Hendrik Kempt

After all conceptual tools are assembled, we can assess how chatbots can be understood as a candidate for social relationships. We reject the idea of anthropomorphism, as the negatives outweigh the positives. We also reject the idea that we relate to machines via mental states or gendered appearance. We discuss the idea of understanding chatbots as instances of the objective spirit but reject this on conceptual grounds. Instead, we can understand chatbots as opening up a similar social category like animals, through a “second domestication.” This allows for a plethora of non-anthropomorphic realizations.

LLMsLarge Language Models (LLM), general-purpose chatbots, and any more specific interactive conversational technology will be more socially relatable than any other technology before. This has thus been the premise under which philosophers have speculated about questions of moral standing, robot rightsRobot rights, and other terms to reckon with their different social and moral status. This chapter explores the relationality of these machines via the concept of positionality, investigates the consequences for our relationships with technology, and explores other values relevant in this context.

This chapter turns toward the consequences of human–human relationships when a relational approach to human–machine relationships is adopted. This happens in four ways. First, we enter the debate on robot rights and test the pragmacentrist approach. We find that a relational approach of this kind, similar to Gunkel’s and Coeckelbergh’s positions, defuses many of the issues motivating the robot rights debate. Second, we assess whether human-centered design requires some limitations to this relational approach, which it does. Third, we evaluate the relational real estate and propose to understand conversational AI as “digital persons” to explain issues of responsibility and blame. And fourth, we introduce new terminology to denote the fault lines creating social tension—robophobia and robophilia.

This chapter discusses the constitution of the social as a relational network. We propose to understand social relationships as embedded and consequential relationships. Through a pragmatic approach to assigning descriptors and thereby constituting them, we reject the idea of essentialist interpretations of social categories. The introduction of the concepts of domestication and digitization proves the points by showing that social categories have been expanded to include animals, and that physical presence is not required anymore for social relationships.

In this chapter, we propose a working definition of AI to narrow the inquiry to chatbots. We discuss the requirement of embodiment for constructing AI and reject its importance. After a short review of chatbots development, we discuss whether the current method of AI construction, machine learning and Large Language Models will remain a promising approach. Due to the increasing size of linguistic base models and more refined algorithms for open-domain chatbots, this is affirmative. Lastly, we discuss the Turing test as an estimate for the intelligence of chatbots and reject its relevance.

This chapter begins with a summary of the main findings of most chapters. It points out the argumentation path this project went down, from the introduction of domestication and digitization as key concepts for the research agenda to the robot rights debate and the terminological recommendations. This chapter ends on an encouragement to realize the relational approach by arguing for it in the public square. Same with the animal rights movement, the recognition of human–machine relationships will not happen by itself.

With the publication of chatGPT in December 2022, a new era of artificial dialogue systems and chatbots has emerged. Many established thinkers and figures have published their stance on the development of this technology, not rarely combined with stark language about its transformative potential and potential risks. New terminologies have found entrance to public and philosophical discourse, new regulatory frameworks have been introduced, from small to large scale, and an entire economy has shifted: many apps have sprung into existence using chatGPT and other LLMs as a basis for their services, large corporations take the risks of releasing unfinished language models to undercut the competition. In the meantime, many more stakeholders have entered the economic, regulatory, and social discourses on how to deal with this emerging technology. Taking the latest developments in the field into consideration this completely revised and updated second edition now includes new chapters on the tech-ethics of Large Language Models and LLMs as relational technology. The author considers the consequences of chatbots on human-human relationships, providing analysis on robot rights, human-centered design, and the social tension between robophobes and robophiles. Hendrik Kempt, PhD, is Postdoctoral scholar at the Applied Ethics Group at RWTH Aachen University. He has obtained his PhD from Aachen with works on the ethical concerns of using AI for medical diagnostics and has published several monographs on philosophical challenges regarding emerging technology, including "Synthetic Friends - A Philosophy of Human-Machine Friendship" (2022) and "(Un-)Explainable Technology" (2024), both with Palgrave Macmillan. He is also the author of a number of articles on related topics and the co-editor of "RuPaul's Drag Race and Philosophy" (2019).

In this chapter, we turn towards the ethical benefits of explainable technologies. After discussing several instrumental benefits of explainability, we turn towards the deontologically motivated debate of “right to explanation” and their arguments. In refuting that approach on theoretical and practical grounds, we instead introduce the idea of a “claim to expertise”, which preserves our practices and allows unexplainable technology to be added to these practices without much disruption. In discussion the consequences of unexplainable technology for democratic governance, this chapter provides some reasons why explainability might be useful politically.

In this chapter, we explore under what conditions explanations about technology amount to knowledge, and how this knowledge can be used further. In first determining that the level of granularity is already a purpose-driven one, we establish that there is no explainability without predetermined epistemic of practical interests. In pointing towards the different epistemic statuses of explanations and interpretations, we determine that interpretations do not provide knowledge-based explanations, but still produce some insight that are useful for practical purposes. Lastly, we turn to the question whether unexplainable technology can contribute to our scientific research, considering the results cannot be explained as other scientific knowledge is explained.

This chapter introduces the core philosophical questions of explainability and the innate human quest for explanations. We first critically assess the metaphor of AI as a black box and discuss how much we actually know about the functioning of AI. In entertaining a thought experiment about an unfailing oracle and the scrutiny it would receive today, we uncover practical and epistemic purposes for explanations, and transfer those to explanations of technological behavior. This way, we gather elements of what would make a good explanation for technology, and we set out the task for their epistemological and ethical questions.

Having elaborated on the epistemological conditions and the ethical interplay between explainability and expertise, we have found that explainability does not have a purpose beyond its instrumental value for other purposes. It might even be problematic to insist on explainability too rigorously. Since the applied cases of unexplained technology have demonstrated that explainability is often not the technique required for ethical permissibility, we can elaborate on whether explainability is practically relevant in the first place. Especially considering the emergence of ever-growing foundational models has undercut hopes to meaningfully increase explainability in AI.

This book explores the issue of (un)explainable technology. As we face technologies, mostly autonomous, machine-learned algorithms (AI) that elude a seamless explanation on how they work (“black boxes”), several issues both from an epistemological as well as ethical perspectives emerge. It is thus not surprising that there are plenty of technological attempts in illuminating the black box as well as philosophical efforts in conceptualizing and re- assessing our concepts of an explanation and understanding, as well emerging ethical questions on how to deal with this unexplainable technology. This book thus offers a succinct and comprehensive, opinionated but fair view on the emerging ethical debate on explainability of AI and its relevance for using AI for different more or less sensitive decision-making procedures. As a short book, the goal is to introduce the reader to the issues at hand while also offering normative arguments from different sides that motivate, complicate, and resolve these issues.

This paper explores the role and resolution of disagreements between physicians and their diagnostic AI-based decision support systems. With an ever-growing number of applications for these independently operating diagnostic tools, it becomes less and less clear what a physician ought to do in case their diagnosis is in faultless conflict with the results of the DSS. The consequences of such uncertainty can ultimately lead to effects detrimental to the intended purpose of such machines, e.g. by shifting the burden of proof towards a physician. Thus, we require normative clarity for integrating these machines without affecting established, trusted, and relied upon workflows. In reconstructing different causes of conflicts between physicians and their AI-based tools—inspired by the approach of “meaningful human control” over autonomous systems and the challenges to resolve them—we will delineate normative conditions for “meaningful disagreements”. These incorporate the potential of DSS to take on more tasks and outline how the moral responsibility of a physician can be preserved in an increasingly automated clinical work environment.

The increased presence of medical AI in clinical use raises the ethical question which standard of explainability is required for an acceptable and responsible implementation of AI-based applications in medical contexts. In this paper, we elaborate on the emerging debate surrounding the standards of explainability for medical AI. For this, we first distinguish several goods explainability is usually considered to contribute to the use of AI in general, and medical AI in specific. Second, we propose to understand the value of explainability relative to other available norms of explainable decision-making. Third, in pointing out that we usually accept heuristics and uses of bounded rationality for medical decision-making by physicians, we argue that the explainability of medical decisions should not be measured against an idealized diagnostic process, but according to practical considerations. We conclude, fourth, to resolve the issue of explainability-standards by relocating the issue to the AI’s certifiability and interpretability.

The first truly philosophical exploration of the drag queen in the context of this ground-breaking reality TV showAs RuPaul has said, this is the Golden Age of Drag-and that's chiefly the achievement of RuPaul's Drag Race, which in its eleventh year is more popular than ever, and has now become fully mainstream in its appeal. The show has an irresistible allure for folks of all persuasions and proclivities. Yet serious or philosophical discussion of its exponential success has been rare. Now at last we have RuPaul's Drag Race and Philosophy, shining the light on all dimensions of this amazing phenomenon: theories of gender construction and identity, interpretations of RuPaul's famous quotes and phrases, the paradoxes of reality shows, the phenomenology of the drag queen, and how the fake becomes the truly authentic.Among the topics covered:? How the show's controversial contestant The Vixen was able to proclaim her own truth by challenging the conventional stereotypes of reality TV.? If, as RuPaul has taught us, all social roles are equivalent to drag, what justifies the show's almost exclusive focus on the classical drag queen?? Nietzsche's insight that the self is plural rather than singular, divided rather than unified, helps us construct a new kind of self - the "drag self."? When two fan subcultures meet: The interpenetration of RuPaul's Drag Race and Twin Peaks in "Fire Werk with Me."? The lip-sync is the final challenge of the glamorous gauntlet of RuPaul's Drag Race, a game without rules or clear expectations that forces us to think about the meaning of play.? New technologies of surveillance analyze and classify us in order to fix our identities; we can defend ourselves by creating our own masks, to preserve our privacy and define our own identities.? Drag queens and mall Santas as performances of gender roles.

In this paper, we first classify different types of second opinions and evaluate the ethical and epistemological implications of providing those in a clinical context. Second, we discuss the issue of how artificial intelligent could replace the human cognitive labour of providing such second opinion and find that several AI reach the levels of accuracy and efficiency needed to clarify their use an urgent ethical issue. Third, we outline the normative conditions of how AI may be used as second opinion in clinical processes, weighing the benefits of its efficiency against concerns of responsibility attribution. Fourth, we provide a ‘rule of disagreement’ that fulfils these conditions while retaining some of the benefits of expanding the use of AI-based decision support systems in clinical contexts. This is because the rule of disagreement proposes to use AI as much as possible, but retain the ability to use human second opinions to resolve disagreements between AI and physician-in-charge. Fifth, we discuss some counterarguments.

The production of easy-read and plain language is a challenging task, requiring well-educated experts to write context-dependent simplifications of texts. Therefore, the domain of easy-read and plain language is currently restricted to the bare minimum of necessary information. Thus, even though there is a tendency to broaden the domain of easy-read and plain language, the inaccessibility of a significant amount of textual information excludes the target audience from partaking or entertainment and restricts their ability to live life autonomously. Large language models can solve a vast variety of natural language tasks, including the simplification of standard language texts to easy-read or plain language. Moreover, with the rise of generative models like GPT, easy-read and plain language may be applicable to all kinds of natural language texts, making formerly inaccessible information accessible to marginalized groups like, a.o., non-native speakers, and people with mental disabilities. In this paper, we argue for the feasibility of text simplification and generation in that context, outline the ethical dimensions, and discuss the implications for researchers in the field of ethics and computer science.

Large Language Models are one of the most prominent examples of current uses of AI, and one of the most urgently pursued normative tasks is to make them and their interactive user-interfaces—open-domain chatbots—safe. However, in this paper, we elaborate first on why such a limited view on the permissibility and desirability of their utterances falls conceptually flat, is philosophically insufficient, and leads to severe technological limits. We then propose a positive normative concept, appropriateness, that can provide the required orientation for determining what chatbots may or may not say, and distinguish between technical-discursive, social, and moral appropriateness. Lastly, we explore the requirements for social and conversational contexts to fulfill appropriateness-requirements, and find that positionality is a key element for their position in discourse.

The paper offers a normative analysis of the new Loss & Damage Fund supporting vulnerable countries grappling with climate change-related harms. This fund is primarily financed by affluent nations, often identified as historical polluters. However, the perspective of relational egalitarianism highlights persistent structural injustices in the background of the fund. Addressing them necessitates conceptualizing the fund not merely as an act of cooperative solidarity but as compensation for the consequences of historical and ongoing structural injustices. Properly conceived, the fund manages adverse climate effects and simultaneously tackles their underlying root causes, including the unjust distribution of advantages and power.

Providing healthcare services frequently involves cognitively demanding tasks, including diagnoses and analyses as well as complex decisions about treatments and therapy. From a global perspective, ethically significant inequalities exist between regions where the expert knowledge required for these tasks is scarce or abundant. One possible strategy to diminish such inequalities and increase healthcare opportunities in expert-scarce settings is to provide healthcare solutions involving digital technologies that do not necessarily require the presence of a human expert, e.g., in the form of artificial intelligent decision-support systems (AI-DSS). Such algorithmic decision-making, however, is mostly developed in resource- and expert-abundant settings to support healthcare experts in their work. As a practical consequence, the normative standards and requirements for such algorithmic decision-making in healthcare require the technology to be at least as explainable as the decisions made by the experts themselves. The goal of providing healthcare in settings where resources and expertise are scarce might come with a normative pull to lower the normative standards of using digital technologies in order to provide at least some healthcare in the first place. We scrutinize this tendency to lower standards in particular settings from a normative perspective, distinguish between different types of absolute and relative, local and global standards of explainability, and conclude by defending an ambitious and practicable standard of local relative explainability.