Lorem ipsum dolor sit amet, consectetur adipiscing elit. Curabitur eleifend tortor nec augue pretium
One of the most pressing, and least-understood, issues facing sustainability professionals today is the emergence of generative AI (GenAI), which will come at a huge cost to the environment and society if not developed responsibly.
GenAI differs from traditional AI in its ability to generate original content – rather than just analysing existing data – and is set to transform almost every sector, including energy, healthcare, finance, entertainment, customer service, art and product design.
“This is probably the first technology that completely changes cognitive tasks, and it’s evolving very quickly,” says Antonio Vezzani, economic affairs officer at the UN Trade and Development (UNCTAD) agency.
Indeed, we are in the midst of an ‘AI boom’ following the release of tools such as ChatGPT, with a report co-authored by Vezzani warning that 40% of jobs could be affected within the next decade. “It will replace many jobs and tasks, but also create new ones, so the problem is how we manage the transition,” he says. “There is strong transformative potential for societies and economies, but also a risk of widening inequality.”
UNCTAD’s Technology and Innovation Report 2025 forecasts AI to reach $4.8trn (£3.6trn) in market value by 2033 – which is roughly the size of Germany’s economy – but the benefits could be highly concentrated.
Just 100 firms, mainly in the US and China, account for 40% of global corporate R&D spending, while leading tech giants, such as Apple and Microsoft, are each worth more than the whole GDP of Africa.
Furthermore, 118 countries, mostly in the Global South, are currently absent from major AI governance discussions. There are already divides between developed and developing nations in relation to digital infrastructure, such as internet connectivity, and there’s a risk that AI will further amplify those divides.
“There are three key leverage points: infrastructure, data and skills, which favour developed countries with the most patents, supercomputers, and so on,” Vezzani explains. “AI could also see the competitive advantage of low-cost labour in developing countries become less relevant as a result of automation; transforming global value chains.”
Turning to environmental impacts, research by Goldman Sachs – which is rolling out a GenAI assistant for its bankers, traders and asset managers – suggests that the AI boom will see data-centre power demand grow 160% by 2030 as their carbon emissions double.
A ChatGPT query typically requires nearly 10 times as much electricity to process as a Google search, with the research forecasting data centres to consume 3-4% of the world’s power by the end of the decade, up from 1-2%.
“For years, power demand from data centres has been flat despite a near tripling in workload,” according to the researchers. However, a deceleration in power efficiencies and increased AI demand could result in “electricity growth in the US and Europe not seen in a generation”.
Meanwhile, emissions from AI chip manufacturing – much of which come from mining rare earth elements – skyrocketed more than 4.5 times last year, new research from Greenpeace East Asia indicates.
The region sits at the heart of the chip supply chain, and its heavy reliance on fossil fuels presents another challenge to the sustainability of AI. “Chipmaking is being leveraged to justify new fossil fuel capacity in Taiwan and South Korea – demand that could, and should, be met by renewable energy sources,” explains Katrin Wu, Greenpeace East Asia’s supply chain project lead.
While the increase in electricity demand for data centres is set to drive up emissions, a recent report from the International Energy Agency suggests that the rise could be offset by emission reductions if AI adoption is widespread.
This is backed up by a new white paper from Google, which claims AI could help mitigate 5-10% of global greenhouse gas emissions by 2030 – equivalent to the EU’s total annual emissions.
“That number is much bigger than emissions from data centres, so there’s a huge opportunity here,” says Adam Elman, Google’s director of sustainability for EMEA and ISEP Fellow. “Billions of people use products like Google Maps every day, and we work with partners in governments, cities, companies, startups and NGOs that can use AI for climate action.”
For example, the company’s Solar API platform uses geospatial data and AI to map 472 million buildings around the world. Providing data on roof size and radiation, this enables solar designers, installers and developers to scale up more cheaply and quickly.
AI’s predictive capabilities can also forecast wind output up to 36 hours in advance – helping to dispatch wind energy to grids promptly – and can accurately predict river flooding a week ahead of schedule.
Elman adds: “Google Maps has helped people get around more sustainably for many years, but we now use AI to very quickly calculate the most fuel-efficient driving routes, which has helped avoid 2.9 million metric tonnes worth of emissions in the past two-and-a-half years; equivalent to taking 650,000 fuel-based cars off the road annually.
“We’ve also been working with the aviation sector on contrails, which account for a third of the emissions from flying. Using geospatial data and AI, we’ve been able to cut real-world contrails by 54%.”
Despite the numerous applications for climate action, the technology must decarbonise and become more efficient if it is to have a net-positive impact.
Google owns and operates more than 29 data centres worldwide, and the explosion in AI drove its emissions up 48% between 2019 and 2023 – highlighting the challenge for the company to achieve its 2030 net-zero goal.
“It’s a hugely ambitious goal – 20 years ahead of the Paris Agreement – and it’s something we remain committed to, but access to clean energy is very different around the world, which is why we’ve been so focused working with policymakers and partners in Asia to bring clean energy to the region,” Elman explains.
“We also develop our own tensor processing units – the chips that power AI – and our latest version is twice as efficient as the previous version, which was 67% more efficient than the one before. On the AI models themselves, we have tried and tested practices that have been shown to reduce energy use by 100 times and emissions by 1,000 times.”
However, concerns around the vast quantities of water needed to cool data centres, and their impacts on biodiversity, remain. A study of 8,000 centres by NatureFinance found that 45% are in areas where there is a high risk to the availability of water, while more than 50% are in areas where there is a risk of drought.
“We’re very thoughtful about where we locate our data centres, working with local communities and experts,” Elman says. “We have different solutions we can deploy in terms of whether or not we use water to cool data centres, and we have an overarching goal to replenish 120% of the fresh water we do use.”
Returning to the societal impacts of AI, considerable uncertainty remains around how it will transform lives, and how we can narrow the gap between the inevitable winners and losers that arise from its development under the current technological monopoly.
As regulations and ethical frameworks take shape, developing nations must have a seat at the table to ensure AI serves global progress and to tackle existing biases in models that are trained in the West.
UNCTAD is also calling for an AI equivalent to the global ESG framework, which would act as a public disclosure mechanism to improve accountability.
“We need to find a way to involve big tech companies at the UN or international level, because the rate of technological progress is too fast for governments and policymakers to understand what’s going on,” says Vezzani. “We use the example of ESG because it works and has been taken up by stock markets, and will bring some transparency and accountability to the industry.”
On the one hand, GenAI could be transformative for sustainable development, preventing climate disasters, helping develop new drugs such as malaria vaccines, and vastly improving quality of life and job satisfaction, while adding trillions to global GDP.
On the other, it could further widen inequality where millions, if not billions, of people are left jobless and struggling to find meaning while climate and nature targets slip out of reach.
Vezzani says: “Humans are behind the technology, so it’s up to us to decide where this goes.” As for a pause or slowdown in GenAI development, he simply adds: “Impossible.”
Read UNCTAD's Technology and Innovation Report 2025 here: Technology and Innovation Report 2025: Inclusive artificial intelligence for development | UN Trade and Development (UNCTAD)
Read Google's whitepaper here: The AI opportunity for Europe’s climate goals
Chris Seekings is the Deputy Editor of ISEP’s Transform magazine, which is published biomonthly for ISEP members. Chris’s role involves writing sustainability-related news, features and interviews, as well as helping to plan and manage the magazine’s other day-to-day activities.
