It’s no secret that the building and construction sector account for a huge portion of global carbon emissions, as much as 37% according to a UN report. Finding ways to reduce those emissions is crucial to meeting climate change targets and now an academic consortium is aiming to do that using artificial intelligence (AI).
Led by the Universidad Politécnica de Madrid, the project known as ZEBAI brings together 18 institutes from seven European countries and is backed by €3.8M in funding from the European Union’s Horizon Europe program.
The departments of Engineering Science and Physics at Oxford will focus on characterizing and modelling how materials respond to temperature and mechanical forces, using AI to select locally-sourced building materials suited to specific environmental conditions. This analysis will be used to create a simulation with AI methodologies that can be incorporated into construction design platforms.
“This is a very timely and exciting project where the work that Professor Contera and I are doing with engineering materials at multiple scales will feed into direct practice to maximise impact in the real world,” said professor Antoine Jérusalem in a release from Oxford. “ZEBAI is posed to push forward the frontier of energy-efficiency in future constructions with the state-of-the-art of mechanics and AI.”
Along with Oxford, the other representative demonstrators which will serve as test cases are located in the Netherlands, Spain and the Ukraine. These sites were chosen to evaluate the new methodology’s performance across different climates, use cases and building patterns.
Fighting climate change with AI?
According to the project’s mission statement, “ZEBAI aims to revolutionize zero-emission buildings design through a comprehensive methodology which will incorporate analyses, decision-making processes and holistic evaluations of energy performance, environmental impact, indoor environmental quality, and cost-effectiveness.”
It’s a laudable goal, but the inclusion of artificial intelligence among the tools used to improve sustainability in construction seems somewhat counter to it. While its own carbon footprint is nowhere near that of construction, AI is still a far cry from carbon-neutral. One study projects that NVIDIA’s new AI servers could be consuming as much as 85.4 terawatt-hours annually by 2027, outpacing entire countries, such as Argentina and Sweden. Another study suggests that training a single AI model can emit more than 284 tonnes of CO2. Meanwhile, data centers already account for over 1% of global electricity consumption, much of which still comes from fossil fuels.
This is not to suggest that AI is incompatible with sustainability, nor that it shouldn’t be used to reduce construction’s much larger carbon footprint. It’s not a question of whether AI can solve the problem, but rather whether it’s the best tool to do so. Given the known environmental costs of using AI, when it comes to using it to address climate change in particular, we ought to ensure that we’re factoring those costs in as well.