Vincent C. Müller

The central concept of this edited volume is "blended cognition", the natural skill of human beings for combining constantly different heuristics during their several task-solving activities. Something that was sometimes observed like a problem as “bad reasoning”, is now the central key for the understanding of the richness, adaptability and creativity of human cognition. The topic of this book connects in a significant way with the disciplines of psychology, neurology, anthropology, philosophy, logics, engineering, logics, and AI. In a nutshell: understanding better humans for designing better machines. It contains a Preface by the editors and 12 chapters.

Artificial intelligence (AI) and robotics are digital technologies that will have significant impact on the development of humanity in the near future. They have raised fundamental questions about what we should do with these systems, what the systems themselves should do, what risks they involve, and how we can control these. - After the Introduction to the field (§1), the main themes (§2) of this article are: Ethical issues that arise with AI systems as objects, i.e., tools made and used by humans. This includes issues of privacy (§2.1) and manipulation (§2.2), opacity (§2.3) and bias (§2.4), human-robot interaction (§2.5), employment (§2.6), and the effects of autonomy (§2.7). Then AI systems as subjects, i.e., ethics for the AI systems themselves in machine ethics (§2.8) and artificial moral agency (§2.9). Finally, the problem of a possible future AI superintelligence leading to a “singularity” (§2.10). We close with a remark on the vision of AI (§3). - For each section within these themes, we provide a general explanation of the ethical issues, outline existing positions and arguments, then analyse how these play out with current technologies and finally, what policy consequences may be drawn.

Die Entwicklungen in der Künstlichen Intelligenz (KI) sind spannend. Aber wohin geht die Reise? Ich stelle eine Analyse vor, der zufolge exponentielles Wachstum von Rechengeschwindigkeit und Daten die entscheidenden Faktoren im bisherigen Fortschritt waren. Im Folgenden erläutere ich, unter welchen Annahmen dieses Wachstum auch weiterhin Fortschritt ermöglichen wird: 1) Intelligenz ist eindimensional und messbar, 2) Kognitionswissenschaft wird für KI nicht benötigt, 3) Berechnung (computation) ist hinreichend für Kognition, 4) Gegenwärtige Techniken und Architektur sind ausreichend skalierbar, 5) Technological Readiness Levels (TRL) erweisen sich als machbar. Diese Annahmen werden sich als dubios erweisen.

This book reports on the results of the third edition of the premier conference in the field of philosophy of artificial intelligence, PT-AI 2017, held on November 4 - 5, 2017 at the University of Leeds, UK. It covers: advanced knowledge on key AI concepts, including complexity, computation, creativity, embodiment, representation and superintelligence; cutting-edge ethical issues, such as the AI impact on human dignity and society, responsibilities and rights of machines, as well as AI threats to humanity and AI safety; and cutting-edge developments in techniques to achieve AI, including machine learning, neural networks, dynamical systems. The book also discusses important applications of AI, including big data analytics, expert systems, cognitive architectures, and robotics. It offers a timely, yet very comprehensive snapshot of what is going on in the field of AI, especially at the interfaces between philosophy, cognitive science, ethics and computing.

While it is often said that robotics should aspire to reproducible and measurable results that allow benchmarking, I argue that a focus on benchmarking can be a hindrance for progress in robotics. The reason is what I call the ‘measure-target confusion’, the confusion between a measure of progress and the target of progress. Progress on a benchmark (the measure) is not identical to scientific or technological progress (the target). In the past, several academic disciplines have been led into pursuing only reproducible and measurable ‘scientific’ results – robotics should be careful to follow that line because results that can be benchmarked must be specific and context-dependent, but robotics targets whole complex systems for a broad variety of contexts. While it is extremely valuable to improve benchmarks to reduce the distance be- tween measure and target, the general problem to measure progress towards more intelligent machines (the target) will not be solved by benchmarks alone; we need a balanced approach with sophisticated benchmarks, plus real-life testing, plus qualitative judgment.

The contribution of the body to cognition and control in natural and artificial agents is increasingly described as “off-loading computation from the brain to the body”, where the body is said to perform “morphological computation”. Our investigation of four characteristic cases of morphological computation in animals and robots shows that the ‘off-loading’ perspective is misleading. Actually, the contribution of body morphology to cognition and control is rarely computational, in any useful sense of the word. We thus distinguish (1) morphology that facilitates control, (2) morphology that facilitates perception and the rare cases of (3) morphological computation proper, such as ‘reservoir computing.’ where the body is actually used for computation. This result contributes to the understanding of the relation between embodiment and computation: The question for robot design and cognitive science is not whether computation is offloaded to the body, but to what extent the body facilitates cognition and control – how it contributes to the overall ‘orchestration’ of intelligent behaviour.

The European Association for Cognitive Systems is the association resulting from the EUCog network, which has been active since 2006. It has ca. 1000 members and is currently chaired by Vincent C. Müller. We ran our annual conference on December 08-09 2016, kindly hosted by the Technical University of Vienna with Markus Vincze as local chair. The invited speakers were David Vernon and Paul F.M.J. Verschure. Out of the 49 submissions for the meeting, we accepted 18 a papers and 25 as posters (after double-blind reviewing). Papers are published here as “full papers” or “short papers” while posters are published here as “short papers” or “abstracts”. Some of the papers presented at the conference will be published in a separate special volume on ‘Cognitive Robot Architectures’ with the journal Cognitive Systems Research. - RC, VCM, YS, MV.

The background to this paper is that in our world of massively increasing personal digital data any control over the data about me seems illusionary – informational privacy seems a lost cause. On the other hand, the production of this digital data seems a necessary component of our present life in the industrialized world. A framework for a resolution of this apparent dilemma is provided if by the distinction between (meaningless) data and (meaningful) information. I argue that computational data processing is necessary for many present-day processes and not a breach of privacy, while collection and processing of private information is often not necessary and a breach of privacy. The problem and the sketch of its solution are illustrated in a case-study: supermarket customer cards.

Floridi and Taddeo propose a condition of “zero semantic commitment” for solutions to the grounding problem, and a solution to it. I argue briefly that their condition cannot be fulfilled, not even by their own solution. After a look at Luc Steels' very different competing suggestion, I suggest that we need to re-think what the problem is and what role the ‘goals’ in a system play in formulating the problem. On the basis of a proper understanding of computing, I come to the conclusion that the only sensible ground-ing problem is how we can explain and re-produce the behavioral ability and function of meaning in artificial computational agents

Concordance of the poetic works of Giorgos Seferis which presents all the principal “words” of the texts in an alphabetical list, stating how often each word occurs, giving a precise location and a relevant piece of text for each occurrence. We found ca. 9500 different Greek words in 39000 different occurrences, so our concordance has 50.000 lines of text. The technical procedure required four main steps: text entry and tagging, production of the concordance, correction of the contexts, formatting for print.

Die gegenwärtig unter dem Titel ›Realismus‹ geführten Debatten in der Philosophie befinden sich nach allgemeiner Ansicht in einem Zustand größter Verwirrung, so daß es nützlich erscheint, ein wenig Ordnung in die theoretischen Optionen zu bringen bevor man für die eine oder andere Auffassung Partei ergreift. In der vorliegenden Arbeit wird dafür argumentiert, daß sich ein systematisch zusammenhängendes Zentrum dieser Debatten mit Hilfe des Begriffes der Referenz ordnen läßt. Nach der Analyse einiger klassischer Positionen soll ein Rahmen erstellt werden, innerhalb dessen die Positionen eingeordnet und die zentralen Probleme fruchtbar diskutiert werden können. Zu diesem Zwecke ist es erforderlich, für die Einordnung der Positionen theoretische Kriterien zu benennen, die sich an den Problemen orientieren von denen hier argumentiert wird, das sie zentral seien. Sehr knapp ausgedrückt wird hier die Auffassung vertreten, Realismus sei als metaphysische These aufzufassen, welche eine von menschlichen epistemischen und semantischen Möglichkeiten unabhängige Existenz behauptet; in den gegenwärtigen Debatten typischerweise die Existenz einer Art von Dingen, nicht die eines individuellen Gegenstandes. Diese metaphysische These wird jedoch in der gegenwärtigen Debatte mit semantischen Argumenten untermauert, bzw. angegriffen und jene semantischen Argumente wiederum verwenden epistemische Erwägungen – die Frage betreffend, was man wissen kann und was nicht. Der Beginn der gegenwärtigen Realismusdebatten mit der soeben skizzierten zentralen Stellung semantischer Argumente ist, der hier vertretenen Auffassung zufolge, die Kritik am traditionellen Fregeschen Referenzbegriff durch Kripke und den frühen Putnam um 1970 (Kap. 2). Aus dieser neu zu bewertenden Kritik und der von den Autoren daraus entwickelten externalistischen realistischen Semantik für Artausdrücke läßt sich das erste Kriterium für eine Position in einer Realismusdebatte ableiten und klären (Kap. 3): Hält man die fragliche Art für eine natürliche Art und meint also, sie habe ihren ›Zusammenhalt‹ von Natur aus? Wenn ja, dann ist die fragliche Position eine realistische (das Kriterium ist dank des engen Zusammenhangs von Realismus und Externalismus sowohl hinreichend als auch notwendig). Hält man die Art nicht für eine natürliche Art, ist man Antirealist. Damit geht jeweils not- wendig eine bestimmte Semantik für den auf die Art referierenden Ausdruck einher. Es zeigt sich zugleich, daß es zwei Varianten des Realismus zu unterscheiden gilt, die hier als klassischer bzw. als moderater Realismus bezeichnet werden. Im folgenden (Kap. 4.1) wird argumentiert, daß der Begriff der Wahrheit nicht eigentlich der zentrale Punkt in den in Frage stehenden Realismusdebatten sein sollte, wie vielfach behauptet wurde, sondern seine Brisanz vielmehr aus zugrundeliegenden semantischen Fragen gewinnt die also in der weiteren Aufdeckung der Kriterien die zentrale Rolle spielen müssen. In der Analyse der Kritik an der in Kap. 3 entwickelten Position des ›klassischen Realismus‹ läßt sich ein zweites unterscheidendes Kriterium für Positionen in Realismusdebatten entwickeln: begriffliche Relativität. Nach der Ablehnung von Putnams Auffassungen zu diesem Thema, werden zwei Varianten vorgeschlagen, starke und schwache begriffliche Relativität (Kap. 4.2). Die Anwendung dieses Merkmals zwingt, so wird argumentiert, in einigen Fällen zu einer Kombination einer realistischen Auffassung von Arten als natürliche mit begrifflicher Relativität. Diese mittlere Position zwischen klassischem Realismus und Antirealismus wird als »moderater Realismus« bezeichnet. Der im ersten Kriterium verwendete Begriff der natürlichen Art, und damit die Optionen in den Debatten, wird schließlich mittels einer Dis- kussion des Phänomens der Vagheit noch weiter verdeutlicht (Kap. 4.3). Ab- schließend werden die drei Optionen zusammenfassend dargestellt und ein Versuch unternommen, ihre Fruchtbarkeit für diverse Realismusdebatten anzudeuten.

This paper investigates a problem about freedom of information. Although freedom of information is generally considered desirable, there are a number of areas where there is substantial agreement that freedom of information should be limited. After a certain ordering of the landscape, I argue that we need to add the category of "dangerous" information and that this category has gained a new quality in the context of current information technology, specifically the Internet. This category includes information the use of which would be morally wrong as well as some of what may be called "corrupting" information. Some such information should not be spread at all and some should be very limited in its spread.

This paper investigates the prospects of Rodney Brooks’ proposal for AI without representation. It turns out that the supposedly characteristic features of “new AI” (embodiment, situatedness, absence of reasoning, and absence of representation) are all present in conventional systems: “New AI” is just like old AI. Brooks proposal boils down to the architectural rejection of central control in intelligent agents—Which, however, turns out to be crucial. Some of more recent cognitive science suggests that we might do well to dispose of the image of intelligent agents as central representation processors. If this paradigm shift is achieved, Brooks’ proposal for cognition without representation appears promising for full-blown intelligent agents—Though not for conscious agents.

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

In this paper we address the issue of grounding for experiential concepts. Given that perceptual demonstratives are a basic form of such concepts, we examine ways of fixing the referents of such demonstratives. To avoid ‘encodingism’, that is, relating representations to representations, we postulate that the process of reference fixing must be bottom-up and nonconceptual, so that it can break the circle of conceptual content and touch the world. For that purpose, an appropriate causal relation between representations and the world is needed. We claim that this relation is provided by spatial and object-centered attention that leads to the formation of object files through the function of deictic acts. This entire causal process takes place at a pre-conceptual level, meeting the requirement for a solution to the grounding problem. Finally we claim that our account captures fundamental insights in Putnam’s and Kripke’s work on “new” reference.

In October 2011, the “2nd European Network for Cognitive Systems, Robotics and Interaction”, EUCogII, held its meeting in Groningen on “Autonomous activity in real-world environments”, organized by Tjeerd Andringa and myself. This is a brief personal report on why we thought autonomy in real-world environments is central for cognitive systems research and what I think I learned about it. --- The theses that crystallized are that a) autonomy is a relative property and a matter of degree, b) increasing autonomy of an artificial system from its makers and users is a necessary feature of increasingly intelligent systems that can deal with the real-world and c) more such autonomy means less control but at the same time improved interaction with the system

In this paper I want to propose an argument to support Jerry Fodor’s thesis (Fodor 1983) that input systems are modular and thus informationally encapsulated. The argument starts with the suggestion that there is a “grounding problem” in perception, i. e. that there is a problem in explaining how perception that can yield a visual experience is possible, how sensation can become meaningful perception of something for the subject. Given that visual experience is actually possible, this invites a transcendental argument that explains the conditions of its possibility. I propose that one of these conditions is the existence of a visual module in Fodor’s sense that allows the step from sensation to object-identifying perception, thus enabling visual experience. It seems to follow that there is informationally encapsulated nonconceptual content in visual perception.

Wenn Präsident Kennedy nicht erschossen worden wäre, hätte er dann Nordvietnam bombardiert? Das weiß Gott allein. Oder doch nicht? Weiß wenigstens Er, was Kennedy getan hätte?... Die Jesuiten behaupteten unter anderem, daß viele menschliche Handlungen in dem Sinne frei seien, daß die Ausführenden nicht logisch oder kausal gezwungen seien, sie auszuführen. („Frei“ wird im vorliegenden Aufsatz stets in diesem Sinne verwendet werden.) Wie behält Gott dann die Kontrolle über die menschliche Geschichte? Nicht dadurch, daß Er menschliche Handlungen kausal determiniert, wie die Dominikaner geglaubt zu haben scheinen, sondern indem Er Umstände herbeiführt, von denen Er weiß, daß wir in ihnen freiwillig Seinen Plänen entsprechend handeln werden.

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

We discuss at some length evidence from the cognitive science suggesting that the representations of objects based on spatiotemporal information and featural information retrieved bottomup from a visual scene precede representations of objects that include conceptual information. We argue that a distinction can be drawn between representations with conceptual and nonconceptual content. The distinction is based on perceptual mechanisms that retrieve information in conceptually unmediated ways. The representational contents of the states induced by these mechanisms that are available to a type of awareness called phenomenal awareness constitute the phenomenal content of experience. The phenomenal content of perception contains the existence of objects as separate things that persist in time and time, spatiotemporal information, and information regarding relative spatial relations, motion, surface properties, shape, size, orientation, color, and their functional properties.

Just as AI has moved away from classical AI, human-computer interaction (HCI) must move away from what I call ‘good old fashioned HCI’ to ‘new HCI’ – it must become a part of cognitive systems research where HCI is one case of the interaction of intelligent agents (we now know that interaction is essential for intelligent agents anyway). For such interaction, we cannot just ‘analyze the data’, but we must assume intentions in the other, and I suggest these are largely recognized through resistance to carrying out one’s own intentions. This does not require fully cognitive agents but can start at a very basic level. New HCI integrates into cognitive systems research and designs intentional systems that provide resistance to the human agent.

This is the short version, in French translation by Anne Querrien, of the originally jointly authored paper: Müller, Vincent C., ‘Autonomous killer robots are probably good news’, in Ezio Di Nucci and Filippo Santoni de Sio, Drones and responsibility: Legal, philosophical and socio-technical perspectives on the use of remotely controlled weapons. - - - L’article qui suit présente un nouveau système d’armes fondé sur des robots qui risque d’être prochainement utilisé. À la différence des drones qui sont manoeuvrés à distance mais comportent une part importante de discernement humain, il s’agit de machines programmées pour défendre, attaquer, ou tuer de manière autonome. Les auteurs, philosophes, préfèrent prévenir de leur prochaine diffusion et obtenir des Nations Unies leur régulation. Une campagne internationale propose plutôt leur interdiction.

From a European perspective the US debate about gun control is puzzling because we have no such debate: It seems obvious to us that dangerous weapons need tight control and that ‘guns’ fall under that category. I suggest that this difference occurs due to different habits that generate different attitudes and support this explanation with an analogy to the habits about knives. I conclude that it is plausible that individual knife-people or gun-people do not want tight regulatory legislation—but tight knife and gun legislation is morally obligatory anyway. We need to give up our habits for the greater good.

Review of Joseph LeDoux: Das Netz der Persönlichkeit. Wie unser Selbst entsteht. Walter Verlag, Düsseldorf 2003. 510 Seiten (mit Abbildungen), 39,90 Euro. - Der eine Mensch ist mißtrauisch, der nächste leichtgläubig, diese ist warmherzig, jene kaltschnäuzig. Viele haben Charakter, manche sogar Persönlichkeit. Wie kommt es dazu? In seinem neuen Buch untersucht der Neurowissenschaftler Joseph LeDoux wie unser Selbst entsteht. In dem sehr lesbaren und angenehm übersetzten Werk wird anschaulich und detailliert berichtet, wie sich in unserem Gehirn die Charakteristika eines Individuums ausbilden.

I want to suggest that the major influence of classical arguments for embodiment like "The Embodied Mind" by Varela, Thomson & Rosch (1991) has been a changing of positions rather than a refutation: Cognitivism has found ways to retreat and regroup at positions that have better fortification, especially when it concerns theses about artificial intelligence or artificial cognitive systems. For example: a) Agent-based cognitivism' that understands humans as taking in representations of the world, doing rule-based processing and then acting on them (sense-plan-act) is often limited to conscious decision processes; and b) Purely syntactic cognition is compatible with embodiment, or supplemented by embodiment (e.g. for 'grounding'). While the empirical thesis of embodied cognition ('embodied cognitive science') is true and the practical engineering thesis ('morphological computation', 'cheap design') is often true, the conceptual thesis ('embodiment is necessary for cognition') is likely false - syntax is often enough for cognition, unless grounding is really necessary. I conclude that it has become more sensible to integrate embodiment with traditional approaches rather than "fight for embodiment" or "against cognitivism".

Ποια είναι η πατρίδα κάποιων; Ας δούμε το θέμα λίγο σαν πρόβλημα της φιλοσοφίας της γλώσσας: για ένα κατηγόρημα «Άγγλος» «Γάλλος», «Πορτογάλος», «Βέλγος», «Φλαμανδός», πώς να αποφασίσουμε ποια αντικείμενα (ποιοι άνθρωποι) εμπίπτουν σε ποιο κατηγόρημα; Έχουν τελικά νόημα αυτά τα κατηγορήματα; Η χρήση και κατάχρηση αυτών των κατηγορημάτων έχει αποτελέσει μια από τις κυριότερες πηγές δυστυχίας στη διάρκεια των δύο τελευταίων αιώνων και συνεχίζει να είναι στην ημερήσια διάταξη. Συνήθως εμφανίζεται στα πλαίσια αγώνων για την απόκτηση «ελευθερίας» σε αντιπαράθεση με την «καταπίεση», και σε παγκόσμιο επίπεδο και κοντύτερα μας: στη Μακεδονία, την Καταλονία, την Ιρλανδία, το Κουρδιστάν ... Νομίζω πώς αρκετά από αυτά βασίζονται σε λάθη.

The nature of vagueness is investigated via a preliminary definition and a discussion of the classical sorites paradox ; this is carried further by asking for the origins of vagueness and a critique of several attempts to remove it from language. It is shown that such attempts are ill motivated and doomed for failure since vagueness is not just a matter of ignorance but firmly grounded in epistemic and metaphysical facts. Finally, the philosophical interest of real vagueness is illustrated by the concept of “natural kind”, which is essential to realism/anti-realism debates.