How can society help reduce the energy usage of AI?

Join our social channels to steer AI towards sustainability
scifi datacentre image
A N3XTCODER series

Implementing AI for Social Innovation

Welcome to the N3xtcoder series on Implementing AI for Social Innovation Series.

In this series we are looking at ways in which Artificial Intelligence can be used to benefit society and our planet - in particular the practical use of AI for Social Innovation projects.

In this article we explore:

  • What can Governments and Regulators do? We look at what policy makers can do to help lower the energy consumption of AI?

If you have missed the first 2 articles in the series, here are the links:

Article 1: The carbon footprint of generative AI 

  • Why does AI need so much energy? We give a brief overview of why AI is so energy and resource intensive.

Article 2: How can developers and AI help reduce the energy usage of AI

covering the topics:

  • What can AI developers do? We look at how developers can reduce energy usage when programming AI, and also when setting up AI systems. 
  • What can AI do? We explore how AI applications and tools can also help reduce energy consumption of AI and other digital systems.

But let’s jump straight in:

What can Governments and Regulators do?

While there is a lot AI developers can do to make AI more energy efficient, governments and regulators could make a much bigger impact if they prioritised green(er) AI. As the EU, European governments, and in particular the German Government, are now investing billions of Euros to support AI development, there are steps they could take to make sure that funding supports more energy-friendly AI approaches:

  1. Prioritise and incentivise Green AI research

Governments need to set companies the right incentives and targets; when developing with AI, we should focus on getting results that  minimise the environmental footprint and encourage applications with a positive environmental impact.

So what is happening in Germany? Germany plans to invest nearly €500M in AI research and innovation by 2024. According to the planning document (https://www.bmbf.de/SharedDocs/Downloads/de/2023/230823-executive-summary-ki-aktionsplan.pdf ). Sustainability will be a focus, including in these specific areas:

  • Developing AI as a tool for sustainability in general and putting the focus of AI development on sustainability.
  • Development of platforms for regional government to make fast yet informed decisions on climate measures in cities.
  • Usage of AI for biodiversity research.
  • Establishing AI as a standard tool for scientific research, especially at scientific research institutes outside of universities.
  • Establishing sustainability at the foundation of the German AI sector and fostering cross-disciplinary collaboration networks between climate and environmental scientists, and AI developers

Addressing the environmental impact of AI urgently requires a shift towards Green AI research. Of course optimising algorithms, developing energy-efficient hardware and fine-tuning software are crucial steps in creating a more sustainable AI ecosystem. But it is equally important to consciously foster collaboration between AI researchers and environmental scientists which might lead to better understanding of the impact of AI on the environment, the discovery of innovative solutions, and the development of strategies to mitigate the negative effects of AI. 

  1. Implementing robust regulatory frameworks for sustainability

Regulations can incentivise the adoption of energy-efficient practices and penalise excessive resource usage. Here are two initiatives that could have a major positive impact on AI development:

  • EU  Ecodesign for Sustainable Products Regulation (ESPR)
    Published in March 2022, the ESPR establishes a framework to set ecodesign requirements for some  product groups to improve their circularity, energy performance, and other sustainability aspects. Although software, including AI, largely falls beyond the current scope of ESPR, the forthcoming AI Act might introduce some measures in this area, albeit with limited scope due to the lack of common standards.​ https://www.techpolicy.press/addressing-ai-energy-consumption-why-the-eu-must-embrace-ecodesign-for-software/ 
  • Energieeffizienzgesetz -Energy Efficiency Law, Germany -  EnEfG
    Passed on September 21, 2023, the EnEfG aims to significantly increase energy efficiency requirements for companies and data centres in Germany. It includes obligations for companies to establish energy management systems or environmental management systems, implement energy-saving measures, conduct energy audits, and consider measures for avoiding and reducing waste heat. (https://www.jdsupra.com/legalnews/update-germany-tightens-energy-7136097/
  1. Clear and rigorous energy and environmental impact reporting requirements

As we have seen in the other articles in this series, one of the main challenges in reducing AI’s energy and environmental impact, is accurately measuring the impact of AI models. Governments and regulators could make a huge contribution here by establishing and then enforcing rigorous and meaningful reporting standards on not just AI companies, but all technology companies along the AI supply chain. This should include for:

  • AI developers: accurate reporting on size of overall energy consumption, as well as energy costs for every individual prompt
  • Cloud services: easy-to-read carbon impact reports using standardised measures, including embedded energy and carbon (from manufacturing), and end-of-life assessments
  • GPU and other hardware manufacturers: Energy and carbon footprint reports for all individual components based on standardised measurements, including impact of raw materials

By enforcing environmental reporting standards on all suppliers, as well as AI developers, end users should get a clear picture on the true energy and carbon cost of using AI tools.

  1. Funding Open Source models and Open Data initiatives

Open approaches to technology foster collaboration, more efficient “edge” processes, and also the reusing of energy-intensive processes, such as model training. Governments could support these approaches through guidelines or stipulations on funds, and also by scoring these approaches more highly in public tender processes.

  1. Education and Awareness

Governments can raise awareness (using projects like this AI series) about the environmental impact of AI. Through content, training and events awareness projects could educate developers, researchers, and organisations about best practices for optimising AI models and making best use of resource-efficient AI technologies.

Conclusion and helpful resources 

The intersection of green AI and innovation presents an opportunity to both mitigate the environmental consequences of AI and usher in a new era of responsible and inclusive technological progress. As AI becomes more widely adopted, it is incumbent upon the AI community to prioritise sustainability, ensuring that the benefits of AI innovation are harmoniously aligned with the well-being of our planet.

AI can be used to help lower the energy usage of AI but only if humans instruct it to do so and point it in the right direction. It will be necessary for society and politics to create the right incentives for companies and developers to prioritise green AI development. 

Helpful Resources: https://sustain.algorithmwatch.org/

If you missed the first two parts:

Read on in part 1: The Energy Impact of Artificial Intelligence
Read on in part 2: What can AI and AI developers do to reduce the carbon footprint of AI

Was this article helpful? yes no

Join us in the conversation on various social channels

Join us in the conversation on various social channels. We discuss the latest developments in technology as they happen!

This article has been realised with the help of
Bundesministerium für Wirtschaft und Klimaschutz
NextGenerationEU