AI and ethics最新文献

Artificial intelligence (AI) is increasingly being incorporated into public health strategies for vector-borne disease (VBD) management, offering several advances in surveillance, prediction, and control. At the same time however, the integration of AI technologies raises critical ethical and equity concerns, particularly in regions disproportionately affected by VBDs. Here, we explore seven key ethical and equitable challenges in the use of AI for VBD management: (1) data quality and representativeness, (2) risk of discrimination and inequality reinforcement, (3) transparency and reproducibility, (4) privacy and data protection, (5) cybersecurity, (6) fair and equitable benefit-sharing, and (7) environmental considerations. Within each of these challenges, we highlight how unaddressed ethical and equity issues can exacerbate health disparities and undermine public trust. We then propose actionable pathways forward, including inclusive data governance, transparency-enhancing tools, and environmentally-conscious AI practices. By highlighting how accounting for these ethical and equity concerns during AI development and deployment can further progress towards the United Nations Sustainable Development Goals, we advocate for a more responsible and inclusive approach to AI in VBD management.

Large language models (LLMs) show promise for improving clinical trials in wealthy countries but remain underexplored in low- and middle-income countries (LMICs), where healthcare infrastructure is weaker and resources limited. This article explores opportunities for LLM integration and addresses ethical challenges in LMIC clinical trials through a review of recent literature (2024–2025) on LLMs in healthcare and clinical research, examining adaptation potential for LMICs using thematic analysis to identify ethical issues specific to LMIC contexts, including data control, fairness, and sustainability. LLMs can accelerate protocol development, improve multilingual patient recruitment, streamline regulatory processes, and address data gaps through synthetic records; however, implementation raises concerns about data privacy, community representation, AI transparency, and technological dependence on foreign platforms. While LLMs can enhance clinical trial efficiency and inclusivity in LMICs, successful integration requires locally-adapted models, community-centered ethical oversight, and regional partnerships, with thoughtful implementation potentially democratizing healthcare innovation benefits across Global South populations.

Can artificial agents be moral agents themselves? In this paper, I develop the foundations of reasons-based artificial agents, and I argue that it is superior to some alternatives to AI ethics. I focus on reinforcement-learning agents for concreteness, discussing some methodological issues along the way.

In recent years there has been a growing interest in the notion of AI consciousness—the question of whether artificial intelligences (AIs) can be conscious, and under what conditions this might emerge. This interest extends beyond academia to industry and the media. question of AI consciousness is underpinned by a moral question: should conscious AIs be granted moral standing? Emerging philosophical literature has begun to explore these ideas. We argue that these discussions neglect relevant psychological literature that can inform another element of this question—how our social and cognitive biases may impact our willingness to ascribe moral standing to AIs. In the current paper, we draw on the literature that examines moral consideration for non-human animals, and argue that similar biases limit moral standing for AI.

Can artificial intelligence systems genuinely inquire—or is optimization the limit of their intelligence? This paper argues that the boundary between inquiry and optimization marks the defining criterion of machine intelligence. Drawing on John Dewey’s pragmatist epistemology, I contend that inquiry, in Dewey’s sense, is distinguished by a capacity optimization lacks, the power to reconstruct its own problem space. While optimization assumes a fixed goal and searches for efficient means, inquiry can recognize when its initial framing fails and reconstitute the very situation it seeks to resolve. This capacity for problem-reconstruction, I argue, is not merely a desirable feature of advanced AI but the defining characteristic of genuine intelligence. After reconstructing Dewey’s logic of inquiry, the paper demonstrates its computational implementability. It then examines the ontological status of machine inquiry through enactivist cognitive science (Varela, Thompson, and Rosch, 1991) and Dennett’s concept of “competence without comprehension” (2017). I engage critical objections from Dreyfus and Floridi, arguing that computational systems can perform genuine inquiry in a functionally and epistemically grounded sense. The paper concludes with ethical and existential implications concerning responsibility and alignment. The central claim is this: If we define intelligence as the capacity to navigate genuine novelty and indeterminacy, then AI systems must possess the ability to redefine the very tasks they confront—this is the Deweyan condition for machine inquiry, where intelligence entails reflective adaptation within social and ethical contexts. Today’s AI systems excel at solving predefined tasks yet remain incapable of the creative, reconstructive intelligence that defines genuine inquiry. If we seek to build truly intelligent machines, we must move beyond optimization.

This paper examines the algorithmic formalization of language through a phenomenological lens, engaging Martin Heidegger and Maurice Merleau-Ponty in dialogue with contemporary large language models (LLMs) and related AI systems. Instead of treating computationally modeled language as a neutral medium for information transfer, we argue that both formal logic and data-driven models are historically specific crystallizations of a more primordial field of embodied expression. The idea of the ”unity of language” refers to the dynamic, historically situated field of expressive possibilities within which multiple linguistic systems — natural languages, formal calculi, code, poetic language — emerge, sediment, and transform. Drawing on Merleau-Ponty’s account of embodied speech, we reconstruct language as a living, self-renewing medium whose unity lies in its ongoing capacity to generate new sense. Heidegger’s analysis of technological ”enframing” (Gestell) and his reflections on ”traditional language” then allow us to interpret algorithmic conceptions of language as powerful, but critically informing of the existential risk of reducing speech to optimizable signals within the wider field of linguistic life. We confront these insights with current developments in AI, including LLMs, embodied AI, and enactive or 4E approaches to cognition. We conclude by sketching phenomenologically informed criteria for language technologies that respect expressive openness, relational depth, and the historicity of signifiers, and indicate how such criteria can orient debates in AI ethics.

The use of Conversational AI (CAI) agents within education has seen a rise with the rapid integration of generative AI (GenAI). The generative ability of the application combined with conversational capabilities has enhanced the perceived and actual usefulness of CAI applications. Given this development, it is critical to undertake a comprehensive review to understand the actual application domains, challenges, and efforts within this area. A range of empirical studies as well as reviews have been undertaken in recent years, but the current understanding remains fragmented. To better understand the current state-of-the-art, current trends and future implications of CAI on education, we conducted an umbrella review (UR) to systematically synthesize findings from thirty-four review articles. Articles were collected through a search across five major databases. They were screened using predefined eligibility criteria focusing on CAI agents used across educational domains and contexts. The PRISMA framework for transparent reporting is followed throughout the process and a thematic analysis has been undertaken to analyze the data. The results show that CAI utilization is concentrated in pedagogical applications such as teaching support, psychological engagement, and metacognitive development, while administrative functions, research assistance, and specialized training remain underdeveloped. Technical limitations and concerns with educational impact dominate discussions. Ethically, human-AI relationship concerns persist across all CAI generations, while academic integrity and data privacy represent emerging areas of concern. The review reveals gaps in CAI frameworks: lack of end-to-end design guidance, weak CAI specific usability methods, unclear pedagogical guidance and classroom implementation strategies, and limited AI literacy support. The article concludes by proposing a roadmap for ethical CAI implementation in education and identifying priority areas for future research.

This article examines Indigenous ethics of reciprocity as a normative and epistemological framework that challenges Western linear conceptions of time, economy, and social relations. Drawing from Marcel Mauss’s theory of the gift and Andean traditions, reciprocity is conceived not as mere exchange but as a vital principle of survival that structures human, communal, and cosmic relations. The acts of giving, receiving, and returning are understood as sacred obligations that guarantee continuity, balance, and mutual recognition across generations. Andean temporality, cyclical and rooted in natural and agricultural rhythms, situates reciprocity beyond economic utility and embeds it within cultural and cosmological orders. Historical encounters, such as colonial Christianity, illustrate the adaptive and inclusive nature of reciprocity, which facilitated intercultural coexistence and survival. Within this horizon, wealth is defined as relational abundance, measured by kinship networks and the capacity to fulfill communal obligations. Practices such as ayni (reciprocal labor), minka (collective work), and randi-randi (generalized reciprocity) embody this ethical system. Reciprocity thus emerges as both epistemology and moral system: a categorical imperative that governs enduring relations among humans, nature, and the cosmos. Returning a gift is not optional but a universal moral principle grounded in respect for Indigenous law, often misunderstood within Western frameworks. Despite growing critiques of Artificial Intelligence, non-Western epistemologies remain excluded. AI redefines truth as technical construction, suppressing subjectivity, dissent, and plurality, while fostering algorithmic obedience that undermines political imagination. The ethics of reciprocity offers a counterpoint, demanding the recovery of relational responsibility as a moral and political principle to guide the development of more just, situated, and human technologies.

Artificial intelligence (AI) is a rapidly evolving technology with transformative potential across many fields, including health. Interest in AI is growing globally, including in low- and middle-income countries (LMICs), but few studies have assessed the knowledge, experiences, and ethical practices of innovators in this space. This study aimed to assess the knowledge, experiences, challenges, and ethical practices of researchers and innovators, involved in AI-driven health research, to support the responsible integration of AI into health innovation. A cross-sectional, semi-quantitative online survey was conducted between February and May 2025 among health AI innovators in LMICs, primarily through the Grand Challenges network and its partners. Fifty respondents from 13 countries participated, the majority (92%) emphasized the importance of ethical principles, and 80% perceived AI-related ethical risks as greater than moderate. However, the majority of respondents (74%) were not sure about their competence in identifying and addressing the risks associated with AI practices and the same proportion reported having received no formal training in AI ethics. Respondents identified key challenges such as weak governance, limited ethics capacity, and risks of bias and data misuse. There was strong demand for training on ethics, bias mitigation, explainability, data governance, and privacy. The findings reveal an encouraging level of awareness but emphasize the need for formal training, clearer guidelines, and context-specific frameworks to ensure ethical AI development and deployment in health research.

The U.S. healthcare system is plagued by systemic challenges, including corporate dominance and a profound erosion of public trust in institutions such as the CDC and FDA. This crisis, rooted in inefficiencies, inequities, and a profit-centric model, has left millions without adequate care, exacerbating health disparities and fueling a public health emergency. This paper presents an argument that while artificial intelligence (AI) applications in healthcare governance could potentially remedy this problematic situation, its implementation is fraught with challenges. The article offers an on-point discussion of various governance weaknesses that exist within the U.S. health system, but it also critically examines whether AI technology genuinely has the potential to tackle deep-rooted systematic and institutional failures. AI might mitigate these significant institutional issues in some contexts but could also inadvertently exacerbate them in others if not implemented with rigorous oversight. A particularly unsatisfactory aspect of current proposals is the inadequate and superficial address of data quality and accessibility concerns—which are absolutely essential for the successful implementation of AI. Given that both the data infrastructure and legal frameworks currently in place are insufficient, this paper argues that a more critical and nuanced analysis is required to navigate the practical and ethical challenges facing healthcare transformation. This vision requires dismantling the problematic aspects of a profit-driven model and addressing the moral and structural failures that have left the U.S. lagging behind its peers, with a clear understanding that AI is a tool that requires careful, ethical, and equitable implementation.