BM @ MY LNA Opinion @ My LNA Technology

Sustainable computing strategies boosted by AI-powered energy efficiency optimisation

ByopinionMay 29, 20234 Mins read318

Assoc. Prof. Ts. Dr. Norziana Jamil

By: Assoc. Prof. Ts. Dr. Norziana Jamil

Computing technologies have become an indispensable part of our daily lives. The increasing reliance on computing technologies has made them an essential aspect of our everyday existence. However, this growing demand for technological progress has come with a corresponding surge in energy usage. As evidenced by a report from enerdata.net, electricity consumption per person has risen from 3,900 kWh in 2010 to 4,700 kWh in 2020. Unfortunately, this upward trend in energy consumption has detrimental consequences for the environment, contributing to issues such as global warming and climate change. Therefore, it is essential to optimise energy efficiency in computing strategies to achieve national sustainability. In this article, we will explore the challenges and initiatives taken to optimise energy efficiency in computing strategies.

Energy efficiency refers to using less energy to perform the same task or achieve the same output. In computing, energy efficiency can be achieved by using energy-efficient hardware, algorithms, software, and infrastructure. It is also essential to consider the entire computing infrastructure, including data centres, networks, and cloud services, among others.

The United Nations Sustainable Development Goals (SDGs) call for a reduction in carbon emissions to combat climate change, and energy efficiency is a crucial component of achieving this goal. Based on the Sustainable Development Goals Report 2022, the global greenhouse gas emissions must decline by 43 per cent by 2030 and to net zero by 2050. However, the 2022 UK Parliament Post reported that in 2020, it is estimated that the Information and Communication Technology (ICT) sector consumed approximately 4-6% of the total global electricity generation and the experts predict that energy consumption in the ICT sector is expected to rise by 2030.

The high rise in energy consumption in the ICT sector can be attributed to several factors. It was reported in a research manuscript published in Springer journal that some key contributors to the rising energy consumption includes growing digitalisation, increasing demand for cloud computing, online services and storage, the proliferation of computing devices, the expansion of ICT networks and increasing adoption of Artificial Intelligence (AI) and complex computing workloads.

AI has been a topic of much discourse as it has emerged as a disruptive technology that is transforming the way people work and innovate. While AI plays its role, let us look at how pertinently, the challenges of optimising energy efficiency can also be facilitated by various aspects of AI including Green AI, energy-efficient AI hardware, and algorithmic optimisation.

While optimising energy efficiency in computing strategies is critical for national sustainability, it is also important to balance the potential benefits of emerging technologies with their energy requirements. For example, AI and DL require significant energy consumption to operate, which must be balanced against their potential benefits in energy optimisation. The term “balanced” here refers to finding an equilibrium or trade-off between the energy consumption of AI/DL systems and their potential benefits in energy optimisation. It does not necessarily mean that the energy consumption must be an exact match with the benefits achieved. The idea behind balancing energy consumption and benefits is to ensure that the energy expended in running AI and DL algorithms is justified by the positive impact they have on energy optimisation. The goal is to optimise energy efficiency while achieving desirable outcomes, such as improved resource allocation, reduced energy waste, or enhanced sustainability.

It is the concept of energy rebound effect, a phenomenon where energy efficiency improvements reduce the cost of energy use, making it more affordable and leading to increased consumption, offsetting some or all of the energy savings achieved.

Although AI generates much enthusiasm and creates more avenues for energy efficiency, further research and innovation are necessary to tackle issues such as the energy rebound effect discussed above and potential risks or challenges linked with using these technologies for sustainability purposes. It is important to note that the rebound effect does not negate the importance of energy efficiency. Energy efficiency measures still contribute to energy savings and can have positive environmental and economic impacts. However, understanding and managing the rebound effect are crucial to ensure that efficiency gains lead to meaningful long-term energy and sustainability benefits. It is also crucial to consider how issues like cybersecurity and privacy intersect with the goal of optimising energy efficiency in computing strategies, and what steps can be taken to ensure that these concerns are addressed sustainably.

The discussion would be incomplete without addressing the importance of standardisation. To the best of my knowledge, as of the time of writing this article, there is no universally accepted standard for measuring and reporting energy efficiency in the computing industry. The absence of consistent metrics for measuring energy efficiency makes it difficult to assess the energy performance of different computing strategies accurately. Additionally, variations in workload characteristics and methodologies used to measure energy efficiency contribute to inconsistencies in data and unreliable comparisons. Organisations such as ENERGY STAR and the Green Grid are working to develop standards for measuring energy efficiency in computing. It is hoped that the establishment of world widely accepted standards will provide a valuable reference for researchers and companies to evaluate the effectiveness of their computing strategies in achieving energy efficiency goals.

In this article, we have briefly discussed why optimising energy efficiency in computing strategies is crucial for achieving national sustainability goals and combatting climate change. While emerging technologies like AI offer numerous benefits, it is crucial to address the potential challenges, such as the energy rebound effect, that can arise from their implementation. This necessitates further exploration and investigation of computing strategies to mitigate any adverse environmental impacts. To ensure a sustainable future, fostering a culture of energy conservation and education is essential. Initiatives like the Green Data Center Program by MDEC, the MyHijau Mark administered by GreenTech Malaysia, and the Smart Grid Initiatives by TNB exemplify Malaysia’s commitment to promoting sustainable computing practices. Additionally, programs such as Net Energy Metering by SEDA encourage the adoption of renewable energy sources and energy self-sufficiency.

However, the journey towards sustainable computing does not end here. Continuous research and innovation are needed to address emerging concerns and risks associated with energy-efficient computing. It is crucial to strike a balance between technological advancement and the broader societal and economic implications of our actions. By embracing sustainable computing strategies, investing in energy-efficient technologies, and fostering collaboration among stakeholders, we can pave the way for a greener and more sustainable future. Together, let us drive progress towards a harmonious coexistence of technological advancements, environmental preservation, and societal well-being.

……
^{The author is the Director of Institute of Informatics and Computing in Energy (IICE), Universiti Tenaga Nasional (UNITEN), and may be reached at norziana@uniten.edu.my}