How Singapore taps on sandboxes to drive green AI development

By Si Ying Thian

The ability of the government to “create safe areas to test green AI technology” brings together R&D and industry expertise to accelerate green software innovation, said speakers at the GSF Global Summit Singapore.

Speakers from Tony Blair Institute for Global Change, IMDA, Red Hat, and GitHub at the inaugural GSF Global Summit Singapore on October 9. Image: Infocomm Media Development Authority (IMDA)

As countries launch more artificial intelligence (AI) efforts, they face the dilemma of managing the environmental impact that comes with AI’s high energy requirements.

This is why green software development is gaining traction in the economy. Rather than only neutralising its impact, green software focuses on reducing carbon footprints while enabling innovations.

GovTech Singapore and Infocomm Media Development Authority (IMDA) are among the first government agencies to join the Green Software Foundation (GSF), said the non-profit’s speaker and Accenture's Southeast Asia Technology Sustainability Lead, Arun VT.

“This not only sets up Singapore as a pioneering country [in the green software space] but globally for GSF, it’s an important milestone,” Arun said.

He was speaking at the inaugural GSF Global Summit in Singapore on October 9, organised by IMDA.

Last week saw other GSF summits happening around the world – but the Singapore edition was the only one hosted by a government agency.

“Countries like Singapore are now setting the pace [for green artificial intelligence (AI)], but it’s clear that efforts need to increase around the world,” said Tony Blair Institute for Global Change’s Chief Policy Strategist, Benedict Macon-Cooney.

The institute’s newly launched global report on Greening AI highlighted Singapore’s approach to using sandbox environments to leverage research and development (R&D) expertise to accelerate the development of green AI innovations for industry use.

The R&D expertise includes cutting-edge research technology transfer and capacity-building efforts.

To subscribe to the GovInsider bulletin click here.

Singapore’s sandbox approach

IMDA’s Assistant Chief Executive of BizTech Group, Ong Chen Hui, said that its research and statistics unit found that of the 1,000 companies surveyed, 57 per cent had no initial plans to adopt green software.

50 per cent of the 57 per cent cited that they were unaware of what green software entails and its benefits.

Around 45 per cent of companies also reported a lack of internal expertise to implement green AI efforts.

She also shared some updates regarding the green software trials announced earlier this year, where the agency was working with the private sector to deploy green software and measure the improvements.

Minister of State for Trade and Industry Mr Alvin Tan touring the Sustainable Tropical Data Centre Testbed facility, and was shown the manifold wall in the facility used for smart monitoring for better energy efficiency. Image: NUS

“The likes of AWS and IBM have brought their customers along to join the green software trials and through these trials, we want to increase the awareness of various software and data architecture choices that are available,” she explained.

These efforts are accompanied by technical sharing sessions highlighting the industry’s work in green software, and speakers included Temus, Alibaba Cloud, and more.

“Now to build a sustainable digital future, it's imperative that we raise awareness about software practices and equip developers with the necessary skills implemented,” she emphasised.

Last November, National University of Singapore (NUS) and Nanyang Technological University (NTU) also collaborated to set up the world’s first tropical climate data centre testbed.

The sandbox targets to “bridge the gap between research and practical applications” to develop sustainable data centre innovations, according to an official press release.

Industry buy-in for partnerships

At the summit, global open-source software company, Red Hat, announced plans to set up a Sustainable AI Innovation Centre in Singapore to unify various research and innovation efforts around green software design and operations.

Red Hat’s APAC Chief Technology Officer, Vincent Caldeira, said that Singapore is well-positioned to foster partnerships between the government, academia and industry.

“For me, it starts with a very strong support from government, especially in creating safe areas for us to test the technology very quickly in the industry,” he said, adding that Singapore is an essential part of the firm’s technology strategy to work on AI sustainability efforts.

Red Hat is currently working with other big tech companies, including IBM Research, Intel, Amazon Web Services (AWS) and more, and invites other industry players to be part of the initiative.

The partnerships approach can create “a multiplier effect to advance efforts [in green AI]”, said Macon-Cooney, highlighting different roles played by the government, research and industry.

To subscribe to the GovInsider bulletin click here.

Open-source green software initiatives

There was an emphasis on open-source software development throughout the summit, as open-source can enable the reuse of software components.

On what drives companies to adopt greener practices, IBM TJ Watson Research Centre’s IBM Fellow and Chief Scientist of Sustainable Computing, Tamar Eilam, in a panel pointed to a culture that embraces open-source and reuse.

*A culture that embraces open-source and reuse incentivises companies to adopt green software development practices, said IBM TJ Watson Research Centre's Tamar Eilam. Image: IMDA*

“When you reuse something that already exists, rather than to reinvent the wheel, obviously you save so much energy that you’d otherwise waste in [software] development…”

She explained that software development ends up becoming energy-intensive due to continuous redevelopment process. Tapping on reusable open-source components can remove the need to develop new software or AI models.

In addition, the transparent nature of open-source software means that organisations can confidently verify the AI model’s datasets and source code.

“You want a model that’s already vetted, trusted, and transparent [where] you know what data sets were used to train the model, then fine tune it for your particular needs for your enterprise data, so much energy could be saved using that,” she shared.

Aside from software development, GSF’s Arun highlighted the non-profit’s mission to develop open-source tools for measuring green software.

He shared a few initiatives that can help organisations get started with measuring software’s environmental impact, including the Impact Framework, the SCI ISO specification, and the Green Software Maturity Matrix.

These tools are available on the Github repository.

GSF’s Chairperson, Sanjay Podder, previously shared with GovInsider how GSF is collaborating with governments to drive digital sustainability efforts.