Insights - Digital Sustainability Hub

Insights

Epistemia: When AI Fluency Replaces Grounded Knowledge

In the current landscape of artificial intelligence, large language models generate text that is plausible but not necessarily grounded in verifiable knowledge. This phenomenon raises a critical concern: when users mistake linguistic fluency for understanding, they risk embracing epistemia, a degenerated form of knowledge that prioritizes surface coherence over rigorous verification (Quattrociocchi et al. 2025). This post explores how such epistemic risks arise from misplaced trust in generative AI, why grounding matters, and how recent scholarship and public discourse illuminate pathways toward more responsible and critically aware engagement with AI systems.

January 14, 2026

Insights

China’s “Hundred Model War”: Understanding the Digital Sustainability of Its LLM Strategy

China’s frenetic “Hundred Model War” in the large language model space reveals a unique intersection of state power, technological ambition, and ideological strategy. As dozens of domestic actors race to build powerful LLMs, the country pursues not only technical independence but also narrative sovereignty. This post examines both the promising digital sustainability aspects of China’s LLM ecosystem and its systemic vulnerabilities, especially around governance, resource inefficiency, and ideological alignment.

December 27, 2025

Insights

Shaping AI’s Impact on Work: Why Outcomes Depend on Policy and Context

The labor market effects of AI are complex, and simple narratives of mass job loss or automatic job creation miss the bigger picture. To assess the real impact, we must consider interacting factors such as macroeconomic conditions, institutional frameworks, and the time horizon used for evaluation. This post synthesizes recent empirical and policy oriented work to show why AI’s effects on jobs are context dependent, often heterogeneous across sectors and countries, and amenable to shaping through policy choices. Practical implications follow for employers, educators, and policymakers seeking orderly, fair transitions.

December 22, 2025

Insights

The Real Sustainability Test for Social Media: Rethinking Engagement and Attention

Social media are not just platforms for sharing content, they are complex socio technical systems whose long term viability depends on environmental, social, and economic factors. Evaluating the sustainability of social platforms means looking beyond data center emissions, to include cognitive load, mental health, business model fragility, and governance transparency. Recent empirical work shows measurable harms for some user groups, while other studies highlight large heterogeneity in individual responses. This post synthesizes those findings, and offers pragmatic directions for designers, policymakers, and platform operators who want social media to be resilient, equitable, and sustainable.

December 18, 2025

Insights

Can AI Help Cool the Planet, Without Heating It Up First?

AI offers powerful tools to understand, forecast, and respond to climate risks, from improving weather and wildfire prediction to optimising energy systems and precision agriculture. At the same time, training and operating advanced AI models consumes significant energy and can increase greenhouse gas emissions, unless those systems are governed and engineered with sustainability in mind. Balancing AI’s climate mitigation potential with its carbon footprint requires transparent measurement, careful prioritisation of high leverage applications, and policy frameworks that reward ethical, energy-aware AI practices. This article outlines practical pathways to deploy AI for climate action while minimising unintended environmental harms.

December 11, 2025

Insights

Making Crypto Sustainable: Beyond the Energy Debate

Cryptocurrencies reshaped finance, but their environmental and social costs now press for systemic reform. Reducing energy use is necessary, but insufficient. A truly sustainable crypto ecosystem must address consensus design, device efficiency, e waste, grid integration, governance, and distributional impacts on communities. Recent empirical studies show that Proof of Stake dramatically cuts energy consumption compared with Proof of Work, yet technical fixes must be paired with policy frameworks, transparency, and incentives that reward ecological and social responsibility. The path forward requires aligning economic signals with sustainability goals, and treating blockchains as socio technical systems, not isolated protocols.

December 5, 2025

Insights

The Cost of Deception: Deepfakes and the Fragility of Digital Trust

Deepfakes are no longer a technological curiosity but a structural challenge to digital sustainability. As synthetic media becomes more realistic and easier to produce, societies face a new kind of pollution: misinformation that scales with computation. The environmental cost of generating, distributing, and combating deepfakes is matched by a cognitive and ethical burden. Understanding how these false images and voices undermine trust, consume resources, and erode public discourse is essential if we hope to design a more transparent and responsible digital ecosystem.

December 1, 2025

Insights

Unlearning Machines: Toward Sustainable Forgetting in AI

As artificial intelligence systems grow in scale and power, they also grow in memory. But what happens when machines must forget? “Machine unlearning” is an emerging field that explores how to remove specific data or knowledge from trained models without retraining them from scratch. Beyond privacy protection, it opens a broader discussion about the sustainability of digital intelligence itself. By learning how to forget responsibly, AI systems could become more energy efficient, ethically compliant, and aligned with human values: a crucial step toward more sustainable computing.

November 26, 2025

Insights

Inbox Pollution: How Spam Is Fueling Emissions and Draining Our Attention

Every day, billions of unwanted emails travel across the internet, clogging inboxes and wasting attention. But spam is not only a nuisance, it is also a source of measurable environmental and social damage. From the carbon emissions caused by data transmission and storage to the cognitive overload generated by constant filtering, the global spam ecosystem embodies a largely invisible form of digital pollution. Rethinking spam through the lens of sustainability reveals an urgent need for cleaner communication systems and more responsible digital infrastructure.

November 21, 2025

Insights

Beyond the Buzz: Why an AI Bubble Would Threaten Sustainability

The hype around artificial intelligence is enormous. Corporations, venture capitalists, and governments are pouring vast sums into AI projects, expecting transformative returns. But alongside genuine innovation lies the risk of a speculative bubble, one that could create economic instability, misallocated resources, and long term sustainability problems. As we review lessons from the dot com and cryptocurrency bubbles, it is time to ask whether the current AI fervour might lead not only to a boom, but also to a bust with far reaching consequences for society, business, and the environment.

November 17, 2025

The DSH Observatory