Beyond the buzzword: leveraging AI to decarbonise our buildings
Since the November 2022 launch of ChatGPT, AI has recaptured the world’s imagination. I say recaptured, because we have seen this several times before, dating back to the 1950s when the term “artificial intelligence” was coined. Since the arrival of ChatGPT, other large language models (or LLMs) soon followed, sparking a proverbial LLM arms race between the likes of OpenAI, Google, Anthropic, Meta, DeepSeek and others. By 2025, ChatGPT alone has grown in significance to now boast more than five billion site visits each month.
While there’s a whole lot more to AI than these tools, LLMs have nevertheless dominated technology discourse over recent years. Concurrently, a swelling discussion centering on job displacement and the environmental footprint of energy-intensive LLMs has unfolded. Dario Amodei of Anthropic, for example, has heralded a near-term “white collar bloodbath” that risks spiking unemployment to 10-20 percent.
While the future of AI remains uncertain, I have an aspirational view of how AI can, and should, be leveraged for people, planet and profit.
A big part of our work at Building Atlas involves interfacing with building consultants, and they serve as a vital piece of the decarbonisation retrofit puzzle. What have we learned? The legacy process of assessing buildings for possible retrofits is broken. Buildings will not decarbonise on time. One recent example stands out: a consultant lamented to us that after onboarding a client, undertaking preparation work, driving 3.5 hours to the building, inspecting it, driving 3.5 hours home, and completing a report, the outcome was a polite and generic “sorry, not much that can be done with this one–it’s actually just fine”.
What a waste of the consultant’s time and the asset owner’s money!
When contemplating what role AI might play in improving outcomes for both the consultant and the asset owner in such a situation, a 19th century phenomenon offers a hint.
Despite coal-fired steam engines markedly improving in efficiency at the dawn of the Second Industrial Revolution, English economist William Stanley Jevons observed demand for coal continuing to climb.
How did this happen? Since the relative cost of each lump of coal had effectively decreased, previously uneconomical applications of steam power suddenly became economical.
In the case of our friendly building consultant, decarbonisation will not be achieved with asset-by-asset 7-hour round-trip inspections and reports that take weeks or months to complete. In fact, according to our own estimations, it would take more than a century for the existing 950 UK building consultants to complete reports on the 1.7 million buildings comprising the UK’s non-domestic (i.e. commercial) stock. Given the usual cost of at least £5,000 for a report, that’s an £8.5 billion expense across the building stock—translating to a whole lotta heat pumps and solar panels that could’ve been installed instead!
At Building Atlas, we view this as a key factor affecting the learned helplessness of the commercial real estate industry in this area. Therefore, our job is to make this uneconomical task economical with the steam power of the modern age: artificial intelligence.
Starting at a bird’s eye (portfolio or district) view, we developed Building Atlas to make it possible to shortlist the most promising assets–and retrofit interventions within them–for building consultants and asset owners to focus their time and energy. The fruitless 7-hour round trips can be avoided, redirecting building consultants to the retrofits demanding the most creativity and ingenuity, relegating the rest to AI. This way, building decarbonisation starts with the lowest-hanging fruit and humans are empowered to focus on what they do best.
Instead of building a platform aimed at fueling unemployment, our job is to make human work more impactful–benefiting the bottom line of asset owners, the world’s climate, and the livelihoods of skilled workers.
Olga Khroustaleva is co-founder at Building Atlas.
