AI's Energy Paradox: How can we balance innovation and conservation?

Moreover, the growing demand for GenAI is driving a significant expansion of hyperscale data centers, which use large amounts of water – billions of liters daily. In a water-stressed world, there is a need to start examining water management and planning early. 

Sustainable solutions
 
With the challenges ranging from training to usage to uptake, we need to look at systemic, holistic solutions that tackle the energy problem at its source. 

Most critical is continuing our efforts to supply our grids, and thereby our data centers, with round the clock renewable energy. This means designing renewable energy systems that can provide uninterrupted power, 24 hours a day, seven days a week, without having to rely on weather conditions such as wind or sunshine. By combining multiple renewable energy sources which can together match our growing energy needs, we can take the biggest step toward sustainable AI use.



Beyond infrastructure, the energy consumption of the AI models themselves can be reduced. Creating models that are designed to carry out specific tasks and suit different needs can lessen computational demands, both while those models are being trained and while they draw conclusions from that new data. Writing more efficient computer code can further optimize performance and reduce energy usage. 

Individual behavior has an important role to play here, too: being mindful of the fact that using GenAI has very real costs each time, we can reserve our use of AI for truly complex tasks, rather than turning to ChatGPT to find simple facts, or transcribing meeting notes we will never look at again.

And as with all solutions on the global scale, policy will have an essential role to play. This complex technology is too ubiquitous and too powerful to be left entirely unregulated, challenging governments to react quickly and decisively to incentivize sustainable AI development and hold business accountable.

Steps towards sustainability: the case of Singapore

Encouragingly, countries are already taking up the mantle to respond to GenAI’s environmental challenges, and Singapore is leading the charge with a strong emphasis on collaboration, innovation and practical implementation. 

In its first national AI strategy in 2019, Singapore led the triple-helix model of innovation, encouraging the government, academia and the private sector to collaborate on shared goals. 

Singapore Parliament.

The country's Infocomm Media Development Authority (IMDA) has acted as the connecting agent between researchers and industry. Through initiatives like the Green Computing Funding Initiative (GCFI), IMDA has provided funding and support for research into energy-efficient software which can be translated to practical applications and inform government policies.

Singapore’s government was also willing to act meaningfully on environmental imperatives, with a four-year pause on building new data centers between 2019 and 2022, implementing new environmental regulations once the restriction was lifted. Additionally, initiatives like The Sustainable Tropical Data Center Testbed have incentivized energy efficiency and decarbonization in AI. 

Progress and preservation: a call to action for sustainable AI

We've witnessed the incredible speed at which AI has been adopted and developed, proving our collective capacity to rapidly adapt and change. We must channel this same urgency into our environmental efforts.

At the individual level, we can make conscious choices about our technology consumption and support companies committed to sustainability. At the national level, governments must implement policies that prioritize green AI development and invest in research and infrastructure. Globally, international cooperation is essential to establish standards and guidelines that ensure AI is developed and used responsibly.

Importantly, AI can also play a powerful role in decarbonizing global energy systems and accelerating the transition to net zero. Joint research by ADNOC and Masdar, in collaboration with Microsoft, published in October 2024 found that over 90% of energy executives see a central role for AI in improving energy efficiency and developing new energy solutions. 

By acting decisively and collaboratively, we can harness the power of AI to create a sustainable future for generations to come. With the future of our planet depending on our ability to innovate responsibly, we must take action to make AI more sustainable.