The environmental impact of AI in Social Work

Overview

This briefing explores the emerging ethical challenge of environmental sustainability in the deployment of Artificial Intelligence (AI) within social work. While AI offers potential efficiencies, it carries an environmental impact that must be weighed against its benefits. This resource bridges the gap between technical reports and practice, utilising findings from research and principles from the AI Playbook for the UK Government to support informed decision-making.

Learning outcomes

As a result of engaging with this resource, participants will be able to:

• Understand the environmental impacts, such as energy and water consumption and e-waste, associated with AI development and use, and the potential sustainability opportunities AI brings.
• Identify key principles from the Government AI Playbook for mitigating environmental harm.
• Reflect on the ethical trade-offs between digital innovation and environmental justice.
• Consider questions to ask providers about the sustainability of their AI tools.

Introduction

As the social work sector begins to adopt artificial intelligence (AI) for tasks ranging from case recording to more advance predictive models, a new ethical dimension has emerged: environmental impact. AI infrastructure is resource intensive, and research is starting to measure its environmental impact.

Research on the ‘Emerging use of artificial intelligence in social work education and practice’ highlighted a growing but uneven awareness of this issue among social workers. While 38% of survey respondents expressed concern about the environmental impact of AI, this was rated far lower than other risks such as data privacy or bias.

The concerns about the environmental impact of AI are well founded in research, as are the potential sustainability opportunities. For social work leaders and practitioners committed to social justice - which includes environmental justice - understanding this impact is now a core component of digital literacy.

Key messages from research

Energy consumption

The "computational cost" of AI is significantly higher than traditional computing. The significant computational demands of rapid innovation and more advanced AI models contribute to environmental impacts because of reliance on non-renewable energy sources. While AI currently consumes a small portion of global electricity, its growth trajectory is steep.

• Projected Consumption Surge: Global electricity consumption from data centres is estimated to more than double the current levels by 2030 (1)

• Carbon intensity of training: The development phase of AI models is particularly carbon intensive. Research indicates that the process of training a single large AI model can emit as much carbon as five cars over their entire lifetimes (2).

• Everyday use: While training models gets attention, everyday use accounts for 80-90% of total AI computing power (3).

Water consumption

AI systems require enormous volumes of freshwater for cooling servers, generating electricity, and manufacturing components. This demand is often geographically uneven, exacerbating water stress in vulnerable regions.

• Water intensity of training: Training GPT-3 at a Microsoft data centre in the U.S. consumed more than 4 million litres of water (accounting for cooling and electricity-related use) (3)
• Water Use at Scale: The estimated annual water consumption for everyday use of GPT-4o (user prompting) is projected to be equivalent to filling over 500 Olympic-sized swimming pools (4).

E-Waste

AI relies on specialised computer processing equipment. The equipment requires constant upgrading to meet rising computational demands, leading to a massive increase in electronic waste (5).

• Current E-Waste Output: Generative AI systems currently produce around 2,500 tons of e-waste annually (5).

• Looming E-Waste Crisis: By 2030, e-waste from AI could soar to between 400,000 and 2.5 million tons per year, depending on adoption speed, potentially accounting for 3% to 12% of global e-waste (5).

Sustainability opportunities

Artificial Intelligence (AI) also presents significant environmental opportunities by acting as a powerful tool to accelerate solutions across energy systems, climate monitoring, resource management, and scientific innovation, often leading to measurable reductions in carbon emissions and resource consumption.

• Enabling sustainability: AI offers three capabilities essential for sustainability: the ability to measure, predict, and optimise complex systems; the power to accelerate the development of sustainability solutions; and the capacity to empower the sustainability workforce (6).
• Realised reductions: AI-enabled use cases have already helped organisations reduce Greenhouse Gas (GHG) emissions by 13% and improve power efficiency by 11% over two years (6).
• AI solving AI problems: AI assists in improving the efficiency of resource use, waste management, and the tracking of harmful emissions, water management, and climate and weather forecasting. AI tools can also help reduce material wastage in construction by up to 50% (1).

Concerns that AI could accelerate climate change appear overstated, as do expectations that AI alone will address the issue. Efficiency gains must be balanced against the rebound effect, where increased AI efficiency makes the technology cheaper and easier to use, leading to a surge in overall adoption and, potentially, an intensification of total resource consumption (1).

Implications for practice

For strategic leads and commissioners

The UK Government’s AI Playbook emphasises that sustainability must be built into the procurement and deployment lifecycle. Leaders should not view AI adoption in isolation but as part of their organisation's net-zero strategy.

• Mandate environmental impact assessments: Before procurement, require an assessment of the AI tool's energy footprint. Is the use of AI proportional to the task, or could a simpler, less energy-intensive tool suffice?

• Lifecycle assessment: Evaluate the impact of both hardware and software over their entire lifespan. Ensure that efficiency gains (e.g., time saved in administration) are not negated by the carbon cost of the compute power.

• Carbon offsetting: Where AI use is deemed necessary, ensure that implementation plans include strategies for carbon neutrality, such as offsetting emissions or utilising green cloud providers.

For workforce development and educators

• Sustainability Training: Integrate "digital sustainability" into the curriculum. Practitioners need the vocabulary to challenge the "inevitability" of high-carbon tech.

• Critical digital literacy: Encourage students to ask: "Where is this data stored?" and "What is the energy cost of this query?"

Supporting effective practice

References

1. International Energy Agency (IEA), Energy and AI: World Energy Outlook Special Report, April 2025 [accessed 11 April 2025].

2. Emma Strubell, Ananya Ganesh and Andrew Mccallum, Energy and Policy Considerations for Modern Deep Learning Research. University of Massachusetts Amherst.

3. Hastings Initiative Members, ‘The Hidden Footprint of AI: Climate, Water, and Justice Costs’, The Hastings Center, 21 July 2025, [accessed 13 January 2026].

4. Jegham, Nidhal, and others, 'How Hungry is AI? Benchmarking Energy, Water, and Carbon Footprint of LLM Inference', Cornell University arXiv, 17 June 2024, [accessed 28 November 2025].

5.  'The Hidden Cost of AI: A Looming E-Waste Crisis by 2030', AI2MED, 13 May 2024, [accessed 13 Jan 2025].

6. CapGemini Research Institute (2020). How artificial intelligence can power your climate action strategy.