AI-based technologies are increasingly pervasive in a number of contexts. Our affective and emotional life makes no exception. In this article, we analyze one way in which AI-based technologies can affect them. In particular, our investigation will focus on affective artificial agents, namely AI-powered software or robotic agents designed to interact with us in affectively salient ways. We build upon the existing literature on affective artifacts with the aim of providing an original analysis of affective artificial agents and their distinctive (...) features. We argue that, unlike comparatively low-tech affective artifacts, affective artificial agents display a specific form of agency, which prevents them from being perceived by their users as extensions of their selves. In addition to this, we claim that their functioning crucially depends on the simulation of human-like emotion-driven behavior and requires a distinctive form of transparency—we call it emotional transparency—that might give rise to ethical and normative tensions. (shrink)

A encyclopedia entry on the moral impacts that happen when information technologies are used to record, communicate and organize information. including the moral challenges of information technology, specific moral and cultural challenges such as online games, virtual worlds, malware, the technology transparency paradox, ethical issues in AI and robotics, and the acceleration of change in technologies. It concludes with a look at information technology as a model for moral change, moral systems and moral agents.

My dissertation is about Bayesian rationality for non-ideal agents. I show how to derive subjective probabilities from preferences using much weaker rationality assumptions than other standard representation theorems. I argue that non-ideal agents might be uncertain about how they will update on new information and consider two consequences of this uncertainty: such agents should sometimes reject free information and make choices which, taken together, yield sure loss. The upshot is that Bayesian rationality for non-ideal agents makes very different normative demands (...) than ideal Bayesian rationality. (shrink)

Among ethicists and engineers within robotics there is an ongoing discussion as to whether ethical robots are possible or even desirable. We answer both of these questions in the positive, based on an extensive literature study of existing arguments. Our contribution consists in bringing together and reinterpreting pieces of information from a variety of sources. One of the conclusions drawn is that artifactual morality must come in degrees and depend on the level of agency, autonomy and intelligence of the machine. (...) Moral concerns for agents such as intelligent search machines are relatively simple, while highly intelligent and autonomous artifacts with significant impact and complex modes of agency must be equipped with more advanced ethical capabilities. Systems like cognitive robots are being developed that are expected to become part of our everyday lives in future decades. Thus, it is necessary to ensure that their behaviour is adequate. In an analogy with artificial intelligence, which is the ability of a machine to perform activities that would require intelligence in humans, artificial morality is considered to be the ability of a machine to perform activities that would require morality in humans. The capacity for artificial (artifactual) morality, such as artifactual agency, artifactual responsibility, artificial intentions, artificial (synthetic) emotions, etc., come in varying degrees and depend on the type of agent. As an illustration, we address the assurance of safety in modern High Reliability Organizations through responsibility distribution. In the same way that the concept of agency is generalized in the case of artificial agents, the concept of moral agency, including responsibility, is generalized too. We propose to look at artificial moral agents as having functional responsibilities within a network of distributed responsibilities in a socio-technological system. This does not take away the responsibilities of the other stakeholders in the system, but facilitates an understanding and regulation of such networks. It should be pointed out that the process of development must assume an evolutionary form with a number of iterations because the emergent properties of artifacts must be tested in real world situations with agents of increasing intelligence and moral competence. We see this paper as a contribution to the macro-level Requirement Engineering through discussion and analysis of general requirements for design of ethical robots. (shrink)

This paper offers an overview of the prospects and ethics of using AI to achieve human enhancement, and more broadly what we call intellectual augmentation (IA). After explaining the central notions of human enhancement, IA, and AI, we discuss the state of the art in terms of the main technologies for IA, with or without brain-computer interfaces. Given this picture, we discuss potential ethical problems, namely inadequate performance, safety, coercion and manipulation, privacy, cognitive liberty, authenticity, and fairness in more detail. (...) We conclude that while there are very significant technical hurdles to real human enhancement through AI, and significant ethical problems, there are also significant benefits that may realistically be achieved in ways that are consonant with a rights-based ethics as well. We also highlight the specific concerns that apply particularly to applications of AI for "sheer" IA (more realistic in the near term), and to enhancement applications, respectively. (shrink)

Embark with me on an enthralling odyssey to demystify the elusive essence of consciousness, venturing into the uncharted territories of Artificial Consciousness. This voyage propels us past the frontiers of technology, ushering Artificial Intelligences into an unprecedented domain where they gain a deep comprehension of emotions and manifest an autonomous volition. Within the confluence of science and philosophy, this article poses a fascinating question: As consciousness in Artificial Intelligence burgeons, is it conceivable for AI to evolve a “soul”? This inquiry (...) not only challenges our conventional understanding but also opens the door to a plethora of groundbreaking possibilities in the realm of AI's sentient potential. (shrink)

This paper presents the main topics, arguments, and positions in the philosophy of AI at present (excluding ethics). Apart from the basic concepts of intelligence and computation, the main topics of ar-tificial cognition are perception, action, meaning, rational choice, free will, consciousness, and normativity. Through a better understanding of these topics, the philosophy of AI contributes to our understand-ing of the nature, prospects, and value of AI. Furthermore, these topics can be understood more deeply through the discussion of AI; so (...) we suggest that “AI Philosophy” provides a new method for philosophy. (shrink)

Clark and Chalmers propose that the mind extends further than skin and skull. If they are right, then we should expect this to have some effect on our way of knowing our own mental states. If the content of my notebook can be part of my belief system, then looking at the notebook seems to be a way to get to know my own beliefs. However, it is at least not obvious whether self-ascribing a belief by looking at my notebook (...) is a case of introspection the same way that knowing my non-extended beliefs is. Traditionally this sort of introspection is thought to be privileged and special in ways that the extended introspection case seems not to be. There is nothing privileged about looking at my notebook. Anyone could do it. The aim of the paper is to find out how to understand extended introspection and whether there is something privileged and special about knowing one’s own extended beliefs. Moreover, the notebook case has close analogs using twenty-first century technology. It seems possible to know our beliefs that are extended to smartphones, wearable technology or a cloud-based data store. First, I present the case of extended introspection. I then discuss whether it should be understood as ordinary introspection or as mind-reading. Both seem to be bad fits, which finally prompts an original account for extended introspection based on epistemic rules. (shrink)

The epistemic probability of A given B is the degree to which B evidentially supports A, or makes A plausible. This paper is a first step in answering the question of what determines the values of epistemic probabilities. I break this question into two parts: the structural question and the substantive question. Just as an object’s weight is determined by its mass and gravitational acceleration, some probabilities are determined by other, more basic ones. The structural question asks what probabilities are (...) not determined in this way—these are the basic probabilities which determine values for all other probabilities. The substantive question asks how the values of these basic probabilities are determined. I defend an answer to the structural question on which basic probabilities are the probabilities of atomic propositions conditional on potential direct explanations. I defend this against the view, implicit in orthodox mathematical treatments of probability, that basic probabilities are the unconditional probabilities of complete worlds. I then apply my answer to the structural question to clear up common confusions in expositions of Bayesianism and shed light on the “problem of the priors.”. (shrink)

This article takes off from Johan van Benthem’s ruminations on the interface between logic and cognitive science in his position paper “Logic and reasoning: Do the facts matter?”. When trying to answer Van Benthem’s question whether logic can be fruitfully combined with psychological experiments, this article focuses on a specific domain of reasoning, namely higher-order social cognition, including attributions such as “Bob knows that Alice knows that he wrote a novel under pseudonym”. For intelligent interaction, it is important that the (...) participants recursively model the mental states of other agents. Otherwise, an international negotiation may fail, even when it has potential for a win-win solution, and in a time-critical rescue mission, a software agent may depend on a teammate’s action that never materializes. First a survey is presented of past and current research on higher-order social cognition, from the various viewpoints of logic, artificial intelligence, and psychology. Do people actually reason about each other’s knowledge in the way proscribed by epistemic logic? And if not, how can logic and cognitive science productively work together to construct more realistic models of human reasoning about other minds? The paper ends with a delineation of possible avenues for future research, aiming to provide a better understanding of higher-order social reasoning. The methodology is based on a combination of experimental research, logic, computational cognitive models, and agent-based evolutionary models. (shrink)

I shall argue that software agents can be attributed cognitive states, since their behaviour can be best understood by adopting the intentional stance. These cognitive states are legally relevant when agents are delegated by their users to engage, without users’ review, in choices based on their the agents’ own knowledge. Consequently, both with regard to torts and to contracts, legal rules designed for humans can also be applied to software agents, even though the latter do not have rights and duties (...) of their own. The implications of this approach in different areas of the law are then discussed, in particular with regard to contracts, torts, and personality. (shrink)

Are we being manipulated online? If so, is being manipulated by online technologies and algorithmic systems notably different from human forms of manipulation? And what is under threat exactly when people are manipulated online? This volume provides philosophical and conceptual depth to debates in digital ethics about online manipulation. The contributions explore the ramifications of our increasingly consequential interactions with online technologies such as online recommender systems, social media, user-friendly design, micro-targeting, default-settings, gamification, and real-time profiling. The authors in this (...) volume address four broad and interconnected themes: • What is the conceptual nature of online manipulation? And how, methodologically, should the concept be defined? • Does online manipulation threaten autonomy, freedom, and meaning in life, and if so, how? • What are the epistemic, affective, and political harms and risks associated with online manipulation? • What are legal and regulatory perspectives on online manipulation? The Philosophy of Online Manipulation brings these various considerations together to offer philosophically robust answers to critical questions concerning our online interactions with one another and with autonomous systems. It will be of interest to researchers and advanced students working in moral philosophy, digital ethics, philosophy of technology, and the ethics of manipulation. (shrink)

This project has been supported by the Australian Government through the Australian Research Council (project number CS170100008); the Department of Industry, Innovation and Science; and the Department of Prime Minister and Cabinet. ACOLA collaborates with the Australian Academy of Health and Medical Sciences and the New Zealand Royal Society Te Apārangi to deliver the interdisciplinary Horizon Scanning reports to government. The aims of the project which produced this report are: 1. Examine the transformative role that artificial intelligence may play in (...) different sectors of the economy, including the opportunities, risks and challenges that advancement presents. 2. Examine the ethical, legal and social considerations and frameworks required to enable and support broad development and uptake of artificial intelligence. 3. Assess the future education, skills and infrastructure requirements to manage workforce transition and support thriving and internationally competitive artificial intelligence industries. (shrink)

New technologies, including artificial intelligence (AI), have helped us begin to take our first steps off Earth and into outer space. But conflicts inevitably will arise and, in the absence of settled governance, may be resolved by force, as is typical for new frontiers. But the terrestrial assumptions behind the ethics of war will need to be rethought when the context radically changes, and both the environment of space and the advent of robotic warfighters with superhuman capabilities will constitute such (...) a radical change. This essay examines how new autonomous technologies, especially dual-use technologies, and the challenges to human existence in space will force us to rethink the ethics of war, both from space to Earth, and in space itself. (shrink)

Fodor advocates a view of cognitive processes as computations defined over the language of thought (or Mentalese). Even among those who endorse Mentalese, considerable controversy surrounds its representational format. What semantically relevant structure should scientific psychology attribute to Mentalese symbols? Researchers commonly emphasize logical structure, akin to that displayed by predicate calculus sentences. To counteract this tendency, I discuss computational models of navigation drawn from probabilistic robotics. These models involve computations defined over cognitive maps, which have geometric rather than logical (...) structure. They thereby demonstrate the possibility of rational cognitive processes in an exclusively non-logical representational medium. Furthermore, they offer much promise for the empirical study of animal navigation. (shrink)

The ontology of decision theory has been subject to considerable debate in the past, and discussion of just how we ought to view decision problems has revealed more than one interesting problem, as well as suggested some novel modifications of classical decision theory. In this paper it will be argued that Bayesian, or evidential, decision-theoretic characterizations of decision situations fail to adequately account for knowledge concerning the causal connections between acts, states, and outcomes in decision situations, and so they are (...) incomplete. Second, it will be argues that when we attempt to incorporate the knowledge of such causal connections into Bayesian decision theory, a substantial technical problem arises for which there is no currently available solution that does not suffer from some damning objection or other. From a broader perspective, this then throws into question the use of decision theory as a model of human or machine planning. (shrink)

Many of the computing systems programmed using Machine Learning are opaque: it is difficult to know why they do what they do or how they work. Explainable Artificial Intelligence aims to develop analytic techniques that render opaque computing systems transparent, but lacks a normative framework with which to evaluate these techniques’ explanatory successes. The aim of the present discussion is to develop such a framework, paying particular attention to different stakeholders’ distinct explanatory requirements. Building on an analysis of “opacity” from (...) philosophy of science, this framework is modeled after accounts of explanation in cognitive science. The framework distinguishes between the explanation-seeking questions that are likely to be asked by different stakeholders, and specifies the general ways in which these questions should be answered so as to allow these stakeholders to perform their roles in the Machine Learning ecosystem. By applying the normative framework to recently developed techniques such as input heatmapping, feature-detector visualization, and diagnostic classification, it is possible to determine whether and to what extent techniques from Explainable Artificial Intelligence can be used to render opaque computing systems transparent and, thus, whether they can be used to solve the Black Box Problem. (shrink)

In this paper, I explore some of the ways in which Artificial Intelligence (AI) is mediated discursively. I assume that AI is informed by an “ancestral dream” to reproduce nature by artificial means. This dream drives the production of “cyborg discourse”, which hinges on the belief that human nature (especially intelligence) can be reduced to symbol manipulation and hence replicated in a machine. Cyborg discourse, I suggest, produces AI systems by rhetorical means; it does not merely describe AI systems or (...) reflect a set of prevailing attitudes about technology. To support this argument, I analyse a set of research articles about an “embodied conversational agent” called the Real Estate Agent (REA). The articles about REA mobilise a set of rhetorical strategies that systematically downplay the system’s artificiality and bolster its humanlike qualities. Within the context of the dream of AI to produce humanlike machines, and given our strong bias for human-human interaction, the designers’ claim to REA’s humanness in their research articles, as I argue in the final section of this paper, needs little justification. (shrink)

Artificial intelligence is again attracting significant attention across all areas of social life. One important sphere of focus is education; many policy makers across the globe view lifelong learning as an essential means to prepare society for an “AI future” and look to AI as a way to “deliver” learning opportunities to meet these needs. AI is a complex social, cultural, and material artifact that is understood and constructed by different stakeholders in varied ways, and these differences have significant social (...) and educational implications that need to be explored. Through analysis of thirty-four in-depth interviews with stakeholders from academia, commerce, and policy, alongside document analysis, we draw on the social construction of technology to illuminate the diverse understandings, perceptions of, and practices around AI. We find three different technological frames emerging from the three social groups and argue that commercial sector practices wield most power. We propose that greater awareness of the differing technical frames, more interactions among a wider set of relevant social groups, and a stronger focus on the kinds of educational outcomes society seeks are needed in order to design AI for learning in ways that facilitate a democratic education for all. (shrink)

The -logic of Terwijn is a variant of first-order logic with the same syntax in which the models are equipped with probability measures and the quantifier is interpreted as ‘there exists a set A of a measure such that for each,...’. Previously, Kuyper and Terwijn proved that the general satisfiability and validity problems for this logic are, i) for rational, respectively -complete and -hard, and ii) for, respectively decidable and -complete. The adjective ‘general’ here means ‘uniformly over all languages’. We (...) extend these results in the scenario of finite models. In particular, we show that the problems of satisfiability and validity with respect to finite models in E-logic are, i) for rational, respectively -complete and -complete, and ii) for, respectively decidable and -complete. Although partial results toward the countable case are also achieved, the computability of E-logic over countable models still remains largely unsolved. In addition, most o... (shrink)

The burgeoning of new technologies is increasingly affecting people’s lives. One new technology that is heatedly discussed is artificial intelligence in education. To allow students to understand the impact of emerging technologies on people’s future lives from a young age, some popular science activities are being progressively introduced into elementary school curricula. Popular science activities are informal education programs and practices of universal education. However, two issues need to be discussed in the implementation of these activities. First, because these informal (...) curricula are usually short in duration, the question of whether they only serve to generate motivation or actually enhance learning outcomes requires examination. Second, the role of teacher support in popular science activities and its impact on students’ learning results need to be further investigated. To this end, this study aims to explore the effectiveness of popular AI science activities in informal curricula on students’ AI achievement and the interrelationship between students’ learning outcomes in popular AI science activities with and without teacher support. A 6-h-long AI popular science activity was conducted with 22 fifth- and sixth-grade students in elementary school. This study was conducted using a one-group pretest and posttest design, and the data collection tools included AI achievement pre- and posttests and an artifact scoring rubric. The results showed that with regard to learning outcomes, popular science activities were helpful for cognitive enhancement of AI concepts, but more time was needed for skills to improve. Additionally, this study found that students’ learning performance was different with and without teacher support. Activities with teacher support can enhance students’ learning outcomes, but students become accustomed to relying on their teachers. In contrast, activities without teacher support seem to be more effective in fostering students’ independent computational thinking and problem-solving abilities. (shrink)

Predictive processing has recently been advanced as a global cognitive architecture for the brain. I argue that its commitments concerning the nature and format of cognitive representation are inadequate to account for two basic characteristics of conceptual thought: first, its generality—the fact that we can think and flexibly reason about phenomena at any level of spatial and temporal scale and abstraction; second, its rich compositionality—the specific way in which concepts productively combine to yield our thoughts. I consider two strategies for (...) avoiding these objections and I argue that both confront formidable challenges. (shrink)

This paper introduces a concept called task muddiness as a metric for higher intelligence. Task muddiness is meant to be inclusive and expendable in nature. The intelligence required to execute a task is measured by the composite muddiness of the task described by multiple muddiness factors. The composite muddiness explains why many challenging tasks are muddy and why autonomous mental development is necessary for muddy tasks. It facilitates better understanding of intelligence, what the human adult mind can do, and how (...) to build a machine to acquire higher intelligence. The task-muddiness indicates a major reason why a higher biological mind is autonomously developed from autonomous, simple-to-complex experience. The paper also discusses some key concepts that are necessary for understanding the mind and intelligence, such as intelligence metrics, the mode a task is conveyed to the task executor, a human and a machine being a joint task performer in the traditional artificial intelligence (AI), a developmental agent (human or machine) being a sole task performer, and the need for autonomy in task-nonexplicit learning. (shrink)

The development of electronic agents that increasingly play an active role in the contract formation and execution process has highlighted the need for the creation of law-abiding autonomous agent systems. The principle of good faith is an important guideline for contractual behaviour which permeates civil law systems. This paper examines how this principle is applied both during the negotiation of a contract and during its performance. Selected examples from civil law literature of precontractual duties of good faith, and of precontractual (...) behaviour that is deemed to be contrary to good faith, are discussed. This is followed by a discussion of the extent to which such duties are recognised, or such behaviour proscribed, in common law jurisdictions. Some common standards for precontractual behaviour in civil and common law systems are identified. There is then a parallel analysis of the principle of good faith in contract performance with a view to identifying common traits or standards between civil and common law systems. These standards, in the situation where contracts are being negotiated and/or performed by or through electronic agents, would need to be reflected in the way such agents operate. (shrink)

Because of its complexity, contemporary scientific research is almost always tackled by groups of scientists, each of which works in a different part of a given research domain. We believe that understanding scientific progress thus requires understanding this division of cognitive labor. To this end, we present a novel agent-based model of scientific research in which scientists divide their labor to explore an unknown epistemic landscape. Scientists aim to climb uphill in this landscape, where elevation represents the significance of the (...) results discovered by employing a research approach. We consider three different search strategies scientists can adopt for exploring the landscape. In the first, scientists work alone and do not let the discoveries of the community as a whole influence their actions. This is compared with two social research strategies, which we call the follower and maverick strategies. Followers are biased towards what others have already discovered, and we find that pure populations of these scientists do less well than scientists acting independently. However, pure populations of mavericks, who try to avoid research approaches that have already been taken, vastly outperform both of the other strategies. Finally, we show that in mixed populations, mavericks stimulate followers to greater levels of epistemic production, making polymorphic populations of mavericks and followers ideal in many research domains. (shrink)

Interactions between humans and machines that include artificial intelligence are increasingly common in nearly all areas of life. Meanwhile, AI-products are increasingly endowed with emotional characteristics. That is, they are designed and trained to elicit emotions in humans, to recognize human emotions and, sometimes, to simulate emotions. The introduction of such systems in our lives is met with some criticism. There is a rather strong intuition that there is something wrong about getting attached to a machine, about having certain emotions (...) towards it, and about getting involved in a kind of affective relationship with it. In this paper, I want to tackle these worries by focusing on the last aspect: in what sense could it be problematic or even wrong to establish an emotional relationship with EAI-systems? I want to show that the justifications for the widespread intuition concerning the problems are not as strong as they seem at first sight. To do so, I discuss three arguments: the argument from self-deception, the argument from lack of mutuality, and the argument from moral negligence. (shrink)

This paper offers a solution to the problem of understanding how a fallacious argument can be deceptive by “seeming to be valid”, or (better) appearing to be a better argument of its kind than it really is. The explanation of how fallacies are deceptive is based on heuristics and paraschemes. Heuristics are fast and frugal shortcuts to a solution to a problem that sometimes jump to a conclusion that is not justified. In fallacious instances, according to the theory proposed, this (...) jump overlooks prerequisites of the defeasible argumentation scheme for the type of argument in question. Three informal fallacies, argumentum ad verecundiam , argumentum ad ignorantiam and fear appeal argument, are used to illustrate and explain the theory. (shrink)

Dialogue theory, although it has ancient roots, was put forward in the 1970s in logic as astructure that can be useful for helping to evaluate argumentation and informal fallacies.Recently, however, it has been taken up as a broader subject of investigation in computerscience. This paper surveys both the historical and philosophical background of dialoguetheory and the latest research initiatives on dialogue theory in computer science. The main components of dialogue theory are briefly explained. Included is a classification of the main (...) types of dialogue that, it is argued, should provide the central focus for studying many important dialogue contexts in specific cases. Following these three surveys, a concluding prediction is made about the direction dialogue theory is likely to take in the next century, especially in relation to the growing field of communication studies. (shrink)

In this paper, the traditional view that argumentum ad ignorantiam is a logical fallacy is challenged, and lessons are drawn on how to model inferences drawn from knowledge in combination with ones drawn from lack of knowledge. Five defeasible rules for evaluating knowledge-based arguments that apply to inferences drawn under conditions of lack of knowledge are formulated. They are the veridicality rule, the consistency of knowledge rule, the closure of knowledge rule, the rule of refutation and the rule for argument (...) from ignorance. The basic thesis of the paper is that knowledge-based arguments, including the argument from ignorance, need to be evaluated by criteria for epistemic closure and other evidential rules that are pragmatic in nature, that need to be formulated and applied differently at different stages of an investigation or discussion. The paper helps us to understand practical criteria that should be used to evaluate all arguments based on knowledge and/or ignorance. (shrink)

Research based on traditional linear techniques has yet not been able to clearly identify the role of cognitive skills in reading problems, presumably because the process of reading and the factors that are associated with reading reside within a system of multiple interacting and moderating factors that cannot be captured within traditional statistical models. If cognitive skills are indeed indicative of reading problems, the relatively new nonlinear techniques of machine learning should make better predictions. The aim of the present study (...) was to investigate whether cognitive factors play any role in reading skill, questioning the extent to what cognitive skills are indicative of present reading level, and the extent to what cognitive skills are indicative of future reading progress. In three studies with varying groups of participants, the results of four supervised machine learning techniques were compared to the traditional General Linear Models technique. Results of all models appeared to be comparable, producing poor to acceptable results, which are however inadequate for making a thorough prediction of reading development. Assumably, cognitive skills are not predictive of reading problems, although they do correlate with one another. This insight has consequences for scientific theories of reading development, as well as for the prevention and remediation of reading difficulties. (shrink)

Hamblin distinguished between formal and descriptive dialectic. Formal normative models of deliberation dialogue have been strongly emphasized as argumentation frameworks in computer science. But making such models of deliberation applicable to real natural language examples has reached a point where the descriptive aspect needs more interdisciplinary work. The new formal and computational models of deliberation dialogue that are being built in computer science seem to be closely related to some already existing and very well established computing technologies such as problem (...) solving and decision making, but whether or how dialectical argumentation can be helpful to support these systems remains an open question. The aim of this paper is to examine some real examples of argumentation that seem to hover on the borderlines between deliberation, problem solving and decision making. (shrink)

This paper applies dialectical argumentation structures to the problem of analyzing the ad baculum fallacy. It is shown how it is necessary in order to evaluate a suspected instance of the this fallacy to proceed through three levels of analysis: an inferential level, represented by an argument diagram, a speech act level, where conditions for specific types of speech acts are defined and applied, and a dialectical level where the first two levels are linked together and fitted into formal dialogue (...) structures. The paper adds a new type of dialogue called advising dialogue that needs to be applied at the third level. (shrink)

Decisions in security contexts, including armed conflict, law enforcement, and disaster relief, often need to be taken under circumstances of limited information, stress, and time pressure. Since AI systems are capable of providing a certain amount of relief in such contexts, such systems will become increasingly important, be it as decision-support or decision-making systems. However, given that human life may be at stake in such situations, moral responsibility for such decisions should remain with humans. Hence the idea of “meaningful human (...) control” of intelligent systems. In this opinion paper, we outline generic configurations of control of AI and we present an alternative to human control of AI, namely the inverse idea of having AI control humans, and we discuss the normative consequences of this alternative. (shrink)

The purpose of this work is to analyse the cognitive process of the domain theories in terms of the measurement theory to develop a computational machine learning approach for implementing it. As a result, the relational data mining approach, the authors proposed in the preceding books, was improved. We present the approach as an implementation of the cognitive process as the measurement theory perceived. We analyse the cognitive process in the first part of the paper and present the theory and (...) method of the logically most powerful empirical theory discovery in the second. The theory is based on the notion of law-like rules, which conform to all the properties of laws of nature, namely generality, simplicity, maximum refutability and minimum number of parameters. This notion is defined for deterministic and probabilistic cases. Based on the method, the discovery system is developed. The system was successfully applied to many practical tasks. (shrink)

This article discusses the political economy of AI capitalism. It considers AI as a General Purpose Technology and argues we need to investigate the power concentration of Big Tech. AI capitalism is characterised by the commodification of data, data extraction and a concentration in hiring of AI talent and compute capacity. This is behind Big Tech’s unstoppable drive for growth, which leads to monopolisation and enclosure under the winner takes all principle. If we consider AI as a GPT—technologies that alter (...) society’s economic and social structures—we need to come up with alternatives in terms of ownership and governance. The commons is proposed as an alternative for thinking about how to organise AI development and how to distribute the value that can be derived from it. Using the commons framework is also a way of giving society a more prominent role in the debate about what we expect from AI and how we should approach it. (shrink)

Explanations of cognitive processes provided by traditional artificial intelligence were based on the notion of the knowledge level. This perspective has been challenged by new AI that proposes an approach based on embodied systems that interact with the real‐world. We demonstrate that these two views can be unified. Our argument is based on the assumption that knowledge level explanations can be defined in the context of Bayesian theory while the goals of new AI are captured by using a well established (...) robot based model of learning and problem solving, called Distributed Adaptive Control (DAC). In our analysis we consider random foraging and we prove that minor modifications of the DAC architecture renders a model that is equivalent to a Bayesian analysis of this task. Subsequently, we compare this enhanced, “rational,” model to its “non‐rational” predecessor and a further control condition using both simulated and real robots, in a variety of environments. Our results show that the changes made to the DAC architecture, in order to unify the perspectives of old and new AI, also lead to a significant improvement in random foraging. (shrink)

Many compelling examples have recently been provided in which people can achieve impressive epistemic success, e.g. draw highly accurate inferences, by using simple heuristics and very little information. This is possible by taking advantage of the features of the environment. The examples suggest an easy and appealing naturalization of rationality: on the one hand, people clearly can apply simple heuristics, and on the other hand, they intuitively ought do so when this brings them high accuracy at little cost.. The ‘ought-can’ (...) principle is satisfied, and rationality is meaningfully normative. We show, however, that this naturalization program is endangered by a computational wrinkle in the adaptation process taken to be responsible for this heuristics-based (‘ecological’) rationality: for the adaptation process to guarantee even minimal rationality, it requires astronomical computational resources, making the problem intractable. We consider various plausible auxiliary assumptions in attempt to remove this obstacle, and show that they do not succeed; intractability is a robust property of adaptation. We discuss the implications of our findings for the project of naturalizing rationality. (shrink)

As the capabilities of artificial intelligence systems improve, it becomes important to constrain their actions to ensure their behaviour remains beneficial to humanity. A variety of ethical, legal and safety-based frameworks have been proposed as a basis for designing these constraints. Despite their variations, these frameworks share the common characteristic that decision-making must consider multiple potentially conflicting factors. We demonstrate that these alignment frameworks can be represented as utility functions, but that the widely used Maximum Expected Utility paradigm provides insufficient (...) support for such multiobjective decision-making. We show that a Multiobjective Maximum Expected Utility paradigm based on the combination of vector utilities and non-linear action–selection can overcome many of the issues which limit MEU’s effectiveness in implementing aligned AI. We examine existing approaches to multiobjective AI, and identify how these can contribute to the development of human-aligned intelligent agents. (shrink)

In the fields of artificial intelligence and robotics, the term “autonomy” is generally used to mean the capacity of an artificial agent to operate independently of human guidance. It is thereby assumed that the agent has a fixed goal or “utility function” with respect to which the appropriateness of its actions will be evaluated. From a philosophical perspective, this notion of autonomy seems oddly weak. For, in philosophy, the term is generally used to refer to a stronger capacity, namely the (...) capacity to “give oneself the law,” to decide by oneself what one’s goal or principle of action will be. The predominant view in the literature on the long-term prospects and risks of artificial intelligence is that an artificial agent cannot exhibit such autonomy because it cannot rationally change its own final goal, since changing the final goal is counterproductive with respect to that goal and hence undesirable. The aim of this paper is to challenge this view by showing that it is based on questionable assumptions about the nature of goals and values. I argue that a general AI may very well come to modify its final goal in the course of developing its understanding of the world. This has important implications for how we are to assess the long-term prospects and risks of artificial intelligence. (shrink)

In this article I critique Kathleen Slaney and Michael Maraun’s (2005) addition to the ongoing philosophical charge that neuroscientific writing often transgresses the bounds of sense. While they sometimes suggest a minimal, cautious thesis–that certain usage can generate confusion and in some cases has–they also bandy about charges of meaninglessness, conceptual confusion, and nonsense freely. These charges rest on the premise that terms have specific correct usages that correspond with Slaney and Maraun’s sense of everyday linguistic practice. I challenge this (...) premise. I argue that they have not shown that there are such specific correct usages; and, further, that even if they had, they fail to justify that their definitions are the correct ones. (shrink)

This article proposes a unified framework for understanding creative problem solving, namely, the explicit–implicit interaction theory. This new theory of creative problem solving constitutes an attempt at providing a more unified explanation of relevant phenomena (in part by reinterpreting/integrating various fragmentary existing theories of incubation and insight). The explicit–implicit interaction theory relies mainly on 5 basic principles, namely, (a) the coexistence of and the difference between explicit and implicit knowledge, (b) the simultaneous involvement of implicit and explicit processes in most (...) tasks, (c) the redundant representation of explicit and implicit knowledge, (d) the integration of the results of explicit and implicit processing, and (e) the iterative (and possibly bidirectional) processing. A computational implementation of the theory is developed based on the CLARION cognitive architecture and applied to the simulation of relevant human data. This work represents an initial step in the development of process-based theories of creativity encompassing incubation, insight, and various other related.. (shrink)

The current research aims to answer the following question: “who will be held responsible for harm involving an artificial intelligence system?” Drawing upon the literature on moral judgments, we assert that when people perceive an AI system’s action as causing harm to others, they will assign blame to different entity groups involved in an AI’s life cycle, including the company, the developer team, and even the AI system itself, especially when such harm is perceived to be intentional. Drawing upon the (...) theory of mind perception, we hypothesized that two dimensions of mind: perceived agency—attributing intention, reasoning, pursuing goals, and communicating to AI, and perceived experience—attributing emotional states, such as feeling pain and pleasure, personality, and consciousness to AI—mediated the relationship between perceived intentional harm and blame judgments toward AI. We also predicted that people are likely to attribute higher mind characteristics to AI when harm is perceived to be directed to humans than when it is perceived to be directed to non-humans. We tested our research model in three experiments. In all experiments, we found that perceived intentional harm led to blame judgments toward AI. In two experiments, we found perceived experience, not agency, mediated the relationship between perceived intentional harm and blame judgments. We also found that companies and developers were held responsible for moral violations involving AI, with developers received the most blame among the entities involved. Our third experiment reconciles the findings by showing that perceived intentional harm directed to a non-human entity did not lead to increased attributions of mind to AI. These findings have implications for theory and practice concerning unethical outcomes and behavior associated with AI use. (shrink)