Dissertations / Theses on the topic 'Philosophy of AI'

The work I have presented here pulls together Maori epistemologies as evidenced in the whakapapa knowledge particularly of Ngati Maniapoto to see if and where connection lies with understandings of Maori cultural wellbeing. Whakapapa knowledge is the unbounded collection of theory, observation and experience as seen through Maori eyes. It is intricately connected by whakapapa, a tool for working with and extrapolating understanding and is the common thread that binds hapu, whanau and iwi (O’Regan, 2001). The aim is to investigate contemporary Maori realities with a strong interest in these traditions of wisdom and knowing. The rangahau presented here is of necessity both deconstructive and reconstructive. As a deconstructive project the rangahau seeks to place under the microscope of indigenous gaze the colonial theoretical, ethical, moral and political construction of Maori ways of knowing and being and the ontological orders of western paradigms and non-Maori worldview (Romero-Little, 2006). As a constructive project I am concerned with placing on the record and opening up sites for, but not defining, Maori epistemology as legitimate and ‘tika’ and at the same time putting forward ‘alternate epistemologies’ (Collins, 1991; Lopez, 1998; Smith, 1999; Marsden, 2003) that challenge certified knowledge and critically challenges dominant constructions of the truth as related to knowing. The implications of these explorations of epistemologies for Maori lives, opportunities and experience are also considered. This work argues for the maintenance of Maori cultural identities via whakapapa knowledge using connections to Maori ways of knowing. This includes examination of the effects of coming to terms with, of encountering, coming to terms with and engaging with Maori cultural practices, as well as, processes commonly referred to as ‘culture shock’ (Weaver, 1993) the psychological, emotional and physical responses to the phenomenon of identity reclamation and how these realities can be negotiated. What I found is that Maori knowledge systems are replete with elements that contribute positively to the maintenance of cultural identities and these identities are uniquely and distinctively contextually and culturally relevant. These systems have been and continue to be threatened by the impacts of colonisation and colonial ideologies. The work has found that elders and relevant contexts retain and provide a large volume of knowledge that when engaged with can provide useful insights into living within Maori paradigms that can enhance wellbeing in the present. Maori communities and whanau are under high levels of stress with the pressures of contemporary living and the dis-location from ancestral lands, and the living activities, knowledge sharing opportunities and learning practices they support. This work seeks to offer up solutions via the maintenance, enhancement and advancement of cultural identities as a way for mediating and removing some of the effects of the stresses. The implications are that the continued disconnection of Maori from unique cultural identities informed by whakapapa korero knowledge may serve to weaken important elements and connections to an individual’s and group’s cultural identity, including personal history, stories, land and people. The potential exists for further investigation of how crucial cultural connections that acknowledge contemporary realities and yet support the maintenance of cultural identities with strong and vibrant connections to whakapapa korero knowledge connections might be maintained, enhanced and advanced. Additionally, the work here opens up the space for and advocates for much deeper exploration of distinctive elements of a groups identity through contextually located knowledge in forms such as waiata, purakau, pakiwaitara, whakairo, rongoa, wairua and the many other knowledge forms of tea o Maori to further depths/heights not yet achieved to reclaim (k)new and subjugated knowledge forms. This potential is exciting but there are a range of risks involved (including appropriations of indigenous knowledge) that requires certain minimum standards of knowledge protection such as discerning which knowledge is suitable for public consumption and that which is not. This is most suitably done after receiving guidance from the knowledge holders as to what the appropriate forums for such knowledge might be and analysing risks for abuse, risks of misinterpretation and risks of unintended use that might cause whakama. The enquiry suggested above as being of benefit is of course a deeply personal exploration and ideas of what is appropriate for public consumption and what is not is something that must be explored at the time of enquiry. As in my work here I was asked to include some things and to exclude others as a result of views by the elders that the public consumption of some knowledge they contributed was inappropriate to be shared beyond our korero because it could be perceived in a number of ways, some helpful and some not, for the people concerned, or for different groups of people. The knowledge that has been shared here and that which has not has therefore been discerned.

The thesis investigates how one should determine the appropriate transparency of an information processing system from a receiver perspective. Research in the past has suggested that the model should be maximally transparent for what is labeled as ”high stake decisions”. Instead of motivating the choice of a model’s transparency on the non-rigorous criterion that the model contributes to a high stake decision, this thesis explores an alternative method. The suggested method involves that one should let the transparency depend on how well an explanation of the model’s output satisfies the purpose of an explanation. As a result, we do not have to bother if it is a high stake decision, we should instead make sure the model is sufficiently transparent to provide an explanation that satisfies the expressed purpose of an explanation.

Vilken grundläggande idé utgår vi ifrån när vi talar om förnuft och rationalitet och vilka konsekvenser får detta för synen på individen, individen i samhället samt individen i datasystemet? Uppsaten belyser problematik rörande den icke-situerade individen, vilken inte är en individ utan snarare en skenbar produkt utav homogena värderingar.

<p>Are agency and responsibility solely ascribable to humans? The advent of artificial intelligence (AI), including the development of so-called “affective computing,” appears to be chipping away at the traditional building blocks of moral agency and responsibility. Spurred by the realization that fully autonomous, self-aware, even rational and emotionally-intelligent computer systems may emerge in the future, professionals in engineering and computer science have historically been the most vocal to warn of the ways in which such systems may alter our understanding of computer ethics. Despite the increasing attention of many philosophers and ethicists to the development of AI, there continues to exist a fair amount of conceptual muddiness on the conditions for assigning agency and responsibility to such systems, from both an ethical and a legal perspective. Moral and legal philosophies may overlap to a high degree, but are neither interchangeable nor identical. This paper attempts to clarify the actual and hypothetical ethical and legal situations governing a very particular type of advanced, or “intelligent,” computer system: medical decision support systems (MDSS) that feature AI in their system design. While it is well-recognized that MDSS can be categorized by type and function, further categorization of their mediating effects on users and patients is needed in order to even begin ascribing some level of moral or legal responsibility. I conclude that various doctrines of Anglo legal systems appear to allow for the possibility of assigning specific types of agency – and thus specific types of legal responsibility – to some types of MDSS. Strong arguments for assigning moral agency and responsibility are still lacking, however.</p>

Within the interdisciplinary field of new materialism Karen Barad’s theory of agential realism deconstructs our current euro-western metaphysical perception of the world and our existence within it, to then re-build an understanding based on relatively new findings within quantum physics. In this thesis I try to recreate Barad’s theory to see what ethical consequences might come from it. Together with practical examples within the discourse of today’s social world and our global connectedness I hope to create a better understanding of the impact of our actions and being on our culture and what we call the natural world. Removing the unique agency given to human culture and language to instead, with the help of post-humanistic ideas, add agency as a universal enactment rather than an attribute, we should start to see ourselves as active and real parts of the world-building that is our home. One main question that I see arise in the end is: what does responsibility entail when we all are one and the same?

Made available in DSpace on 2016-04-29T14:23:49Z (GMT). No. of bitstreams: 1 Luciana Akemi Nakabayashi.pdf: 390381 bytes, checksum: a5f04f61e556536db3c6dfab97227bb0 (MD5) Previous issue date: 2009-05-12<br>Coordenação de Aperfeiçoamento de Pessoal de Nível Superior<br>The current research has as central theme investigate the concept of intelligence, and specifically of Artificial Intelligence (AI), derived from the Computer Science and its effect on research of Philosophy of Mind. It tries to consider and to understand such perspective to the light of Technoscience and Cybernetics, in view of the concepts that look for to understanding the human mind by the imitation of its behavior, applying the concepts to the investigation and simulation of dialogues: called Chatterbots. The research stars from a methodological and bibliographic study diverse from the concepts of Artificial Intelligence (AI) and the concepts of Philosophy of Mind, focusing on the prominence of the issue in its interdisciplinary aspects. Discusses the prospects consolidated in the community about the issue, especially the approaches of Searle (1984, 1992, 1997), Gardner (1994, 1995) and, in the Brazilian landscape, from Teixeira (1990, 1994, 1995, 1998, 2000, 2004, 2008), focusing on relations between the concepts of intent, brain, mental and cognitive experiment, as well as some trends of criticism and defense of its formal limits. Drawing up of previous searches on the subject and methodology that merges computing elements and philosophy of mind, analyzes the conceptual experiment of the Chinese Room of Searle (in aspects of syntax and semantics). From this experiment and its results, it presents the prospect of Artificial Intelligence (AI) investigation as belonging to the nature of cognition, supported by the theory of cognition, particularly for phenomena such as categorization and identification of objects, problem solving, decision and conscience. The search culminates in the analysis of the concept of Artificial Intelligence (AI) in Philosophy of Mind, proposed as a noematic element, being constituted as prerequisite for the production of scientific knowledge, implemented in the areas of Intelligence Technology and Digital Design. Applies the concepts and achievements to the definition and extension of the concept of hypertext, taking it as a computational mechanism capable of structuring hierarchical dialogues, indexed, so as to the formulation of Robot Primo (2001) and Roth Coelho (2001). Finally, the assumptions investigated are shown in results observed in the so-called Chatterbots in the area of Artificial Intelligence (AI), highlighting its characteristics and its importance in the current context of its computer utilization in cyberspace<br>A presente pesquisa tem como tema central investigar o conceito de Inteligência e, especificamente, de Inteligência Artificial (IA), derivado das Ciências da Computação e sua repercussão nas pesquisas da Filosofia da Mente. Procura conceituar e entender tal perspectiva à luz da Tecnociência e da Cibernética, tendo em vista os conceitos que buscam compreender a mente humana pela imitação de seu comportamento, aplicando os conceitos à investigação e simulação de diálogos: os chamados chatterbots. A pesquisa parte de um estudo metodológico e bibliográfico diversificado dos conceitos de Inteligência Artificial (IA) e das concepções de Filosofia da Mente, enfocando a proeminência do tema em seus aspectos interdisciplinares. Discute as perspectivas consolidadas na comunidade acerca do tema, especialmente as abordagens de Searle (1984, 1992, 1997), Gardner (1994, 1995) e, no panorama brasileiro, de Teixeira (1990, 1994, 1995, 1998, 2000, 2004, 2008), incidindo nas relações entre os conceitos de intencionalidade, cérebro, experimento mental e cognitivismo, bem como algumas tendências críticas e de contestação de seus limites formais. Valendo-se de pesquisas anteriores sobre o tema e de metodologia que mescla elementos computacionais e de filosofia da mente, analisa o experimento conceitual do quarto chinês de Searle (em seus aspectos de sintaxe e semântica). A partir deste experimento e seus resultados, apresenta a perspectiva da investigação da Inteligência Artificial (IA), como pertencente à natureza da cognição, apoiada na teoria da cognição, nomeadamente de fenômenos como categorização e identificação de objetos, resolução de problemas, decisão e consciência. A pesquisa culmina na análise do conceito de Inteligência Artificial (IA), na Filosofia da Mente, proposto como um elemento noemático, constituindo-se como condição para a produção do conhecimento científico, aplicado este nas áreas de Tecnologia da Inteligência e Design Digital. Aplica os conceitos e resultados alcançados à definição e extensão do conceito de hipertexto, tomando-o como um mecanismo computacional capaz de estruturar diálogos hierarquizados, indexados, ao modo da formulação de Robot de Primo (2001) e Roth Coelho (2001). Por fim, os pressupostos investigados são apresentados em resultados observados nos chamados chatterbots na área de Inteligência Artificial (IA), destacando suas características e sua importância no atual contexto de sua utilização computacional no ciberespaço

Dans le cadre de la maintenance prédictive du constructeur automobile Renault, cette thèse vise à fournir des solutions à faible coût énergétique pour la détection non supervisée d'anomalies sur des séries temporelles. Avec l'évolution récente de l'automobile, de plus en plus de données sont produites et doivent être traitées par des algorithmes d'apprentissage automatique. Ce traitement peut être effectué dans le cloud ou directement à bord de la voiture. Dans un tel cas, la bande passante du réseau, les coûts des services cloud, la gestion de la confidentialité des données et la perte de données peuvent être économisés. L'intégration d'un modèle d'apprentissage automatique dans une voiture est un défi car elle nécessite des modèles frugaux en raison des contraintes de mémoire et de calcul. Dans ce but, nous étudions l'utilisation de réseaux de neurones impulsionnels (SNN) pour la detection d'anomalies, la prédiction et la classification sur des séries temporelles. Les performances et les coûts énergétiques des modèles d'apprentissage automatique sont évalués dans un scénario Edge à l'aide de modèles matériels génériques qui prennent en compte tous les coûts de calcul et de mémoire. Pour exploiter autant que possible l'activité neuronale parcimonieuse des SNN, nous proposons un modèle avec des connexions peu denses et entraînables qui consomme la moitié de l'énergie de sa version dense. Ce modèle est évalué sur des benchmarks publics de détection d'anomalies, un cas d'utilisation réel de détection d'anomalies sur les voitures de Renault Alpine, les prévisions météorologiques et le dataset Google Speech Command. Nous comparons également ses performances avec d'autres modèles d'apprentissage automatique existants. Nous concluons que, pour certains cas d'utilisation, les modèles SNN peuvent atteindre les performances de l'état de l'art tout en consommant 2 à 8 fois moins d'énergie. Pourtant, d'autres études devraient être entreprises pour évaluer ces modèles une fois embarqués dans une voiture. Inspirés par les neurosciences, nous soutenons que d'autres propriétés bio-inspirées telles que l'attention, l'activité parcimonieuse, la hiérarchie ou la dynamique des assemblées de neurons pourraient être exploités pour obtenir une meilleure efficacité énergétique et de meilleures performances avec des modèles SNN. Enfin, nous terminons cette thèse par un essai à la croisée des neurosciences cognitives, de la philosophie et de l'intelligence artificielle. En plongeant dans les difficultés conceptuelles liées à la conscience et en considérant les mécanismes déterministes de la mémoire, nous soutenons que la conscience et le soi pourraient être constitutivement indépendants de la mémoire. L'objectif de cet essai est de questionner la nature de l'humain par opposition à celle des machines et de l'IA<br>In the context of the predictive maintenance of the car manufacturer Renault, this thesis aims at providing low-power solutions for unsupervised anomaly detection on time-series. With the recent evolution of cars, more and more data are produced and need to be processed by machine learning algorithms. This processing can be performed in the cloud or directly at the edge inside the car. In such a case, network bandwidth, cloud services costs, data privacy management and data loss can be saved. Embedding a machine learning model inside a car is challenging as it requires frugal models due to memory and processing constraints. To this aim, we study the usage of spiking neural networks (SNNs) for anomaly detection, prediction and classification on time-series. SNNs models' performance and energy costs are evaluated in an edge scenario using generic hardware models that consider all calculation and memory costs. To leverage as much as possible the sparsity of SNNs, we propose a model with trainable sparse connections that consumes half the energy compared to its non-sparse version. This model is evaluated on anomaly detection public benchmarks, a real use-case of anomaly detection from Renault Alpine cars, weather forecasts and the google speech command dataset. We also compare its performance with other existing SNN and non-spiking models. We conclude that, for some use-cases, spiking models can provide state-of-the-art performance while consuming 2 to 8 times less energy. Yet, further studies should be undertaken to evaluate these models once embedded in a car. Inspired by neuroscience, we argue that other bio-inspired properties such as attention, sparsity, hierarchy or neural assemblies dynamics could be exploited to even get better energy efficiency and performance with spiking models. Finally, we end this thesis with an essay dealing with cognitive neuroscience, philosophy and artificial intelligence. Diving into conceptual difficulties linked to consciousness and considering the deterministic mechanisms of memory, we argue that consciousness and the self could be constitutively independent from memory. The aim of this essay is to question the nature of humans by contrast with the ones of machines and AI

After a short introduction concerning the problem of "Artificial Intelligence" (AI) the work continues with a summary of the state of the art.Thereafter, it goes on to profile four different basic scientific views of human beings and AI: symbolism, connectionism, biologism and physicalism. The emphasis is on the elucidation of anthropologically relevant statements to intelligence, spirit, thinking, perception, will, consciousness, self-consciousness, feelings and life.It is demonstrated that the basic views referred to represent greatly abbreviated and distorted pictures of human beings. Theories that do not go beyond the quantifiable level cannot adequately encompass the nature of relevant concepts and capabilities. That is above all because of the fact that generally a philosophical materialism is advocated, which considers the existence of intellectual substances impossible. For this reason a philosophical critique is necessary. The position of moderate and critical realism is advocated, whose anthropological statements are secured by epistemological and metaphysical investigations.The work comes to the conclusion that human beings cannot be understood symbolistically, connectionistically, biologistically or physicalistically. Man is a physical-intellectual entity, endowed with reason, a living social being. He is formed and led by his intellectual and therefore immortal soul, which gives him uniqueness, irreplaceability and the value of personhood. He is capable of thinking and thus of objective, abstract perception, and therefore is intelligent. Humans have an unfettered will, which, led by mental perception, is to be directed toward the good. They are moreover, through reflection, self-conscious. Humans live an intellectually determined life, which essentially differs, despite biological similarity, from that of animals and cannot possibly, due to its substantial superiority, have developed from animal life.All substantial anthropological abilities (such as intelligence, will, consciousness etc.) presuppose spirit. Because it is not within the power of human beings to create a simple substance such as spirit, a thinking, perceptive, intelligent, willing, self-conscious, sentient living being can at best be only technically imitated, modelled or simulated but never be reproduced, copied or created. The relationship of humans to AI is thus determined by an insuperable difference between their natures.

In pop culture, artificial intelligences (AI) are frequently portrayed as worthy of moral personhood, and failing to treat these entities as such is often treated as analogous to racism. The implicit condition for attributing moral personhood to an AI is usually passing some form of the "Turing Test", wherein an entity passes if it could be mistaken for a human. I argue that this is unfounded under any moral theory that uses the capacity for desire as the criteria for moral standing. Though the action-based theory of desire ensures that passing a rigourous enough version of the Turing Test would be sufficient for moral personhood, that theory has unacceptable results when used in moral theory. If a desire-based moral theory is to be made defensible, it must use a phenomenological account of desire, which would make the Turing Test fail to track the relevant property.<br>October 2015

<p>This dissertation argues for a number of theses related to responsibility, norm psychology, and personhood. Although most of the papers argue for “standalone” theses, in the sense that their truth does not depend the truth of the others, the five papers collectively illustrate a broader view of humans as (a) responsible agents who are (b) self-governing and (c) equipped with a capacity for norms, and whose agency (d) centers on dynamic responsiveness to corrective feedback. Drawing on this broader picture, the dissertation sheds light on ethical questions about our social practices and technologies, as well as descriptive questions about the nature of substance use disorder. </p> <p>Most centrally, the dissertation argues that forward-looking considerations are relevant for responsibility, not merely because the consequences of our responsibility practices are desirable, but primarily because of a connection which I argue exists between relationships, norms, and learning. On the view I defend, an agent is a responsible agent only if she can learn from being held responsible, so as to regulate herself according to norms of which she presently falls short. I argue that, if it were not for the capacity of humans to learn from <i>social corrective feedback</i>, such as normative responses like praise and blame, humans would be unable to participate in norm-governed relationships and communities. It is in virtue of their participation in these relationships and communities that humans are subject to interpersonal norms, such that they can fulfill or violate these norms and be praiseworthy or blameworthy for doing so. So, without the kind of learning that makes participation in these relationships a possibility, humans could never be praiseworthy or blameworthy for anything that they do. </p> <p>The dissertation also argues that human norm psychology has implications for how we should relate to “social robots”—artificial agents designed to participate in relationships with humans. I argue that, like humans, social robots should be equipped with a capacity to recognize and respond to normative feedback. Lastly, the dissertation resists a common narrative about addiction as being a form of akrasia in which agents act against their own better judgment. While this is certainly a central aspect of many cases of addiction, I argue that it fails to appreciate the ways in which addiction sometimes interacts with a person’s identity and goals, especially in cases where the agent believes that the things she values would not be feasible if she did not continue to engage in addictive behavior.</p>