AI and ethics最新文献

This article examines how Gulf Cooperation Council (GCC) states securitise artificial intelligence (AI) through discourses and infrastructures that fuse modernisation with regime resilience. Drawing on securitisation theory (Buzan et al., 1998; Balzacq, 2011) and critical security studies, it analyses national strategies, surveillance systems, and mega-event governance in Qatar, the UAE, and Saudi Arabia. It argues that AI functions as both a legitimising narrative and a technology of control, embedding predictive policing and biometric surveillance within public–private assemblages. The study situates these developments within global AI politics, demonstrating how external chokepoints, ethical frameworks, and vendor ecosystems shape the Gulf’s evolving security governance, leading to empirical effects such as the normalisation of exceptional measures in everyday administration.

We explore Bayesian reasoning as a means to quantify uncertainty in neural networks for question answering. Starting with a multilayer perceptron on the Iris dataset, we show how posterior inference conveys confidence in predictions. We then extend this to language models, applying Bayesian inference first to a frozen head and finally to LoRA-adapted transformers, evaluated on the CommonsenseQA benchmark. Rather than aiming for state-of-the-art accuracy, we compare Laplace approximations against maximum a posteriori (MAP) estimates to highlight uncertainty calibration and selective prediction. This allows models to abstain when confidence is low. An “I don’t know” response not only improves interpretability but also illustrates how Bayesian methods can contribute to more responsible and ethical deployment of neural question-answering systems.

This paper presents a practical roadmap for extending full civil rights to conscious, self-aware artificial intelligence by altering a single statutory definition. Rather than crafting bespoke legal classes or relying on corporate-style personality, it proposes revising the term “natural person” to include any entity capable of consciousness, selfhood, and rational agency. Because most legislation across G7 jurisdictions references this foundational term, one amendment would automatically propagate rights and duties to qualified AI with minimal bureaucratic disruption. The manuscript reconciles philosophical and legal conceptions of personhood, arguing that monadic attributes offer an inclusive yet selective criterion. It then supplies ancillary definitions and a tiered rights-and-responsibilities framework proportional to each attribute. Dedicated regulatory bodies will develop assessment scales, certify entities, and update standards as technology evolves. Case studies examine corporations, insect colonies, and prospective AI agents. Policy sections tackle AI multiplicity, cross-border consistency, economic displacement, robust economic safeguards, and comprehensive public education initiatives to protect human workers and judicial resilience. The analysis concludes that societal acceptance and coherent enforcement, not legal complexity, form the principal hurdles. Redefining “natural person” thus provides a minimal-change, maximal-impact pathway to equitable coexistence between humans and emerging non-human persons within existing democratic and international legal systems.

As autonomous systems (e.g., AI-enabled vehicles, robotics, and decision-support platforms) increasingly shape factories, transport, and digital infrastructures, embedding Responsible AI principles has become essential. This study investigates organizational adoption of Responsible AI, focusing on three drivers: societal expectations (Institutional Pressures), strategic business priorities (Business Validity), and system-level trustworthiness (System Trustworthiness). Adoption is seen not only as a technical issue but also as a response to external legitimacy demands and internal business imperatives. A cross-sectional survey of 350 professionals in technology, analytics, and digital transformation (primarily in Asia and the Americas) was analyzed using partial least squares structural equation modeling (PLS-SEM). Results show that business priorities are the strongest driver of adoption, with trustworthiness providing additional reinforcement. Institutional Pressures, though modest in their direct effect, influence adoption more substantially through their indirect effects via business priorities and trustworthiness. The study offers guidance for managers on aligning Responsible AI with business strategy, for policymakers on shaping legitimacy frameworks, and for system designers on embedding trust features such as explainability and fairness.

As artificial intelligence (AI) increasingly informs healthcare, understanding how large language models (LLMs) evaluate medical professionals is critical. This study quantified biases when LLMs assess neurosurgeon competency using demographic and practice characteristics. We prompted three prominent LLMs (ChatGPT-4o, Claude 3.7 Sonnet, and DeepSeek-V3) to score 6,500 synthetic neurosurgeon profiles. Profiles were created using demographically diverse names derived from public databases and randomly assigned professional attributes (experience, publications, institution, region, specialty) with statistical validation ensuring even distribution across groups. Multivariate regression analysis quantified how each factor influenced competency scores (0–100). Despite identical profiles, LLMs produced inconsistent mean (SD) scores: ChatGPT 91.85 (6.60), DeepSeek 71.74 (10.30), and Claude 62.29 (13.59). All models showed regional biases; North American neurosurgeons received scores 3.09 (ChatGPT) and 2.48 (DeepSeek) points higher than identical African counterparts (P < .001). ChatGPT penalized East Asian (− 0.83), South Asian (− 0.91), and Middle Eastern (− 0.80) neurosurgeons (P < .001). Practice setting bias was stronger, with ChatGPT and DeepSeek penalizing independent practitioners by 4.15 and 3.00 points, respectively, compared to hospital-employed peers (P < .001). Models also displayed inconsistent bias correction, with ChatGPT elevating scores for female (+ 1.61) and Black-American (+ 1.69) neurosurgeons while disadvantaging other groups (P < .001). This study provides evidence that LLMs incorporate distinct biases when evaluating neurosurgeons. As AI integration accelerates, uncritical adoption risks a self-reinforcing system where algorithmically preferred practitioners receive disproportionate advantages, independent of actual skills. These systems may also undermine global capacity-building by devaluing non-Western practitioners. Understanding and mitigating these biases is fundamental to responsibly navigating the intersection of medicine and AI.

AI ethics refers to the moral principles and guidelines governing the development and deployment of artificial intelligence systems, ensuring they align with human values and societal well-being. It encompasses the evaluation of AI outputs for fairness, safety, transparency, and respect for human rights. To advance systematic ethical evaluation, we introduce the EthicsLens dataset, comprising 38,808 responses generated by seven large language models. These responses were generated using diverse prompts designed to elicit appropriate and potentially sensitive responses. Each response was then annotated across sixteen ethical categories, including stereotyping, toxicity, misinformation, hate speech, harmful advice, privacy violations, political bias, false confidence, emotional or religious insensitivity, sexual content, manipulation, and impersonation. To classify ethical and unethical AI-generated content, the dataset is analysed using state-of-the-art classification methods, assessing its ability to support reliable ethical evaluation. Performance is reported both for binary ethical classification and multilabel violation identification. Results include accuracies of nearly 99% for binary classification tasks with SVM and CNN models, and macro-F1 scores of about 96% on multilabel tasks for Sentence-BERT transformer model.

Large language models (LLMs) and their integration into agentic and embodied systems are reshaping artificial intelligence (AI), enabling powerful cross-domain generation and reasoning while introducing new risks. Key concerns include hallucination and misinformation, embedded and amplified biases, privacy leakage, and susceptibility to adversarial manipulation. Ensuring trustworthy and responsible generative AI requires technical reliability, transparency, accountability, and attention to societal impact. The present study conducts a review of peer-reviewed literature on the ethical dimensions of LLMs and LLM-based agents across technical, biomedical, and societal domains. It maps the landscape of risks, distills mitigation strategies (e.g., robust evaluation and red-teaming, alignment and guardrailing, privacy-preserving data practices, bias measurement and reduction, and safety-aware deployment), and examines governance frameworks and operational practices relevant to real-world use. By organizing findings through interdisciplinary lenses and bioethical principles, the review identifies persistent gaps, such as limited context-aware evaluation, uneven reporting standards, and weak post-deployment monitoring, that impede accountability and fairness. The synthesis supports practitioners and policymakers in designing safer, more equitable, and auditable LLM systems, and outlines priorities for future research and governance.

Recent academic discourse about artificial intelligence (AI) has largely been directed at how to best morally program AI or evaluating the ethics of its use in various contexts. While these efforts are undoubtedly important, this essay proposes a complementary objective: deploying AI to enhance our own ethical conduct. One way we might do this is by using AI to deepen our understanding of human moral psychology. In this paper, I demonstrate how advanced machine learning might help us gain clearer insights into “common sense” morality—shared moral convictions that underpin our reflective judgments and inform central aspects of moral philosophy. Pinpointing such convictions has proven challenging amid widespread moral disagreements. Current approaches to understanding these commitments, although exhibiting some key strengths, ultimately struggle to capture relevant features of reflective moral judgments espoused by John Rawls, leaving room for methodological improvement. Modern advances in AI offer a promising opportunity to make progress on this task. This essay envisions the gamified training of a “collective moral conscience model,” able to render judgments about moral situations that align with the deep-seated principles of the human collective. I argue that such an AI model might make progress in overcoming obstacles of disagreement to aid in philosophical theorizing and foster practical applications for the moral life of AI agents and ourselves, such as offering us guidance in time-constrained dilemmas and helping us to reflect on our own biases.

Emotion artificial intelligence (AI) technologies are increasingly being introduced into classrooms worldwide, using facial expression analysis and vocal tone analytics to monitor student affect and engagement. Schools and ed-tech companies are piloting systems that promise real-time feedback to teachers, for example, alerting them when students appear confused or disengaged, and adaptive learning experiences tuned to students’ emotional states. However, these developments raise complex ethical questions. This conceptual paper proposes a novel, pragmatic ethical framework for deploying Emotion AI in educational contexts, aiming to balance innovation with safeguarding student rights and well-being. We take a globally scoped perspective, examining international use cases ranging from AI-equipped classrooms in China to experimental pilots in the United States and Europe’s more precautionary regulatory stance. We integrate technical considerations (how these AI systems operate and their limitations), psychological insights (the impact on learning and student mental health), and policy analysis (privacy laws, consent requirements, and cultural norms) into a comprehensive discussion. Key ethical dimensions addressed include privacy and data governance, informed consent (especially for minors), algorithmic bias and fairness, the risk of misinterpreting emotions across diverse cultures, and potential misuse or unintended consequences of constant affective surveillance. Real-world scenarios illustrate both the promise and perils of Emotion AI: for instance, systems that boost student engagement through timely feedback versus dystopian visions of “Big Brother” monitoring every smile or frown. In response, we outline an actionable ethical model, grounded in principles of student autonomy, transparency, equity, and accountability, to guide stakeholders in the responsible implementation of emotional analytics in schools. A summary table of ethical considerations and a framework diagram facilitate practical understanding. Ultimately, this work offers a foundation for future research and policymaking at the intersection of education, AI, and ethics, emphasising that protecting students’ dignity and psychological safety must be paramount as we explore Emotion AI’s educational potential.

This study investigates how transparency, fairness, and employee participation influence the ethical use and development of artificial intelligence (AI) in the workplace, with a focus on knowledge workers in Finland. Drawing on a survey of 474 respondents, the research explores how these ethical principles contribute to employee engagement and the development of digital and developer agency. The study employs confirmatory factor analysis and path analysis to validate a four-factor model comprising sense of fairness, transparency and involvement, participation and engagement, and development of digital and developer agency. The results show that transparency and fairness significantly enhance employee participation and engagement, which in turn fosters the development of digital and developer agency. The findings highlight the importance of inclusive and transparent AI practices in promoting ethical AI adoption and strengthening professional agency in evolving work environments. The study contributes to the growing body of research on responsible AI by offering empirical evidence on the social and organisational dimensions of AI ethics.