ANTALYA, Turkey – The sustainability community at the ongoing UN climate conference COP31 in Turkey is echoing the calls of some tech leaders to slow down the development of artificial intelligence (AI), as global superpowers remain locked in a race to dominate the emerging field.Tech moguls such as Anthropic’s chief executive Dario Amodei have called for an AI slowdown, saying more oversight is needed to prevent loss of control.But observers at the climate summit have a different concern, saying that the expansion of data centres to power the AI boom could increase fossil fuel use, strain power grids, and raise electricity prices for people.At COP31, a pledge on AI was on Nov 16 launched by host country Turkey for countries and companies to sign. The pledge urges signatories to seek the technology’s potential in finding climate solutions, as well as assess its growing environmental risks.Turkey said that the pledge had received 80 endorsements, though it did not specify those who had. The Straits Times has asked the Ministry of Sustainability and the Environment if Singapore had signed the pledge.UN climate chief Simon Stiell said at the launch of the pledge that AI offers both “great potential” and “great peril” for the battle against global warming.“We cannot ask people to accept higher bills, more planet heating pollution and deeper inequality today in exchange for promises of a better tomorrow. AI benefits must outweigh its growing costs,” Stiell said.He called on companies to step up and align their business plans with climate goals and urged them to power their date centres with renewable energy.Competing for electricityDiscussions on energy are a key focus at this year’s climate talks.The Turkish presidency of COP31 has set an electrification target for countries to agree on at the conference. The aim is to increase the share of electricity in global energy demand from 20% currently to 35% by 2035 through electrification.Electrification can help to cut carbon emissions, when processes that rely on fossil fuels are tweaked to electricity generated from low-carbon sources instead. This could be in the form of switching from internal combustion engine vehicles to electric cars to reduce consumption of petroleum.Alex de Vries-Gao, a researcher at Vrije Universiteit Amsterdam told The Straits Times that the development of AI should be slowed down if its growth cannot be accommodated without increasing the use of fossil fuels or putting further pressure on electricity grids.This is especially the case if data centres are built before renewable energy resources come online, said Vries-Gao, who studies the environmental impacts of digital infrastructure.“Building data centres first and promising to address the consequences later is not a sustainable approach,” he said.As more data centres are built, there will be a higher demand for electricity. Grid capacity, which refers to how much electricity a network can reliably deliver to users, will also need to be improved.Vries-Gao said: “Using more electricity is not necessarily a bad thing if it replaces fossil fuels in transport, heating or industry. The same cannot automatically be said for AI.“Data centres also need electricity and grid capacity, meaning their expansion can compete with these other uses. We therefore need to ask whether this growth helps the energy transition or gets in its way.”Unchecked AI data centre expansion threatens to derail the global energy transition, said Xueying Wu, Co-Lead of Global Big Tech campaign at Greenpeace.She noted that a report by International Energy Agency (IEA) showed that global data centre electricity consumption will more than double by 2030.Philip Newell, Climate Action Against Disinformation coalition’s COP31 Actual Intelligence Unit lead, said that the growth of data centres would not need to slow down if tech companies committed to tap more wind and solar power, backed by batteries, rather than relying on fossil fuels to power them. The coalition is a global one involving over 50 climate and anti-disinformation organisations.The increase in electricity demand could also hurt consumers’ wallets by driving energy prices up, experts at COP31 told ST.Consumers are already seeing the strain, said Nidhi Upadhyaya, the deputy director of global policy and finance at the Atlantic Council’s Climate Resilience Center. The Council is a US-based think-tank.The AI infrastructure construction boom, which led to rising electricity bills, was among the main concerns for voters during the Nov 3 US midterm elections that saw a Democratic victory in Congress.With the growing resistance to data centres in the US, South-east Asia will need to assess the implications of investments in these areas, said non-profit research organisation Oil Change International’s senior partnership and outreach officer for its Asia programme Hikmat Soeriatanuwijaya.In South-east Asia, the data centre growth is emerging as a new source of electricity demand, according to IEA. The data centres in the region currently consume more than 13 terawatt-hour of electricity, with consumption expected to nearly triple by 2035, according to IEA’s Nov 9 report. This 13TWh is equivalent to about a quarter of Singapore’s total electricity consumption in 2024.Vries-Gao noted that even before accounting for the AI push, the region already needs substantial investment in its energy infrastructure to deliver electricity.While data centres could bring in investments for clean energy to run them, it risks increasing fossil fuel dependence if the demand for clean energy grows faster than supply, he said.“Governments need to consider who benefits, who pays for the additional infrastructure, and what happens to local electricity and water supplies. More investment is not automatically a better outcome,” he said.More guardrails neededIndependent climate innovation agency Climate-KIC’s chief executive Kirsten Dunlop said that the expansion of data centres should slow down to allow developers to build infrastructure that is more sustainable.Upadhyaya from Atlantic Council’s Climate Resilience Center, added that it will be critical to ensure that the cost of building data centres is borne by the companies setting up the data centres, not by consumers.Grid capacity also needs to be re-evaluated and ensure that the infrastructure is being retrofitted, upgraded or rebuilt to be more resilient, she said.She also called for guardrails governing where the data centres are built, the impacts they will have on the neighbourhood and governance on how AI is going to be used.
AI boom could increase reliance on fossil fuels, strain grids: Observers at COP31
Full Article
Original Source
Read the full article at Straitstimes →KhanList aggregates and links to publicly available news content. We do not host full articles from third-party sources. Always verify important information with original sources.