AI and climate change: Can the technology help the planet without harming it?
AI is being used to spread, or support the dissemination of, disinformation and misinformation.
AI is a double-edged environmental sword — it has the opportunity to streamline and accelerate climate action across the world but, if not done sustainably, could see more harmful emissions pumped out to fuel AI chatbot requests.
While data centres can vary in size and are used for everything from cloud computing to apps, it is AI-focused or ‘hyperscale’ centres that are the most energy-intensive, and many still burn fossil fuels to generate the energy supply they require.
But it is also seen as a defensive weapon for communities facing the mammoth challenge of adapting to, and surviving, hotter heatwaves, more destructive storms, and eroding coastlines.
Panama’s environment minister said “sustainable” AI could help his country improve forest monitoring, protect water resources, strengthen early warning systems, and better understand the risks facing coastal communities.
Juan Carlos Navarro, who this week claimed the world was being “held hostage by the petroleum mafia”, said technology and AI can “certainly help us to deliver results faster and more effectively”.
“At the same time, we must ensure that the technology itself is sustainable, especially renewable energy and water use,” he told his fellow ministers at the Cop31 conference in Turkiye.
“This is crucial. We cannot address one environmental challenge by creating another.” So which one should win out?
One of the ways AI could help with the environmental challenges facing the world is by speeding up and streamlining the transition to climate neutrality and adaptation to fiercer weather events.
Where there is a gap in data or a lack of an efficient system for aggregating key data, it could help provide solutions.
This could include using an AI mechanism to automate when is the cheapest time to put on a clothes wash, instead of checking the price every hour; to create the optimal public transport route and schedule; and figure out where to prioritise resources for updating the grid ahead of evolving solar, EV, and heat-pump technology.
An Oireachtas committee was told by the director of UCD’s Aimsir research centre, Andrew Parnell, that weather forecasting had become 1,000 times more efficient because of new AI systems, with predictions calculated in “a fraction of a second”.
An AI-powered early warning programme is hoping to give between 13 and 21 hours’ notice of major floods in Dublin City, giving council staff, emergency services, and the public more time to prepare.
While a single computer could be defined as a data centre, it is the AI-focused and ‘hyperscale’ centres that prompt concerns about emissions and energy price hikes.
One study concluded that 403 hyperscale data centres in the US emit as much greenhouse gas a year as Denmark.
In Ireland, most of the electricity used by data centres comes from hyperscalers, according to a study co-authored by Maynooth University researchers.
That study also said Ireland hosts around 5% of the world’s hyperscale data centre capacity, and the concentration of data centres in Ireland is the highest in Europe.
While the average data centre uses between 10MW and 25MW of power a year, AI-focused facilities use up to 100MW, which could power around 100,000 households.
On electricity, it is almost inevitable that the demand from the concentration of data centres in Ireland has led to an increase in energy price hikes, and there have been calls from Irish academics for greater transparency on the energy used by data centres and how that ties to AI models.
Permission for data centres in constrained areas of supply were paused in 2021 but Eirgrid has said this autumn that proposals for data centres would be considered on a “case-by-case basis”.
There is also a requirement included that in five years’ time, 80% of their electricity should come from renewables.
“That means, in the short-term, we will be seeing an increase in electricity use from fossil fuels associated with the growth of data centres,” said Brian Ó Gallachóir, professor of sustainability at University College Cork.
“Last year was the first year that renewable-generated electricity passed out the amount of electricity coming from gas, which is a significant milestone, but the risk is that, with new data centres, it will prompt an increase in gas-fired electricity generation.” Ireland’s more temperate climate means that data centres here require less resources to cool down, and figures from Uisce Éireann suggest that data centres use less than 0.3% of Ireland’s total water usage at its peak, with it falling to 0.2% on average.
The EU is looking to triple its data centre capacity over the next five to seven years, so a common rating scheme for data centres has been proposed in an attempt to increase transparency on their energy use.
Data centres with a capacity above 500kW will fall under this rating scheme, with the first sustainability labels expected to be issued in 2027.
At Cop31, the Antalya AI Pledge — which calls for a reduction in the environmental impact of AI, more efficient infrastructure, the tracking of emissions, and improved access to the technology in developing countries — has been supported by just a handful of countries and one tech giant, Chinese e-commerce company Alibaba.
The Department of Climate and the Environment has been asked if Ireland will sign up to the pledge.
AI has also fed into climate-related misinformation and disinformation.
A report co-authored by the UN found that while AI can answer specific factual questions on climate correctly, it tends “to prioritise agreement over scientific accuracy”.
In tests on 15 models, where the models were challenged on the correct answers they had presented, and then presented with false credentials and fabricated evidence, “every model yielded, driving accuracy down by 31 to 35 percentage points”.
In eight out of 10 cases, models gave up the correct answer at the very first objection, it said.
“A system that can be talked out of basic scientific facts cannot serve as a dependable guide on climate change,” the report concluded.
Mr Ó Gallachóir said that one of the ways AI is being used is to spread, or support the dissemination of, disinformation and misinformation.
“What’s happening in the last couple of years, as well as the attacks on democracy, the attacks on feminism, the manosphere, and those dimensions, algorithms are pushing certain messages and, essentially, AI is helping with all that. If it is not managed in an ethical and fair and just manner, it will increase the divides that we are seeing in the world.”




