Henry Shevlin

In light of recent breakneck pace in machine learning, questions about whether near-future artificial systems might be conscious and possess moral status are increasingly pressing. This paper argues that as matters stand these debates lack any clear criteria for resolution via the science of consciousness. Instead, insofar as they are settled at all, it is likely to be via shifts in public attitudes brought about by the increasingly close relationships between humans and AI users. Section 1 of the paper I briefly lays out the current state of the science of consciousness and its limitations insofar as these pertain to machine consciousness, and claims that there are no obvious consensus frameworks to inform public opinion on AI consciousness. Section 2 examines the rise of conversational chatbots or Social AI, and argues that in many cases, these elicit strong and sincere attributions of consciousness, mentality, and moral status from users, a trend likely to become more widespread. Section 3 presents an inconsistent triad for theories that attempt to link consciousness, behaviour, and moral status, noting that the trends in Social AI systems will likely make the inconsistency of these three premises more pressing. Finally, Section 4 presents some limited suggestions for how consciousness and AI research communities should respond to the gap between expert opinion and folk judgment.

The science of consciousness has made great strides in recent decades. However, the proliferation of competing theories makes it difficult to reach consensus about artificial consciousness. While for purely scientific purposes we might wish to adopt a ‘wait and see’ attitude, we may soon face practical and ethical questions about whether, for example, agents artificial systems are capable of suffering. Moreover, many of the methods used for assessing consciousness in humans and even non-human animals are not straightforwardly applicable to artificial systems. With these challenges in mind, I propose that we look for ecumenical heuristics for artificial consciousness to enable us to make tentative assessments of the likelihood of consciousness arising in different artificial systems. I argue that such heuristics should have three main features: they should be (i) intuitively plausible, (ii) theoretically-neutral, and (iii) scientifically tractable. I claim that the concept of general intelligence — understood as a capacity for robust, flexible, and integrated cognition and behavior — satisfies these criteria and may thus provide the basis for such a heuristic, allowing us to make initial cautious estimations of which artificial systems are most likely to be conscious.

There is growing interest in machine ethics in the question of whether and under what circumstances an artificial intelligence would deserve moral consideration. This paper explores a particular type of moral status that the author terms psychological moral patiency, focusing on the epistemological question of what sort of evidence might lead us to reasonably conclude that a given artificial system qualified as having this status. The paper surveys five possible criteria that might be applied: intuitive judgments, assessments of intelligence, the presence of desires and autonomous behavior, evidence of sentience, and behavioral equivalence. The author suggests that despite its limitations, the latter approach offers the best way forward, and defends a variant of that, termed the cognitive equivalence strategy. In short, this holds that an artificial system should be considered a psychological moral patient to the extent that it possesses cognitive mechanisms shared with other beings such as nonhuman animals whom we also consider to be psychological moral patients.

Despite there being little consensus on what intelligence is or how to measure it, the media and the public have become increasingly preoccupied with the concept owing to recent accomplishments in machine learning and research on artificial intelligence (AI). Governments and corporations are investing billions of dollars to fund researchers who are keen to produce an ever‐expanding range of artificial intelligent systems. More than 30 countries have announced such research initiatives over the past 3 years 1. For example, the EU Commission pledged to increase the investment in AI research to €1.5 billion by 2020 (from €500 million in 2017), while China has committed $2.1 billion towards an AI technology park in Beijing alone 1. This global investment in AI is astonishing and prompts several questions: What are the true possibilities and limitations of AI? What do AI researchers and developers mean by “intelligence”? How does this compare to the everyday concept of intelligence and how the term is other branches of cognitive science? And can machine learning produce anything that is truly “intelligent”?

This paper presents an overview of the risks and benefits of Social AI, understood as conversational AI systems that cater to human social needs like romance, companionship, or entertainment. Section 1 of the paper provides a brief history of conversational AI systems and introduces conceptual distinctions to help distinguish varieties of Social AI and pathways to their deployment. Section 2 of the paper adds further context via a brief discussion of anthropomorphism and its relevance to assessment of human-chatbot relationships. Section 3 of the paper provides a survey of potential and in some cases demonstrated harms associated with user interactions with Social AI systems. Finally, Section 4 discusses how the benefits and harms of Social AI can best be addressed, with a primary focus on how frameworks from AI ethics can inform their development.

We review publications in automated scientific discovery using deep learning, with the aim of shedding light on problems with strong connections to philosophy of science, of physics in particular. We show that core issues of philosophy of science, related, notably, to the nature of scientific theories; the nature of unification; and of causation loom large in scientific deep learning. Therefore, advances in deep learning could, and ideally should, have impact on philosophy of science, and vice versa. We suggest lines of further research, and highlight the role ‘theory-based’ AI could have in future developments of the field.

Most people will grant that we bear special moral obligations toward at least some nonhuman animals that we do not bear toward inanimate objects like stones, mountains, or works of art. These moral obligations are plausibly grounded in the fact that many if not all nonhuman animals share important psychological states and capacities with us, such as consciousness, suffering, and goal-directed behavior. But which of these states and capacities are really critical for a creature’s possessing moral status, and how can we determine which animals do in fact have them? In this paper, I examine three main approaches to answering these questions. First are what I term consciousness-based approaches that tackles these questions by first asking which animals are conscious. Second are affective-state approaches that focus on identifying behavioural and physiological signatures of states like pain, fear, and stress. Finally, I consider what I call preference-based approaches whose focus is on the question of which organisms have robust motivational states. I examine the prospects and challenges—both theoretical and empirical—faced by these seemingly contrasting methodologies. I go on to suggest that there are reasons why, despite challenges, we should be robustly committed to the project of identifying psychological grounds of moral status. I conclude by suggesting we should also take seriously the idea of pluralism about moral status, according to which each of these approaches might be capable of providing independent grounds for moral consideration.

Most scientific theories of consciousness are challenging to apply outside the human case insofar as non‐human systems (both biological and artificial) are unlikely to implement human architecture precisely, an issue I call the specificity problem. After providing some background on the theories of consciousness debate, I survey the prospects of four approaches to this problem. I then consider a fifth solution, namely the theory‐light approach proposed by Jonathan Birch. I defend a modified version of this that I term the modest theoretical approach, arguing that it may provide insights into challenging cases that would otherwise be intractable.

The concept of creativity is a central one in folk psychological explanation and has long been prominent in philosophical debates about the nature of art, genius, and the imagination. The scientific investigation of creativity in humans is also well established, and there has been increasing interest in the question of whether the concept can be rigorously applied to non-human animals. In this paper, I argue that such applications face serious challenges of both a conceptual and methodological character, reflecting deep controversies within both philosophy and psychology concerning how to define and apply the concept of creativity. After providing a brief review of some of the leading theories of creativity and discussing some of the strongest putative cases of creative intelligence in non-human animals, I examine some of the more worrisome difficulties faced by attempts to use these theories to answer the question of whether animals are truly creative. I conclude by examining how we might overcome them, and suggest that one approach worth taking seriously is to adopt what I term a Strong Rejectionist view of creativity, eschewing use of the term entirely in the scientific study of comparative cognition.

There is compelling evidence that pain experience is influenced by cognitive states. We explore one specific form of such influence, namely placebo analgesia, and examine its relevance for the cognitive penetration debate in philosophy of mind. We single out as important a form of influence on experience that we term radical cognitive penetration, and argue that some cases of placebo analgesia constitute compelling instances of this phenomenon. Still, we urge caution in extrapolating from this to broader conclusions about cognitive penetration in perceptual experience. Instead, we suggest that the cognitive penetration of pain raises distinctive psychological, epistemological, and ethical issues.

One influential philosophical account of desire treats it as a species of propositional attitude, possessing broadly the same kinds of content as belief while differing in direction of fit. However, this arguably neglects more basic forms of desire. It seems an open possibility, for example, that animals that lack propositional attitudes might still have simple desires mediated by sensations like hunger and thirst. In this essay, I will argue the case for the existence of these basic desires, and suggest a strategy for understanding them. I begin by briefly discussing some different approaches to the psychological ascription of desires, and examine some candidate forms of behaviour that might be constitutive of a creature's having a basic desire. I go on to argue that a form of behaviour known as motivational trade-off offers a particularly promising framework for attributions of basic desire.

In debates in philosophy and cognitive science concerning short-term memory mechanisms and perceptual experience, most discussion has focused on the working memory and the various forms of sensory memory such as iconic memory. In this paper, I present a summary of some evidence for a proposed further form of memory termed conceptual short-term memory. I go on to outline some of the ways in which this additional distinctive sort of short-term memory might be of relevance to ongoing philosophical debates, specifically in relation to questions about high-level perceptual phenomenology, the relationship between consciousness and reportability, and the boundary between cognition and perception. I conclude that conceptual short-term memory offers a promising new direction of research and philosophical investigation.