AI futures

The facts about global warming should alarm every business leader. Before the Industrial Revolution, emissions of manmade greenhouse gasses (GHGs) were almost negligible, but began growing steadily as industrialization spread.

Since the 1950s, however, emissions have soared, and now stand at 58 gigatons (58 billion tons): that's the equivalent of 580,000 fully laden aircraft carriers of planet-heating gas. Carbon dioxide (CO2) emissions alone have grown from roughly five gigatons in 1950 to 36 gigatons today, with that process accelerated by the digital and personal computing revolutions in the 1980s and 90s.

GHGs trap heat in the atmosphere, which is why the planet has warmed by about one degree Celsius since the Industrial Revolution. That might seem like a minor change, but most scientists agree that it is essential to keep the increase below 1.5 degrees to avoid a catastrophic deterioration of climate and weather systems, early signs of which we can already see.

CO2 emissions are forecast to remain stable but high for the foreseeable future, but emissions of other GHGs, such as methane and nitrous oxide, are predicted to rise by 30 percent by the middle of this century. This is one reason why overall GHG emissions are forecast to hit 62 gigatons by 2030 alone (the equivalent of adding 40,000 more aircraft carriers of greenhouse gasses this decade).

So, what can we do about it? The problem is that the focus is often on the obvious polluters and carbon emitters – industries such as international travel, food production, or fossil fuels, which are slow to change. But this leaves most organizations unaware of how much energy they are wasting themselves: a problem that can be fixed more quickly.

Tackling energy wastage is critical to hitting Net Zero, because 72% of all manmade GHGs are by-products of energy production. Thirty-one percent of carbon emissions come from generating electricity and heat alone: that's more than twice the total for transportation, including air travel and petrol cars.

All of which leads us to the key fact: 20% of all GHG emissions come from buildings, with the heating, ventilation, and air conditioning (HVAC) systems of office blocks, warehouses, and large retail environments being among the worst offenders.

Malls, out-of-town shopping centres and mega-warehouses might seem like a good idea, but they waste vast amounts of electricity: about 35% of the energy they consume, according to Canadian autonomous artificial intelligence provider Brainbox AI.

It might not be the sexiest application of the technology, but the Montreal-based company is focused on making existing HVAC systems smarter through autonomous AI and machine learning. Why? Because this is where the maximum impact can be achieved in the shortest time – and for the least money.

CEO Sam Ramadori explains:

These are massive global problems, but the traditional way [of solving them] requires a lot of capital. And, worse, you need highly specialized teams to execute a deep retrofit of a building. So, today, perhaps only one percent of buildings get retrofitted a year.

And that's your entire pool of talent: if you think you’re going to be able to do 20% of buildings a year, or even 10%, that's a virtual impossibility because those engineers and specialists don't exist in big enough numbers.

So, our thinking was we can use our capability in autonomous AI, machine learning, and data analysis to really make a difference. Put that self-learning and self-deciding capability in a building and you can change things materially.

The company's AI can be applied to, and control, existing HVAC systems, with the intention of turning them from dumb consumers of energy into adaptive, responsive installations.

Inefficient, wasteful buildings are an urgent problem to tackle because – unlike most cars, for example – they might be around for decades or even centuries, he says:

The vast majority of buildings are still going to be here in 50, or 80, years’ time. You can't just swap them all out, so you need a system that can learn about each building in a cost-effective way. And that's AI.

A 30-storey tower might have 200 to 400 individual components of its HVAC system, from the little pump or fan to the big boiler in the basement. And 40 to 50% of a building's energy consumption goes into heating and cooling it.

So, if you have something that can tweak how those components work every minute of every day to make them more efficient, well… autonomous decision-making with AI can provide that. And we believe this technology can be deployed at scale to make a real impact.

He adds:

Some energy usage patterns in a building are more predictable – lighting, elevators, and so on. But the usage of an HVAC system can swing dramatically: from a day when it's 20 degrees and not sunny, to another when it's 32 degrees and your HVAC is operating at a dramatically different level.

So, not only are we tackling the biggest piece of energy consumption in a building, but we are also tackling by far the most variable, which would otherwise create inefficiencies in how it uses electricity. And importantly, that inefficiency then puts pressure on the energy grid.

This is a critical point: wasting energy in one building rapidly scales up to wasting it in many, and then to inefficient companies, towns, cities, authorities, regions, and nations. In turn, that means the electricity grid – some of which still comprises fossil-fuel sources, of course – is put under tremendous strain, while often heating or cooling empty rooms.

At a time of soaring energy bills, stamping out waste on that scale becomes an economic imperative, as well as an environmental one.

So, what is Brainbox AI's strategy for the future? Ramadori explains:

The next big evolution for us is the concept of grid-interactive buildings. As our energy systems themselves become more complex and we bring more renewables onstream, we face the new problem of intermittency of production. So, we have to make our users of energy much more flexible too, relative to how the grid is at any time with its available resources.

This is why our goal is bringing AI to this capability: imagine having 100 or 500 buildings that are behaving together in a city, to provide the energy grid with the flexibility it needs so it can take on more and more renewables. That way it doesn't matter to the grid so much if it's sunny or it's cloudy.

We're excited about bringing the same technology that we're using to optimize a single building to, say, 500, so they can cooperate in terms of how they're consuming the grid's energy. That's grid-interactive buildings.

So, the major challenge facing innovators like Brainbox AI would seem to be procurement at scale: persuading companies, then towns, cities, authorities, and governments to get behind the broad concept. But this – as any socially conscious entrepreneur knows – is the spanner in the works of global ambition. And, arguably, of global change. He says:

The public procurement system is horrendous for any game-changing innovation. Of course, it is designed to avoid fraud and ensure fairness and fiduciary duty over taxpayers’ money. But it is not friendly to innovation at all, especially if you are doing ‘first in the world’ stuff in a competitive market. So, it really doesn't fit into the public procurement model.

I speak as CEO of a climate technology company, but it's the same no matter what your technology is. Public procurement is hard and, as start-ups, we have our own pressures too. We have to move fast and show progress just as quickly. If I can sell to a private client in two months and the public procurement process takes 12, then we would die on the vine trying to go the public procurement route. That's just the reality.

Even if the leader of a city, or the president or prime minister of a country, had genuine ambitions in this field, the moment they start trying to influence public procurement, it's a dangerous precedent. It just doesn't work. Even the most ambitious cities still have rules that ensure an entire ecosystem of public procurement.

Within broader public procurement, there may be space for pilot programmes. But to have real scale, to take the 5,000 buildings or whatever number an authority manages, owns, or occupies, we're not there yet as a society. So, we have to focus on areas that will move fastest, which then become proof-points of success.

One route for doing this may be the retail sector, he explains:

When we think about buildings and total square footage, our minds naturally go to pictures of London, New York, and Tokyo: all the tall, shiny buildings. But the reality is that the massive amount of square footage comes from all the medium-sized and small buildings, many of which are retailers.

So, when we think about size, one far outstrips the other in terms of square footage that we need to optimize. And many retailers have made public commitments about sustainability, and so are under huge pressure to show progress.

There are other advantages of working with that sector, he continues:

Fortunately, many retail stores are broadly similar. So, when we're deploying, we could do 500 stores of the same chain far faster than we could do 500 other units. That combination makes it a very interesting sector in which to make an impact fast – literally pulling tons of carbon out of the air or stopping it from being emitted.

Great news. But how concerned is Brainbox AI about the carbon footprint of the artificial intelligence sector itself? Picture millions or billions of people using processor-intensive AI and machine-learning models in the cloud, via huge, energy-guzzling data centres – which also need to be cooled.

He responds:

It's one thing when customers are using AI to decide what movies to watch. But in our case, where we're using it to reduce the energy consumed by a building, the energy saved versus the energy used is multiple times different. We're not using one kilowatt to save two kilowatts; we're probably using one to save 1,000.

We live in a world where we need to decarbonize buildings that already exist, and there are millions upon millions of them on the planet."

Not the most glamorous or eye-catching use of AI, but one that is many times more beneficial to the planet than using AIs like ChatGPT or Stable Diffusion to xput creative professionals out of work. This is the kind of deployment that AI ought to be used for, but the challenge of getting everyone to understand what companies like Brainbox AI are doing is massive, let alone persuading them to take positive action.

Perhaps double-digit inflation and ruinously expensive energy in some countries may, unintentionally, be an immediate spur for change.