1. Reflecting on possible cases can be very valuable in differing ways. Sometimes it makes clear a consequence of a theory, a consequence that then plays an important role in debates about the theory. Utilitarians who favour maximising average happiness confront utilitarians who favour maximising total happiness with possible cases where there are enormously many sentient beings whose lives are barely worth living. Sometimes reflecting on possible cases serves to clarify a doctrine. Classical versions of consequentialism value equity for its (...) good results; they don't value it per se. What this means is that in cases where we have two actions alike in the value of their consequences but differing in that one delivers a more equitable distribution, classical consequentialism treats the two actions as morally equivalent. Sometimes reflections on possible cases reveal relationships that are not immediately obvious. Take Galileo's famous thought experiment directed against the view that the heavier a body is the faster it falls in a vacuum. He describes a possible case where the only way to retain the view requires embracing independently implausible views about how various ways two bodies are connected to each other make a big difference to how fast the joined bodies fall. The thought experiment reveals a connection that might otherwise escape notice.This reminds us that we should not expect a uniform answer to the question, What role do thought experiments play? However, it is conventional wisdom that very often their role is to make trouble for one or another proffered conceptual analysis. Take two thought experiments that have been found especially compelling by the analytical philosophical community: Edmund Gettier's examples of true, justified belief, where the beliefs are true by some kind of fluke, and Ned Block's example of a being that makes ‘intelligent’ responses to the challenges the …. (shrink)

“Triviality arguments” against functionalism in the philosophy of mind hold that the claim that some complex physical system exhibits a given functional organization is either trivial or has much less content than is usually supposed. I survey several earlier arguments of this kind, and present a new one that overcomes some limitations in the earlier arguments. Resisting triviality arguments is possible, but requires functionalists to revise popular views about the “autonomy” of functional description.

Quine’s thesis of the indeterminacy of translation has puzzled the philosophical community for several decades. It is unquestionably among the best known and most disputed theses in contemporary philosophy. Quine’s classical argument for the indeterminacy thesis, in his seminal work Word and Object, has even been described by Putnam as “what may well be the most fascinating and the most discussed philosophical argument since Kant’s Transcendental Deduction of the Categories” (Putnam, 1975a: p. 159).

Eliminative functionalism is the view that mental attributes, of humans and other machines, consist ultimately in behavioural abilities or dispositions. Hence, ‘Strong AI’: if a machine consistently acts as if it were fully conscious, then conscious it is. From these assumptions, optimistic futurists have derived a variety of remarkable visions of our ‘post-human’ future; from widely-recognised ‘robot rights’ to ‘mind uploading’, immortality, ‘apotheosis’ and beyond. It is argued here, however, that eliminative functionalism is false; for at least on our present (...) knowledge, the subjectively qualitative characteristics of conscious experience are neither deducible from, nor logically required to generate, the performance of any sort of overtly ‘intelligent’, or indeed, characteristically human behaviour. Thus, a machine could easily be designed to report awareness of phenomenal qualities, without necessarily possessing them; and Alan Turing’s ‘Imitation Game’ test for artificial thinking is unable to determine whether or not a machine is sentient. An alternative test is proposed, in which the machine is asked phenomenological questions under conditions designed to detect any form of cheating—whilst also, potentially revealing evidence for the occurrence of genuine qualitative experience. (shrink)

The claim has often been made that passing the Turing Test would not be sufficient to prove that a computer program was intelligent because a trivial program could do it, namely, the “Humongous-Table (HT) Program”, which simply looks up in a table what to say next. This claim is examined in detail. Three ground rules are argued for: (1) That the HT program must be exhaustive, and not be based on some vaguely imagined set of tricks. (2) That the HT (...) program must not be created by some set of sentient beings enacting responses to all possible inputs. (3) That in the current state of cognitive science it must be an open possibility that a computational model of the human mind will be developed that accounts for at least its nonphenomenological properties. Given ground rule 3, the HT program could simply be an “optimized” version of some computational model of a mind, created via the automatic application of program-transformation rules [thus satisfying ground rule 2]. Therefore, whatever mental states one would be willing to impute to an ordinary computational model of the human psyche one should be willing to grant to the optimized version as well. Hence no one could dismiss out of hand the possibility that the HT program was intelligent. This conclusion is important because the Humongous-Table Program Argument is the only argument ever marshalled against the sufficiency of the Turing Test, if we exclude arguments that cognitive science is simply not possible. (shrink)

Using empirical evidence to attack intuitions can be epistemically dangerous, because various of the complaints that one might raise against them (e.g., that they are fallible; that we possess no non-circular defense of their reliability) can be raised just as easily against perception itself. But the opponents of intuition wish to challenge intuitions without at the same time challenging the rest of our epistemic apparatus. How might this be done? Let us use the term “hopefulness” to refer to the extent (...) to which we possess a good capacity for the detection and correction of the errors of any fallible source of evidence. I argue that we should not trust putative sources of evidence that are substantially lacking in hopefulness (even if they are basically reliable), and that we are indeed already operating under such a norm in our ordinary and scientific practices. I argue further that the philosophical practice of the appeal to intuitions is, in these terms, badly hopeless... (shrink)

The aim of this paper is to present a certain kind of argumentation against the idea of the Turing test and to discuss the issue of its first formulation. Ned Block, with his idea of “Aunt Bubbles” argument, is thought of as a founding father of CCSC, but we present the results of our bibliographical researches which clearly show that the first formulation of CCSC should be ascribed to Polish writer and philosopher Stanisław Lem.

In 1950, Alan Turing proposed his eponymous test based on indistinguishability of verbal behavior as a replacement for the question "Can machines think?" Since then, two mutually contradictory but well-founded attitudes towards the Turing Test have arisen in the philosophical literature. On the one hand is the attitude that has become philosophical conventional wisdom, viz., that the Turing Test is hopelessly flawed as a sufficient condition for intelligence, while on the other hand is the overwhelming sense that were a machine (...) to pass a real live full-fledged Turing Test, it would be a sign of nothing but our orneriness to deny it the attribution of intelligence. The arguments against the sufficiency of the Turing Test for determining intelligence rely on showing that some extra conditions are logically necessary for intelligence beyond the behavioral properties exhibited by an agent under a Turing Test. Therefore, it cannot follow logically from passing a Turing Test that the agent is intelligent. I argue that these extra conditions can be revealed by the Turing Test, so long as we allow a very slight weakening of the criterion from one of logical proof to one of statistical proof under weak realizability assumptions. The argument depends on the notion of interactive proof developed in theoretical computer science, along with some simple physical facts that constrain the information capacity of agents. Crucially, the weakening is so slight as to make no conceivable difference from a practical standpoint. Thus, the Gordian knot between the two opposing views of the sufficiency of the Turing Test can be cut. (shrink)

A fundamental problem in artificial intelligence is that nobody really knows what intelligence is. The problem is especially acute when we need to consider artificial systems which are significantly different to humans. In this paper we approach this problem in the following way: we take a number of well known informal definitions of human intelligence that have been given by experts, and extract their essential features. These are then mathematically formalised to produce a general measure of intelligence for arbitrary machines. (...) We believe that this equation formally captures the concept of machine intelligence in the broadest reasonable sense. We then show how this formal definition is related to the theory of universal optimal learning agents. Finally, we survey the many other tests and definitions of intelligence that have been proposed for machines. (shrink)

This paper addresses a problem about epistemic warrant. The problem is posed by philosophical arguments for externalism about the contents of thoughts, and similarly by philosophical arguments for architecturalism about thinking, when these arguments are put together with a thesis of first person authority. In each case, first personal knowledge about our thoughts plus the kind of knowledge that is provided by a philosophical argument seem, together, to open an unacceptably ‘non-empirical’ route to knowledge of empirical facts. Furthermore, this unwelcome (...) prospect of transferring a ‘non-empirical’ warrant from premises about our own mental states and about philosophical theory to a conclusion about external environment or internal architecture seems to depend upon little more than the possibility of knowledge by inference. (The use of the scare-quoted term ‘non-empirical’ is explained a couple of paragraphs further on.). (shrink)

Probabilism is committed to two theses: 1) Opinion comes in degrees—call them degrees of belief, or credences. 2) The degrees of belief of a rational agent obey the probability calculus. Correspondingly, a natural way to argue for probabilism is: i) to give an account of what degrees of belief are, and then ii) to show that those things should be probabilities, on pain of irrationality. Most of the action in the literature concerns stage ii). Assuming that stage i) has been (...) adequately discharged, various authors move on to stage ii) with varied and ingenious arguments. But an unsatisfactory response at stage i) clearly undermines any gains that might be accrued at stage ii) as far as probabilism is concerned: if those things are not degrees of belief, then it is irrelevant to probabilism whether they should be probabilities or not. In this paper we scrutinize the state of play regarding stage i). We critically examine several of the leading accounts of degrees of belief: reducing them to corresponding betting behavior (de Finetti); measuring them by that behavior (Jeffrey); and analyzing them in terms of preferences and their role in decision-making more generally (Ramsey, Lewis, Maher). We argue that the accounts fail, and so they are unfit to subserve arguments for probabilism. We conclude more positively: ‘degree of belief’ should be taken as a primitive concept that forms the basis of our best theory of rational belief and decision: probabilism. (shrink)

Fodor argued that in contrast to input systems which are informationally encapsulated, general intelligence is unencapsulated and hence non-modular; for this reason, he suggested, prospects for understanding it are not bright. It is argued that an additional property, primitive functionality, is required for non-modularity. A functionally primitive computational model for quantifier scoping, limited to some scoping influences, is then motivated, and an implementation described. It is argued that only such a model can be faithful to intuitive scope preferences. But it (...) is also argued that an extended model which includes all scoping influences is a hopeless prospect from a developmental perspective. Fodor's views are concluded to have some independent support: quantifier scoping is a mental ability parasitic on general intelligence that is non-modular though in a revised sense, warranting pessimism about our ability of model it. (shrink)

The idea of a folk theory has played many important roles in much recent philosophy. To do the work they are designed for, they need to be both internal features of agents who possess them, and yet scrutable without the full resources of empirical cognitive science. The worry for the theorist of folk theories, is that only one of these desiderata is met in each plausible conception of a folk theory. This paper outlines a third conception that meets them both.1.

Hilary Putnam has argued that computational functionalism cannot serve as a foundation for the study of the mind, as every ordinary open physical system implements every finite-state automaton. I argue that Putnam's argument fails, but that it points out the need for a better understanding of the bridge between the theory of computation and the theory of physical systems: the relation of implementation. It also raises questions about the class of automata that can serve as a basis for understanding the (...) mind. I develop an account of implementation, linked to an appropriate class of automata, such that the requirement that a system implement a given automaton places a very strong constraint on the system. This clears the way for computation to play a central role in the analysis of mind. (shrink)

Over the past thirty years, it is been common to hear the mind likened to a digital computer. This essay is concerned with a particular philosophical view that holds that the mind literally is a digital computer (in a specific sense of “computer” to be developed), and that thought literally is a kind of computation. This view—which will be called the “Computational Theory of Mind” (CTM)—is thus to be distinguished from other and broader attempts to connect the mind with computation, (...) including (a) various enterprises at modeling features of the mind using computational modeling techniques, and (b) employing some feature or features of production-model computers (such as the stored program concept, or the distinction between hardware and software) merely as a guiding metaphor for understanding some feature of the mind. This entry is therefore concerned solely with the Computational Theory of Mind (CTM) proposed by Hilary Putnam [1961] and developed most notably for philosophers by Jerry Fodor [1975, 1980, 1987, 1993]. The senses of ‘computer’ and ‘computation’ employed here are technical; the main tasks of this entry will therefore be to elucidate: (a) the technical sense of ‘computation’ that is at issue, (b) the ways in which it is claimed to be applicable to the mind, (c) the philosophical problems this understanding of the mind is claimed to solve, and (d) the major criticisms that have accrued to this view. (shrink)

In this section, we will start with an influential attempt to define `intelligence', and then we will move to a consideration of how human intelligence is to be investigated on the machine model. The last part of the section will discuss the relation between the mental and the biological.

What psychological and philosophical significance should we attach to recent efforts at computer simulations of human cognitive capacities? In answering this question, I find it useful to distinguish what I will call "strong" AI from "weak" or "cautious" AI. According to weak AI, the principal value of the computer in the study of the mind is that it gives us a very powerful tool. For example, it enables us to formulate and test hypotheses in a more rigorous and precise fashion. (...) But according to strong AI, the computer is not merely a tool in the study of the mind; rather, the appropriately programmed computer really is a mind, in the sense that computers given the right programs can be literally said to. (shrink)

This paper provides a survey of philosophical discussion of the "the Turing Test". In particular, it provides a very careful and thorough discussion of the famous 1950 paper that was published in Mind.

The standard interpretation of the imitation game is defended over the rival gender interpretation though it is noted that Turing himself proposed several variations of his imitation game. The Turing test is then justified as an inductive test not as an operational definition as commonly suggested. Turing's famous prediction about his test being passed at the 70% level is disconfirmed by the results of the Loebner 2000 contest and the absence of any serious Turing test competitors from AI on the (...) horizon. But, reports of the death of the Turing test and AI are premature. AI continues to flourish and the test continues to play an important philosophical role in AI. Intelligence attribution, methodological, and visionary arguments are given in defense of a continuing role for the Turing test. With regard to Turing's predictions one is disconfirmed, one is confirmed, but another is still outstanding. (shrink)

The Turing Test, originally proposed as a simple operational definition of intelligence, has now been with us for exactly half a century. It is safe to say that no other single article in computer science, and few other articles in science in general, have generated so much discussion. The present article chronicles the comments and controversy surrounding Turing's classic article from its publication to the present. The changing perception of the Turing Test over the last fifty years has paralleled the (...) changing attitudes in the scientific community towards artificial intelligence: from the unbridled optimism of 1960's to the current realization of the immense difficulties that still lie ahead. I conclude with the prediction that the Turing Test will remain important, not only as a landmark in the history of the development of intelligent machines, but also with real relevance to future generations of people living in a world in which the cognitive capacities of machines will be vastly greater than they are now. (shrink)

Turing''s test has been much misunderstood. Recently unpublished material by Turing casts fresh light on his thinking and dispels a number of philosophical myths concerning the Turing test. Properly understood, the Turing test withstands objections that are popularly believed to be fatal.

Qualia inversion thought experiments are ubiquitous in contemporary philosophy of mind. The most popular kind is one or another variant of Locke's hypothetical case of.

A phenomenal concept is the concept of a particular type of sensory or perceptual experience, where the notion of experience is understood phenomenologically. A recent and increasingly influential idea in philosophy of mind suggests that reflection on these concepts will play a major role in the debate about conscious experience, and in particular in the defense of physicalism, the thesis that psychological truths supervene on physical truths. According to this idea—I call it the phenomenal concept strategy —phenomenal concepts are importantly (...) different from other concepts, and arguments against physicalism fatally neglect to take this difference into account. This paper divides the phenomenal concept strategy into a number of different versions, and argues that no version of the strategy is successful. The paper ends by contrasting the phenomenal concept strategy with a rival strategy—I call it the missing concept strategy . I suggest that the missing concept strategy presents a more plausible response to the issues about physicalism and experience. (shrink)

Frank Jackson champions the cause of conceptual analysis as central to philosophical inquiry. In recent years conceptual analysis has been undervalued and widely misunderstood, suggests Jackson. He argues that such analysis is mistakenly clouded in mystery, preventing a whole range of important questions from being productively addressed. He anchors his argument in discussions of specific philosophical issues, starting with the metaphysical doctrine of physicalism and moving on, via free will, meaning, personal identity, motion, and change, to ethics and the philosophy (...) of color. In this way the book not only offers a methodological program for philosophy, but also casts new light on some much-debated problems and their interrelations. (shrink)

The notion of intelligence is relevant to several fields of research, including cognitive and comparative psychology, neuroscience, artificial intelligence, and philosophy, among others. However, there is little agreement within and across these fields on how to characterise and explain intelligence. I put forward a behavioural, operational characterisation of intelligence that can play an integrative role in the sciences of intelligence, as well as preserve the distinctive explanatory value of the notion, setting it apart from the related concepts of cognition and (...) rationality. Finally, I examine a popular hypothesis about the underpinnings of intelligence: the capacity to manipulate internal representations of the environment. I argue that the hypothesis needs refinement, and that so refined, it applies only to some forms of intelligence. (shrink)

Over the last decade, AI models of language and word meaning have been dominated by what we might call a statistics-of-occurrence, strategy: these models are deep neural net structures that have been trained on a large amount of unlabeled text with the aim of producing a model that exploits statistical information about word and phrase co-occurrence in order to generate behavior that is similar to what a human might produce, or representations that can be probed to exhibit behavior similar to (...) what a human might produce (meaningsemblant behavior). Examples of what we can call Statistics-of-Occurrence Models (SOMs) include: Word2Vec (CBOW and Skip-Gram), BERT, GPT-3, and, most recently, ChatGPT. Increasingly, there have been suggestions that such systems have semantic understanding, or at least a proto-version of it. This paper argues against such claims. I argue that a necessary condition for a system to possess semantic understanding is that it function in ways that are causally explainable by appeal to its semantic properties. I then argue that SOMs do not plausibly satisfy this Functioning Criterion. Rather, the best explanation of their meaning-semblant behavior is what I call the Statistical Hypothesis: SOMs do not themselves function to represent or produce meaningful text; they just reflect the semantic information that exists in the aggregate given strong correlations between word placement and meaningful use. I consider and rebut three main responses to the claim that SOMs fail to meet the Functioning Criterion. The result, I hope, is increased clarity about why and how one should make claims about AI systems having semantic understanding. (shrink)

How all philosophical explanations of human consciousness and the fundamental structure of the cosmos are bizarre—and why that’s a good thing Do we live inside a simulated reality or a pocket universe embedded in a larger structure about which we know virtually nothing? Is consciousness a purely physical matter, or might it require something extra, something nonphysical? According to the philosopher Eric Schwitzgebel, it’s hard to say. In The Weirdness of the World, Schwitzgebel argues that the answers to these fundamental (...) questions lie beyond our powers of comprehension. We can be certain only that the truth—whatever it is—is weird. Philosophy, he proposes, can aim to open—to reveal possibilities we had not previously appreciated—or to close, to narrow down to the one correct theory of the phenomenon in question. Schwitzgebel argues for a philosophy that opens. According to Schwitzgebel’s “Universal Bizarreness” thesis, every possible theory of the relation of mind and cosmos defies common sense. According to his complementary “Universal Dubiety” thesis, no general theory of the relationship between mind and cosmos compels rational belief. Might the United States be a conscious organism—a conscious group mind with approximately the intelligence of a rabbit? Might virtually every action we perform cause virtually every possible type of future event, echoing down through the infinite future of an infinite universe? What, if anything, is it like to be a garden snail? Schwitzgebel makes a persuasive case for the thrill of considering the most bizarre philosophical possibilities. (shrink)

From artistic performances in the visual arts and in music to motor control in gymnastics, from tool use to chess and language, humans excel in a variety of skills. On the plausible assumption that skillful behavior is a visible manifestation of intelligence, a theory of intelligence—whether human or not—should be informed by a theory of skills. More controversial is the question as to whether, in order to theorize about intelligence, we should study certain skills in particular. My target is the (...) view that only a particular class/kind of skill (i.e., ‘theoretical’, or ‘intellectual’ skills, versus ‘practical’, or ‘embodied’ skills) manifests intelligence, or especially does so. I call this view ‘Intelligence Elitism’. Intelligence Elitism is pervasive in popular culture, in academic culture, as well as in psychometrics. It has, arguably, a long pedigree in philosophy (though explicit arguments for it are rare). But it is a substantive view which, if false, can promote unfairness. In this article, I defend Intelligence Socialism—the claim that intelligent behavior is everywhere skillful behavior is. I go on to isolate the best case for the further Elitist claim that, even if all skillful behavior is intelligent in some modest sense of ‘intelligence’, there is a principled difference in intelligence between theoretical, or intellectual, skills, on one side, and practical, or embodied skills, on the other, and I show it wanting. (shrink)

Theories of rational decision making often abstract away from computational and other resource limitations faced by real agents. An alternative approach known as resource rationality puts such matters front and center, grounding choice and decision in the rational use of finite resources. Anticipated by earlier work in economics and in computer science, this approach has recently seen rapid development and application in the cognitive sciences. Here, the theory of rationality plays a dual role, both as a framework for normative assessment (...) and as a source of scientific hypotheses about how mental processes in fact work. The latter project, often called rational analysis, depends for its success on a fine-grained characterization of the computational problem facing a decision maker, which may in turn depend on realistic assumptions about what the relevant agent is like. As a consequence, resource rationality involves a delicate, but often fruitful interplay between the normative and the descriptive. (shrink)

The present paper looks at one of the most thorough articles on the intelligence of GPT, research conducted by engineers at Microsoft. Although there is a great deal of value in their work, I will argue that, for familiar philosophical reasons, their methodology, ‘Black-box Interpretability’ is wrongheaded. But there is a better way. There is an exciting and emerging discipline of ‘Inner Interpretability’ (also sometimes called ‘White-box Interpretability’) that aims to uncover the internal activations and weights of models in order (...) to understand what they represent and the algorithms they implement. In my view, a crucial mistake in Black-box Interpretability is the failure to appreciate that how processes are carried out matters when it comes to intelligence and understanding. I can’t pretend to have a full story that provides both necessary and sufficient conditions for being intelligent, but I do think that Inner Interpretability dovetails nicely with plausible philosophical views of what intelligence requires. So the conclusion is modest, but the important point in my view is seeing how to get the research on the right track. Towards the end of the paper, I will show how some of the philosophical concepts can be used to further refine how Inner Interpretability is approached, so the paper helps draw out a profitable, future two-way exchange between Philosophers and Computer Scientists. (shrink)

Among the many common criticisms of the Turing test, a valid criticism concerns its scope. Intelligence is a complex and multi-dimensional phenomenon that will require testing using as many different formats as possible. The Turing test continues to be valuable as a source of evidence to support the inductive inference that a machine possesses a certain kind of intelligence and when interpreted as providing a behavioural test for a certain kind of intelligence. This paper raises the novel criticism that the (...) Turing test represents an example of Goodhart’s Law operating in the field of artificial intelligence. As one measure towards the goal of creating genuinely intelligent machines, the Turing test must not be confused with the goal itself. Moreover, the Turing test ought to be augmented such that through its use additional evidence could be secured to support the strong inference that a machine, were it to pass the Turing Test, could think like a human. (shrink)

1. 1.1. Suffering is bad. This is why, ceteris paribus, there are strong moral reasons to prevent suffering. Moreover, typically, those moral reasons are stronger when the amount of suffering at st...

Humans can attribute beliefs to others. However, it is unknown to what extent this ability results from an innate biological endowment or from experience accrued through child development, particularly exposure to language describing others' mental states. We test the viability of the language exposure hypothesis by assessing whether models exposed to large quantities of human language display sensitivity to the implied knowledge states of characters in written passages. In pre‐registered analyses, we present a linguistic version of the False Belief Task (...) to both human participants and a large language model, GPT‐3. Both are sensitive to others' beliefs, but while the language model significantly exceeds chance behavior, it does not perform as well as the humans nor does it explain the full extent of their behavior—despite being exposed to more language than a human would in a lifetime. This suggests that while statistical learning from language exposure may in part explain how humans develop the ability to reason about the mental states of others, other mechanisms are also responsible. (shrink)

This compilation of references includes all references for the knowledge-how chapters included in Bengson & Moffett's edited volume. The volume and the compilation of references may serve as a good starting point for people who are unfamiliar with the philosophical literature on knowledge-how.

David Chalmers has recently developed a novel strategy of refuting external world skepticism, one he dubs the structuralist solution. In this paper, I make three primary claims: First, structuralism does not vindicate knowledge of other minds, even if it is combined with a functionalist approach to the metaphysics of minds. Second, because structuralism does not vindicate knowledge of other minds, the structuralist solution vindicates far less worldly knowledge than we would hope for from a solution to skepticism. Third, these results (...) suggest that the problem of external world skepticism should perhaps be construed as two different problems, since the problem might turn out to require two substantively different solutions, one for knowledge of the kind that is not dependent on other minds and one for knowledge that is. (shrink)

Alan Turing’s 1950 imitation game has been widely understood as a means for testing if an entity is intelligent. Following a series of papers by Diane Proudfoot, I offer a socio-technological interpretation of Turing’s paper and present an alternative way of understanding both the imitation game and Turing’s concept of intelligence. Turing, I claim, saw intelligence as a social concept, meaning that possession of intelligence is a property determined by society’s attitude toward the entity. He realized that as long as (...) human society held a prejudiced attitude toward machinery—seeing machines a priori as mindless objects—machines could not be said to be intelligent, by definition. He also realized, though, that if humans’ a priori, chauvinistic attitude toward machinery changed, the existence of intelligent machines would become logically possible. Turing thought that such a change would eventually occur: He believed that when scientists overcome the technological challenge of constructing sophisticated machines that could imitate human verbal behavior—i.e., do well in the imitation game—humans’ prejudiced attitude toward machinery will have altered in such a way that machines could be said to be intelligent. The imitation game, for Turing, was not an intelligence test, but a technological aspiration whose realization would likely involve a change in society’s attitude toward machines. (shrink)

Computational complexity is a discipline of computer science and mathematics which classifies computational problems depending on their inherent difficulty, i.e. categorizes algorithms according to their performance, and relates these classes to each other. P problems are a class of computational problems that can be solved in polynomial time using a deterministic Turing machine while solutions to NP problems can be verified in polynomial time, but we still do not know whether they can be solved in polynomial time as well. A (...) solution for the so-called NP-complete problems will also be a solution for any other such problems. Its artificial-intelligence analogue is the class of AI-complete problems, for which a complete mathematical formalization still does not exist. In this chapter we will focus on analysing computational classes to better understand possible formalizations of AI-complete problems, and to see whether a universal algorithm, such as a Turing test, could exist for all AI-complete problems. In order to better observe how modern computer science tries to deal with computational complexity issues, we present several different deep-learning strategies involving optimization methods to see that the inability to exactly solve a problem from a higher order computational class does not mean there is not a satisfactory solution using state-of-the-art machine-learning techniques. Such methods are compared to philosophical issues and psychological research regarding human abilities of solving analogous NP-complete problems, to fortify the claim that we do not need to have an exact and correct way of solving AI-complete problems to nevertheless possibly achieve the notion of strong AI. (shrink)

The last 5 years have seen a series of remarkable achievements in deep-neural-network-based artificial intelligence research, and some modellers have argued that their performance compares favourably to human cognition. Critics, however, have argued that processing in deep neural networks is unlike human cognition for four reasons: they are (i) data-hungry, (ii) brittle, and (iii) inscrutable black boxes that merely (iv) reward-hack rather than learn real solutions to problems. This article rebuts these criticisms by exposing comparative bias within them, in the (...) process extracting some more general lessons that may also be useful for future debates. (shrink)

I show that some of the most initially attractive routes of refuting epistemological solipsism face serious obstacles. I also argue that for creatures like ourselves, solipsism is a genuine form of external world skepticism. I suggest that together these claims suggest the following morals: No proposed solution to external world skepticism can succeed which does not also solve the problem of epistemological solipsism. And, more tentatively: In assessing proposed solutions to external world skepticism, epistemologists should explicitly consider whether those solutions (...) extend to knowledge of other minds. Finally, and also tentatively: epistemological solipsism warrants more philosophical attention than it currently enjoys. (shrink)

Artificial Neural Networks have reached “grandmaster” and even “super-human” performance across a variety of games, from those involving perfect information, such as Go, to those involving imperfect information, such as “Starcraft”. Such technological developments from artificial intelligence (AI) labs have ushered concomitant applications across the world of business, where an “AI” brand-tag is quickly becoming ubiquitous. A corollary of such widespread commercial deployment is that when AI gets things wrong—an autonomous vehicle crashes, a chatbot exhibits “racist” behavior, automated credit-scoring processes (...) “discriminate” on gender, etc.—there are often significant financial, legal, and brand consequences, and the incident becomes major news. As Judea Pearl sees it, the underlying reason for such mistakes is that “... all the impressive achievements of deep learning amount to just curve fitting.” The key, as Pearl suggests, is to replace “reasoning by association” with “causal reasoning” —the ability to infer causes from observed phenomena. It is a point that was echoed by Gary Marcus and Ernest Davis in a recent piece for theNew York Times: “we need to stop building computer systems that merely get better and better at detecting statistical patterns in data sets—often using an approach known as ‘Deep Learning’—and start building computer systems that from the moment of their assembly innately grasp three basic concepts: time, space, and causality.” In this paper, foregrounding what in 1949 Gilbert Ryle termed “a category mistake”, I will offer an alternative explanation for AI errors; it is not so much that AI machinery cannot “grasp” causality, but that AI machinery (qua computation) cannot understand anything at all. (shrink)

Disertační práce se zabývá problematikou připisování myšlení jiným entitám, a to pomocí imitační hry navržené v roce 1950 britským filosofem Alanem Turingem. Jeho kritérium, známé v dějinách filosofie jako Turingův test, je podrobeno detailní analýze. Práce popisuje nejen původní námitky samotného Turinga, ale především pozdější diskuse v druhé polovině 20. století. Největší pozornost je věnována těmto kritikám: Lucasova matematická námitka využívající Gödelovu větu o neúplnosti, Searlův argument čínského pokoje konstatující nedostatečnost syntaxe pro sémantiku, Blockův návrh na použití brutální síly pro (...) řešení imitační hry, Frenchova teorie subkognitivních informací a Michieho skepticismus ohledně možnosti umělého vědomí. Závěr práce zachycuje současný stav recepce Turingova testu a představuje pokusy o jeho praktickou realizaci, například v každoroční soutěži o Loebnerovu cenu. Autor práce zastává názor, že ani po více než šedesáti letech od uveřejnění Turingova paradigmatického eseje stále neexistují žádné vážné důvody pro zamítnutí jeho tvrzení. Tradiční komputační funkcionalismus možná není ideální teorií vysvětlující činnost myslí a jako slibnější se může jevit vývoj v neurálních vědách, ale Turingův test je přesto užitečným a snad i jediným nástrojem pro detekci inteligence u lidmi vytvořených strojů. (shrink)

The goal of the paper is to develop and propose a general model of the state space of AI. Given the breathtaking progress in AI research and technologies in recent years, such conceptual work is of substantial theoretical interest. The present AI hype is mainly driven by the triumph of deep learning neural networks. As the distinguishing feature of such networks is the ability to self-learn, self-learning is identified as one important dimension of the AI state space. Another dimension is (...) recognized as generalization, the possibility to go over from specific to more general types of problems. A third dimension is semantic grounding. Our overall analysis connects to a number of known foundational issues in the philosophy of mind and cognition: the blockhead objection, the Turing test, the symbol grounding problem, the Chinese room argument, and use theories of meaning. It shall finally be argued that the dimension of grounding decomposes into three sub-dimensions. And the dimension of self-learning turns out as only one of a whole range of “self-x-capacities” (based on ideas of organic computing) that span the self-x-subspace of the full AI state space. (shrink)

In the 70 years since Alan Turing’s ‘Computing Machinery and Intelligence’ appeared in Mind, there have been two widely-accepted interpretations of the Turing test: the canonical behaviourist interpretation and the rival inductive or epistemic interpretation. These readings are based on Turing’s Mind paper; few seem aware that Turing described two other versions of the imitation game. I have argued that both readings are inconsistent with Turing’s 1948 and 1952 statements about intelligence, and fail to explain the design of his game. (...) I argue instead for a response-dependence interpretation. This interpretation has implications for Turing’s view of free will: I argue that Turing’s writings suggest a new form of free will compatibilism, which I call response-dependence compatibilism. The philosophical implications of rethinking Turing’s test go yet further. It is assumed by numerous theorists that Turing anticipated the computational theory of mind. On the contrary, I argue, his remarks on intelligence and free will lead to a new objection to computationalism. (shrink)

Turing used the expression “emotional” in three distinct ways: to state his philosophical theory of the concept of intelligence, to classify arguments for and against the possibility of machine intelligence, and to describe the education of a “child machine”. The remarks on emotion include several of the most important philosophical claims. This paper analyses these remarks and their significance for current research in Artificial Intelligence.

Objections to the computational theory of cognition, inspired by twentieth century phenomenology, have tended to fixate on the embodiment and embeddedness of intelligence. In this paper I reconstruct a line of argument that focusses primarily on the abstract nature of scientific models, of which computational models of the brain are one sort. I observe that the critique of scientific abstraction was rather commonplace in the philosophy of the 1920s and 30s and that attention to it aids the reading of The (...) Organism ([1934] 1939) by the neurologist Kurt Goldstein. With this background in place, we see that some brief but spirited criticisms of cybernetics by two later thinkers much influenced by Goldstein, Georges Canguilhem (1963) and Maurice Merleau-Ponty (1961), show continuity with the earlier discussions of abstraction in science. (shrink)

Many debates in philosophy focus on whether folk or scientific psychological notions pick out cognitive natural kinds. Examples include memory, emotions and concepts. A potentially interesting type of kind is: kinds of mental representations (as opposed, for example, to kinds of psychological faculties). In this chapter we outline a proposal for a theory of representational kinds in cognitive science. We argue that the explanatory role of representational kinds in scientific theories, in conjunction with a mainstream approach to explanation in cognitive (...) science, suggest that representational kinds are multi-level. This is to say that representational kinds’ properties cluster at different levels of explanation and allow for intra- and inter-level projections. (shrink)