Despite its centrality in the philosophy of cognitive science, there has been little prior philosophical work engaging with the notion of representation in contemporary NLP practice. This paper attempts to fill that lacuna: drawing on ideas from cognitive science, I introduce a framework for evaluating the representational claims made about components of neural NLP models, proposing three criteria with which to evaluate whether a component of a model represents a property and operationalising these criteria using probing classifiers, a popular (...) analysis technique in NLP (and deep learning more broadly). The project of operationalising a philosophically-informed notion of representation should be of interest to both philosophers of science and NLP practitioners. It affords philosophers a novel testing-ground for claims about the nature of representation, and helps NLPers organise the large literature on probing experiments, suggesting novel avenues for empirical research. (shrink)

This textbook presents a concise, accessible and engaging first introduction to deep learning, offering a wide range of connectionist models which represent the current state-of-the-art. The text explores the most popular algorithms and architectures in a simple and intuitive style, explaining the mathematical derivations in a step-by-step manner. The content coverage includes convolutional networks, LSTMs, Word2vec, RBMs, DBNs, neural Turing machines, memory networks and autoencoders. Numerous examples in working Python code are provided throughout the book, and the code (...) is also supplied separately at an accompanying website. Topics and features: introduces the fundamentals of machine learning, and the mathematical and computational prerequisites for deep learning; discusses feed-forward neural networks, and explores the modifications to these which can be applied to any neural network; examines convolutional neural networks, and the recurrent connections to a feed-forward neural network; describes the notion of distributed representations, the concept of the autoencoder, and the ideas behind language processing with deep learning; presents a brief history of artificial intelligence and neural networks, and reviews interesting open research problems in deep learning and connectionism. This clearly written and lively primer on deep learning is essential reading for graduate and advanced undergraduate students of computer science, cognitive science and mathematics, as well as fields such as linguistics, logic, philosophy, and psychology. (shrink)

As modern deep networks become more complex, and get closer to human‐like capabilities in certain domains, the question arises as to how the representations and decision rules they learn compare to the ones in humans. In this work, we study representations of sentences in one such artificial system for natural language processing. We first present a diagnostic test dataset to examine the degree of abstract composable structure represented. Analyzing performance on these diagnostic tests indicates a (...) lack of systematicity in representations and decision rules, and reveals a set of heuristic strategies. We then investigate the effect of training distribution on learning these heuristic strategies, and we study changes in these representations with various augmentations to the training set. Our results reveal parallels to the analogous representations in people. We find that these systems can learn abstract rules and generalize them to new contexts under certain circumstances—similar to human zero‐shot reasoning. However, we also note some shortcomings in this generalization behavior—similar to human judgment errors like belief bias. Studying these parallels suggests new ways to understand psychological phenomena in humans as well as informs best strategies for building artificial intelligence with human‐like language understanding. (shrink)

The emerging contemporary natural philosophy provides a common ground for the integrative view of the natural, the artificial, and the human-social knowledge and practices. Learning process is central for acquiring, maintaining, and managing knowledge, both theoretical and practical. This paper explores the relationships between the present advances in understanding of learning in the sciences of the artificial, natural sciences, and philosophy. The question is, what at this stage of the development the (...) inspiration from nature, specifically its computational models such as info-computation through morphological computing, can contribute to machine learning and artificial intelligence, and how much on the other hand models and experiments in machine learning and robotics can motivate, justify, and inform research in computational cognitive science, neurosciences, and computing nature. We propose that one contribution can be understanding of the mechanisms of ‘learning to learn’, as a step towards deep learning with symbolic layer of computation/information processing in a framework linking connectionism with symbolism. As all natural systems possessing intelligence are cognitive systems, we describe the evolutionary arguments for the necessity of learning to learn for a system to reach humanlevel intelligence through evolution and development. The paper thus presents a contribution to the epistemology of the contemporary philosophy of nature. (shrink)

The contract review process can be a costly and time-consuming task for lawyers and clients alike, requiring significant effort to identify and evaluate the legal implications of individual clauses. To address this challenge, we propose the use of natural language processing techniques, specifically text classification based on deontic tags, to streamline the process. Our research question is whether natural language processing techniques, specifically dense vector embeddings, can help semi-automate the contract review process and reduce (...) time and costs for legal professionals reviewing deontic modalities in contracts. In this study, we create a domain-specific dataset and train both baseline and neural network models for contract sentence classification. This approach offers a more efficient and cost-effective solution for contract review, mimicking the work of a lawyer. Our approach achieves an accuracy of 0.90, showcasing its effectiveness in identifying and evaluating individual contract sentences. (shrink)

This paper presents an in-depth literature review on the driving forces and barriers for achieving operational excellence through artificial intelligence. Artificial intelligence is a technological concept spanning operational management, philosophy, humanities, statistics, mathematics, computer sciences, and social sciences. AI refers to machines mimicking human behavior in terms of cognitive functions. The evolution of new technological procedures and advancements in producing intelligence for machines creates a positive impact on decisions, operations, strategies, and management incorporated (...) in the production process of goods and services. Businesses develop various methods and solutions to extract meaningful information, such as big data, automatic production capabilities, and systematization for business improvement. The progress in organizational competitiveness is apparent through improvements in firm’s decisions, resulting in increased operational efficiencies. Innovation with AI has enabled small businesses to reduce operating expenses and increase revenues. The focused literature review reveals the driving forces for achieving operational excellence through AI are improvement in computing abilities of machines, development of data-based AI, advancements in deep learning, cloud computing, data management, and integration of AI in operations. The barriers are mainly cultural constraints, fear of the unknown, lack of employee skills, and strategic planning for adopting AI. The current paper presents an analysis of articles focused on AI adoption in production and operations. We selected articles published between 2015 and 2020. Our study contributes to the literature reviews on operational excellence, artificial intelligence, driving forces for AI, and AI barriers in achieving operational excellence. (shrink)

Sentence boundary detection (SBD) represents an important first step in natural language processing since accurately identifying sentence boundaries significantly impacts downstream applications. Nevertheless, detecting sentence boundaries within legal texts poses a unique and challenging problem due to their distinct structural and linguistic features. Our approach utilizes deep learning models to leverage delimiter and surrounding context information as input, enabling precise detection of sentence boundaries in English legal texts. We evaluate various deep learning models, (...) including domain-specific transformer models like LegalBERT and CaseLawBERT. To assess the efficacy of our deep learning models, we compare them with a state-of-the-art domain-specific statistical conditional random field (CRF) model. After considering model size, F1-score, and inference time, we identify the Convolutional Neural Network Model (CNN) as the top-performing deep learning model. To further enhance performance, we integrate the features of the CNN model into the subsequent CRF model, creating a hybrid architecture that combines the strengths of both models. Our experiments demonstrate that the hybrid model outperforms the baseline model, achieving a 4% improvement in the F1-score. Additional experiments showcase the superiority of the hybrid model over SBD open-source libraries when confronted with an out-of-domain test set. These findings underscore the importance of efficient SBD in legal texts and emphasize the advantages of employing deep learning models and hybrid architectures to achieve optimal performance. (shrink)

Wittgenstein’s ideas are a common ground for developers of Natural Language Processing (NLP) systems and linguists working on Language Acquisition and Mastery (LAM) models (Mills 1993; Lowney, Levy, Meroney and Gayler 2020; Skelac and Jandrić 2020). In recent years, we have witnessed a fast development of NLP systems capable of performing tasks as never before. NLP and LAM have been implemented based on deep learning neural networks, which learn concepts representation from rough data, but (...) are nonetheless very effective in tasks such as question answering, textual entailment, and translation (Devlin et al. 2019; Kitaev, Cao and Klein 2019; Wang et al. 2019). In this paper, I will debate some Wittgensteinian concepts that impact the architectures of many NLP deep-learning systems. I will focus, in particular, on the attempt to build a specific kind of architecture to model a private language. The discussion, I think, helps extract philosophical assumptions leading the research and development of AI systems capable of language modelling. In this paper, I will address some of the main features of NLP systems2 used for word embedding and one proposal to manipulate through a neural network a form of private language (Lawney et al. 2020). In §2, I will reconstruct the complex path of the private language argument (PLA). In §3, I will discuss connectivist language models and introduce notions about NPL systems' architecture. An overview of this kind of model is helpful to introduce the work of Lowney, Levy, Meroney and Gayler (2020). They submit that their model can respond to the issues raised by Wittgenstein in the notorious Private Language Argument (PLA). This argument unexpectedly turned out to be relevant not only for the philosophy of language but also for NLP and LAM modellers. I will describe the language game concept in NLP, how it is embedded, and its role in inductive systems development. This central concept in Wittgenstein’s work is relevant to describe the role of context in understanding the meanings of words. In §4, I present the Wittgensteinian main concepts in play in the connectivist paradigm. I argue that the connectionist theoretical framework can better catch the dependency of word meaning on context. There seems to be a correlation between Wittgenstein's invitation to look, an invitation to dismiss the aim of theorizing about languages, and the absence of theory-ladenness in deep learning technologies involved in NLP. In §5, I criticize Lowney and colleagues’ claim, whose model does not successfully capture Wittgenstein’s Beetle in the Box case. Moreover, I argue that even if we can distinguish a strong and a weak definition of a private language, Wittgenstein’s argument also holds for deep-learning models, and his worries are still a good guide for NLP developers. (shrink)

The emerging contemporary natural philosophy provides a common ground for the integrative view of the natural, the artificial, and the human-social knowledge and practices. Learning process is central for acquiring, maintaining, and managing knowledge, both theoretical and practical. This paper explores the relationships between the present advances in understanding of learning in the sciences of the artificial (deep learning, robotics), natural sciences (neuroscience, cognitive science, biology), and philosophy (philosophy (...) of computing, philosophy of mind, natural philosophy). The question is, what at this stage of the development the inspiration from nature, specifically its computational models such as info-computation through morphological computing, can contribute to machine learning and artificial intelligence, and how much on the other hand models and experiments in machine learning and robotics can motivate, justify, and inform research in computational cognitive science, neurosciences, and computing nature. We propose that one contribution can be understanding of the mechanisms of ‘learning to learn’, as a step towards deep learning with symbolic layer of computation/information processing in a framework linking connectionism with symbolism. As all natural systems possessing intelligence are cognitive systems, we describe the evolutionary arguments for the necessity of learning to learn for a system to reach humanlevel intelligence through evolution and development. The paper thus presents a contribution to the epistemology of the contemporary philosophy of nature. (shrink)

This series will include monographs and collections of studies devoted to the investigation and exploration of knowledge, information and data-processing systems of all kinds, no matter whether human, (other) animal or machine. Its scope is intended to span the full range of interests from classical problems in the philosophy of mind and phi losophical psychology through issues in cognitive psychology and socio biology (concerning the mental capabilities of other species) to ideas related to artificial intelligence (...) and computer science. While primary emphasis will be placed upon theoretical, conceptual and epistemologi cal aspects of these problems and domains, empirical, experimental and methodological studies will also appear from time to time. Among the most challenging and difficult projects within the scope of artificial intelligence is the development and implementation of com puter programs suitable for processing natural language. Our purpose in compiling the present volume has been to contribute to the foundations of this enterprise by bringing together classic papers devoted to crucial problems involved in understanding natural language, which range from issues of formal syntax and logical form to those of possible-worlds and situation semantics. The book begins with a comprehensive introduc tion composed by Jack Kulas, the senior editor of this work, which pro vides a systematic orientation to this complex field, and ends with a selected bibliography intended to promote further research. If our efforts assist others in dealing with these problems, they will have been worthwhile. J. H. F. (shrink)

Presupposing no familiarity with the technical concepts of either philosophy or computing, this clear introduction reviews the progress made in AI since the inception of the field in 1956. Copeland goes on to analyze what those working in AI must achieve before they can claim to have built a thinking machine and appraises their prospects of succeeding. There are clear introductions to connectionism and to the language of thought hypothesis which weave together material from philosophy, (...) class='Hi'>artificial intelligence and neuroscience. John Searle's attacks on AI and cognitive science are countered and close attention is given to foundational issues, including the nature of computation, Turing Machines, the Church-Turing Thesis and the difference between classical symbol processing and parallel distributed processing. The book also explores the possibility of machines having free will and consciousness and concludes with a discussion of in what sense the human brain may be a computer. (shrink)

Presupposing no familiarity with the technical concepts of either philosophy or computing, this clear introduction reviews the progress made in AI since the inception of the field in 1956. Copeland goes on to analyze what those working in AI must achieve before they can claim to have built a thinking machine and appraises their prospects of succeeding.There are clear introductions to connectionism and to the language of thought hypothesis which weave together material from philosophy, artificial (...) intelligence and neuroscience. John Searle's attacks on AI and cognitive science are countered and close attention is given to foundational issues, including the nature of computation, Turing Machines, the Church-Turing Thesis and the difference between classical symbol processing and parallel distributed processing. The book also explores the possibility of machines having free will and consciousness and concludes with a discussion of in what sense the human brain may be a computer. (shrink)

Museum collection databases contain echoes of encounter between colonial collectors (broadly defined) and Indigenous people from around the world. The moment of acquisition—when an item passed out of a community and into the hands of the collector—often included multilingual acts of translation. An artist may have shared the Indigenous name of the object, or the terms associated with its origin and use. Late nineteenth and twemtieth century museum registrars would in turn transcribe this information from field logs into museum catalogs. (...) Over time, these catalog entries were transformed into digital records within collections managements systems (e.g., EMu, PastPerfect, etc.). As a result of this 150-year process, today’s museum collection databases are riddled with Indigenous words and descriptions, scattered across various metadata fields. They may include Native place-names, family names or vocabulary terms that, when translated, extend far beyond the categories ascribed by museum collection managers. These instances of Indigenous description may also serve as a crucial bridge for reconnecting source communities with items of particular interest to their cultural heritage and linguistic preservation efforts. Aiming to enhance the accessibility of Indigenous languages contained in the metadata of cultural heritage collections, this paper explores applications of machine learning methodologies to identify Indigenous terms present in museum catalogs. Specifically, we discuss methods that incorporate the Google Cloud Language Identification Service to detect A:shiwi (Pueblo of Zuni) language terms through a case study of metadata records from the two largest natural history museums in the USA. We utilize an elimination mechanism to exclude specific languages (e.g., English and Spanish) at the word and phrase levels to detect A:shiwi terms. Our approach outperforms conventional methods in terms of accuracy, recall, precision, and F1-scores. This method can be used to confront the “Digital Heap” of cultural heritage records across institutions to improve the discoverability of Indigenous languages in metadata descriptions and reconnect source communities with items of cultural patrimony. (shrink)

Modeling law search and retrieval as prediction problems has recently emerged as a predominant approach in law intelligence. Focusing on the law article retrieval task, we present a deep learning framework named LamBERTa, which is designed for civil-law codes, and specifically trained on the Italian civil code. To our knowledge, this is the first study proposing an advanced approach to law article prediction for the Italian legal system based on a BERT (Bidirectional Encoder Representations from Transformers) (...) class='Hi'>learning framework, which has recently attracted increased attention among deep learning approaches, showing outstanding effectiveness in several natural language processing and learning tasks. We define LamBERTa models by fine-tuning an Italian pre-trained BERT on the Italian civil code or its portions, for law article retrieval as a classification task. One key aspect of our LamBERTa framework is that we conceived it to address an extreme classification scenario, which is characterized by a high number of classes, the few-shot learning problem, and the lack of test query benchmarks for Italian legal prediction tasks. To solve such issues, we define different methods for the unsupervised labeling of the law articles, which can in principle be applied to any law article code system. We provide insights into the explainability and interpretability of our LamBERTa models, and we present an extensive experimental analysis over query sets of different type, for single-label as well as multi-label evaluation tasks. Empirical evidence has shown the effectiveness of LamBERTa, and also its superiority against widely used deep-learning text classifiers and a few-shot learner conceived for an attribute-aware prediction task. (shrink)

Artificial intelligency can bring speed and reliability to drug discovery process. It represents an additional intelligence, which in any case can replace the strategic and logic creative insight of the medicinal chemist who remains the architect and molecule master designer. In terms of drug design, artificial intelligency, deep learning machines, and other revolutionary technologies will match with the medicinal chemist’s natural intelligency, but for sure never go beyond. This manuscript tries to assess the impact (...) of the artificial intelligency on drug discovery today. (shrink)

A concise but informative overview of AI ethics and policy. -/- Artificial intelligence, or AI for short, has generated a staggering amount of hype in the past several years. Is it the game-changer it's been cracked up to be? If so, how is it changing the game? How is it likely to affect us as customers, tenants, aspiring homeowners, students, educators, patients, clients, prison inmates, members of ethnic and sexual minorities, and voters in liberal democracies? Authored by experts (...) in fields ranging from computer science and law to philosophy and cognitive science, this book offers a concise overview of moral, political, legal and economic implications of AI. It covers the basics of AI's latest permutation, machine learning, and considers issues such as transparency, bias, liability, privacy, and regulation. -/- Both business and government have integrated algorithmic decision support systems into their daily operations, and the book explores the implications for our lives as citizens. For example, do we take it on faith that a machine knows best in approving a patient's health insurance claim or a defendant's request for bail? What is the potential for manipulation by targeted political ads? How can the processes behind these technically sophisticated tools ever be transparent? The book discusses such issues as statistical definitions of fairness, legal and moral responsibility, the role of humans in machine learning decision systems, “nudging” algorithms and anonymized data, the effect of automation on the workplace, and AI as both regulatory tool and target. (shrink)

With the ever-growing accessibility of case law online, it has become challenging to manually identify case law relevant to one’s legal issue. In the Netherlands, the planned increase in the online publication of case law is expected to exacerbate this challenge. In this paper, we tried to predict whether court decisions are cited by other courts or not after being published, thus in a way distinguishing between more and less authoritative cases. This type of system may be used to process (...) the large amounts of available data by filtering out large quantities of non-authoritative decisions, thus helping legal practitioners and scholars to find relevant decisions more easily, and drastically reducing the time spent on preparation and analysis. For the Dutch Supreme Court, the match between our prediction and the actual data was relatively strong (with a Matthews Correlation Coefficient of 0.60). Our results were less successful for the Council of State and the district courts (MCC scores of 0.26 and 0.17, relatively). We also attempted to identify the most informative characteristics of a decision. We found that a completely explainable model, consisting only of handcrafted metadata features, performs almost as well as a less well-explainable system based on all text of the decision. (shrink)

Today, artificial intelligence is actively mastering natural languages, becoming an interlocutor and partner of human in various aspects of activity. However, the symbolic approach, which implies the transfer of rules and logic, has failed, the number of rules and exceptions of the language does not allow its formalization, so modern «deep learning» of artificial neural networks involves an independent search for patterns in extensive databases. During training, artificial intelligence puts a word (...) into a sentence so that the syntagmatic relationships are as close as possible to those of the target word in the base, taking into account both the semantic relationships of words and the relationships between words in the sequence of presentation. The «language» of information technologies is digital. During natural language processing, words are represented in vector form as a sequence of numbers. The idea of representing words mathematically is familiar to people and is usually associated with logical consistency. Visualization of the position of words in a multidimensional space created by artificial intelligence demonstrates a number of patterns, obvious semantic and syntactic relationships, but the essence of other relationships between words is not obvious. The mathematical representation of words, created by artificial intelligence, can allow you to look at the language from a new non-human point of view. (shrink)

The book’s core argument is that an artificial intelligence that could equal or exceed human intelligence—sometimes called artificial general intelligence (AGI)—is for mathematical reasons impossible. It offers two specific reasons for this claim: Human intelligence is a capability of a complex dynamic system—the human brain and central nervous system. Systems of this sort cannot be modelled mathematically in a way that allows them to operate inside a computer. In supporting their claim, the authors, Jobst (...) Landgrebe and Barry Smith, marshal evidence from mathematics, physics, computer science, philosophy, linguistics, and biology, setting up their book around three central questions: What are the essential marks of human intelligence? What is it that researchers try to do when they attempt to achieve "artificial intelligence" (AI)? And why, after more than 50 years, are our most common interactions with AI, for example with our bank’s computers, still so unsatisfactory? Landgrebe and Smith show how a widespread fear about AI’s potential to bring about radical changes in the nature of human beings and in the human social order is founded on an error. There is still, as they demonstrate in a final chapter, a great deal that AI can achieve which will benefit humanity. But these benefits will be achieved without the aid of systems that are more powerful than humans, which are as impossible as AI systems that are intrinsically "evil" or able to "will" a takeover of human society. (shrink)

The developing academic field of machine ethics seeks to make artificial agents safer as they become more pervasive throughout society. In contrast to computer ethics, machine ethics is concerned with the behavior of machines toward human users and other machines. This study aims to use an action-based ethical theory founded on the combinational aspects of deontological and teleological theories of ethics in the construction of an artificial moral agent (AMA).,The decision results derived by the AMA are (...) acquired via fuzzy logic interpretation of the relative values of the steady-state simulations of the corresponding rule-based fuzzy cognitive map (RBFCM).,Through the use of RBFCMs, the following paper illustrates the possibility of incorporating ethical components into machines, where latent semantic analysis (LSA) and RBFCMs can be used to model dynamic and complex situations, and to provide abilities in acquiring causal knowledge.,This approach is especially appropriate for data-poor and uncertain situations common in ethics. Nonetheless, to ensure that a machine with an ethical component can function autonomously in the world, research in artificial intelligence will need to further investigate the representation and determination of ethical principles, the incorporation of these ethical principles into a system’s decision procedure, ethical decision-making with incomplete and uncertain knowledge, the explanation for decisions made using ethical principles and the evaluation of systems that act based upon ethical principles.,To date, the conducted research has contributed to a theoretical foundation for machine ethics through exploration of the rationale and the feasibility of adding an ethical dimension to machines. Further, the constructed AMA illustrates the possibility of utilizing an action-based ethical theory that provides guidance in ethical decision-making according to the precepts of its respective duties. The use of LSA illustrates their powerful capabilities in understanding text and their potential application as information retrieval systems in AMAs. The use of cognitive maps provides an approach and a decision procedure for resolving conflicts between different duties.,This paper suggests that cognitive maps could be used in AMAs as tools for meta-analysis, where comparisons regarding multiple ethical principles and duties can be examined and considered. With cognitive mapping, complex and abstract variables that cannot easily be measured but are important to decision-making can be modeled. This approach is especially appropriate for data-poor and uncertain situations common in ethics. (shrink)

The article discusses the process of “conceptual borrowing”, according to which, when a new discipline emerges, it develops its technical vocabulary also by appropriating terms from other neighbouring disciplines. The phenomenon is likened to Carl Schmitt’s observation that modern political concepts have theological roots. The authors argue that, through extensive conceptual borrowing, AI has ended up describing computers anthropomorphically, as computational brains with psychological properties, while brain and cognitive sciences have ended up describing brains and minds computationally and informationally, as (...) biological computers. The crosswiring between the technical languages of these disciplines is not merely metaphorical but can lead to confusion, and damaging assumptions and consequences. The article ends on an optimistic note about the self-adjusting nature of technical meanings in language and the ability to leave misleading conceptual baggage behind when confronted with advancement in understanding and factual knowledge. (shrink)

The application of deep learning methods to problems in natural language processing has generated significant progress across a wide range of natural language processing tasks. For some of these applications, deep learning models now approach or surpass human performance. While the success of this approach has transformed the engineering methods of machine learning in artificial intelligence, the significance of these achievements for the modelling of human (...) class='Hi'>learning and representation remains unclear. Deep Learning and Linguistic Representation looks at the application of a variety of deep learning systems to several cognitively interesting NLP tasks. It also considers the extent to which this work illuminates our understanding of the way in which humans acquire and represent linguistic knowledge. Key Features: combines an introduction to deep learning in AI and NLP with current research on Deep Neural Networks in computational linguistics. is self-contained and suitable for teaching in computer science, AI, and cognitive science courses; it does not assume extensive technical training in these areas. provides a compact guide to work on state of the art systems that are producing a revolution across a range of difficult natural language tasks. (shrink)

Deep Learning has been widely used for tackling challenging natural language processing tasks over the recent years. Similarly, the application of Deep Neural Networks in legal analytics has increased significantly. In this survey, we study the early adaptation of Deep Learning in legal analytics focusing on three main fields; text classification, information extraction, and information retrieval. We focus on the semantic feature representations, a key instrument for the successful application of deep (...) learning in natural language processing. Additionally, we share pre-trained legal word embeddings using the word2vec model over large corpora, comprised legislations from UK, EU, Canada, Australia, USA, and Japan among others. (shrink)

Artificial intelligence is now widely accessible and already being used by healthcare researchers throughout various stages in the research process, such as assisting with systematic reviews, supporting data collection, facilitating data analysis and drafting manuscripts for publication. The most common AI tools used are forms of generative AI such as ChatGPT, Claude and Gemini. Generative AI is a type of AI that can generate human-like text, audio, videos, code and images based on text-based prompts inputted by a human (...) user. Generative AI is trained on large amounts of data, and the outputs are sophisticated and can be indistinguishable from a response from a skilled human. In this article, we outline several AI applications that can be used in healthcare research, examining their benefits, limitations and outline best practices for maintaining research integrity and ethical standards. (shrink)

Transformer-based language models (TLMs) have widely been recognized to be a cutting-edge technology for the successful development of deep-learning-based solutions to problems and applications that require natural language processing and understanding. Like for other textual domains, TLMs have indeed pushed the state-of-the-art of AI approaches for many tasks of interest in the legal domain. Despite the first Transformer model being proposed about six years ago, there has been a rapid progress of this technology at (...) an unprecedented rate, whereby BERT and related models represent a major reference, also in the legal domain. This article provides the first systematic overview of TLM-based methods for AI-driven problems and tasks in the legal sphere. A major goal is to highlight research advances in this field so as to understand, on the one hand, how the Transformers have contributed to the success of AI in supporting legal processes, and on the other hand, what are the current limitations and opportunities for further research development. (shrink)

This is a digest of how various researchers in biology and astrobiology have explored questions of what defines living organisms—definitions based on functions or structures observed in organisms, or on systems terms, or on mathematical conceptions like closure, chirality, quantum mechanics and thermodynamics, or on biosemiotics, or on Darwinian evolution—to clarify the field and make it easier for endeavors in artificial intelligence to make progress. Current ideas are described to promote work between astrobiologists and computer scientists, each concerned (...) with living organisms. A four-parameter framework is presented as a scaffold that is later developed into what machines lack to be considered alive: systems, evolution, energy and consciousness, and includes Jagers operators and the idea of dual closure. A novel definition of consciousness is developed which describes mental objects both with and without communicable properties, and this helps to clarify how consciousness in machines may be studied as an emergent process related to choice functions in systems. A perspective on how quantization, acting on nucleic acids, sets up natural limits to system behavior is offered as a partial address to the problem of biogenesis. (shrink)

Philosophers have recently focused on critical, epistemological challenges that arise from the opacity of deep neural networks. One might conclude from this literature that doing good science with opaque models is exceptionally challenging, if not impossible. Yet, this is hard to square with the recent boom in optimism for AI in science alongside a flood of recent scientific breakthroughs driven by AI methods. In this paper, I argue that the disconnect between philosophical pessimism and scientific optimism is driven by (...) a failure to examine how AI is actually used in science. I show that, in order to understand the epistemic justification for AI-powered breakthroughs, philosophers must examine the role played by deep learning as part of a wider process of discovery. The philosophical distinction between the 'context of discovery' and the 'context of justification' is helpful in this regard. I demonstrate the importance of attending to this distinction with two cases drawn from the scientific literature, and show that epistemic opacity need not diminish AI's capacity to lead scientists to significant and justifiable breakthroughs. (shrink)

This book provides a framework for thinking about foundational philosophical questions surrounding machine learning as an approach to artificial intelligence. Specifically, it links recent breakthroughs in deep learning to classical empiricist philosophy of mind. In recent assessments of deep learning's current capabilities and future potential, prominent scientists have cited historical figures from the perennial philosophical debate between nativism and empiricism, which primarily concerns the origins of abstract knowledge. These empiricists were generally (...) faculty psychologists; that is, they argued that the active engagement of general psychological faculties-such as perception, memory, imagination, attention, and empathy-enables rational agents to extract abstract knowledge from sensory experience. This book explains a number of recent attempts to model roles attributed to these faculties in deep neural network based artificial agents by appeal to the faculty psychology of philosophers such as Aristotle, Ibn Sina (Avicenna), John Locke David Hume, William James, and Sophie de Grouchy. It illustrates the utility of this interdisciplinary connection by showing how it can provide benefits to both philosophy and computer science: computer scientists can continue to mine the history of philosophy for ideas and aspirational targets to hit on the way to more robustly rational artificial agents, and philosophers can see how some of the historical empiricists' most ambitious speculations can be realized in specific computational systems. (shrink)

An original quantum foundations concept of a deep learning computational Universe is introduced. The fundamental information of the Universe (or Triuniverse)is postulated to evolve about itself in a Red, Green and Blue (RGB) tricoloured stable self-mutuality in three information processing loops. The colour is a non-optical information label. The information processing loops form a feedback-reinforced deep learning macrocycle with trefoil knot topology. Fundamental information processing is driven by ψ-Epistemic Drive, the Natural appetite (...) for information selected for advantageous knowledge. From its substrate of Mathematics, the knotted information processing loops determine emergent Physics and thence the evolution of super-emergent Life (biological and artificial intelligence). RGB-tricoloured information is processed in sequence in an Elemental feedback loop (R), then an Operational feedback loop (G), then a Structural feedback loop (B) and back to an Elemental feedback loop (R), and so on around the trefoil in deep learning macrocycles. It is postulated that hierarchical information correspondence from Mathematics through Physics to Life is mapped and conserved within each colour. The substrate of Mathematics has RGB-tricoloured feedback loops which are respectively Algebra (R), Algorithms (G) and Geometry (B). In Mathematics, the trefoil macrocycle is Algebraic Algorithmic Geometry and its correlation system is a Tensor Neural Knot Network enabling Qutrit Entanglement. Emergent Physics has corresponding RGB-tricoloured feedback loops of Quantum Mechanics (R), Quantum Deep Learning (G) and Quantum Geometrodynamics(B). In Physics, the trefoil macrocycle is Quantum Intelligent Geometrodynamics and its correlation system is Quantum Darwinism. Super-emergent Life has corresponding RGB-tricoloured loops of Variation (R), Selection (G) and Heredity (B). In the evolution of Life, the trefoil macrocycle is Variational Selective Heredity and its correlation ecosystem is Darwin’s ecologically “Entangled Bank”. (shrink)

Historically, the dispute between empiricists and nativists in philosophy and cognitive science has concerned human and animal minds (Margolis and Laurence in Philos Stud: An Int J Philos Anal Tradit 165(2): 693-718, 2013, Ritchie in Synthese 199(Suppl 1): 159–176, 2021, Colombo in Synthese 195: 4817–4838, 2018). But recent progress has highlighted how empiricist and nativist concerns arise in the construction of artificial systems (Buckner in From deep learning to rational machines: What the history of philosophy (...) can teach us about the future of artificial intelligence. Oxford University Press.). This paper uses nativism and empiricism to address questions about the nature of artificial intelligence and its trajectory. It begins by defining the nativism/empiricism debate in terms of the generality of a system. Nativist systems have initial states with domain-specific features; empiricist systems have initial states with only domain-general features. With the debate framed in this way, it then explores a variety of arguments for nativism and empiricism in AI. These arguments revolve around two different questions which must be distinguished: whether nativism the only possible approach to developing human-level AI (HLAI); and whether nativism is the most practical approach to developing HLAI. On the first question, it argues that nativism is quite clearly not the only possible approach to developing HLAI, as is sometimes suggested. It argues that existing arguments for the necessity of nativism are unconvincing, because they analogize from poverty of the stimulus arguments about humans, while AIs often have access to much more data than humans. Then it argues that the case of evolution gives us a compelling argument against nativism. On the second, practical question, the paper argues that there is a tradeoff between the advantages of encoding innate machinery directly, and the advantages of evolving or learning it. However, as the past decade has shown, empiricism is a much more viable path to greater capability levels, given the ‘bitter lesson’ (Sutton in _Reinforcement Learning: An Introduction_. MIT press.) that encoding the ‘correct’ knowledge in AI systems is perennially outperformed by more empiricist methods that leverage large-scale data and computation. (shrink)

Wittgenstein’s Private Language Argument holds that language requires rule-following, rule following requires the possibility of error, error is precluded in pure introspection, and inner mental life is known only by pure introspection, thus language cannot exist entirely within inner mental life. Fodor defends his Language of Thought program against the Private Language Argument with a dilemma: either privacy is so narrow that internal mental life can be known outside of introspection, or so broad that computer (...) language serves as a counter-example. I suggest that the developing field of artificial intelligence (deep learning neural networks) tends to vitiate Fodor’s defense and hence vindicate the Private Language Argument. The first horn of Fodor’s dilemma requires language to encompass genuinely internal mental life, i.e. non-projected intentional states, which are not exhibited in classical machine learning but only by deep learning neural networks (artificial intelligence). Such networks act as black boxes, however, whose state cannot be understood by tracking the changes in their supervenience bases without shared context, and that shared context introduces the possibility of error. The language of artificial intelligence is not private. (shrink)

The success of deep learning in natural language processing raises intriguing questions about the nature of linguistic meaning and ways in which it can be processed by natural and artificial systems. One such question has to do with subword segmentation algorithms widely employed in language modeling, machine translation, and other tasks since 2016. These algorithms often cut words into semantically opaque pieces, such as ‘period’, ‘on’, ‘t’, and ‘ist’ in ‘period|on|t|ist’. The (...) system then represents the resulting segments in a dense vector space, which is expected to model grammatical relations among them. This representation may in turn be used to map ‘period|on|t|ist’ (English) to ‘par|od|ont|iste’ (French). Thus, instead of being modeled at the lexical level, translation is reformulated more generally as the task of learning the best bilingual mapping between the sequences of subword segments of two languages; and sometimes even between pure character sequences: ‘p|e|r|i|o|d|o|n|t|i|s|t’ → ‘p|a|r|o|d|o|n|t|i|s|t|e’. Such subword segmentations and alignments are at work in highly efficient end-to-end machine translation systems, despite their allegedly opaque nature. The computational value of such processes is unquestionable. But do they have any linguistic or philosophical plausibility? I attempt to cast light on this question by reviewing the relevant details of the subword segmentation algorithms and by relating them to important philosophical and linguistic debates, in the spirit of making artificial intelligence more transparent and explainable. (shrink)

A wide-ranging history of the intellectual developments that produced the modern idea of the algorithm. Bringing together the histories of mathematics, computer science, and linguistic thought, Language and the Rise of the Algorithm reveals how recent developments in artificial intelligence are reopening an issue that troubled mathematicians long before the computer age. How do you draw the line between computational rules and the complexities of making systems comprehensible to people? Here Jeffrey M. Binder offers a compelling (...) tour of four visions of universal computation that addressed this issue in very different ways: G. W. Leibniz's calculus ratiocinator; a universal algebra scheme Nicolas de Condorcet designed during the French Revolution; George Boole's nineteenth-century logic system; and the early programming language ALGOL, whose name is short for algorithmic language. These episodes show that symbolic computation has repeatedly become entangled in debates about the nature of communication. To what extent can meaning be controlled by individuals, like the values of a and b in algebra, and to what extent is meaning inevitably social? By attending to this long-neglected question, we come to see that the modern idea of the algorithm is implicated in a long history of attempts to maintain a disciplinary boundary separating technical knowledge from the languages people speak day to day. Machine learning, in its increasing dependence on words, now places this boundary in jeopardy, making its stakes all the more urgent to understand. The idea of the algorithm is a levee holding back the social complexity of language, and it is about to break. This book is about the flood that inspired its construction. (shrink)

The growing advancement of Artificial Intelligence models based on deep learning and the consequent popularization of large language models (LLMs), such as ChatGPT, place the academic community facing unprecedented dilemmas, in addition to corroborating questions involving research activities and human beings. In this work, Content Analysis was chosen as the object of study, an important technique for analyzing qualitative data and frequently used among Brazilian researchers. The objective of this work was to compare the process (...) of categorization by themes carried out by human researchers on material from the Educational area, at the end of 2022, with the help of the ATLAS.TI software and the same process of categorization by themes, now carried out by the version 2023 of the ATLAS.TI software, carried out in an automated way by Artificial Intelligence, for the same material. Through lexical and semantic analysis, the themes used in the two processes were compared, observing similarities and differences between the two proposals, and verifying the consistency between the two analyses. The main result of this study is the difficulty of Artificial Intelligence in carrying out a broader categorization, eliminating specificities that were considered irrelevant by the authors for understanding the material. It is concluded, therefore, that the automated categorization process proposed by ATLAS.TI through Natural Language Processing (NLP) is consistent from a lexical point of view but is still insufficient from a semantic aspect. (shrink)

This article engages the anthropological assumptions underlying the apprehensions and promises associated with language in artificial intelligence (AI). First, we present the contours of two rivalling paradigms for assessing artificial language generation: a holistic-enactivist theory of language and an informational theory of language. We then introduce two language generation models – one presently in use and one more speculative: Firstly, the transformer architecture as used in current large language models, such as (...) the GPT-series, and secondly, a model for 'autonomous machine intelligence' recently proposed by Yann LeCun, which involves not only language but a sensory-motor interaction with the world. We then assess the language capacity of these models from the perspectives of the two rivalling language paradigms. Taking a holistic-enactivist stance, we then argue that there is currently no reason to assume a human-comparable language capacity in LLMs and, further, that LeCun's proposed model does not represent a significant step toward artificially generating human language because it still lacks essential features that underlie the linguistic capacity of humans. Finally, we suggest that proponents of these rivalling interpretations of LLMs should enter into a constructive dialogue and that this dialogue should continuously involve further empirical, conceptual, and theoretical research. (shrink)

In his book “Minds and Computers: An Introduction to the Philosophy of Artificial Intelligence”, Matt Carter presents a comprehensive exploration of the philosophical questions surrounding artificial intelligence (AI). Carter argues that the development of AI is not merely a technological challenge but fundamentally a philosophical one. He delves into key issues like the nature of mental states, the limits of introspection, the implications of memory decay, and the functionalist framework that allows for the possibility of (...) AI. Carter contrasts functionalism with reductive materialism, highlighting how the former accommodates the concept of artificial minds. He also emphasizes the significance of computationalism, which combines functionalism with computational theory to provide a robust explanation for mental processes. The book further discusses formal systems, the role of register machines in computation, and the inherent challenges in AI research, particularly in developing natural language processing systems. Carter’s work underscores the limitations of current computational models of cognition while remaining hopeful that advances in neuroscience may lead to stronger AI systems in the future. (shrink)

This three-volume set, LNCS 14325-14327 constitutes the thoroughly refereed proceedings of the 20th Pacific Rim Conference on Artificial Intelligence, PRICAI 2023, held in Jakarta, Indonesia, in November 2023. The 95 full papers and 36 short papers presented in these volumes were carefully reviewed and selected from 422 submissions. PRICAI covers a wide range of topics in the areas of social and economic importance for countries in the Pacific Rim: artificial intelligence, machine learning, natural (...) language processing, knowledge representation and reasoning, planning and scheduling, computer vision, distributed artificial intelligence, search methodologies, etc. (shrink)

The article describes the possibilities of using and modifying existing machine learning technologies in the field of natural language processing for the purpose of designing a system for automatically generating control and test tasks (CTT). The reason for such studies was the limitations in generating theminimumrequired amount ofCTtomaintain student engagement in game-based learning formats, such as quizzes, and others. These limitations are associated with the lack of time resources among training professionals for manual generation (...) of tests. The article discusses the applied research of the Large Language Model (LLM) and Generative pre-trained transformer (GPT) technologies for the development of a system for automatic generation of tests for the purpose of its implementation in the BoxBattle gamified learning platform. The result of such applied research can be a system for automatic generation of tests, which will reduce the time for developing tests. As a result, this will allow teachers to free up time to implement a personalized approach to teaching and develop students’soft skills. (shrink)

Some philosophers and machine learning experts have speculated that superintelligent Artificial Intelligences (AIs), if and when they arrive on the scene, will wrestle away power from humans, with potentially catastrophic consequences. Dan Hendrycks has recently buttressed such worries by arguing that AI systems will undergo evolution by natural selection, which will endow them with instinctive drives for self-preservation, dominance and resource accumulation that are typical of evolved creatures. In this paper, we argue that this argument is (...) not compelling as it stands. Evolutionary processes, as we point out, can be more or less Darwinian along a number of dimensions. Making use of Peter Godfrey-Smith’s framework of Darwinian spaces, we argue that the more evolution is top-down, directed and driven by intelligent agency, the less paradigmatically Darwinian it becomes. We then apply the concept of “domestication” to AI evolution, which, although theoretically satisfying the minimal definition of natural selection, is channeled through the minds of fore-sighted and intelligent agents, based on selection criteria desirable to them (which could be traits like docility, obedience and non-aggression). In the presence of such intelligent planning, it is not clear that selection of AIs, even selection in a competitive and ruthless market environment, will end up favoring “selfish” traits. In the end, however, we do agree with Hendrycks’ conditionally: If superintelligent AIs end up “going feral” and competing in a truly Darwinian fashion, reproducing autonomously and without human supervision, this could pose a grave danger to human societies. (shrink)

This paper explores two important ways in which the practice of close reading differs from the technique of natural language processing, the use of computer programming to decode, process, and replicate messages within a human language. It does so in order to highlight distinctive features of close reading that are not replicated by natural language processing. The first point of distinction concerns the nature of the meaning generated in each case. While natural (...) language processing proceeds on the principle that a text’s meaning can be deciphered by applying the rules governing the language in which the text is written, close reading is premised on the idea that this meaning lies in the interplay that the text prompts within readers. While the semantic theory of meaning upon which natural language processing programs are based is often taken for granted today, I draw from phenomenological and hermeneutic theories, particularly Wolfgang Iser and Hans-Georg Gadamer, to explain why a different theory of meaning is necessary for understanding the meaning generated by close reading. Second, while natural language processing programs are considered successful when they generate what epistemologists call true beliefs about a text, I argue that close reading aims first and foremost at the development, not of true belief, but of understanding. To develop this distinction, I draw from recent scholarship on the epistemology of education, including work by Duncan Pritchard, to explain how understanding differs from true belief and why attainment of the latter is less educationally significant than the former. (shrink)

The first issue of Artificial Intelligence and Law journal was published in 1992. This paper offers some commentaries on papers drawn from the Journal’s third decade. They indicate a major shift within Artificial Intelligence, both generally and in AI and Law: away from symbolic techniques to those based on Machine Learning approaches, especially those based on Natural Language texts rather than feature sets. Eight papers are discussed: two concern the management and use (...) of documents available on the World Wide Web, and six apply machine learning techniques to a variety of legal applications. (shrink)

The field of artificial intelligence (AI) can be characterized as the investigation of computational systems that exhibit intelligent behavior (including algorithms and models used in these systems). The emphasis is not so much on understanding (human) cognitive processes as on producing models, algorithms, and systems that are capable of apparently intelligent behavior by whatever means available. The idea of AI has had a long history that can be traced all the way back to, for example, Leibniz. The idea (...) was furthered through the development, early in this century, of mathematical logic, cybernetics, and information theory, all of which contributed to AI theorizing. Alan Turing's bold claim that machines can think, advanced in his famous 1950 paper, and the “Turing test” he proposed therein as a means to test the intelligence of machines predated and prompted the formation of AI. Six years later, the Dartmouth conference in 1956, the participants of which included Allen Newell, Herbert Simon, Marvin Minsky, and John McCarthy, inaugurated AI as an academic discipline. This contemporary version of AI was facilitated by the invention and rapidly widening use of digital computers. AI embodies an extremely diverse set of ideas that are often mutually conflicting (as in the case of symbolism versus connectionism), and its relationship with other fields studying cognition (such as psychology and linguistics) has been a close but uneasy one. (shrink)

The notion of computation has changed the world more than any previous expressions of knowledge. However, as know-how in its particular algorithmic embodiment, computation is closed to meaning. Therefore, computer-based data processing can only mimic life’s creative aspects, without being creative itself. AI’s current record of accomplishments shows that it automates tasks associated with intelligence, without being intelligent itself. Mistaking the abstract for the concrete has led to the religion of “everything is an output of computation”—even the humankind (...) that conceived the computer. The hypostatized role of computers explains the increased dependence on them. The convergence machine called deep learning is only the most recent form through which the deterministic theology of the machine claims more than what it actually is: extremely effective data processing. A proper understanding of complexity, as well as the need to distinguish between the reactive nature of the artificial and the anticipatory nature of the living are suggested as practical responses to the challenges posed by machine theology. (shrink)

ABSTRACT I argue that this examination and appreciation for the shift to abductive reasoning should be extended to the intersection of neuroscience and novel brain-computer interfaces too. This paper highlights the implications of applying abductive reasoning to personalized implantable neurotechnologies. Then, it explores whether abductive reasoning is sufficient to justify insurance coverage for devices absent widespread clinical trials, which are better applied to one-size-fits-all treatments. INTRODUCTION In contrast to the classic model of randomized-control trials, often with a large number of (...) subjects enrolled, precision medicine attempts to optimize therapeutic outcomes by focusing on the individual.[i] A recent publication highlights the strengths and weakness of both traditional evidence-based medicine and precision medicine.[ii] Plus, it outlines a tension in the shift from evidence-based medicine’s inductive reasoning style (the collection of data to postulate general theories) to precision medicine’s abductive reasoning style (the generation of an idea from the limited data available).[iii] The paper’s main example is the application of precision medicine for the treatment of cancer.[iv] I argue that this examination and appreciation for the shift to abductive reasoning should be extended to the intersection of neuroscience and novel brain-computer interfaces too. As the name suggests, brain-computer interfaces are a significant advancement in neurotechnology that directly connects someone’s brain to external or implanted devices.[v] Among the various kinds of brain-computer interfaces, adaptive deep brain stimulation devices require numerous personalized adjustments to their settings during the implantation and computation stages in order to provide adequate relief to patients with treatment-resistant disorders. What makes these devices unique is how adaptive deep brain stimulation integrates a sensory component to initiate the stimulation. While not commonly at the level of sophistication as self-supervising or generative large language models,[vi] they currently allow for a semi-autonomous form of neuromodulation. This paper highlights the implications of applying abductive reasoning to personalized implantable neurotechnologies. Then, it explores whether abductive reasoning is sufficient to justify insurance coverage for devices absent widespread clinical trials, which are better applied to one-size-fits-all treatments.[vii] ANALYSIS I. The State of Precision Medicine in Oncology and the Epistemological Shift While a thorough overview of precision medicine for the treatment of cancer is beyond the scope of this article, its practice can be roughly summarized as identifying clinically significant characteristics a patient possesses (e.g., genetic traits) to land on a specialized treatment option that, theoretically, should benefit the patient the most.[viii] However, in such a practice of stratification patients fall into smaller and smaller populations and the quality of evidence that can be applied to anyone outside these decreases in turn.[ix] As inductive logic helps to articulate, the greater the number of patients that respond to a particular therapy the higher the probability of its efficacy. By straying from this logical framework, precision medicine opens the treatment of cancer to more uncertainty about the validity of these approaches to the resulting disease subcategories.[x] Thus, while contemporary medical practices explicitly describe some treatments as “personalized”, they ought not be viewed as inherently better founded than other therapies.[xi] A relevant contemporary case of precision medicine out of Norway focuses on the care of a patient with cancer between the ventricles of the heart and esophagus, which had failed to respond to the standard regimen of therapies over four years.[xii] In a last-ditch effort, the patient elected to pay out-of-pocket for an experimental immunotherapy (nivolumab) at a private hospital. He experienced marked improvements and a reduction in the size of the tumor. Understandably, the patient tried to pursue further rounds of nivolumab at a public hospital. However, the hospital initially declined to pay for it given the “lack of evidence from randomised clinical trials for this drug relating to this [patient’s] condition.”[xiii] In rebuttal to this claim, the patient countered that he was actually similar to a subpopulation of patients who responded in “open‐label, single arm, phase 2 studies on another immune therapy drug” (pembrolizumab).[xiv] Given this interpretation of the prior studies and the patient’s response, further rounds of nivolumab were approved. Had the patient not had improvements in the tumor’s size following a round of nivolumab, then pembrolizumab’s prior empirical evidence in isolation would have been insufficient, inductively speaking, to justify his continued use of nivolumab.[xv] The case demonstrates a shift in reasoning from the traditional induction to abduction. The phenomenon of ‘cancer improvement’ is considered causally linked to nivolumab and its underlying physiological mechanisms.[xvi] However, “the weakness of abductions is that there may always be some other better, unknown explanation for an effect. The patient may for example belong to a special subgroup that spontaneously improves, or the change may be a placebo effect. This does not mean, however, that abductive inferences cannot be strong or reasonable, in the sense that they can make a conclusion probable.”[xvii] To demonstrate the limitations of relying on the abductive standard in isolation, commentators have pointed out that side effects in precision medicine are hard to rule out as being related to the initial intervention itself unless trends from a group of patients are taken into consideration.[xviii] As artificial intelligence (AI) assists the development of precision medicine for oncology, this uncertainty ought to be taken into consideration. The implementation of AI has been crucial to the development of precision medicine by providing a way to combine large patient datasets or a single patient with a large number of unique variables with machine learning to recommend matches based on statistics and probability of success upon which practitioners can base medical recommendations.[xix] The AI is usually not establishing a causal relationship[xx] – it is predicting. So, as AI bleeds into medical devices, like brain-computer interfaces, the same cautions about using abductive reasoning alone should be carried over. II. Responsive Neurostimulation, AI, and Personalized Medicine Like precision medicine in cancer treatment, computer-brain interface technology similarly focuses on the individual patient through personalized settings. In order to properly expose the intersection of AI, precision medicine, abductive reasoning, and implantable neurotechnologies, the descriptions of adaptive deep brain stimulation systems need to deepen.[xxi] As a broad summary of adaptive deep brain stimulation, to provide a patient with the therapeutic stimulation, a neural signal, typically referred to as a local field potential,[xxii] must first be detected and then interpreted by the device. The main adaptive deep brain stimulation device with premarket approval, the NeuroPace Responsive Neurostimulation system, is used to treat epilepsy by detecting and storing “programmer-defined phenomena.”[xxiii] Providers can optimize the detection settings of the device to align with the patient’s unique electrographic seizures as well as personalize the reacting stimulation’s parameters.[xxiv] The provider adjusts the technology based on trial and error. One day machine learning algorithms will be able to regularly aid this process in myriad ways, such as by identifying the specific stimulation settings a patient may respond to ahead of time based on their electrophysiological signatures.[xxv] Either way, with AI or programmers, adaptive neurostimulation technologies are individualized and therefore operate in line with precision medicine rather than standard treatments based on large clinical trials. Contemporary neurostimulation devices are not usually sophisticated enough to be prominent in AI discussions where the topics of neural networks, deep learning, generative models, and self-attention dominate the conversation. However, implantable high-density electrocorticography arrays (a much more sensitive version than adaptive deep brain stimulation systems use) have been used in combination with neural networks to help patients with neurologic deficits from a prior stroke “speak” through a virtual avatar.[xxvi] In some experimental situations, algorithms are optimizing stimulation parameters with increasing levels of independence.[xxvii] An example of neurostimulation that is analogous to the use of nivolumab in Norway surrounds a patient in the United States who was experiencing both treatment-resistant OCD and temporal lobe epilepsy.[xxviii]Given the refractory nature of her epilepsy, implantation of an adaptive deep brain stimulation system was indicated. As a form of experimental therapy, her treatment-resistant OCD was also indicated for the off-label use of an adaptive deep brain stimulation set-up. Another deep brain stimulation lead, other than the one implanted for epilepsy, was placed in the patient’s right nucleus accumbens and ventral pallidum region given the correlation these nuclei had with OCD symptoms in prior research. Following this, the patient underwent “1) ambulatory, patient-initiated magnet-swipe storage of data during moments of obsessive thoughts; (2) lab-based, naturalistic provocation of OCD-related distress (naturalistic provocation task); and (3) lab-based, VR [virtual reality] provocation of OCD-related distress (VR provocation task).”[xxix] Such signals were used to identify when to deliver the therapeutic stimulation in order to counter the OCD symptoms. Thankfully, following the procedure and calibration the patient exhibited marked improvements in their OCD symptoms and recently shared her results publicly.[xxx] In both cases, there is a similar level of abductive justification for the efficacy of the delivered therapy. In the case study in which the patient was treated with adaptive deep brain stimulation, they at least had their neural activity tested in various settings to determine the optimum parameters for treatment to avoid them being based on guesswork. Additionally, the adaptive deep brain stimulation lead was already placed before the calibration trials were conducted, meaning that the patient had already taken on the bulk of the procedural risk before the efficacy could be determined. Such an efficacy test could have been replicated in the first patient’s cancer treatment, had it been biopsied and tested against the remaining immunotherapies in vitro. Yet, in the case of cancer with few options, one previous dose of a drug that appeared to work on the patient may justify further doses. However, as the Norwegian case presents, corroboration with known responses to a similar drug (from a clinical trial) could be helpful to validate the treatment strategy. (It should be noted that both patients were resigned to these last resort options regardless of the efficacy of treatment.) There are some elements of inductive logic seen with adaptive deep brain stimulation research in general. For example, abductively the focus could be that patient X’s stimulation parameters are different from patient Y’s and patient Z’s. In contrast, when grouped as subjects who obtained personalized stimulation, patients X, Y, and Z demonstrate an inductive aspect to this approach’s safety and/or efficacy. The OCD case holds plenty of abductive characteristics in line with precision medicine’s approach to treating cancer and as more individuals try the method, there will be additional data. With the gradual integration of AI into brain-computer interfaces in the name of efficacy, this reliance on abduction will continue, if not grow, over time. Moving forward, if a responsive deep brain stimulation treatment is novel and individualized (like the dose of nivolumab) and there is some other suggestion of efficacy (like clinical similarities to other patients in the literature), then it may justify insurance coverage for the investigative intervention, absent other unrelated reasons to deny it. III. Ethical Implications and Next Steps While AI’s use in oncology and neurology is not yet as prominent as its use in other fields (e.g., radiology), it appears to be on the horizon for both.[xxxi] AI can be found in both the functioning of the neurotechnologies as well as the implementation of precision medicine. The increasing use of AI may serve to further individualize both oncologic and neurological therapies. Given these implications and the handful of publications cited in this article, it is important to have a nuanced evaluation of how these treatments, which heavily rely on abductive justification, ought to be managed. The just use an abductive approach may be difficult as AI infused precision medicine is further pursued. At baseline, such technology relies on a level of advanced technology literacy among the general public and could exclude populations who lack access to basic technological infrastructure or know-how from participation.[xxxii] Even among nations with adequate infrastructure, as more patients seek out implantable neurotechnologies, which require robust healthcare resources, the market will favor patient populations that can afford this complex care.[xxxiii] If patients already have the means to pay for an initial dose/use of a precision medicine product out of pocket, should insurance providers be required to cover subsequent treatments?[xxxiv] That is, if a first dose of a cancer drug or a deep brain stimulator over its initial battery life is successful, patients may feel justified in having the costs of further treatments covered. The Norwegian patient’s experience implies there is a precedent for the idea that some public insurance companies ought to cover successful cancer therapies, however, insurance companies may not all see themselves as obligated to cover neurotechnologies that rely on personalized settings or that are based on precision/abductive research more than on clinical trials. CONCLUSION The fact that the cases outlined above rely on abductive style of reasoning implies that there may not be as strong a justification for coverage by insurance, as they are both experimental and individualized, when compared to the more traditional large clinical trials in which groups have the same or a standardized protocol (settings/doses). If a study is examining the efficacy of a treatment with a large cohort of patients or with different experimental groups/phases, insurance companies may conclude that the resulting symptom improvements are more likely to be coming from the devices themselves. A preference for inductive justification may take priority when ruling in favor of funding someone’s continued use of an implantable neurostimulator. There are further nuances to this discussion surrounding the classifications of these interventions as research versus clinical care that warrant future exploration, since such a distinction is more of a scale[xxxv] than binary and could have significant impacts on the “right-to-try” approach to experimental therapies in the United States.[xxxvi] Namely, given the inherent limitations of conducting large cohort trials for deep brain stimulation interventions on patients with neuropsychiatric disorders, surgically innovative frameworks that blend abductive and inductive methodologies, like with sham stimulation phases, have traditionally been used.[xxxvii] Similarly, for adaptive brain-computer interface systems, if there are no large clinical trials and instead only publications that demonstrate that something similar worked for someone else, then, in addition to the evidence that the first treatment/dose worked for the patient in question, the balance of reasoning would be valid and arguably justify insurance coverage. As precision approaches to neurotechnology become more common, frameworks for evaluating efficacy will be crucial both for insurance coverage and for clinical decision making. ACKNOWLEDGEMENT This article was originally written as an assignment for Dr. Francis Shen’s “Bioethics & AI” course at Harvard’s Center for Bioethics. I would like to thank Dr. Shen for his comments as well as my colleagues in the Lázaro-Muñoz Lab for their feedback. - [i] Jonathan Kimmelman and Ian Tannock, “The Paradox of Precision Medicine,” Nature Reviews. Clinical Oncology 15, no. 6 (June 2018): 341–42, https://doi.org/10.1038/s41571-018-0016-0. [ii] Henrik Vogt and Bjørn Hofmann, “How Precision Medicine Changes Medical Epistemology: A Formative Case from Norway,” Journal of Evaluation in Clinical Practice 28, no. 6 (December 2022): 1205–12, https://doi.org/10.1111/jep.13649. [iii] David Barrett and Ahtisham Younas, “Induction, Deduction and Abduction,” Evidence-Based Nursing 27, no. 1 (January 1, 2024): 6–7, https://doi.org/10.1136/ebnurs-2023-103873. [iv] Vogt and Hofmann, “How Precision Medicine Changes Medical Epistemology,” 1208. [v] Wireko Andrew Awuah et al., “Bridging Minds and Machines: The Recent Advances of Brain-Computer Interfaces in Neurological and Neurosurgical Applications,” World Neurosurgery, May 22, 2024, S1878-8750(24)00867-2, https://doi.org/10.1016/j.wneu.2024.05.104. [vi] Mark Riedl, “A Very Gentle Introduction to Large Language Models without the Hype,” Medium (blog), May 25, 2023, https://mark-riedl.medium.com/a-very-gentle-introduction-to-large-language-models-without-the-hype-5f67941fa59e. [vii] David E. Burdette and Barbara E. Swartz, “Chapter 4 - Responsive Neurostimulation,” in Neurostimulation for Epilepsy, ed. Vikram R. Rao (Academic Press, 2023), 97–132, https://doi.org/10.1016/B978-0-323-91702-5.00002-5. [viii] Kimmelman and Tannock, 2018. [ix] Kimmelman and Tannock, 2018. [x] Simon Lohse, “Mapping Uncertainty in Precision Medicine: A Systematic Scoping Review,” Journal of Evaluation in Clinical Practice 29, no. 3 (April 2023): 554–64, https://doi.org/10.1111/jep.13789. [xi] Kimmelman and Tannock, “The Paradox of Precision Medicine.” [xii] Vogt and Hofmann, 1206. [xiii] Vogt and Hofmann, 1206. [xiv] Vogt and Hofmann, 1206. [xv] Vogt and Hofmann, 1207. [xvi] Vogt and Hofmann, 1207. [xvii] Vogt and Hofmann, 1207. [xviii] Vogt and Hofmann, 1210. [xix] Mehar Sahu et al., “Chapter Three - Artificial Intelligence and Machine Learning in Precision Medicine: A Paradigm Shift in Big Data Analysis,” in Progress in Molecular Biology and Translational Science, ed. David B. Teplow, vol. 190, 1 vols., Precision Medicine (Academic Press, 2022), 57–100, https://doi.org/10.1016/bs.pmbts.2022.03.002. [xx] Stefan Feuerriegel et al., “Causal Machine Learning for Predicting Treatment Outcomes,” Nature Medicine 30, no. 4 (April 2024): 958–68, https://doi.org/10.1038/s41591-024-02902-1. [xxi] Sunderland Baker et al., “Ethical Considerations in Closed Loop Deep Brain Stimulation,” Deep Brain Stimulation 3 (October 1, 2023): 8–15, https://doi.org/10.1016/j.jdbs.2023.11.001. [xxii] David Haslacher et al., “AI for Brain-Computer Interfaces,” 2024, 7, https://doi.org/10.1016/bs.dnb.2024.02.003. [xxiii] Burdette and Swartz, “Chapter 4 - Responsive Neurostimulation,” 103–4; “Premarket Approval (PMA),” https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpma/pma.cfm?id=P100026. [xxiv] Burdette and Swartz, “Chapter 4 - Responsive Neurostimulation,” 104. [xxv] Burdette and Swartz, 126. [xxvi] Sean L. Metzger et al., “A High-Performance Neuroprosthesis for Speech Decoding and Avatar Control,” Nature 620, no. 7976 (August 2023): 1037–46, https://doi.org/10.1038/s41586-023-06443-4. [xxvii] Hao Fang and Yuxiao Yang, “Predictive Neuromodulation of Cingulo-Frontal Neural Dynamics in Major Depressive Disorder Using a Brain-Computer Interface System: A Simulation Study,” Frontiers in Computational Neuroscience 17 (March 6, 2023), https://doi.org/10.3389/fncom.2023.1119685; Mahsa Malekmohammadi et al., “Kinematic Adaptive Deep Brain Stimulation for Resting Tremor in Parkinson’s Disease,” Movement Disorders 31, no. 3 (2016): 426–28, https://doi.org/10.1002/mds.26482. [xxviii] Young-Hoon Nho et al., “Responsive Deep Brain Stimulation Guided by Ventral Striatal Electrophysiology of Obsession Durably Ameliorates Compulsion,” Neuron 0, no. 0 (October 20, 2023), https://doi.org/10.1016/j.neuron.2023.09.034. [xxix] Nho et al. [xxx] Nho et al.; Erik Robinson, “Brain Implant at OHSU Successfully Controls Both Seizures and OCD,” OHSU News, accessed March 3, 2024, https://news.ohsu.edu/2023/10/25/brain-implant-at-ohsu-successfully-controls-both-seizures-and-ocd. [xxxi] Awuah et al., “Bridging Minds and Machines”; Haslacher et al., “AI for Brain-Computer Interfaces.” [xxxii] Awuah et al., “Bridging Minds and Machines.” [xxxiii] Sara Green, Barbara Prainsack, and Maya Sabatello, “The Roots of (in)Equity in Precision Medicine: Gaps in the Discourse,” Personalized Medicine 21, no. 1 (January 2024): 5–9, https://doi.org/10.2217/pme-2023-0097. [xxxiv] Green, Prainsack, and Sabatello, 7. [xxxv] Robyn Bluhm and Kirstin Borgerson, “An Epistemic Argument for Research-Practice Integration in Medicine,” The Journal of Medicine and Philosophy: A Forum for Bioethics and Philosophy of Medicine 43, no. 4 (July 9, 2018): 469–84, https://doi.org/10.1093/jmp/jhy009. [xxxvi] Vijay Mahant, “‘Right-to-Try’ Experimental Drugs: An Overview,” Journal of Translational Medicine 18 (June 23, 2020): 253, https://doi.org/10.1186/s12967-020-02427-4. [xxxvii] Michael S. Okun et al., “Deep Brain Stimulation in the Internal Capsule and Nucleus Accumbens Region: Responses Observed during Active and Sham Programming,” Journal of Neurology, Neurosurgery & Psychiatry 78, no. 3 (March 1, 2007): 310–14, https://doi.org/10.1136/jnnp.2006.095315. (shrink)

This article offers a critical inquiry of contemporary neural network models as an instance of machine learning, from an interdisciplinary perspective of AI studies and performativity. It shows the limits on the architecture of these network systems due to the misemployment of ‘natural’ performance, and it offers ‘context’ as a variable from a performative approach, instead of a constant. The article begins with a brief review of machine learning-based natural language processing systems (...) and continues with a concentration on the relevant model of recurrent neural networks, which is applied in most commercial research such as Facebook AI Research. It demonstrates that the logic of performativity is not brought into account in all recurrent nets, which is an integral part of human performance and languaging, and it argues that recurrent network models, in particular, fail to grasp human performativity. This logic works similarly to the theory of performativity articulated by Jacques Derrida in his critique of John L. Austin’s concept of the performative. Applying Jacques Derrida’s work on performativity, and linguistic traces as spatially organized entities that allow for this notion of performance, the article argues that recurrent nets fall into the trap of taking ‘context’ as a constant, of treating human performance as a ‘natural’ fix to be encoded, instead of performative. Lastly, the article applies its proposal more concretely to the case of Facebook AI Research’s Alice and Bob. (shrink)

The COVID-19 pandemic and its related policies (e.g., stay at home and social distancing orders) have increased people’s use of digital technology, such as social media. Researchers have, in turn, utilized artificial intelligence to analyze social media data for public health surveillance. For example, through machine learning and natural language processing, they have monitored social media data to examine public knowledge and behavior. This paper explores the ethical considerations of using artificial (...) class='Hi'>intelligence to monitor social media to understand the public’s perspectives and behaviors surrounding COVID-19, including potential risks and benefits of an AI-driven approach. Importantly, investigators and ethics committees have a role in ensuring that researchers adhere to ethical principles of respect for persons, beneficence, and justice in a way that moves science forward while ensuring public safety and confidence in the process. (shrink)

During the past decade, a small but rapidly growing number of Law&Tech scholars have been applying algorithmic methods in their legal research. This Article does it too, for the sake of saving disclosure regulation failure: a normative strategy that has long been considered dead by legal scholars, but conspicuously abused by rule-makers. Existing proposals to revive disclosure duties, however, either focus on the industry policies (e.g. seeking to reduce consumers’ costs of reading) or on rulemaking (e.g. by simplifying linguistic intricacies). (...) But failure may well depend on both. Therefore, this Article develops a `comprehensive approach', suggesting to use computational tools to cope with linguistic and behavioral failures at both the enactment and implementation phases of disclosure duties, thus filling a void in the Law & Tech scholarship. Specifically, it outlines how algorithmic tools can be used in a holistic manner to address the many failures of disclosures from the rulemaking in parliament to consumer screens. It suggests a multi-layered design where lawmakers deploy three tools in order to produce optimal disclosure rules: machine learning, natural language processing, and behavioral experimentation through regulatory sandboxes. To clarify how and why these tasks should be performed, disclosures in the contexts of online contract terms and privacy online are taken as examples. Because algorithmic rulemaking is frequently met with well-justified skepticism, problems of its compatibility with legitimacy, efficacy and proportionality are also discussed. (shrink)

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

Whether it was John Searle’s Chinese Room argument (Searle, 1980) or Roger Penrose’s argument of the non-computable nature of a mathematician’s insight – an argument that was based on Gödel’s Incompleteness theorem (Penrose, 1989), we have always had skeptics that questioned the possibility of realizing strong Artificial Intelligence (AI), or what has become known by Artificial General Intelligence (AGI). But this new book by Landgrebe and Smith (henceforth, L&S) is perhaps the strongest argument ever made against (...) strong AI. It is a very extensive review of what building a mind essentially amounts to drawing on insights and results from biology, physics, linguistics, computability, philosophy, and mathematics. (shrink)

For over a decade John Searle's ingenious argument against the possibility of artificial intelligence has held a prominent place in contemporary philosophy. This is not just because of its striking central example and the apparent simplicity of its argument. As its appearance in Scientific American testifies, it is also due to its importance to the wider scientific community. If Searle is right, artificial intelligence in the strict sense, the sense that would claim that mind can (...) be instantiated through a formal program of symbol manipulation, is basically wrong. No set of formal conditions can provide us with the characteristic feature of mind which is the intentionally of its mental contents. Formally regarded, such intentionally is an irreducible primitive. It cannot be analyzed into non-intentional (purely syntactic, symbolic) components. This paper will argue that this objection is based on a misunderstanding. Intentionality is not simply something given which is incapable of further analysis. It only appears so when we mistakenly abstract it from time. When we regard its temporal structure, it shows itself as a rule-governed, synthetic process, one capable of being instantiated both by machines and men. (shrink)