The Ethics of Artificial Intelligence Conference

Abstracts (in alphabetical order) 

Professor BRADLEY Adam (Lingnan University)

AI Alignment VS AI Ethical Treatment: Ten Challenges (with Bradford Saad)

A morally acceptable course of AI development should avoid two dangers: creating unaligned AI systems that pose a threat to humanity and mistreating AI systems that merit moral consideration in their own right. This paper argues these two dangers interact and that if we create AI systems that merit moral consideration, simultaneously avoiding both of these dangers would be extremely challenging. While our argument is straightforward and supported by a wide range of pretheoretical moral judgments, it has far-reaching moral implications for AI development. Although the most obvious way to avoid the tension between alignment and ethical treatment would be to avoid creating AI systems that merit moral consideration, this option may be unrealistic and is perhaps fleeting. So, we conclude by offering some suggestions for other ways of mitigating mistreatment risks associated with alignment.

Professor CAPPELEN Herman (The University of Hong Kong)

Engineering “Empathy”: How AI is Reshaping Our Moral Language

This talk examines the ethical and linguistic implications of developing AI systems that mimic empathy and friendliness, on the assumption that current AI lacks the emotional and cognitive capacities for genuine empathy and friendliness. The first part addresses ethical questions: Does simulating empathy and friendliness amount to deception? Are current AI development and marketing practices by companies tantamount to lying? The second part explores the linguistic implications: Could the widespread use of AI systems that simulate empathy and friendliness lead to shifts in the meanings of these terms? Is this a case of conceptual engineering, and if so, what are the potential consequences of such linguistic evolution? The third part draws on insights from the second part to challenge the assumption that current AI systems lack the capacity for empathy and friendliness.

Dr. HIPÓLITO Inês (Macquarie University)

The Human Roots of AI

This paper rejects the notion of AI as a neutral technology, instead framing it as an intrinsically social tool shaped by cultural practices and power dynamics. We argue that once AI is understood as a product of specific cultural and epistemic communities, the critical question becomes how these AI tools differentially impact various social groups. Our analysis reveals that AI's social nature leads to uneven distributions of benefits and risks across society. This perspective underscores the necessity of diverse cultural input in AI development and the importance of considering varied societal impacts to ensure more equitable and inclusive outcomes in AI design and implementation.

Professor LAZAR Seth (The Australian National University)

Evaluating LLM Ethical Competence

Existing approaches to evaluating LLM ethical competence place too much emphasis on the verdicts—of permissibility and impermissibility—that they render. But ethical competence doesn’t consist in one’s judgments conforming to those of a cohort of crowdworkers. It consists in being able to identify morally relevant features, prioritise among them, associate them with reasons and weave them into a justified conclusion. We identify the limitations of existing evals for ethical competence, provide an account of moral reasoning that can ground better alternatives, and discuss the practical—and philosophical—implications if LLMs ultimately do prove to be adept moral reasoners.

Professor O’NEILL Elizabeth (Eindhoven University of Technology)

_{Artificial Moral Discourse and the Future of Human Morality}

Many publicly accessible large language model (LLM)-based chatbots readily and flexibly generate outputs that look like moral assertions, advice, praise, expression of moral emotions, and other morally-significant communications. We can call this phenomenon “artificial moral discourse.” In the first part of this talk, I supply a characterization of artificial moral discourse. In the second part of the talk, I make a preliminary case for the claim that artificial moral discourse is likely to influence human norms and values in ways that past technologies have not. Namely, I propose that regular interaction with LLM-based chatbots can influence human morality via mechanisms that resemble modes of social influence on morality, such as influence via advice and testimony, influence via example, and influence via norm enforcement. Such influence could be orchestrated by humans seeking to advance particular worldviews or it could be exerted without any humans having intended the chatbot to have such an influence. I sketch what some of these paths of influence might look like. In concluding, I suggest some research questions for further empirical, technical, and philosophical investigation on how artificial moral discourse may influence human morality and what the ethical implications of that influence may be.

Professor SHEVLIN Henry (University of Cambridge)

Consciousness, Anthropomimesis, and Artificial Intelligence

Recent rapid progress in the development of Large Language Models has resulted in artificial systems with increasingly humanlike social and linguistic capabilities, and a growing number of AI tools explicitly cater towards meeting users’ social needs. This anthropomimetic turn in AI is likely to have significant ramifications for the future of human science and human society. In this talk, I examine three such possible impacts, as follows. First, I consider the immediate ethical questions prompted by social AI, including risks of social and emotional deskilling and harms to user well-being. Second, I outline ways in which increasingly anthropomimetic AI agents may make certain AI safety challenges more serious. Finally, I explore how deeper human-AI relationships may inform key debates in cognitive science concerning the psychological capacities of AI systems themselves.

Mr. TSE Yip Fai (recently Princeton University)

AI Alignment: The Case for Including Animals

AI alignment research is an emerging field of research concerned with developing machine intelligence. It is often defined in many different ways. Some hold that AI is "aligned" if and only if the AI is aligned with human interests, whereas others require it to be aligned with human intentions, and others again, with human values. Accordingly, AI alignment research has to date focused on these forms of alignment. But AI systems can benefit or harm nonhuman animals too, in significant ways, and therefore could be affected by efforts in AI alignment. In this talk, I argue for the position that nonhuman animals ought to be included in the project of AI alignment, at least in the sense of aligning AI with animals’ interests. I also attempt to respond to a few counterarguments raised against his position.

Professor VOLD Karina (University of Toronto)

The Many Value-Alignment Problems

The term ‘value alignment’ frequently enters into contemporary discussions around the ethics and safety of artificial intelligence. Sometimes this is phrased as the value-alignment problem, sometimes the goal of value-alignment, and sometimes the value-alignment research community, and so on. In this talk I argue that there are many value-alignment problems; I pull apart at least six different problems that get wrapped up under one title. I further want to argue that these problems, for the most part, cannot realistically be solved. Hence, it is most useful to think of value-alignment as a responsible design process, not a problem to be solved.

Professor YAO Xin (Lingnan University)

Ethical Risks and Challenges of Artificial Intelligence

As the rapid development and applications of artificial intelligence (AI), AI ethics has become an increasingly important interdisciplinary research area. This talk has two major parts. The first part gives a brief overview of AI ethics, starting from the basic concepts of ethics, applied ethics, technology ethics, information ethics to AI ethics, to the current status of the AI ethics research. We summarise current AI ethics research at three levels, i.e., at the individual, societal (group) and environmental levels. Some of the major research topics in AI ethics are highlighted. The second part of this talk will zoom in a couple of most important ethical issues in AI, e.g., fairness and explainability. We will highlight the challenges of defining fairness and explainability precisely, and the multi-objective nature of these ethical issues. Then some potential technical solutions to improve fairness and explainability are presented although not in any technical details. It is argued that many ethical considerations are inherently hard to quantify and cannot be measured by any single metric. A multi-objective approach is needed.

For enquiries, please call 2616 7445 or e-mail to [email protected].