Artificial intelligence is often presented as a technological ally in the fight against climate change, capable of optimizing electricity grids, reducing industrial waste, improving transport systems and helping governments manage scarce resources. But a new analysis argues that this optimistic story may be concealing a deeper problem: many systems promoted as “sustainable AI” are being developed within the same economic structures that drive environmental destruction and social inequality. Rather than asking only how artificial intelligence can be made safer, fairer or more energy-efficient, the authors say policymakers should confront a more fundamental question: is a particular AI system necessary in the first place?

Writing in AI & Society, Paul Schütze of Osnabrück University and Benedetta Brevini of the University of Sydney argue that current responsible AI governance often creates the appearance of action without imposing meaningful limits on the technology’s environmental and social costs. Their paper describes this phenomenon as “placebo change”: policies, voluntary commitments and ethical principles suggest that risks are being managed, while the underlying forces driving those risks remain intact. The authors propose replacing the dominant model of responsible AI with what they call “critical AI governance,” an approach that would judge AI against democratically defined social and ecological goals before allowing systems to be developed or deployed.

The controversy begins with the phrase “sustainable AI,” which contains two very different ideas. The first is AI for sustainability: using machine-learning systems to reduce emissions, improve energy efficiency, monitor ecosystems or optimize transport. The second is the sustainability of AI itself: accounting for the electricity, water, minerals, data infrastructure, labor and electronic waste required to build and operate these systems. The paper argues that public discussion has focused heavily on the first idea while often treating the second as a secondary technical problem. That imbalance can make an AI application appear environmentally beneficial even when its full lifecycle imposes substantial costs elsewhere.

Modern AI systems depend on a large physical and social infrastructure that is easy to overlook when the technology is described as software. Training and running large models requires data centers filled with specialized processors, cooling equipment and networking hardware. Those facilities consume electricity and, in many regions, significant quantities of water. The hardware depends on mined materials and global supply chains, while the production and maintenance of datasets may involve poorly paid or invisible labor. At the end of a device’s useful life, discarded servers and electronic components add to the growing stream of e-waste. From this perspective, the authors contend, AI’s environmental footprint is not simply a design flaw that engineers can eliminate through better code. It is connected to the economic system in which the technology is produced and expanded.

This distinction challenges a powerful assumption known as technological solutionism: the belief that complex political and social problems can be solved primarily by introducing a more advanced tool. An AI system can sort waste, detect equipment failures or forecast energy demand, but it cannot decide what a city should value. For example, a traffic-management system could be optimized for faster car journeys, safer cycling, cleaner air, lower fuel consumption, shorter commutes or improved public health. These objectives may conflict, and selecting among them requires political judgment and public agreement. An algorithm can calculate how to pursue a chosen goal, but it cannot provide the democratic decision about which goal is worth pursuing.

The paper describes the result as a form of displacement. An AI-based efficiency gain in one location may shift environmental burdens to another through energy use, mining, water consumption or waste disposal. Even when an application has a relatively small direct footprint, it may help preserve “business as usual” by creating confidence that technological progress is addressing the climate crisis. This is the authors’ central meaning of placebo change: an intervention appears transformative because it produces visible activity, reports, dashboards or efficiency improvements, but it does not alter the economic and political causes of unsustainability. In the most troubling version of this argument, the language of sustainability becomes a form of ethics washing, allowing institutions to advertise responsibility while continuing to expand resource-intensive systems.

The researchers direct particular attention to the role of soft law in AI governance. Soft law includes voluntary company commitments, industry standards, codes of conduct, best-practice frameworks and certification programs. Unlike hard law, it generally lacks one or more of three features: legally binding obligations, precise requirements and an independent authority empowered to interpret and enforce the rules. Such instruments can be useful when organizations genuinely want to cooperate, but they cannot reliably compel action when compliance conflicts with commercial interests. The authors argue that soft law can therefore create the social impression that AI is under control, reducing pressure for stronger regulation even when companies face no meaningful penalties for failing to meet environmental targets.

The European Union’s AI Act illustrates the tension, according to the analysis. The Act is a binding legislative instrument in many areas, but the paper argues that its treatment of AI’s environmental impact retreats into voluntary governance. The legislation refers to codes of conduct for measuring and minimizing environmental effects, while certain provisions require documentation of energy and computational resource use for general-purpose AI systems. Yet these requirements do not establish binding limits on emissions, water use, material consumption or total lifecycle impacts. Reporting what a system consumes is not the same as requiring companies to reduce that consumption. The authors describe this gap as a regulatory vacuum: environmental sustainability is acknowledged in legal language, but companies are not given enforceable minimum standards or sanctions for failing to lower their footprint.

Corporate sustainability reporting illustrates why disclosure alone may be insufficient. The paper cites a 2024 statement by a member of the European Parliament that Google would not provide a specific AI disclosure in a recent sustainability report, despite having done so two years earlier. The authors also point to Google’s 2025 environmental report, which recorded a 51 percent increase in total carbon dioxide emissions in 2024 compared with 2019. Microsoft, they note, reported emissions 23.4 percent higher than in 2020. These figures do not establish that AI alone caused the increases, and the paper does not present a new emissions dataset. They do, however, highlight the limits of governance that emphasizes documentation without requiring absolute reductions or making companies disclose comparable information about particular AI systems.

Critical AI governance would begin before the conventional regulatory process, the authors argue. Instead of assuming that AI development is inevitable and then attempting to mitigate its consequences, governments would first ask whether a proposed system advances human well-being, social justice and ecological stability. That could mean prohibiting applications judged unnecessary or harmful, rather than merely requiring them to publish risk assessments. It would also mean challenging the idea that innovation and economic growth should automatically take priority. The authors emphasize that this is not a call for rejecting every form of automation or withdrawing from technology. It is a demand to treat technological development as a political choice rather than as an unavoidable future to which society must adapt.

Their proposal rests on three broad principles. The first is to question the ideologies and economic imperatives that drive AI expansion, including the pressure to become an “AI leader” and the assumption that more computational power necessarily represents progress. The second is to prioritize social and political solutions over technological fixes, addressing the roots of climate change, inequality and resource consumption rather than delegating those problems to algorithms. The third is to invest in public digital infrastructure and independent research. Publicly controlled infrastructure could reduce dependence on a small number of powerful technology companies, while independent universities, civil-society groups and researchers could support scrutiny that is not shaped by corporate incentives.

This model would also shift who gets to decide which technologies are built. Communities affected by data centers, extraction, automated workplaces and algorithmic decisions would have a stronger role in defining acceptable uses. Democratic participation would not be limited to commenting on systems after companies had already designed them. It would include deciding whether particular systems are desirable, what social goals they should serve and whether the environmental cost is justified. The authors argue that questions about ownership, labor, supply chains, data control and the relationship between technology companies and states must be part of AI governance, because these conditions determine who benefits from AI and who absorbs its costs.

The paper is conceptual rather than experimental: no datasets were generated or analyzed, and it does not offer a complete legislative blueprint. Instead, Schütze and Brevini identify what they see as a structural weakness in responsible AI frameworks. If governance starts from the assumption that AI will exist and should be made more trustworthy, they argue, it may never reach the prior question of whether a system should be developed at all. Their warning is deliberately provocative because it targets the most familiar promise in the AI debate—that better design and better oversight will always be enough. As the technology spreads into energy, transport, medicine, government and everyday communication, the authors say, the crucial test of responsibility may not be whether AI can be made more efficient, but whether society is willing to say no when its costs outweigh its benefits.

At the heart of this investigation lies the understanding that AI’s environmental impact extends well beyond energy consumption and carbon footprints. The report emphasizes the intricate supply chains and physical systems supporting AI: sprawling data centers, semiconductor fabrication, cooling mechanisms, and resources extracted for critical minerals. These components introduce significant water withdrawals, land use for energy infrastructure, and the escalating challenge of electronic waste management. In doing so, the report marks a crucial shift from the conventional carbon-centric discussions toward a holistic environmental perspective.

The scale of AI’s operational energy demands is staggering. Projections estimate that data centers, the backbone of AI computing, will consume 448 terawatt-hours of electricity in 2025—an amount equivalent to the national consumption of France, ranking them as the 11th largest global electricity user if considered a country. Notably, AI workloads account for roughly 20% of this power use, a share predicted to rise to 40% by 2030. Should current growth trajectories persist, the energy consumption attributed to AI could nearly triple by 2030, corresponding to around 945 terawatt-hours annually and equating to nearly 3% of worldwide electricity usage. This prodigious demand alone could sustain the energy needs of 1.3 billion people living in Sub-Saharan Africa for over five years—a demographic particularly vulnerable to energy scarcity.

Beyond energy, the water footprint of AI infrastructure poses an underappreciated risk to global freshwater resources. Data centers currently utilize an estimated 9.3 trillion liters of water, sufficing for the drinking requirements of the global population for approximately 1.6 years. The report underscores that water withdrawals, especially in arid or depleted regions, can severely stress aquatic ecosystems and groundwater reserves, even when some of this water is eventually returned. Moreover, land requirements for electricity generation related to AI’s growth are poised to surpass 14,000 square kilometers by 2030, roughly the size of Northern Ireland, presenting additional challenges for land management and biodiversity conservation.

Training state-of-the-art AI models such as ChatGPT-5 demands colossal energy inputs, consuming around 100 gigawatt-hours of electricity—comparable to the annual residential energy consumption of 770,000 individuals in Sub-Saharan Africa. The corresponding water and land footprints—1 billion liters and 1.5 square kilometers respectively—highlight the significant spatial and resource components embedded within AI’s developmental phase. However, the report pivots attention toward the AI’s ubiquitous daily use, which far exceeds the energy footprint of training alone. For instance, ChatGPT processes roughly 2.5 billion prompts daily, translating into annual electricity use of about 383 gigawatt-hours and water consumption sufficient for half a million people’s domestic needs annually, reflecting the enormous cumulative resource drain of AI services.

The environmental cost per AI interaction varies significantly by technology and usage context. For example, Google handles approximately 5 trillion search queries each year, where a traditional search requires around 0.3 watt-hours, but AI-enhanced generative searches inflate this figure to up to 3 watt-hours—a tenfold increase. Additionally, AI-generated video content emerges as a looming environmental crisis. A single high-resolution video clip may demand more than 415 watt-hours of energy, outstripping the energy required for producing hundreds of static AI-generated images. Given that energy requirements rise quadratically with resolution and frame count, the burgeoning prevalence of AI video generation could rapidly escalate infrastructure strain.

Crucially, the report explores the intricate trade-offs between carbon, water, and land footprints in AI energy sourcing. Transitioning from coal to bioenergy production can reduce carbon emissions by an average of 72%, yet simultaneously inflates water consumption more than thirtyfold and enlarges land use by a factor of one hundred. This nuance dismantles simplistic narratives around “green” or “renewable-powered” data centers and compels stakeholders to weigh multifaceted environmental impacts in energy procurement and infrastructure siting. The geographic variance in electricity supply further complicates the notion of universal sustainability metrics.

The environmental and social implications extend deeply into the realm of mineral extraction and electronic waste. AI infrastructure relies on minerals often mined under conditions that disproportionately harm communities in the Global South, exacerbating environmental degradation and social injustices. By 2030, AI-related hardware waste could reach 2.5 million metric tons annually—equivalent to discarding a quarter of a million Eiffel Towers—posing severe challenges for hazardous material management and pollution control. The report calls for robust lifecycle governance spanning from resource acquisition through responsible disposal to mitigate these burdens on vulnerable populations.

Disparities in AI infrastructure distribution exacerbate global inequalities. Currently, 90% of specialized AI cloud infrastructure capacity is concentrated in just two countries—the United States and China—with only 32 nations worldwide hosting such facilities at all. The vast majority of over 150 countries remain dependent consumers of AI services, bearing metal extraction and e-waste costs disproportionately while reaping scant strategic benefits. This digital divide manifests not only as an economic disparity but as an environmental justice concern demanding urgent attention and coordinated global action.

Ireland stands as a cautionary exemplar of the perils of unregulated AI infrastructure growth. Data centers now consume 21% of the country’s total metered electricity—a sharp rise from 5% in 2015—exceeding the energy used by all urban households combined. The national grid operator’s decision to pause new data center approvals until 2028 encapsulates the critical need for integrative energy planning and sustainable infrastructure development, highlighting the risks that other nations might encounter without proactive governance.

The report presents a compelling call to action and a roadmap for responsible AI governance framed around six foundational principles: transparency in environmental impact reporting; efficiency engineered at the design phase; equity and environmental justice considerations; lifecycle accountability; international collaboration; and sustainable use practices. It addresses varied stakeholders—from governments integrating AI into energy and land-use policy, to industry prioritizing footprint-aware model development, to users selecting appropriate computational scales—emphasizing governance as a collective, multilevel imperative.

Finally, the report recognizes user interface design and behavioral choices as potent instruments for environmental stewardship. For instance, adopting a “concise mode” in AI interactions, which avoids unnecessary politeness or verbosity, can reduce token output by 30%, saving significant electricity—estimated at 87 to 98 gigawatt-hours annually. This reduction parallels the residential energy usage of 760,000 individuals in Sub-Saharan Africa, illustrating how seemingly small efficiency gains in user interactions and product defaults can cascade into substantial sustainability dividends.

In its starkest summary, UNU-INWEH’s report declares that AI’s environmental footprint is neither fixed nor inevitable; it is the product of cumulative engineering, usage, and policy decisions rooted in physical realities. Confronting AI’s rapid expansion with holistic, transparent, and just frameworks offers the only viable path to ensuring that technological progress advances human well-being within planetary boundaries. Without systemic and cooperative stewardship, the opportunity for AI to be a force for sustainable innovation risks being eclipsed by escalating environmental costs and intensifying inequalities.

Subject of Research: Environmental impacts of AI infrastructure and usage, including energy, carbon, water, land footprints, and associated social justice concerns.

Article Title: Environmental Cost of AI’s Energy Use: Carbon, Water and Land Footprints

News Publication Date: 2026

Web References:
https://unu.edu/inweh/collection/environmental-cost-of-AIs-Enrgy-Use-Carbon-water-and-land-footprints

References:
Aczel, M., Chamanara, S., Matin, M., Farsi, A., Marwala, T., Madani, K. (2026). Environmental Cost of AI’s Energy Use: Carbon, Water and Land Footprints. United Nations University Institute for Water, Environment and Health (UNU-INWEH), Richmond Hill, Ontario, Canada. doi: 10.53328/INR26RMA002

Image Credits: United Nations University Institute for Water, Environment and Health (UNU-INWEH)

Keywords

Artificial intelligence, AI energy consumption, carbon emissions, water footprint, land footprint, environmental justice, data centers, AI infrastructure, e-waste, sustainable AI, mineral extraction, global digital divide

The energy demands of AI are staggering; training a single large language model requires the equivalent yearly electricity for powering around 120 homes. Every AI-generated image consumes energy comparable to charging a smartphone, painting a vivid picture of the hidden environmental costs embedded in our digital convenience. Considering that approximately 85% of global energy production is still sourced from fossil fuels, the environmental ramifications of unchecked AI use are immense. The study, led by Cheng “Chris” Chen of OSU’s College of Liberal Arts, underscores the urgency of addressing AI’s energy footprint from a user behavior perspective.

Chen points out a glaring gap in existing AI systems: the lack of transparent communication regarding their environmental impacts. “Most AI platforms prioritize speed and output quality over ecological awareness,” he notes, “which means users rarely grasp the real-world consequences of their interactions with AI.” This opacity hampers users’ ability to make informed, environmentally responsible decisions, often leading them to unintentionally exacerbate energy consumption through repetitive or redundant AI queries.

The research explores the concept of “design friction,” which involves deliberately embedding small hurdles or pauses into software workflows to encourage users to think before proceeding. Two varieties were tested: action-based friction and cue-based friction. Action-based friction required users to actively search for existing images and specify detailed parameters before generating a new AI image. This intervention was found to make users more reflective and motivated toward eco-friendly AI usage, suggesting that deliberate user engagement techniques can foster sustainable digital habits.

On the other hand, cue-based friction, characterized by persuasive environmental messaging presented during AI use, had a more nuanced effect. While it increased users’ trust in the system, it did not significantly shift their intentions toward responsible AI use. This suggests that awareness alone is not enough to change behavior; actionable design elements that slow down the interaction and invoke deliberate decision-making are more effective.

The study’s findings resonate deeply in the context of AI’s rapid proliferation across industries and consumer applications. High-performance computing potentially accounting for 20% of global energy consumption by 2030 starkly illustrates the stakes involved. Chen emphasizes that such software design strategies for prompting user reflection could serve as vital tools in curtailing unnecessary AI usage and diminishing its cumulative environmental toll.

From a technical perspective, implementing design friction modifies usual user interface patterns by adding “speed bumps”—small interaction costs that interrupt the default consumption flow. These interruptions compel users to reconsider their needs and evaluate if generating new AI outputs is truly necessary, hence avoiding redundant computational energy expenditure. This principle leverages behavioral science by aligning ecological responsibility with moment-to-moment design cues.

Chen also articulates clear guidelines for responsible AI use, advocating leveraging these systems only when no comparably effective alternatives exist, meticulously avoiding duplicated efforts across AI projects, and consciously closing AI tools once goals have been met. The subtle encouragement to save generated outputs prevents the reinvention of similar AI creations and further conserves energy.

Crucially, the study reveals a disconnect between users’ understanding of AI’s environmental costs and their everyday usage. Chen highlights that the seemingly innocuous request for a “happy panda eating bamboo shoots” image entails tangible energy expenses that users typically do not factor into their decision-making. Increasing user awareness through design friction could bridge this gap and foster more sustainable interactions with AI platforms.

The implications of this research extend beyond energy savings; they invite a fundamental rethinking of human-computer interaction in the AI era. As AI systems become increasingly autonomous and efficient, embedding reflective pauses may be one of the few ways to preserve human intentionality in digital consumption. By encouraging users to intervene thoughtfully, these design principles could help transform AI usage from impulsive convenience into conscious collaboration with technological resources.

Moreover, the findings underscore the societal responsibility of AI developers and platform designers in balancing usability with sustainability. Prioritizing output speed and quality without addressing environmental impact neglects the broader consequences of AI’s digital footprint. Integrating design friction could become a standard practice in interface design, blending user experience optimization with ecological stewardship.

As digital technologies reshape how we create, work, and communicate, the environmental costs of AI stand out as a pressing challenge necessitating immediate attention. This study by Chen and collaborators from the University of Illinois and the University of Virginia offers a promising path forward by harnessing design psychology to cultivate environmental mindfulness—transforming how we engage with artificial intelligence for the betterment of both society and the planet.

Subject of Research: People

Article Title: Pausing for Reflection: How Design Friction Shapes Environmentally Responsible Artificial Intelligence Use and Trust

News Publication Date: 1-May-2026

Web References:
https://journals.sagepub.com/doi/10.1177/10755470261434438

References:
Chen, C., et al. (2026). Pausing for Reflection: How Design Friction Shapes Environmentally Responsible Artificial Intelligence Use and Trust. Science Communication. DOI: 10.1177/10755470261434438

Keywords

Artificial Intelligence, Environmental Impact, Energy Consumption, Design Friction, User Behavior, Sustainable Computing, Human-Computer Interaction, Eco-friendly AI, Digital Sustainability, Behavioral Science, AI Ethics, Energy Efficiency