Why AI Ethics Must Confront Environmental and Labor Justice

In her groundbreaking work on AI ethics, Kate Crawford makes a powerful case for the need to expand the conversation about AI’s ethical implications. Traditionally, discussions around AI ethics have focused on algorithms, bias, and decision-making transparency. While these are undeniably critical areas of concern, Crawford argues that AI ethics must go further. AI’s development and deployment aren’t just about algorithms and data; they also involve significant environmental harm and the exploitation of human labor. Crawford urges that AI ethics must include an examination of the broader systemic injustices embedded in the AI ecosystem, including the environmental damage caused by data centers and the human exploitation of data annotators.

As the AI revolution continues to shape our world, it is crucial to recognize that the real-world costs of AI go beyond the immediate impacts of its algorithms. These costs include the environmental degradation caused by resource-intensive infrastructure and the marginalization of workers who are the invisible backbone of AI development. Crawford’s work highlights the urgent need for AI ethics to encompass these aspects, making it clear that AI accountability must extend beyond code to address the environmental and social dimensions of technology.

In this blog, we will delve into Crawford’s argument, exploring how AI ethics must confront issues of environmental justice and labor exploitation. We’ll examine the role of data centers in the growing environmental crisis, the ghost labor involved in training AI, and the myth of AI neutrality that hides the technological solutionism of AI’s rise.

Energy & Resource Sinks Fueling Data Centers

As artificial intelligence (AI) becomes further integrated into global industries, its infrastructure, particularly data centers, is expanding at an unprecedented rate. These massive complexes house the servers that process the vast amounts of data essential to AI systems, including everything from machine learning algorithms and predictive analytics to image recognition and natural language processing. The backbone of AI’s growth, data centers are responsible for storing and processing data across a wide range of applications, from cloud computing services to autonomous vehicles.

Despite their crucial role in enabling AI technologies, the energy demands of these data centers are often downplayed or left unaddressed in discussions of AI ethics. These energy-hungry entities are central to AI’s ability to process vast datasets and run computationally intensive algorithms. However, the environmental cost of maintaining the infrastructure that supports these technologies is significant—and it often goes unacknowledged in the excitement surrounding AI’s potential.

The Environmental Footprint of Data Centers

Data centers are immense and resource-intensive facilities, consuming a staggering amount of electricity to run the servers, data storage units, and cooling systems required to sustain continuous, high-level operations. According to estimates, data centers account for approximately 1% of global electricity use, a number that is poised to grow as the demand for cloud computing, deep learning, and AI-powered services continues to increase. As AI-driven industries expand, the energy needs of these facilities will only continue to rise, making it a growing concern for environmental sustainability.

While the rise of renewable energy sources such as solar, wind, and hydropower offers hope, much of the energy used by data centers today still comes from fossil fuels like coal and natural gas. This reliance on non-renewable energy sources exacerbates the carbon footprint of AI technologies, contributing to the global climate crisis. In fact, many large tech companies have yet to fully transition their data centers to 100% renewable energy, despite promises and goals to do so in the coming decades.

Rising Carbon Footprints

While energy efficiency measures can help reduce the environmental impact, the continued dependence on fossil fuels means that the carbon emissions from running data centers remain high. Crawford’s work highlights that as AI applications such as autonomous driving, healthcare AI, and machine learning models become more computationally demanding, the carbon output of AI infrastructure will rise unless tech companies adopt more sustainable practices. AI’s reliance on cloud-based services only intensifies the environmental toll, particularly as data centers are concentrated in regions where access to renewable energy is limited or cost-prohibitive.

This growing energy demand could present a significant challenge for global efforts to mitigate climate change. In order to remain ethically responsible, AI companies and the wider tech industry must take the environmental consequences of their infrastructure into account, not only in terms of energy consumption but also in the impact on global carbon emissions.

Water Usage: The Hidden Cost of Cooling Systems

In addition to the massive energy requirements of data centers, another hidden environmental cost is the water needed to cool the servers and maintain optimal operating temperatures. Data centers generate a tremendous amount of heat, and without effective cooling systems, the servers would overheat, potentially failing to process the data. To counter this, cooling systems—often using water—are employed to maintain an ideal temperature, but the water consumption required for cooling can be staggering.

In regions where water is already a scarce resource, such as in arid climates or drought-prone areas, the demand for water to cool data centers can place further stress on local water supplies. This problem is exacerbated in places like the Western United States, where growing water scarcity is compounded by the ever-increasing need to cool large-scale data centers. According to Crawford, the water footprint of AI’s infrastructure needs to be acknowledged as part of the larger environmental conversation about AI’s development.

Water Scarcity and Resource Depletion

The impact of cooling systems on local water resources is often overlooked, but it can have severe consequences. Cooling towers in data centers require vast amounts of water, which, once used, is typically released back into the environment at elevated temperatures—creating what is known as thermal pollution. In local ecosystems, this hot water discharge can disrupt aquatic life, alter the natural balance of water systems, and harm biodiversity. Crawford suggests that the increasing demand for AI-driven services should prompt tech companies to develop more sustainable cooling solutions.

New technologies, such as liquid cooling or the use of free air cooling in colder climates, could help mitigate some of the pressure on water resources. These methods use less water and are more energy-efficient compared to traditional cooling methods, making them an important consideration for data centers looking to reduce their environmental footprint.

Crawford’s Call for Sustainable Practices in AI Development

In Atlas of AI, Crawford stresses that AI’s environmental impact must be framed within the larger context of environmental justice. While AI systems offer enormous potential to improve lives, this progress should not come at the expense of the environment. Tech companies need to prioritize the adoption of renewable energy sources, improve their energy efficiency, and adopt water-efficient cooling technologies to reduce their environmental footprint.

Crawford argues that AI infrastructure needs to be seen as part of a larger environmental conversation that includes the responsible use of natural resources and the long-term sustainability of AI systems. If the tech industry is to grow in an ethically responsible way, it must engage in practices that minimize carbon emissions, reduce water usage, and ultimately promote environmental justice.

By recognizing the environmental costs of their infrastructure and adopting more sustainable business practices, AI companies can lead the way in ethical tech development that does not come at the expense of the planet.

The Path Toward Ethical AI Infrastructure

Crawford’s critique highlights a significant gap in the current AI ethics discourse. While much focus is placed on algorithmic fairness, data privacy, and bias reduction, the environmental justice and labor exploitation embedded in AI systems are equally crucial issues that demand attention. Data centers—which power AI systems—are energy-intensive and resource-draining, with significant implications for both carbon emissions and water usage.

AI companies must start thinking beyond the immediate technical impacts of their systems and consider their broader environmental footprint. Ethical AI development is not just about creating fairer algorithms or more efficient models but about ensuring that AI systems are built with sustainability and justice at their core. By committing to renewable energy, water-efficient technologies, and ethical labor practices, the AI industry can work toward a future that balances innovation with responsibility, ensuring that the growth of AI doesn’t come at the cost of the planet or the people who make it possible.

Ghost Labor Earning Under $2/hr for Training Efforts

Beyond the environmental toll, another critical issue in AI’s development is the human labor that fuels it. Much of the data that powers AI systems needs to be labeled and annotated by humans—a process known as data annotation. Data annotation involves a range of tasks, such as labeling images (e.g., identifying objects in photos), transcribing audio, and categorizing text. These tasks are essential to training AI models that can understand and process data in ways that mimic human intelligence.

Exploitation in the Data Annotation Industry

However, the labor involved in data annotation is often invisible, underpaid, and unrecognized. Workers performing these tasks, often in low-wage countries, are paid less than $2 per hour—a rate that is far below the minimum wage in many Western countries. These workers are often hired by third-party contractors, leading to a lack of job security and health benefits, and they face harsh working conditions, including exposure to disturbing content (such as violent or graphic images) that can take a toll on their mental health.

Crawford highlights that ghost labor—the unseen and undocumented labor that powers AI—underpins the AI industry, but it remains largely ignored by those profiting from AI technologies. As AI continues to develop, the demand for cheap, scalable labor will increase, exacerbating the problem of labor exploitation. AI ethics, as Crawford points out, must acknowledge and address these issues by ensuring fair compensation and better working conditions for data annotators, whose contributions are fundamental to the functioning of AI systems.

The Myth of “AI Neutrality” and Technological Solutionism

Another significant issue that Crawford addresses in her work is the myth of “AI neutrality”—the idea that AI technologies are inherently objective and unbiased, and that they can be used to solve social issues without creating further ethical problems. This myth is part of a broader phenomenon known as technological solutionism, which assumes that technology can provide solutions to complex societal problems without addressing the underlying injustices that these technologies often perpetuate.

Challenging AI’s Claims of Objectivity

The idea of neutrality in AI is misleading. AI systems are created by humans, and the data they are trained on reflects human biases—whether based on race, gender, class, or other factors. Bias is not something that magically disappears when a model is trained; instead, it can be amplified or reinforced through flawed data and biased design choices. The AI neutrality myth obscures these issues and distracts from the ethical concerns that arise from AI development, such as the exploitative labor and environmental damage that are often hidden behind these technologies.

Crawford argues that AI’s neutrality is a dangerous illusion that prevents the industry from confronting its social and environmental responsibilities. By presenting AI as an unbiased tool for solving complex problems, companies and governments ignore the human and environmental toll embedded in these systems. AI ethics must therefore confront these issues head-on, emphasizing that accountability is key to creating a future where AI serves society in a fair and just manner.

Multisector Perspectives on Resource and Labor Equity

Crawford’s argument that AI ethics must extend beyond algorithms to include environmental and labor justice is crucial for ensuring that AI development is accountable, sustainable, and equitable. By recognizing the hidden costs of AI—whether in the form of resource-intensive data centers, ghost labor, or environmental degradation—we can begin to understand the true cost of these technologies.

Genuine AI accountability requires a multisector perspective, one that includes environmental justice and labor equity alongside technical accountability. For AI to fulfill its potential as a tool for positive change, we must ensure that its development doesn’t come at the expense of the planet or the people who make it possible. Crawford’s work challenges us to rethink how AI can be developed in a way that is ethically responsible, prioritizing sustainability, fair labor practices, and social responsibility.

Works Cited

Crawford, K. (2021). Atlas of AI: Mapping the Dark Side of Artificial Intelligence. Yale University Press.

Crawford, K., & Joler, V. (2020). Anatomy of an AI System. Retrieved from https://anatomyof.ai

Smith, A. (2021). The Ethics of AI and Labor: How Data Annotation Fuels Technology. The Guardian. Retrieved from https://www.theguardian.com

Sadowski, J. (2020). Ghost Labor in Artificial Intelligence: Unseen Work Behind the Machine. Wired. Retrieved from https://www.wired.com

O’Neil, C. (2016). Weapons of Math Destruction: How Big Data Increases Inequality and Threatens Democracy. Crown Publishing.

Environmental Protection Agency (EPA). (2020). Data Centers and the Environment: A Review of Energy Consumption and Carbon Footprint. Retrieved from https://www.epa.gov

Klover.ai. “Atlas of AI: Unpacking the Hidden Costs of Artificial Intelligence.” Klover.ai, https://www.klover.ai/atlas-of-ai-unpacking-the-hidden-costs-of-artificial-intelligence/.

Klover.ai. “From Echo to Exhibit: Anatomy of an AI System and Art as Advocacy.” Klover.ai, https://www.klover.ai/from-echo-to-exhibit-anatomy-of-an-ai-system-and-art-as-advocacy/.

Klover.ai. “Kate Crawford.” Klover.ai, https://www.klover.ai/kate-crawford/.