The AI alternatives to air conditioning

The UK logged 40 days above 30°C this summer, beating the previous record of 34, according to the Met Office. Across Europe, the EuropMOMO mortality tracking network estimates more than 32,000 excess deaths between mid-July and mid-August alone. AC sales rose in Europe nearly 60% year on year in the first half of this year, with portable unit sales up more than 200% . 

It’s an understandable response – but it isn’t sustainable. 

The Air Conditioning Paradox 

Air conditioning cools a room by pumping heat outside. Do that at scale and you heat the whole city: night-time temperatures in cities with lots of AC units rise by more than 1°C, which drives even more demand for cooling. Researchers have measured this loop directly in Phoenix and Barcelona. 

Cooling buildings already produces about a gigatonne of CO2 a year and uses more electricity worldwide than Africa consumes for everything else combined. A 2026 study in Nature Communications predicted AC-related emissions could nearly triple by 2050. Grids strain hardest exactly when heatwaves hit. In cities with high numbers of AC units like New York, evening electricity demand spikes 90% above normal in summer, compared with 25% in France where fewer households have AC. 

Cooling access is also unequal. The countries that need it most often have the least of it, and closing that gap with more conventional AC could add up to 146 billion tonnes of extra emissions by 2050. 

None of this means air conditioning is bad. It saved an estimated 58,000 lives in the US between 2010 and 2020, and better-managed cooling paired with cleaner power could save many more, worldwide. For many people, it’s currently the only thing standing between them and heatstroke. 

The problem is treating AC as the entire strategy when there are other tools and approaches for cooling. Advances in AI tools  are already lowering temperatures in different ways: some without any mechanical cooling at all, some by reaching people AC never reaches, and some by making the cooling we already use less wasteful. 

Cooling Streets and Buildings, Not Just Rooms 

The simplest way to reduce air conditioning usage is to stop the city getting so hot in the first place. 

Seoul, Singapore and Tampere use AI trained on satellite images and infrastructure data to build detailed heat maps showing exactly where a city is hottest. That means reflective roof coatings and new trees go where they’ll reliably cool things down, instead of being spread evenly across a city. In Boston, a tool called TreeTect uses similar imagery to track the health of the tree canopy and flag where planting is most needed. 

This is proving to be an effective heat reducing measure. Since 2016 Medellín has planted more than 8,000 trees along AI-selected routes and recorded a 2°C drop in temperature across the whole city. Studies across Europe show urban greenery cools cities by just over 1°C on average, and up to 2.9°C in the best cases. 

Reaching the People Air Conditioning Leaves Behind 

Outdoor workers, pregnant women without reliable power, and elderly people living alone are among the most at-risk groups in the face of extreme heat. A wave of interventions has been developed directly to protect them. 

In South Korea, Samsung and the government link workers’ smartwatches to site temperature and humidity data. If a worker’s heart rate and the conditions cross a danger threshold, the watch triggers a rest alert or notifies their manager.  

Trials of similar wearables among farmworkers, who are 35 times more likely to die of heat-related illness, found they reliably track the hydration and heart-rate signs of heat stress, replacing generic advice with something closer to a personal warning system. At Emory University and Georgia Tech, researchers are building an AI-powered patch specifically for farmworkers, designed to catch deterioration before someone collapses. 

In a similar vein, an EU-funded project, sends personalised heat alerts to vulnerable groups across the Mediterranean, including outdoor workers in Greece and Spain and pregnant women in Cyprus and Italy. Separately, AI voice agents from Hippocratic AI are already calling vulnerable patients during heatwaves on behalf of health providers like EmblemHealth. 

Buying Time Before the Heat Arrives 

Most funding goes to technology that manages a crisis once it’s already happening rather than to technology that stops it building up in the first place. Forecasting is the exception, thanks to advancements in machine learning. 

Researchers at the Euro-Mediterranean Centre on Climate Change have built a model that predicts severe heat events up to seven weeks ahead. That’s enough time for governments, farmers, grid operators and hospitals to prepare, rather than scramble once a heatwave has already hit. 

Smarter Cooling
Some AI tools aren’t an alternative to cooling, but rather are making it more efficient. 

UK Power Networks tested more than 5,000 small AI-controlled devices across east London that automatically reroute power and manage voltage during heatwaves. During a July heatwave the devices cut minor outages by 40%. In Barcelona, an AI system called RESPIRA adjusts the metro’s ventilation fans in real time based on temperature and humidity, cutting energy use by a quarter while keeping trains comfortable. It’s now expanding to cover 94% of all metro passengers.

Portland runs a similar system above ground, using sensors to automatically increase watering in parks and reroute traffic to cut pollution during extreme heat. 

These aren’t fixes – but help mitigate the knock-on effects of air conditioning while cleaner measures gather pace.  

What’s Coming Next 

Additional ideas still in the early stages of development are a further source of optimism. Digital twins, essentially virtual models of a city, let planners test whether a new park or cool roof will actually lower the temperature before  a full-on investment to build it out.  

At the University of Texas at Austin, researchers have built a paint that reflects heat so effectively it cut surface temperatures by up to 20°C in tests, using no energy at all. And in early academic trials, sensors are being tested that can spot frailty or heat-related decline in elderly people living alone, without cameras or wearables, alerting carers before a heat emergency develops. 

While these solutions aren’t quite ready to scale yet – it is a helpful indication of where cooling is heading next. 

The Gap 

The technologies mentioned above are already deployed and already making a difference. Seoul’s heat maps send real budgets to real hotspots. Medellín’s trees delivered a measured 2°C drop. Barcelona’s metro cools itself without running its fans flat out all day. A construction worker in South Korea has a watch on his wrist connected to a government platform. None of these innovations are radically new, they just needed someone to decide they are worth building and investing in. 

What’s still missing is what was missing after this summer’s wildfires: getting a solution that works in one city, or one pilot, into the hands of those who need it most before the next heatwave hits.  Air conditioning will stay part of the answer, and for many  people it’s the only solution that exists today. But installing more of it, without building alternatives that don’t strain the grid or price out the people who need cooling most, just makes the same problem bigger. 

The Conduit’s work scouting solutions in health, nature and urban resilience is asking the same question in each area: not whether the technology exists, but what is needed to build the coalitions and unlock funding to get it where it needs to be – fast.   

Katie McPhee is The Conduit’s AI for Good Expert in Residence

—