In his presentation at The AI Summit at the Singapore Expo on May 30, Greenie Web CEO Ian Chew detailed the environmental impact of digital technology, including the use of Generative AI. Digital carbon emissions from online activities—from sending a WhatsApp message to generating a prompt using ChatGPT—are becoming a “massive present-day issue” as they contribute 3.7 per cent of Global Greenhouse Gas (GHG) emissions, almost as large as the aviation industry.
He also highlighted that this is a “rapidly scaling problem” as the number is expected to grow to 14 per cent by 2040.
“[During the COVID-19 pandemic] what happened was that every single country in the world digitised exponentially: The elderly, the young people who had not used digital devices from the beginning, did it for the first time. This will be a problem with a 300 per cent increase by 2040,” Chew explained.
To tackle them, there are two solutions that the CEO proposed: Digital decarbonisation and sustainable digitalisation.
“These are actually two sides of the same coin. When we think about digital decarbonisation, we are thinking about how we can write the wrongs of the past. When we look at legacy systems and AI models, how can we make that more efficient and continue using it?” Chew said. “When we think of sustainable digitalisation, we think of all the new possibilities. For example, how can we create a low-carbon mobile application?”
He proposed the SWUP framework for developers to help achieve these two goals. This includes monitoring the animation speed of one’s site or app, page weight budget, user optimisation, and eco-personal digital habits.
An example of such work that Greenie Web had done included a low-carbon site for the elderly, which showcased how the company fine-tuned a platform to the users’ needs to ensure an operation with lower carbon emissions.
On the sustainability of AI
On the same day, a panel discussion featuring speakers from the tech industry and social policy institutions discussed the sustainability aspects of AI technology.
According to Kenddrick Chan, Senior Policy Analyst, Policy and Politics at Tony Blair Institute for Global Change, pointed out, “When people talk about AI … the natural impulse is to think about AI for sustainability. Can AI do climate modelling for me? Can they do calculations of parametric risk insurance for climate disasters?”
“What we hear less is the other side of the coin which is not so much about ‘AI for sustainability’, but the ‘sustainability of AI’ itself.”
Chew, who also spoke at the panel, stressed the need for a fundamental change in the business model of data consumption. “If you look at the cloud service providers nowadays, one of the biggest issues is that they are incentivised to increase data consumption. The more services you rent, the more data you use, the more you will pay.”
The panel’s one key issue was that “we cannot manage what we do not measure.” But Chan pointed out that at the moment, there was no standardised way to measure. This was heightened by the complexity of analysing a tech platform’s carbon emissions: For example, a data centre might use electricity from various sources, including an eco-friendly one such as a solar panel.
“But one thing that becomes clear with this measurement problem is that it is coming to the forefront of policymakers’ attention, which is rightly so.”
The AI Summit was held in conjunction with the fourth edition of ATxEnterprise, which concluded on May 31. According to an official statement, the event welcomed over 22,000 attendees from 110 countries and regions.
One of the event’s key highlights was the announcement of the Singapore government’s Digital Enterprise Blueprint (DEB). Aiming to accelerate digital transformation in the country, DEB includes empowering small- and medium-sized enterprises (SMEs) to adopt AI innovation.
The post On the sustainability of AI: Why measuring digital carbon emissions is key to a greener future appeared first on e27.