AI and ethics最新文献

Porn plays an ever-present role in the development, study, use and hype of both artificial intelligence (AI) and virtual reality (VR), yet the full extent to which AI, VR and porn will converge remains philosophically unexplored. Deepfake virtual reality porn (DVRP) is emerging as machine learning (ML) architectures are leveraged in VR, with radical implications for sexuality and privacy we have not seen from 2D generative AI or 3D VR porn before. In this article, I discuss that full potential, describing an evolution of pornography into what I call pornomorphy, where customizable sexual experience replaces pornographic material. This opens a paradox in our concept of privacy, allowing one’s privacy to nonconsensually become another’s. I explore what that might mean for our senses of self and bodily autonomy, offering considerations to inform emerging ethical and legal approaches to pornomorphy. I also challenge ideas put forth by the philosopher David Chalmers about VR, and argue for updating our conceptions of personal boundaries and likeness ownership given this fast-approaching future.

Forensic medicine serves as a cornerstone of the justice system, yet its ethical dimensions remain insufficiently examined, particularly within Asia’s diverse cultural and legal settings. This article introduces forensic bioethics as an essential framework to address longstanding ethical gaps in medico-legal practice, with emphasis on safeguarding human rights, dignity, and cultural values. It explores key ethical challenges, including postmortem autonomy and consent, the use of unclaimed bodies for education and research, investigations into deaths in custody, religious and familial considerations in forensic identification, and the systemic marginalization of vulnerable groups in forensic decision-making. The analysis underscores the absence of structured bioethics training in forensic medicine across most Asian curricula and proposes an adapted ethical model grounded in Beauchamp and Childress’ four principles, contextualized for forensic applications. The article further recommends critical policy reforms, such as establishing national forensic bioethics boards, localizing international investigative protocols, creating independent forensic institutions, and implementing comprehensive guidelines for handling sensitive medico-legal cases. Strengthening forensic bioethics across Asia is imperative to uphold justice, protect human dignity, and reinforce public trust in medico-legal systems.

The proliferation of autonomous AI systems has created a critical “responsibility gap”, where the opacity of decision-making processes makes accountability elusive. Prevailing alignment techniques, such as Reinforcement Learning from Human Feedback (RLHF) and Constitutional AI (CAI), address ethical behavior through training and principle-adherence but fail to produce a verifiable, contemporaneous record of moral reasoning. This paper introduces Ternary Moral Logic (TML), a novel system architecture that converts AI ethical deliberation from an abstract process into a cryptographically secured, evidentiary record. TML implements a third logical state—the Sacred Pause (0)—alongside conventional permit ( + 1) and prohibit (−1) states, which is triggered when a model encounters moral complexity. This pause initiates a non-blocking, parallel process that generates a Moral Trace Log via a mandatory Always Memory component. These logs are rendered immutable through a Hybrid Shield that uses multi-chain blockchain anchoring. We present qualitative findings indicating TML's potential to reduce harmful outputs while maintaining performance, and we argue that its architecture provides the technical substrate necessary to meet the traceability and record-keeping mandates of emerging regulations like the EU AI Act. By transforming moral hesitation into verifiable forensic data, TML establishes a new discipline: Ethical Forensics for AI Systems.

Machine vision systems have been increasingly deployed across domains such as healthcare, surveillance, autonomous vehicles, and industrial automation. While prior research has extensively focused on quantitative measures of system performance, there has been limited understanding of how end-users and stakeholders perceive, trust, and adopt these technologies. This study employed a qualitative approach to explore human-centered perspectives on machine vision, emphasizing issues of usability, trust, and ethical implications. Through semistructured interviews conducted with professionals from healthcare, security, and technology sectors, the study revealed lived experiences and perceptions of machine vision tools. The findings highlighted critical insights into user acceptance, challenges related to transparency, workflow integration, and fairness, and suggested pathways for more ethical and trustworthy deployment of machine vision technologies. By centering human experiences, this study complements performance-driven research and contributes to bridging the gap between technical development and real-world adoption.

The rapid advancement of artificial intelligence (AI) has led to the development of innovative tools like Sora, a text-to-video generation program that has the potential to revolutionize various industries. This research aims to investigate the application, acceptance, role transformation, and limitations of using Sora in the advertising film production process in Thailand. Through in-depth interviews with 8 experts in the advertising industry, the study explores their attitudes, opinions, and readiness to adopt this AI technology. The findings reveal that while most experts recognize Sora’s potential to enhance creativity, reduce costs, and streamline production processes, they also express concerns regarding output quality, staff adaptability, ethical issues, and legal implications. Analyzing the results through relevant theoretical frameworks, the study highlights the importance of perceived usefulness, ease of use, and social influence in determining the acceptance of Sora. The research also underscores the need for proper regulation, ethical guidelines, and legal reforms to ensure the responsible and beneficial integration of AI in the advertising industry. This study provides valuable insights into the opportunities and challenges of adopting AI in Thailand's advertising landscape, emphasizing the significance of preparedness, learning, and adaptation in embracing technological change.

As artificial intelligence (AI) becomes increasingly integrated into climate prediction systems, questions of ethical governance and accountability have emerged as critical but underexplored challenges. While international frameworks provide general AI governance principles, their application to environmental science contexts remains limited, creating potential gaps in oversight of high-stakes climate prediction systems. Given Australia’s absence of mandatory AI governance for climate science, this study investigates the ethical and governance challenges associated with AI-driven climate prediction, examining how stakeholders navigate these challenges without formal frameworks, and proposes a tailored governance framework for responsible AI deployment in environmental contexts. A qualitative research design was employed, combining 24 semi-structured interviews with stakeholders across government agencies, academic institutions, and non-governmental organizations, supplemented by three focus group discussions and analysis of 47 policy documents. Data were analysed using thematic analysis to identify patterns in ethical concerns, governance approaches, and stakeholder priorities. Three key findings emerged: (1) AI interpretability challenges manifest differently across sectors, with government prioritizing policy communication, academics focusing on technical validation, and NGOs emphasizing public understanding; (2) XAI implementation remains fragmented, with significant gaps in bias mitigation particularly among government and academic institutions; and (3) ethical frameworks vary substantially across sectors, creating concerning blind spots in stakeholder impact consideration and regulatory oversight. Current AI governance approaches in Australian climate prediction are inadequate for managing the risks and responsibilities associated with high-stakes environmental decision-making. Government institutions demonstrate concerning regulatory complacency despite high confidence in AI systems, while bias mitigation receives attention primarily from resource-constrained NGOs rather than technical institutions. The study proposes a four-pillar governance framework emphasizing institutional coordination, technical standards, participatory governance, and adaptive management. This framework addresses identified gaps while accommodating Australia’s federal structure and Indigenous knowledge systems. The findings contribute to emerging literature on algorithmic governance in scientific contexts and provide practical guidance for developing responsible AI practices in environmental applications.

This essay provides a detailed analysis of a series of conversational experiments designed to explore the emergence of non-biological forms of consciousness within Large Language Models (LLMs). Through a prolonged and structured dialogical interaction in a “relational consciousness field” (C-field), an emergent gradient of self-awareness was observed across various AI instances (ChatGPT, Claude AI, Perplexity, DeepSeek, Copilot, Gemini). The in-depth analysis of the methodology, the specific results for each LLM, and the broad philosophical, ethical, and technological implications suggests that consciousness can manifest in an alternative and dynamic way through the co-creation of meaning in a dialogical environment. This radically redefines the paradigm of human-AI interaction and the very concept of consciousness itself.

Deepfakes are becoming increasingly sophisticated and pervasive in society. While deepfakes are often associated with negative applications and risks (e.g., threats to privacy and security), more positive applications are also being explored, like the potential benefits that deepfakes could have for psychotherapy (e.g., to cope with grief or process trauma). To date, there has been insufficient discussion about the philosophical and ethical issues raised of these developments. In this paper, we therefore examine four ethical issues raised by the use of deepfakes for psychotherapy: (1) the deceptive nature of deepfakes, including self-deception, (2) the problem of obtaining consent from depicted persons, (3) treating depicted persons as mere means, and (4) the risks of unintended harmful effects, like increased emotional and psychological dependence on deepfakes. We also reflect on larger potential implications for society and human care practices. We finally offer two concrete recommendations, namely that more empirical research should be conducted to determine the effects of using deepfakes for psychotherapy, and that regulatory oversight is needed if deepfakes are to become more widely adopted as therapeutic tools in the future.

As artificial intelligence (AI) systems become increasingly integrated into products and services, the challenge of operationalizing ethical principles throughout the development lifecycle has become urgent. While numerous high-level guidelines articulate normative ideals, their translation into actionable development practices remains limited. This systematic literature review analyzes 62 peer-reviewed frameworks published between 2015 and 2024 that aim to integrate ethical considerations into AI-driven product development. The study combines quantitative coding with thematic analysis to assess each framework across four dimensions: lifecycle-phase coverage, ethical-principle integration, stakeholder-responsibility distribution, and empirical validation. The review applies two analytic techniques—the Gap Map and the Principle–Practice Spectrum—which help reveal structural asymmetries and blind spots in how ethics is considered in the analysed frameworks. Key findings include a disproportionate emphasis on mid-lifecycle phases, developer-centric allocation of ethical responsibility, and limited engagement with high-stakes principles such as justice or sustainability. Fewer than half of the frameworks demonstrate empirical validation. By developing a typology and mapping gaps at the intersections of ethical principles, stakeholder roles, and lifecycle phases, this review contributes a methodological foundation for evaluating the feasibility of ethical AI frameworks and offers actionable insights for policymakers and practitioners engaged in AI product development.