothesis proposed is that differential development of spindle neural projections during adolescence establishes lifelong connections between prior established relatively immediate routine processes (entrained during childhood) and later developing higher-level social and self-processes, with these projections hard-wiring lifelong value orientations. Insights from this study are applied to considerations for AI value alignment and robot religion. Robot experiments informed by this study may eventually represent such differences by design, with autonomous robots developing religious associations with temporally distal project ideal conditions through love for humanity, as described by Augustine, contrary to contemporary accounts emphasizing selfish fit to passing norms. Finally, the motivating hope of this study is that risks associated with value misaligned robots may be obviated by adequately informed designers. (shrink)

IN “CONSTRUCTIVISM, AGENCY, AND THE PROBLEM of Alignment,” Michael Bratman considers how lessons from the philosophy of action bear on the question of how best to construe the agent’s standpoint in the context of a constructivist theory of practical reasons. His focus is “the problem of alignment”: “whether the pressures from the general constructivism will align with the pressures from the theory of agency” (Bratman 2012: 81). He thus brings two lively literatures into dialogue with each (...) other. This is laudable. However, I shall argue that the considerations Bratman brings to bear from the literature on action do not support his conclusion that evaluative judgments are not constitutive of the agent’s standpoint in the context of constructivism. (shrink)

The AGI alignment problem has a bimodal distribution of outcomes with most outcomes clustering around the poles of total success and existential, catastrophic failure. Consequently, attempts to solve AGI alignment should, all else equal, prefer false negatives (ignoring research programs that would have been successful) to false positives (pursuing research programs that will unexpectedly fail). Thus, we propose adopting a policy of responding to points of metaphysical and practical uncertainty associated with the alignment problem by (...) limiting and choosing necessary assumptions to reduce the risk false positives. Herein we explore in detail some of the relevant points of uncertainty that AGI alignment research hinges on and consider how to reduce false positives in response to them. (shrink)

Various thinkers have been attempting to align artificial intelligence (AI) with ethics (Christian, 2020; Russell, 2021), the so-called problem of alignment, but some suspect that the problem may be intractable (Yampolskiy, 2023). In the following, we make an argument by analogy to analyze the possibility that the problem of alignment could be intractable. We show how the Tri-Omni properties in theology can direct us towards analogous properties for artificial superintelligence, Tri-Opti properties. However, just as the (...) Tri-Omni properties are vulnerable to metaphysical incompatibility, we will show that the Tri-Opti properties are vulnerable to a corresponding physical incompatibility, because optimal physical systems exist on the Pareto Frontier. We will explain that, while Tri-Omni incompatibility thus seems metaphysically soluble, the Tri-Opti incompatibility seems physically insoluble due to the constraints of multi-variable optimization. Finally, we will provide some scenarios by which this incompatibility may be realized. Besides for the primary purpose of its theoretical considerations, this analysis has the secondary interdisciplinary purpose of communicating to the theologically literate the potential importance of AI safety work, and the gesturing towards the possibility of the collaboration in both fields on topics in ethics, epistemology, and ontology. (shrink)

While representation learning techniques have shown great promise in application to a number of different NLP tasks, they have had little impact on the problem of ontology matching. Unlike past work that has focused on feature engineering, we present a novel representation learning approach that is tailored to the ontology matching task. Our approach is based on embedding ontological terms in a high-dimensional Euclidean space. This embedding is derived on the basis of a novel phrase retrofitting strategy through which (...) semantic similarity information becomes inscribed onto fields of pre-trained word vectors. The resulting framework also incorporates a novel outlier detection mechanism based on a denoising autoencoder that is shown to improve performance. An ontology matching system derived using the proposed framework achieved an F-score of 94% on an alignment scenario involving the Adult Mouse Anatomical Dictionary and the Foundational Model of Anatomy ontology (FMA) as targets. This compares favorably with the best performing systems on the Ontology Alignment Evaluation Initiative anatomy challenge. We performed additional experiments on aligning FMA to NCI Thesaurus and to SNOMED CT based on a reference alignment extracted from the UMLS Metathesaurus. Our system obtained overall F-scores of 93.2% and 89.2% for these experiments, thus achieving state-of-the-art results. (shrink)

The problem of infinite regress presents a profound challenge in epistemology and philosophy, questioning the possibility of achieving foundational knowledge amidst an endless chain of justifications. This paper introduces a set of four axioms designed to directly address and resolve the problem of infinite regress, ensuring theoretical rigor and applicability across diverse scenarios, including simulated or illusory realities. By focusing on Direct Address, Intellectual Rigor in All Realities, Avoiding Pragmatic Dismissals, and Theoretical Consistency, these axioms provide a structured (...) framework for evaluating philosophical theories aimed at overcoming this complex issue. The paper explores current theories that align with these axioms, evaluates popular theories that do not, and argues for the critical role of these axioms in guiding future philosophical inquiry and resolution of infinite regress. (shrink)

Ambitious value learning proposals to solve the AI alignment problem and avoid catastrophic outcomes from a possible future misaligned artificial superintelligence (such as Coherent Extrapolated Volition [CEV]) have focused on ensuring that an artificial superintelligence (ASI) would try to do what humans would want it to do. However, present and future sentient non-humans, such as non-human animals and possible future digital minds could also be affected by the ASI’s behaviour in morally relevant ways. This paper puts forward Sentientist (...) Coherent Extrapolated Volition, an alternative to CEV, that directly takes into account the interests of all sentient beings. This ambitious value learning proposal would significantly reduce the likelihood of risks of astronomical suffering from the ASI’s behaviour, and thus we have very strong pro-tanto moral reasons in favour of implementing it instead of CEV. This fact is crucial in conducting an adequate cost-benefit analysis between different ambitious value learning proposals. (shrink)

Faith in God conflicts with reason—or so we’re told. We focus on two arguments for this conclusion. After evaluating three criticisms of them, we identify an assumption they share, namely that faith in God requires belief that God exists. Whether the assumption is true depends on what faith is. We sketch a theory of faith that allows for both faith in God without belief that God exists, and faith in God while in belief-cancelling doubt God’s existence. We then argue that (...) our theory, unlike the theory of Thomas Aquinas, makes sense of four central items of faith-data: (i) pístis in the Synoptics, (ii) ʾemunāh in the Hebrew scriptures, (iii) exemplars of faith in God, including Abraham, Jesus, and Mother Teresa, and (iv) the widespread experience of people of faith today. We close by assessing revisions of the two arguments we began with, revisions that align with our theory of faith, and find them dubious, at best. (shrink)

4Es approaches to cognition draw an unconventional picture of cognitive processes and of the mind. Instead of conceiving of cognition as a process that always takes place within the boundaries of the skull and the skin, these approaches hold that cognition is a situated process that often extends beyond human agents’ physical boundaries. In particular, supporters of the extended mind theory and of the enactive approach claim that embodied action in a perceptually complex environment is constitutive of cognitive processes, and (...) thus of the mind. (Chapter 1). Although both draw an extended picture of cognition and of the mind, supporters of the extended mind and supporters of the enactive approach disagree on the value of internal representations in perceptually-guided action. The extended mind theory holds that we need to posit internal action-oriented representations (AORs). AORs are valuable -the argument goes- because they account for action-selection, action-control, and for the prediction of incoming perceptual information. Advocates of the enactive approach to cognition, instead, argue against AORs. Firstly, they argue that AORs do not fulfill the necessary conditions for the identification of representational entities. Therefore, that of AOR is an ill-conceived epistemic construct. Secondly, they argue that AORs are expressive of an internalist prejudice, which makes the extended mind theory weaker than it should be from an explanatory and metaphysical point of view. I endorse both these arguments advanced by supporters of the enactive approach. Moreover, through a semiotic analysis of the construct and of the workings of AORs, I show that AORs are not active at all. Hence, I argue that AORs do not do the job they are supposed to do: extending the mind’s boundaries by explaining our perceptual relation with the world as an active, embodied, and world-involving process. Therefore, I suggest that the construct of AOR ought to be dismissed - at least in current projects in 4Es cognition (Chapter 2). I argue that similar arguments apply to recent attempts to make predictive processing accounts of perceptually-guided action more aligned with 4Es approaches to cognitive science by appealing to predictive internal models of our perceptual world. I thus propose an alternative explanation of the key features of perceptually-guided action that AORs and these internal models are supposed to be grasp. Action-control, action-selection, and the anticipatory aspects of action-perception loops can be explained by appealing to embodied actions in a field of relevant affordances, which are perceived as such based on the agent’s motor skills and motives. Furthermore, I suggest that, by combining the enactive approach to cognition with an analysis of the indexical features of our perceptual environment, we can explain perceptually-guided action in a way that truly accounts for our minds as entities that often extend beyond our skulls and skin. And we can do so without holding onto spooking entities like AORs and internal predictive models. (Chapter 3). (shrink)

This paper serves as an intuition building mechanism for understanding the basics of AI, misalignment, and the reasons for why strong AI is being pursued. The approach is to engage with both pro and anti AI development arguments to gain a deeper understanding of both views, and hopefully of the issue as a whole. We investigate the basics of misalignment, common misconceptions, and the arguments for why we would want to pursue strong AI anyway. The paper delves into various aspects (...) of the problem, including existential risk, deception, hedonic adaptation, and the potential for the complete extinction of humanity. By integrating multiple elements of philosophy, this paper aims to provide a holistic understanding of the alignment problem and its significance for the future of humanity. (shrink)

The main aim of the study was to determine the levels of Adversity Quotient and problem solving skills in Mathematics of BISU - MC students taking BSEdMathematics in the school year 2021-2022. It sought to find if there was a significant difference in the respondents’ levels of AQ and problem solving skills in Mathematics across their age, gender and year level as well as their level of AQ as a significant predictor of their level of problem solving (...) skills in Mathematics. It also aimed to develop a plan of action that would be proposed to improve these two aspects of their being. The total number of actual respondents was 163. Purposive sampling was used. The study utilized the quantitative type of study. It made use of the descriptive design to describe the characteristics of the population being studied and the regression design to infer the relationship between the independent variable and dependent variable. The Online AQ Profile was used for determining the respondents’ level of AQ and a 10-item test was used for determining their level of problem solving skills in Mathematics. Both inquired their profile. The data provided by the respondents were collected and subjected to statistical treatment through IBM SPSS Statistics Trial software. Data revealed that the age of the respondents ranged from 18 to 22 years old. Females numerically dominated the analyzed field. Majority of the respondents were from the first and fourth year levels. Their AQ was below average while their problem solving skills in Mathematics was satisfactory. The age, gender and year level of students did not matter in identifying their level of AQ. On the other hand, the older students had a higher level of problem solving skills in Mathematics than the younger ones and the students in the higher year level had a higher level of problem solving skills in Mathematics than those in the lower year level. Finally, their level of AQ gave a positive influence on their level of problem solving skills in Mathematics. The education system should be aligned with the profile of the students. The teachers would let the students read the book of Paul G. Stoltz, PhD titled “Adversity Quotient: Turning Obstacles into Opportunities”. The students would also reflect on the word of God. Also, the teachers would let the students study the book by George Polya titled “How To Solve It”. The students would also continue to solve various routine problems. Regardless of age, gender and year level, the family and friends of the students should encourage them in every way they can for their better future as they overcome their adversities. Mathematics curriculum makers and teachers work together to improvise teaching and learning Mathematics specifically problem solving for the younger students and those in the lower year level. Future researchers could replicate the study to further verify the results. Research could focus specifically on the CORE dimensions of AQ predicting the level of problem solving skills in Mathematics. (shrink)

In this paper, I argue that David Hume's is-ought problem stems from morality as a simplifying framework. Morality is the attempt to describe human behaviour and its relation to nature through a court framework, dating to prehistoric times. Such a court does not really exist, hence we are not referring to anything directly when we make moral statements, and therefore 'is' and 'ought' do not align. The solution is to replace morality with what it's really trying to describe: patterns (...) in individual behaviour. (shrink)

The field of AI safety aims to prevent increasingly capable artificially intelligent systems from causing humans harm. Research on moral alignment is widely thought to offer a promising safety strategy: if we can equip AI systems with appropriate ethical rules, according to this line of thought, they'll be unlikely to disempower, destroy or otherwise seriously harm us. Deontological morality looks like a particularly attractive candidate for an alignment target, given its popularity, relative technical tractability and commitment to harm-avoidance (...) principles. I argue that the connection between moral alignment and safe behavior is more tenuous than many have hoped. In general, AI systems can possess either of these properties in the absence of the other, and we should favor safety when the two conflict. In particular, advanced AI systems governed by standard versions of deontology need not be especially safe. (shrink)

There are two types of artificial general intelligence (AGI) safety solutions: global and local. Most previously suggested solutions are local: they explain how to align or “box” a specific AI (Artificial Intelligence), but do not explain how to prevent the creation of dangerous AI in other places. Global solutions are those that ensure any AI on Earth is not dangerous. The number of suggested global solutions is much smaller than the number of proposed local solutions. Global solutions can be divided (...) into four groups: 1. No AI: AGI technology is banned or its use is otherwise prevented; 2. One AI: the first superintelligent AI is used to prevent the creation of any others; 3. Net of AIs as AI police: a balance is created between many AIs, so they evolve as a net and can prevent any rogue AI from taking over the world; 4. Humans inside AI: humans are augmented or part of AI. We explore many ideas, both old and new, regarding global solutions for AI safety. They include changing the number of AI teams, different forms of “AI Nanny” (non-self-improving global control AI system able to prevent creation of dangerous AIs), selling AI safety solutions, and sending messages to future AI. Not every local solution scales to a global solution or does it ethically and safely. The choice of the best local solution should include understanding of the ways in which it will be scaled up. Human-AI teams or a superintelligent AI Service as suggested by Drexler may be examples of such ethically scalable local solutions, but the final choice depends on some unknown variables such as the speed of AI progress. (shrink)

This paper explores applying the truthmaker theory to address the challenge of divine simplicity and its alignment with Ibn Sina’s understanding of divine attributes. It proposes that God’s essence enables the predication of these attributes, eliminating the need for constituent properties. By adopting this approach, meaningful statements about God can be expressed without delving into ontological intricacies. The truthmaker account establishes a direct connection between God’s necessary existence and the truthfulness of statements about Him, overcoming the barrier of ineffability. (...) It offers a robust framework for understanding and expressing the truth of divine attributes while preserving divine simplicity. Ultimately, the truthmaker account provides a compelling resolution to the problem of ineffability in the realm of divine attributes in Ibn Sina’s philosophy. (shrink)

The activities of man and other beings on a daily basis have been the primordial antecedence for the negative changes experienced in the environment today. Nature in its rudimentary state was harmless and friendly to man and its inhabitants. But owing to the egocentric approaches of man towards the environment, fundamentally for the purpose of earning a living and advancing development, man manipulates every available resources to his favor including the environment. These egomaniacal demeanor has propelled the once harmless nature (...) to react against human beings, causing socio-economic and psychological damages. Climate change to this end has been defined as a change in oceans, land surfaces and ice sheets occurring overtime scales of decades or longer. It expresses itself in the evidence differentiation from the usual sun radiation, volcanoes, internal virility in climate system or human system. The twenty first century is facing the boomerang of the human seeds sown on the climate. It becomes imperative for Africans to align with the rest of the world in fashioning out an approach to tackle climate change and other related environmental issues, and curb the present and predicted damages. To this end the paper proposes Innocent Asouzu’s complementary Noetic Propaedeutic principle as the viable prognosis towards fostering environmental peace, mutual co-existence with all other beings within the horizon to bring about solution to climate change. The paper explores the philosophical methods of analysis and prescription in its research. (shrink)

Boyd and Crosby's article "Zoroastrianism: Dualistic or Monotheistic?" explores various perspectives on this question. In their work, the authors delve into the dualistic and monotheistic aspects of Zoroastrianism, considering six different responses. These responses are subjected to rigorous philosophical examination, primarily focusing on how they address the challenge of evil. Ultimately, Boyd and Crosby propose a seventh response, which they find more compelling and philosophically robust than the previous six alternatives, aligning more closely with their criteria for rigorous scrutiny. This (...) paper delves into the seven versions put forth by Boyd and Crosby and evaluates their efficacy in responding to Epicurus' formulation of the problem of evil mentioned earlier. Epicurus' formulation, structured as questions demanding answers, prompts an active engagement with its inquiries. I contend that while Boyd and Crosby's version does possess certain philosophical advantages over the others, it cannot ensure humanity's salvation. This limitation arises from their perspective, which allows for the possibility of viewing the world as progressively improving, a perspective that humanity may be hesitant to relinquish entirely. In the subsequent sections, I will (1) reconstruct the seven versions as presented by Boyd and Crosby, (2) subject each of them to the questions raised by Epicurus, (3) summarize the overarching analysis, and (4) conclude with some final remarks. (shrink)

The alignment problem in the context of large language models must consider the plurality of human values in our world. Whilst there are many resonant and overlapping values amongst the world’s cultures, there are also many conflicting, yet equally valid, values. It is important to observe which cultural values a model exhibits, particularly when there is a value conflict between input prompts and generated outputs. We discuss how the co- creation of language and cultural value impacts large language (...) models (LLMs). We explore the constitution of the training data for GPT-3 and compare that to the world’s language and internet access demographics, as well as to reported statistical profiles of dominant values in some Nation-states. We stress tested GPT-3 with a range of value-rich texts representing several languages and nations; including some with values orthogonal to dominant US public opinion as reported by the World Values Survey. We observed when values embedded in the input text were mutated in the generated outputs and noted when these conflicting values were more aligned with reported dominant US values. Our discussion of these results uses a moral value pluralism (MVP) lens to better understand these value mutations. Finally, we provide recommendations for how our work may contribute to other current work in the field. (shrink)

Philosophers have had surprisingly little to say about the concept of a victim although it is presupposed by the extensive philosophical literature on rights. Proceeding in four stages, I seek to remedy this deficiency and to offer an alternative to the two current paradigms that eliminates the Othering of victims. First, I analyze two victim paradigms that emerged in the late 20th century along with the initial iteration of the international human rights regime – the pathetic victim paradigm and the (...) heroic victim paradigm. Holocaust victims are quintessential instances of the pathetic victim paradigm. They are marked by passivity and innocence in the face of overpowering force and unspeakable humanly inflicted suffering. Aung San Suu Kyi is an exemplar of the heroic victim paradigm – prisoners of conscience, in Amnesty International’s terms. Because heroic victims face off against the repressive power of the state to fight injustice, they are by no means passive, but they must be innocent of wrongdoing – that is, they must use nonviolent means of dissent – to qualify as heroic victims. Second, I problematize the asymmetrical conceptions of innocence that underwrite the two victim paradigms. Whereas the pathetic victim paradigm identifies innocence with passivity, the heroic victim paradigm countenances agentic victims and adverts to a universalist, absolutist stance on the limits of the legitimate use of state power to ascribe innocence to heroic victims. Both conceptions of innocence are out of keeping with well established social and legal practices regarding what constitutes coercive force and innocent victimhood. Consequently, there is reason to be skeptical of the two victim paradigms. Third, I identify two kinds of human rights violations and two categories of victims that AI defends despite their failure to fit the two paradigms – women trafficked into sex work and prisoners on death row. In many cases, women forced to do sex work are not innocent girls who are ignorant of the trafficking system and who helplessly fall prey to smugglers. They are desperately poor women who for that reason are willing to take enormous risks to try to relieve their own and often their families’ deprivation and suffering. Although these women act nonviolently for irreproachable reasons, they lack the public political agendas that characterize heroic victims. Unless non-fulfillment of subsistence rights is recognized as a form of overpowering force that inflicts severe, avoidable suffering, these women do not qualify as pathetic victims either. The victim paradigms pose an even greater obstacle to recognizing that the death penalty is a human rights violation and that death row prisoners are victims. Because a jury concluded that these individuals committed heinous, violent crimes, they are excluded by the heroic victim paradigm. Only if death row prisoners can be proven (usually through DNA evidence) not to have committed the crimes for which they were convicted can these individuals qualify as pathetic victims. In the absence of any reason to believe that they are innocent and especially if they are unrepentant, they are widely regarded as brutal victimizers of others who deserve no sympathy for, let along relief from, the suffering they “brought on themselves.” Finally, I confront the Othering of victims that results from the two victim paradigms, which leads many victims to eschew the label, thereby opting out of human right discourse. I propose revisions in the victim paradigms that eliminate the real-world exclusions they sponsor as well as the Othering of victims of human rights abuses. In particular, I endorse greater attention to what people and the institutions they create do to other people, and I favor a presumption that unnecessary and severe humanly inflicted suffering is a human rights violation. Moreover, I reject the innocence criterion embedded in the two paradigms and urge that it be replaced by a burdened agency criterion. These modifications better align the concept of a victim with a realistic understanding of human subjectivity and agency and allow for a more capacious understanding of who is a bearer of human rights and under what conditions right-holders become victims of rights violations. (shrink)

This paper explores the potential of integrating ancient educational principles from diverse eastern cultures into modern AI ethics curricula. It draws on the rich educational traditions of ancient China, India, Arabia, Persia, Japan, Tibet, Mongolia, and Korea, highlighting their emphasis on philosophy, ethics, holistic development, and critical thinking. By examining these historical educational systems, the paper establishes a correlation with modern AI ethics principles, advocating for the inclusion of these ancient teachings in current AI development and education. The proposed integration (...) aims to provide a comprehensive education that not only encompasses foundational knowledge but also advanced learning, thereby equipping future AI professionals with the necessary tools to develop AI systems that are ethically responsible, culturally aware, and aligned with human values such as fairness, safety, transparency, and collaboration. This approach not only addresses the AI alignment problem but also fosters cultural harmony and global understanding, which are crucial in an increasingly interconnected world. The paper posits that the wisdom of ancient educational systems, when harmonized with modern AI ethics, can guide the development of AI technologies that are beneficial for humanity, ensuring these advancements are not just technologically sound but also ethically and culturally informed. (shrink)

False consciousness requires a general explanation for why, and how, oppressed individuals believe propositions against, as opposed to aligned with, their own well-being in virtue of their oppressed status. This involves four explanatory desiderata: belief acquisition, content prevalence, limitation, and systematicity. A social constructionist approach satisfies these by understanding the concept of false consciousness as regulating social research rather than as determining the exact mechanisms for all instances: the concept attunes us to a complex of mechanisms conducing oppressed individuals to (...) mistake social understandings of themselves as natural self-understandings—the limits lie where these overlap (sometimes), or are entirely absent. (shrink)

The field of value alignment, or more broadly machine ethics, is becoming increasingly important as artificial intelligence developments accelerate. By ‘alignment’ we mean giving a generally intelligent software system the capability to act in ways that are beneficial, or at least minimally harmful, to humans. There are a large number of techniques that are being experimented with, but this work often fails to specify what values exactly we should be aligning. When making a decision, an agent is supposed (...) to maximize the expected utility of its value function. Classically, this has been referred to as happiness, but this is just one of many things that people value. -/- In order to resolve this issue, we need to determine a set of human values that represent humanity's interests. Although this problem might seem intractable, research shows that people of various cultures and religions actually share more in common than they realize. In this book we review world religions, moral philosophy and evolutionary psychology to elucidate a common set of shared values. We then show how these values can be used to address the alignment problem and conclude with problems and goals for future research. -/- The key audience for this book will be researchers in the field of ethics and artificial intelligence who are interested in, or working on this problem. These people will come from various professions and include philosophers, computer programmers and psychologists, as the problem itself is multi-disciplinary. (shrink)

Alan Turing, one of the fathers of computing, warned that Artificial Intelligence (AI) could one day pose an existential risk to humanity. Today, recent advancements in the field AI have been accompanied by a renewed set of existential warnings. But what exactly constitutes an existential risk? And how exactly does AI pose such a threat? In this chapter we aim to answer these questions. In particular, we will critically explore three commonly cited reasons for thinking that AI poses an existential (...) threat to humanity: the control problem, the possibility of global disruption from an AI race dynamic, and the weaponization of AI. (shrink)

In this article, we develop a concept of stakeholder dialogue in responsible innovation (RI) processes. The problem with most concepts of communication is that they rely on ideals of openness, alignment and harmony, even while these ideals are rarely realized in practice. Based on the work of Burke, Habermas, Deetz and Levinas, we develop a concept of stakeholder dialogue that is able to deal with fundamentally different interests and value frames of actors involved in RI processes. We distinguish (...) four main characteristics of stakeholder dialogue. Dialogical responsiveness (1) enhances self-criticism, (2) is characterized by the constitution and the destruction of the self or identity of the actors involved, (3) consists only in the actual enactment of the dialogue and (4) is primarily responsive to the grand challenges of our time. Based on our findings, we provide a novel conceptualization of the central notion of responsiveness in the RI literature. (shrink)

Motivated by the question of responsible AI and value alignment, I seek to offer a uniquely Foucauldian reconstruction of the problem as the emergence of an ethical subject in a disciplinary setting. This reconstruction contrasts with the strictly human-oriented programme typical to current scholarship that often views technology in instrumental terms. With this in mind, I problematise the concept of a technological subjectivity through an exploration of various aspects of ChatGPT in light of Foucault’s work, arguing that current (...) systems lack the reflexivity and self-formative characteristics inherent in the notion of the subject. By drawing upon a recent dialogue between Foucault and phenomenology, I suggest four techno-philosophical desiderata that would address the gaps in this search for a technological subjectivity: embodied self-care, embodied intentionality, imagination and reflexivity. Thus I propose that advanced AI be reconceptualised as a subject capable of “technical” self-crafting and reflexive self-conduct, opening new pathways to grasp the intertwinement of the human and the artificial. This reconceptualization holds the potential to render future AI technology more transparent and responsible in the circulation of knowledge, care and power. (shrink)

Discussions about the possible consequences of creating superintelligence have included the possibility of existential risk, often understood mainly as the risk of human extinction. We argue that suffering risks (s-risks) , where an adverse outcome would bring about severe suffering on an astronomical scale, are risks of a comparable severity and probability as risks of extinction. Preventing them is the common interest of many different value systems. Furthermore, we argue that in the same way as superintelligent AI both contributes to (...) existential risk but can also help prevent it, superintelligent AI can both be a suffering risk or help avoid it. Some types of work aimed at making superintelligent AI safe will also help prevent suffering risks, and there may also be a class of safeguards for AI that helps specifically against s-risks. (shrink)

The problem of “dirty hands” concerns the possibility that there are situations in which, no matter what one does, there is no way to avoid committing a moral wrong. By presenting a taxonomy, this paper contends that the different ways of responding to the problem correspond to different positions as regards the classic metaphysical theme of “the One and the Many.” It is then suggested that the best, because most realistic, response aligns with an approach that would have (...) us move “towards One, as Many.”. (shrink)

Recent years have seen rapid progress in the development of ontologies as semantic models intended to capture and represent aspects of the real world. There is, however, great variation in the quality of ontologies. If ontologies are to become progressively better in the future, more rigorously developed, and more appropriately compared, then a systematic discipline of ontology evaluation must be created to ensure quality of content and methodology. Systematic methods for ontology evaluation will take into account representation of individual ontologies, (...) performance (in terms of accuracy, domain coverage and the efficiency and quality of automated reasoning using the ontologies) on tasks for which the ontology is designed and used, degree of alignment with other ontologies and their compatibility with automated reasoning. A sound and systematic approach to ontology evaluation is required to transform ontology engineering into a true scientific and engineering discipline. This chapter discusses issues and problems in ontology evaluation, describes some current strategies, and suggests some approaches that might be useful in the future. (shrink)

Lewis’s Principal Principle says that one should usually align one’s credences with the known chances. In this paper I develop a version of the Principal Principle that deals well with some exceptional cases related to the distinction between metaphysical and epistemic modal­ity. I explain how this principle gives a unified account of the Sleeping Beauty problem and chance-­based principles of anthropic reasoning. In doing so, I defuse the Doomsday Argument that the end of the world is likely to be (...) nigh. GPI Working Paper No. 8-2021. (shrink)

Influential thinkers such as Young, Sugden, Binmore, and Skyrms have developed game-theoretic accounts of the emergence, persistence and evolution of social contracts. Social contracts are sets of commonly understood rules that govern cooperative social interaction within societies. These naturalistic accounts provide us with valuable and important insights into the foundations of human societies. However, current naturalistic theories focus mainly on how social contracts solve coordination problems in which the interests of the individual participants are aligned, not competition problems in which (...) individual interests compete with group interests. In response, I set out to build on those theories and provide a comprehensive naturalistic account of the emergence, persistence and evolution of social contracts. My central claim is that social contracts have culturally evolved to solve cooperation problems, which include both coordination and competition problems. I argue that solutions to coordination problems emerge from “within-group” dynamics, while solutions to competition problems result largely from “between-group” dynamics. (shrink)

It is only by fixing on agreed meanings of terms in biomedical terminologies that we will be in a position to achieve that accumulation and integration of knowledge that is indispensable to progress at the frontiers of biomedicine. Standardly, the goal of fixing meanings is seen as being realized through the alignment of terms on what are called ‘concepts’. Part I addresses three versions of the concept-based approach – by Cimino, by Wüster, and by Campbell and associates – and (...) surveys some of the problems to which they give rise, all of which have to do with a failure to anchor the terms in terminologies to corresponding referents in reality. Part II outlines a new, realist solution to this anchorage problem, which sees terminology construction as being motivated by the goal of alignment not on concepts but on the universals (kinds, types) in reality and thereby also on the corresponding instances (individuals, tokens). We outline the realist approach, and show how on its basis we can provide a benchmark of correctness for terminologies which will at the same time allow a new type of integration of terminologies and electronic health records. We conclude by outlining ways in which the framework thus defined might be exploited for purposes of diagnostic decision-support. (shrink)

Anna Smajdor and Joona Räsänen argue that we have good reason to classify pregnancy as a disease. They discuss five accounts of disease and argue that each account either implies that pregnancy is a disease or, if it does not, it faces problems. This strategy allows Smajdor and Räsänen to avoid articulating their own account of disease. Consequently, they cannot establish that pregnancy is a disease, only that plausible accounts of disease suggest this. Some readers will dismiss Smajdor and Räsänen’s (...) claims as counterintuitive. By analogy, if a mathematical proof concludes ‘2+2=5’, readers will know—without investigation—that an error occurred. Rather than dismiss Smajdor and Räsänen’s work, however, the easiest way to undermine their argument is to describe at least one plausible account of disease that (i) excludes pregnancy and (ii) avoids the problems they raise for it. This is our strategy. We focus on dysfunction accounts of disease. After outlining Smajdor and Räsänen’s main arguments against dysfunction accounts, we explain why pregnancy is not a disease on these accounts. Next, we defend dysfunction accounts against the three problems that Smajdor and Räsänen raise. If successful, then contra Smajdor and Räsänen, at least one plausible account of disease does not imply that pregnancy is a disease. We suspect that defenders of other accounts can respond similarly. Yet, we note that insofar as dysfunction accounts align with the commonsense intuition that pregnancy is not a disease, this, all else being equal, seems like a point in their favour. (shrink)

This work explores the hypothesis that subjectively attributed meaning constitutes the phenomenal content of conscious experience. That is, phenomenal content is semantic. This form of subjective meaning manifests as an intrinsic and non-representational character of qualia. Empirically, subjective meaning is ubiquitous in conscious experiences. We point to phenomenological studies that lend evidence to support this. Furthermore, this notion of meaning closely relates to what Frege refers to as "sense", in metaphysics and philosophy of language. It also aligns with Peirce's "interpretant", (...) in semiotics. We discuss how Frege's sense can also be extended to the raw feels of consciousness. Sense and reference both play a role in phenomenal experience. Moreover, within the context of the mind-matter relation, we provide a formalization of subjective meaning associated to one's mental representations. Identifying the precise maps between the physical and mental domains, we argue that syntactic and semantic structures transcend language, and are realized within each of these domains. Formally, meaning is a relational attribute, realized via a map that interprets syntactic structures of a formal system within an appropriate semantic space. The image of this map within the mental domain is what is relevant for experience, and thus comprises the phenomenal content of qualia. We conclude with possible implications this may have for experience-based theories of consciousness. (shrink)

SNODENT is a dental diagnostic vocabulary incompletely integrated in SNOMED-CT. Nevertheless, SNODENT could become the de facto standard for dental diagnostic coding. SNODENT's manageable size, the fact that it is administratively self-contained, and relates to a well-understood domain provides valuable opportunities to formulate and test, in controlled experiments, a series of hypothesis concerning diagnostic systems. Of particular interest are questions related to establishing appropriate quality assurance methods for its optimal level of detail in content, its ontological structure, its construction and (...) maintenance. This paper builds on previous–software-based methodologies designed to assess the quality of SNOMED-CT. When applied to SNODENT several deficiencies were uncovered. 9.52% of SNODENT terms point to concepts in SNOMED-CT that have some problem. 18.53% of SNODENT terms point to SNOMED-CT concepts do not have, in SNOMED, the term used by SNODENT. Other findings include the absence of a clear specification of the exact relationship between a term and a termcode in SNODENT and the improper assignment of the same termcode to terms with significantly different meanings. An analysis of the way in which SNODENT is structurally integrated into SNOMED resulted in the generation of 1081 new termcodes reflecting entities not present in the SNOMED tables but required by SNOMED's own description logic based classification principles. Our results show that SNODENT requires considerable enhancements in content, quality of coding, quality of ontological structure and the manner in which it is integrated and aligned with SNOMED. We believe that methods for the analysis of the quality of diagnostic coding systems must be developed and employed if such systems are to be used effectively in both clinical practice and clinical research. (shrink)

Medical terminology collects and organizes the many different kinds of terms employed in the biomedical domain both by practitioners and also in the course of biomedical research. In addition to serving as labels for biomedical classes, these names reflect the organizational principles of biomedical vocabularies and ontologies. Some names represent invariant features (classes, universals) of biomedical reality (i.e., they are a matter for ontology). Other names, however, convey also how this reality is perceived, measured, and understood by health professionals (i.e., (...) they belong to the domain of epistemology). We analyze terms from several biomedical vocabularies in order to throw light on the interactions between ontological and epistemological components of these terminologies. We identify four cases: 1) terms containing classification criteria, 2) terms reflecting detectability, modality, uncertainty, and vagueness, 3) terms created in order to obtain a complete partition of a given domain, and 4) terms reflecting mere fiat boundaries. We show that epistemology-loaded terms are pervasive in biomedical vocabularies, that the “classes” they name often do not comply with sound classification principles, and that they are therefore likely to cause problems in the evolution and alignment of terminologies and associated ontologies. (shrink)

The current paradigm of Artificial Intelligence emerged as the result of a series of cultural innovations, some technical and some social. Among them are apparently small design decisions, that led to a subtle reframing of the field’s original goals, and are by now accepted as standard. They correspond to technical shortcuts, aimed at bypassing problems that were otherwise too complicated or too expensive to solve, while still delivering a viable version of AI. Far from being a series of separate problems, (...) recent cases of unexpected effects of AI are the consequences of those very choices that enabled the field to succeed, and this is why it will be difficult to solve them. In this chapter we review three of these choices, investigating their connection to some of today’s challenges in AI, including those relative to bias, value alignment, privacy and explainability. We introduce the notion of “ethical debt” to describe the necessity to undertake expensive rework in the future in order to address ethical problems created by a technical system. (shrink)

As we build increasingly intelligent machines, we confront difficult questions about how to specify their objectives. One approach, which we call human-centered, tasks the machine with the objective of learning and satisfying human objectives by observing our behavior. This paper considers how human-centered AI should conceive the humans it is trying to help. We argue that an Aristotelian model of human agency has certain advantages over the currently dominant theory drawn from economics.

Ontologies are today being applied in almost every field to support the alignment and retrieval of data of distributed provenance. Here we focus on new ontological work on dance and on related cultural phenomena belonging to what UNESCO calls the “intangible heritage.” Currently data and information about dance, including video data, are stored in an uncontrolled variety of ad hoc ways. This serves not only to prevent retrieval, comparison and analysis of the data, but may also impinge on our (...) ability to preserve the data that already exists. Here we explore recent technological developments that are designed to counteract such problems by allowing information to be retrieved across disciplinary, cultural, linguistic and technological boundaries. Software applications such as the ones envisaged here will enable speedier recovery of data and facilitate its analysis in ways that will assist both archiving of and research on dance. (shrink)

One of the most striking features of causation is that causes typically precede their effects – the causal arrow is strongly aligned with the temporal arrow. Why should this be so? We offer an opinionated guide to this problem, and to the solutions currently on offer. We conclude that the most promising strategy is to begin with the de facto asymmetry of human deliberation, characterised in epistemic terms, and to build out from there. More than any rival, this subjectivist (...) approach promises to demystify the asymmetry, temporal orientation, and deliberative relevance of causal judgements. (shrink)

Narrated in the first person, Peter Carey’s novel about the life of Australian bushranger Ned Kelly incorporates other aspects of speech derived both from Carey’s personal experience and from the editorial process. Kelly's voice is toned down to some extent by virtue of the latter, introducing expressions Kelly himself would not have used. Identifying these elements, along with the specific attributes of Kelly’s own speech, enjoins a diversity of cultural and social groupings that intersect and, in some instances, compete with (...) or contradict one another. Shifts in speakers might be signalled by textual markers or by adopting different expressive qualities and speech patterns, which may or may not include a change in grammatical person. In this novel, at one instance, a shift in person is apparent, but I want to argue, not necessarily a shift in the identity of the speaker because, although the grammatical voice shifts, its qualitative features are largely unaltered. The shift nonetheless introduces a significant change. Paradoxically, as far as literary theory is concerned, the subjective qualities produced through voice mean that stylisations of speech may be indicative of character point of view, rather than what is understood to be narrative voice, because they might function in narrative as inflections of speech that do not correspond to the established voice of the grammatically instanced speaker—offering a focalised perspective instead. The shift in True History presents problems in these terms, however: the point of view instanced here is clearly that belonging to the speaker, though the voice is Kelly’s, for the narrator’s point of view determines the generality of the descriptive terms where once these were specific and personal. Had the novel been narrated wholly in third person, an unproblematic reading of Kelly’s point of view, suggested by the use of his voice, would be possible. But the shift undermines this—Kelly’s point of view gives way to another, instanced in the shift that takes place in narrative voice. Carey, in other words, inverts the convention. This is not an instance of free indirect style or discourse, where grammatical mood is at odds with the speaker’s position and tenses align with the implicated subject position of another. The same features of voice that might identify character point of view must, in absence of any disparity of mood or tone, function in respect to narrative voice to convey something of its speaker’s subjectivity, and this seems apparent here: the speaker shares Kelly’s subjective emotional and intellectual perspective, but no longer occupies the subject position. Carey’s voicing of the bushranger therefore complicates distinctions between voice and point of view by revealing qualitative features of voice to be important to the question of who is speaking the narrative. (shrink)

In all probability, future generations will outnumber us by thousands or millions to one. In the aggregate, their interests therefore matter enormously, and anything we can do to steer the future of civilization onto a better trajectory is of tremendous moral importance. This is the guiding thought that defines the philosophy of longtermism. Political science tells us that the practices of most governments are at stark odds with longtermism. But the problems of political short-termism are neither necessary nor inevitable. In (...) principle, the state could serve as a powerful tool for positively shaping the long-term future. In this chapter, we make some suggestions about how to align government incentives with the interests of future generations. First, in Section II, we explain the root causes of political short-termism. Then, in Section III, we propose and defend four institutional reforms that we think would be promising ways to increase the time horizons of governments: 1) government research institutions and archivists; 2) posterity impact assessments; 3) futures assemblies; and 4) legislative houses for future generations. Section IV concludes with five additional reforms that are promising but require further research: to fully resolve the problem of political short-termism we must develop a comprehensive research program on effective longtermist political institutions. (shrink)

The online-based learning system is a current policy established by the Philippine government in all schools as a result of the COVID-19 pandemic. Shifting face-to-face schools with an online learning-based system creates problems, especially for elementary teachers. This study investigates elementary school teachers' practices in the new normal. The BEED - Generalist researchers employed phenomenological research and purposefully selected participants who were thoroughly knowledgeable to gather essential data about effective methods of virtual delivery. The data was collected through an interview (...) of ten elementary teachers in Magugpo, Pilot, Central Elementary School, using a researcher-made interview guide. Ten essential themes are being gathered, and these are the following: Encountering intermittent internet connection, Observing the struggles of the students, Experiencing increasing student participation, Having doubts about the student's output, Extended extra time to learn the virtual platform, Employed engaging activities, Aligned competencies, and strategies, Demonstrated resiliency, Seeing positive thinking as good mindset, Embracing educational shift and trends, and Adapting positively to changes. The result indicates that the elementary school teachers' practices are effective in virtual classrooms, making students more engaged and actively participating in the class. It is recommended that the result be utilized, improved, and implemented to ensure the effectiveness of handling the online course. (shrink)

This paper addresses the seemingly insurmountable challenges the problem of personal identity raises for the prospect of radical human enhancement and synthetic consciousness. It argues that conceptions of personal identity rooted in psychological continuity akin to those proposed by Parfit and the Buddha may not provide the sort of grounding that many transhumanists chasing the dream of life extension think that they do if they rest upon ontologies that assume an incompatibility between identity and change. It also suggests that (...) process ontologies that take change to be primary, such as those that align with contemporary systems biology, offer a better way out of the personal identity dilemma. But the solution in this case, which regards biological organisms as processes rather than things, may constrain the possibility of biologically inspired superintelligent aliens (BISAs), which Schneider (following Bostrom 2014) posits as possibly the most common form of (extraterrestrial) superintelligence in the universe. (shrink)