Front Matter

(Cover, Contents, Editorial, Foreword; pp. III.i – III.v)

Sustaining the Status Quo? Rethinking AI for Sustainability through Power and Infrastructure

Harry Collins (pp. 131 – 141)

Artificial Intelligence (AI) is increasingly presented as a tool for addressing environmental crises, yet there remains significant debate over what should count as genuine AI for Sustainability. I argue that dominant accounts remain too narrow because they focus on environmental outputs while neglecting the political and social conditions through which those outputs are produced. Drawing on Falk and van Wynsberghe’s framework and Lehuedé’s analysis of data infrastructure in Chile, I show that sustainability claims can obscure power concentration, extractive relations, and territorial exclusion. I propose a revised AI for Sustainability framework that adds conditions on benefit and harm distribution, community engagement, stronger carbon reporting and energy-efficiency improvement, and the evaluation of alternative social practices. To make the framework operational, I work through community engagement and impact reconciliation in detail, demonstrate an example application against a real-world AI project, and argue that the framework should be used through external, public assessment rather than self-certification.

Choosing the Path of Most Resistance: Metacognitive AI Literacy and Mitigating Epistemic Risk

Iman Khwaja (pp. 142 – 153)

This paper reimagines metacognition as both an individual tool and a component of AI (Artificial Intelligence) literacy, that acts as a critical pathway to mitigating the epistemic threats posed by AI tools. These threats are explored as the erosion of autonomy, critical thought and creative abilities in individuals, as a result of cognitive dependencies on AI tools.

Metacognition, a strategy involving the active preservation of cognitive ability in individuals, is currently a subtle component of many literacy approaches in varied educational contexts. This paper positions it as a practice that could play a meaningful role in the trajectory of AI development in tandem with the consequent changes to our collective epistemic and cognitive experiences. Specifically, the aspects of self-restraint, critical evaluation and epistemic investigation proposed by traditional metacognitive strategy are reconceptualized as potentially critical aspects of AI literacy efforts.

Drawing on literature that explores AI deskilling, epistemic risk, which Babui describes as “the likelihood that one’s claims inaccurately represent the world” (2019), and digital literacy, this paper proposes a framework for users, developers and deployers to consider applying metacognitive strategy to AI literacy efforts, as a means to mitigate the epistemic risks posed by user dependencies on AI tools.

Risk, Rights, and Regulatory Convergence: The Turkish Artificial Intelligence Law Proposal as a Model for Emerging-Economy AI Governance and the Prospects for a TRIPS-Style International Framework

Zeki Emre Kurt (pp. 154 – 166)

The global AI regulatory landscape has converged architecturally around risk-based tiering, mandatory ethics obligations, prohibitions on unacceptable practices, conformity assessment, supervisory enforcement, and turnover-linked sanctions. Yet this convergence is structural rather than substantive: identical terminology performs divergent doctrinal functions across jurisdictions. Most interventions worldwide remain at the proposal or non-binding policy-framework stage.

This article takes Turkey’s Artificial Intelligence Law Proposal (Esas No. 2/2234, submitted 25 June 2024) as its central case study—an eight-article principle-based framework statute that mirrors the EU AI Act’s turnover-linked sanctions while deferring substantive risk classification and enforcement detail to secondary legislation. Three theses are advanced. First, a universal common core exists, but it is a floor of architectural principles, not a ceiling of identical rules. Second, for upper-middle-income economies with constrained regulatory capacity, a principle-based framework statute is defensible where full advanced-economy transposition is institutionally infeasible—provided a credible secondary-legislation programme accompanies it. Third, the TRIPS Agreement’s minimum-standards-plus-flexibility architecture offers a structural template for a development-sensitive international AI governance instrument, though such a regime risks reinforcing global inequality unless accompanied by genuine capacity-building and technology-transfer commitments.

Resistance as a Technomoral Virtue for Our Times

Nitya Mandyam (pp. 167 – 176)

This paper critically examines differing responses to the harms of Artificial Intelligence (AI) and explores the ethical dimensions of resisting AI. It draws on case studies of three distinct AI-related harms – bias in language models, the environmental impact of large AI models, and job displacement due to automation, to define what constitutes an act of resistance. Applying a virtue ethical framework to the question of resistance, I posit that resistance is a technomoral virtue in our times and can act as a creative, generative, and dynamic principle in producing technologies that engender human flourishing.

The Epistemic Frontier of Capitalism: A Theory of Epistemic Regression through Artificial Intelligence and Misinformation

James Rice (pp. 177 – 187)

This article advances a single thesis about contemporary capitalism: under the pressure of frontier artificial intelligence and platform-mediated misinformation, accumulation, growth, and knowledge production has entered a phase in which outcomes are increasingly shaped by the privatization and strategic destabilization of our shared epistemic commons at the technological frontier. These, I define as consensus-based standards of scientific evidence, categories or criteria for a priori truth, and collaborative channels through which public reasons are disputed and agreed upon, these are facts and features of our politics that capitalism itself requires to coordinate expectations, channel investment, and establish and maintain political legitimacy. Artificial intelligence automates tasks and enhances productivity; but it also industrializes cognitive work itself, enabling the large-scale privatization of decision-making authority and the concentration of power, assets, and skills. AI defines our era, I suggest, but also threatens to disenfranchise a multitude of thinkers and creators, capable of contributing immense value.

Misinformation, on the other hand, threatens our sense of belonging in knowledge-based communities of practice, through the infusion of divisive, polarizing, and invasive falsehoods, intended to seed doubt, confusion, and anarchy. Misinformation, I argue, is the strategic manipulation of shared reality, often working at odds with reasonable intuition, and therefore deeply dangerous when undetected. The paper develops this thesis through two tightly linked case studies — AI and misinformation as strategies operating with the purpose of extreme capital accumulation funnelled to an elite subset of our technocratic society — and argues that capitalism now profits from corroding the epistemic and intellectual preconditions of its own vitality through information warfare.

Metacognitive Conversational AI: A Design Framework for Reflective Well-Being

Tianyi Yu (pp. 188 – 198)

Conversational artificial intelligence systems are increasingly embedded in everyday cognitive and relational tasks, while their optimisation for immediacy and solution delivery raises questions about how they shape users’ reflective capacities. Drawing on metacognitive theory, this paper proposes a conceptual framework for Metacognitive Conversational AI, which structures interaction design around three mechanisms: Reflection Activation, Structured Scaffolding, and Transfer and Retention. This framework translates metacognitive processes into interface-level design principles intended to preserve users’ evaluative participation rather than fully externalise it.

Using design fiction methodology, two illustrative scenarios, grammar correction and interpersonal conflict resolution, demonstrate how modest adjustments to conversational sequencing can shift interaction from direct answer substitution toward structured reflective engagement. The paper argues that such design choices may have implications for regulatory flexibility and sustained well-being. It further examines structural, economic, and ethical tensions, including usability trade-offs, user agency, and cultural variation in conceptions of reflection.

By reframing conversational AI as a cognitive environment rather than merely a problem-solving tool, this work contributes a theoretically grounded design framework for integrating metacognitive support into human–AI interaction.