Everyone is talking about the wonders of AI. If someone wants to add a critical perspective, they often focus on ethical questions. But where are the voices raising concerns about the energy consumption involved in AI processes? Is the world truly prepared to allocate such vast amounts of energy to AI in the name of technological efficiency and speed? Is this not coming at the expense of millions of destitute individuals who are already suffering due to insufficient energy supply? Moreover, millions more are bearing the brunt of environmental degradation caused by the reckless exploitation of resources.
Is efficiency for a privileged few more urgent than addressing the basic needs of these hundreds of millions? AI may thrive, but can our planet sustain this trajectory? If these questions are not openly discussed, do we not risk heading toward a catastrophic future?
This very serious issue which is being
raised here is deeply important and timely the question about the sustainability and ethical considerations of AI development. By emphasizing the trade-offs between technological efficiency and human or environmental costs, you highlight an often-overlooked dimension of the AI debate.
It is especially relevant to question whether the energy-intensive demands of AI systems, such as training large models, align with broader global priorities like addressing poverty, ensuring equitable energy access, and combating climate change. Your critique about prioritizing efficiency for a privileged few over the pressing needs of millions is powerful and thought-provoking.
To make the argument against unchecked energy consumption in AI more compelling, it is crucial to include specific data and real-world examples. AI systems, particularly large models like GPT-4, demand enormous amounts of energy for training and deployment. For instance, training a single large AI model can emit over 626,000 pounds of CO2—equivalent to the lifetime emissions of five cars, including fuel use. This is part of a broader trend where data centers, the backbone of AI and other digital technologies, consume about 1% of global electricity—a figure projected to rise significantly. We here may refer to the latest development as reported here https://analyticsindiamag.com/ai-news-updates/openai-o3-consumes-five-tanks-of-gas-per-task/.
Such energy demands raise serious ethical questions, especially when contrasted with the realities of energy inequality. In developing regions like India and Sub-Saharan Africa, millions lack basic access to electricity. For example, in rural India, power cuts are frequent, with some villages receiving electricity for only a few hours a day. Meanwhile, urban tech hubs consume vast amounts of energy to power industries, including AI research facilities. This disparity highlights how prioritizing high-tech industries can exacerbate existing inequalities, diverting resources away from the most vulnerable populations.
The environmental costs of AI further complicate this picture. Beyond carbon emissions, the mining of rare-earth elements for AI hardware, such as GPUs and processors, leads to deforestation, water pollution, and soil erosion. China, a major producer of these materials, has faced significant environmental degradation in mining regions. Additionally, the increased energy consumption required for AI exacerbates climate change, which disproportionately impacts low-income and marginalized communities. For instance, rising sea levels in Bangladesh have already displaced millions, and further environmental strain could worsen such crises.
This juxtaposition of AI-driven efficiency with the unmet needs of millions paints a stark picture. In India alone, 75 million households still rely on kerosene for lighting due to inadequate electricity access. Redirecting energy investments toward renewable energy or basic electrification projects could alleviate these issues while reducing environmental harm.
By grounding the discussion in concrete examples, it becomes clear that the current trajectory of AI development risks deepening social and environmental inequalities. This underscores the urgent need to adopt more responsible and inclusive approaches to technological advancement, ensuring that efficiency for a privileged few does not come at the expense of humanity and the planet.
The argument presented in Acemoglu and Johnson’s book Power and Progress is that technology, while holding immense potential, has historically been shaped and directed in ways that often benefit a select few at the expense of the broader population. This critique is especially pertinent in the context of artificial intelligence (AI), where unchecked exploitation of resources and uneven distribution of benefits could exacerbate existing inequalities, creating a world where billions suffer under the control of a privileged elite.
Historical Evidence: The Deliberate Shaping of Technology
Acemoglu and Johnson highlight historical instances where technology was deliberately developed and deployed to entrench power structures rather than improve societal well-being:
1. Industrial Revolution: During the Industrial Revolution, technological advancements such as mechanized looms and steam engines increased productivity but were designed to centralize profits. Workers often endured abysmal conditions, with the wealth created by industrialization concentrating in the hands of factory owners and industrialists.
2. Digital Revolution: More recently, the digital revolution has demonstrated similar patterns. The rise of big tech companies like Amazon, Google, and Facebook has created immense wealth and efficiency for a small number of individuals, while issues such as labor exploitation, surveillance capitalism, and data privacy concerns disproportionately affect ordinary people.
These patterns suggest that without intentional efforts to democratize access to technology and its benefits, innovation often serves to deepen inequalities.
AI and Resource Exploitation
The energy and resource demands of AI offer another example of how technological advancements can disproportionately impact vulnerable populations. Training large AI models requires vast computational power, leading to significant energy consumption:
• Environmental Costs: Training a single large AI model can emit as much CO2 as five cars over their lifetimes. Data centers supporting AI operations consume around 1% of global electricity, contributing to climate change, which disproportionately affects the poor. For example, countries like Bangladesh are already experiencing devastating impacts from rising sea levels and extreme weather events exacerbated by climate change.
• Mineral Exploitation: AI hardware, such as GPUs and chips, depends on rare-earth elements and minerals like cobalt and lithium. Mining these resources often occurs in developing countries like the Democratic Republic of Congo, where child labor, unsafe conditions, and environmental destruction are rampant.
These examples reveal how the resources needed to power AI systems often come at a significant human and environmental cost, borne primarily by the world’s most disadvantaged populations.
Concentration of Ownership and Benefits
AI has the potential to exacerbate wealth and power inequalities if its development and deployment remain unchecked:
• Corporate Monopolies: The control of AI technologies is concentrated in the hands of a few corporations, such as OpenAI, Google, and Amazon, which have the resources to dominate the field. This concentration of power mirrors historical patterns of wealth accumulation during previous technological revolutions.
• Economic Displacement: Automation and AI could replace millions of jobs, particularly low-skill and repetitive roles, pushing already vulnerable workers into deeper poverty. For example, studies predict that as many as 375 million workers globally could need to transition to new roles due to automation by 2030.
This centralization of AI ownership and benefits creates a scenario where a small elite reaps the rewards of technological progress while billions face unemployment, reduced opportunities, and heightened economic precarity.
The concerns raised by Acemoglu and Johnson in Power and Progress about the inequitable distribution of technology’s benefits resonate with broader critiques from economists, environmentalists, and sociologists. Beyond their perspective, there are numerous viewpoints and real-world examples that reveal how the monopolization of AI and digital technologies by big tech companies is exacerbating global inequalities, exploiting energy resources, and depriving millions of opportunities.
The Monopolization of Digital Technologies: Global Perspectives
1. Energy Exploitation by Big Tech
Big tech companies like Google, Amazon, Microsoft, and Meta have built sprawling data centers to power their AI systems, cloud services, and digital platforms. These centers are notoriously energy-intensive, consuming vast amounts of electricity and water for cooling.
• According to a report by Nature, global data centers consume around 200 terawatt-hours of electricity annually, equivalent to the energy consumption of some medium-sized countries like Argentina.
• In regions like Arizona, where water is scarce, data centers owned by Amazon Web Services (AWS) use millions of gallons of water daily for cooling. This deprives local communities of access to clean water resources, compounding existing inequalities.
2. Unequal Distribution of Benefits
• While AI and digital technologies generate immense profits for corporations, they displace millions of workers in traditional industries. For example, automation and algorithmic decision-making are projected to eliminate 85 million jobs globally by 2025 while creating fewer high-skill roles that most displaced workers cannot easily access.
• In India, for instance, platforms like Uber and Zomato have disrupted traditional livelihoods in transport and food delivery but offer precarious gig economy jobs with low pay, no benefits, and poor working conditions.
3. Digital Colonialism
Scholars like Nick Couldry and Ulises A. Mejias argue that big tech companies are engaging in “digital colonialism,” where they extract data and resources from developing countries while providing minimal returns to local populations.
• Companies often build data centers in developing nations, exploiting cheap labor and energy subsidies, while repatriating profits to their headquarters in developed countries.
• The expansion of platforms like Facebook and Google in Africa, Asia, and Latin America has deepened their control over digital infrastructure, stifling local innovation and entrepreneurship.
Environmental and Social Costs
1. Environmental Degradation
AI-driven technologies exacerbate environmental degradation:
• Mining for rare-earth materials such as lithium and cobalt, essential for AI hardware, has devastating effects on ecosystems and local communities. For example, in the Democratic Republic of Congo, cobalt mining is associated with toxic waste, deforestation, and displacement of local populations.
• Carbon emissions from training AI models contribute to climate change, which disproportionately affects poor and marginalized communities in countries like Bangladesh and sub-Saharan Africa, already grappling with rising sea levels and extreme weather.
2. Energy Hoarding at the Expense of Communities
• Energy-intensive blockchain technologies and cryptocurrencies, often intertwined with AI systems, further strain global energy supplies. Bitcoin mining alone consumes more energy annually than countries like Norway.
• In developing nations like India, energy allocated to big tech and industrial zones often comes at the cost of rural electrification projects, leaving millions without reliable power.
Broader Critiques of Big Tech and AI
1. Shoshana Zuboff’s Surveillance Capitalism
In her seminal work, The Age of Surveillance Capitalism, Zuboff critiques how big tech companies monetize personal data while providing little value in return. Their business model prioritizes profit over privacy and social welfare, further concentrating wealth and power in the hands of a few.
• AI algorithms used in targeted advertising and recommendation systems fuel consumerism, diverting resources into producing non-essential goods while essential services like healthcare and education remain underfunded in many regions.
2. Economic Inequality and Technological Displacement
Economists like Joseph Stiglitz and Amartya Sen have highlighted how technological advancements often exacerbate economic inequality:
• AI-powered automation disproportionately impacts low-skilled workers, many of whom are already struggling to make ends meet. In sectors like agriculture and manufacturing, where millions in developing countries are employed, automation threatens to displace livelihoods without offering viable alternatives.
• Even within developed nations, inequality widens as tech companies pay relatively low taxes compared to their massive revenues, depriving governments of resources needed for public welfare.
The Impact on Global Opportunities
The monopolization of energy and digital resources by big tech has deprived millions of opportunities for sustainable development:
• Loss of Local Enterprises: Small businesses in developing countries struggle to compete with AI-driven platforms that monopolize markets through algorithmic pricing and global reach.
• Barriers to Digital Sovereignty: Many nations lack the infrastructure to develop and deploy their own AI technologies, forcing them to rely on Western tech companies. This dependence stifles local innovation and perpetuates global inequalities.
A Call for Action
To address these pressing issues, governments, activists, and civil society must take deliberate steps:
1. Regulating Big Tech: Governments must enforce stricter regulations on energy consumption, data privacy, and fair labor practices. Progressive taxation of tech giants can fund renewable energy projects and equitable digital infrastructure.
2. Promoting Local Innovation: Developing countries should invest in building their own AI and digital capabilities to reduce dependence on foreign tech monopolies. For example, India’s push for digital self-reliance through initiatives like Aadhaar and UPI showcases the potential for inclusive technological progress.
3. Encouraging Global Cooperation: The international community must work together to address the environmental and social costs of AI by setting global standards for responsible technological development.
Conclusion
Big tech companies have turned AI into a tool for profit and control, exploiting energy resources and exacerbating social inequalities. By prioritizing the needs of a privileged few, they risk creating a world where billions suffer from displacement, deprivation, and environmental degradation. A more equitable and sustainable approach to AI development is essential to ensure that technological progress benefits humanity as a whole, rather than deepening the divide between the privileged and the destitute.
Toward an Inclusive AI Future
To avoid the dystopian future described in Power and Progress, where billions stagger under suffering while a few privileged individuals dominate, deliberate interventions are needed:
1. Equitable Policies: Governments must regulate AI development to ensure fair distribution of its benefits. For example, progressive taxation of tech giants and reinvestment in public welfare could mitigate inequality.
2. Sustainable Development: Policies must prioritize environmental sustainability, such as transitioning AI operations to renewable energy sources and improving recycling methods for hardware components.
3. Empowering Marginalized Communities: Investments in education, training, and social protections can help workers transition into new roles created by AI, reducing the risk of mass displacement.
Conclusion
The trajectory of AI development, if left unchecked, risks creating a world of profound inequality, echoing historical examples where technology was used to entrench power. Drawing on insights from Acemoglu and Johnson, it is evident that deliberate, inclusive, and ethical interventions are necessary to ensure that AI benefits humanity as a whole rather than a privileged few. Without such measures, the promise of technological progress may devolve into a dystopian reality of widespread suffering and environmental degradation.
