Artificial intelligence (AI) is the latest trend in the smart building world, with promises to make it faster and easier than ever to optimize building system performance, especially for indoor air quality (IAQ) management. AI has the power to transform how we analyze, apply, and track air quality data to ensure our spaces are doing what they’re supposed to do: promote the health and well-being of our occupants.
The industry is actively trying to find the “holy grail” of AI-driven automation that will simultaneously analyze IAQ data, assess system performance, diagnose problems, and prescribe automated solutions to fix those issues.
This vision is incredibly exciting, and while I do think we’ll get there one day, the reality is that certain caveats exist that can limit what we are capable of achieving today.
THREE CAVEATS TO USING AI IN TODAY’S IAQ STRATEGIES
There’s no question that AI will become a core part of how we manage and optimize IAQ in our spaces. However, there are three factors to keep in mind when we think about realizing this vision:
1. The Lack of Reliable Data — Or Any Data at All
Before the holy grail of AI can work its magic, it has to have perfectly reliable, consistent, and accurate data from every part of our buildings, from smarter components like air quality sensors to more legacy components like HVAC dampers. Many buildings today lack this type of data or any data at all. Most will collect basic energy usage and electrical data and call it a day. For those that do have air quality sensors, many of those sensors aren’t being used properly and, as a result, are unable to collect the quality and amount of data that the AI needs.
Without data, there is no input for the AI to work with and no insights to be made. If the data exists but is low-quality (like in many buildings today), the AI still won’t be able to deliver the right insights because it doesn’t have the full picture. To put it simply, bad input equals bad output. There’s no point in investing in an AI-driven solution for IAQ until every part of your building is capable of providing this perfect stream of data.
But, even if you have all of these components sending perfect data to the cloud, the AI still won’t be able to “see” everything going on because building operations are not 100% automated. For example, HVAC could be performing sub-optimally because someone screwed a damper shut (this is something I’ve seen in countless buildings), but the AI wouldn’t be able to see that and would likely point you in the wrong direction to diagnose the problem.
2. Automation Isn’t a Silver Bullet
AI-driven automation is an incredible tool for IAQ optimization, but automation alone cannot fix everything. Many aspects of IAQ still need human intervention, such as changing air filters, removing a chemical product that’s off-gassing volatile organic compounds (VOCs), or removing an object from the air duct, like a shoe — this is a true story. Not only is the AI incapable of removing the shoe, but it also can’t recognize that a shoe is in there in the first place because it’s not part of the data.
So, even if you’re feeding the AI-perfect data on every part of your building and getting the highest quality insights (problem #1 solved), the AI is still incapable of going out to change your filters or clean out your air ducts. The key is to automate everything that can be automated, and then provide clear instructions for humans to manage the rest.
3. Not All Solutions Are Designed to Deliver Value
Because AI is so popular in the smart building world, many tech companies are simply trying to ride the wave and rack up sales instead of creating a product that delivers real value to the customer. A perfect example of this is a building software I saw recently with a chatbot that uses emojis to describe how the building is performing. Flashy features like these may seem exciting at first, but the novelty quickly wears off, and what you’re left with is a solution that doesn’t make it any easier to understand or optimize building performance.
What these solutions should be focusing on is AI’s ability to make incredible connections and discoveries in IAQ data that would otherwise be impossible or extremely tedious for us to do ourselves—not using gimmicks to present data in a “new” way.
THE FUTURE OF AI IN IAQ IS PACKED WITH POTENTIAL
Despite the caveats that exist today, the future of AI in IAQ management holds a ton of promise. Even now, there are so many potential applications for AI in this space, such as:
1. Simplifying Data Analysis
One of the top challenges with air quality optimization today is the sheer volume of data that we have to analyze. You get hundreds of thousands of data points per day from just one sensor—imagine the mountains of data you’d collect each day with dozens of sensors throughout a building. Analyzing this data requires a certain level of expertise in air quality readings and applications. Not everyone understands metrics like parts per billion (ppb) or total volatile organic compounds (TVOC).
This is where AI can be incredibly useful. The algorithms can rapidly crunch the numbers to turn millions of raw data points into digestible and actionable insights that are easily understood by everyone involved in a building’s IAQ strategy.
2. Moving From Prescriptive to Predictive Management
IAQ management today is highly prescriptive, meaning we look at our data to find the issue, diagnose it, and prescribe a solution. The challenge with this approach is that, by the time we are able to detect a problem, it’s likely already started impacting the quality of the air our occupants are breathing.
AI will make it possible to move IAQ management from a prescriptive to a predictive model. Once we have reliable, high-quality data to train the AI on, it can start to forecast air quality patterns and implement proactive solutions. For example, the AI can look at building performance data from the past several months and see that meeting rooms have higher CO2 levels in the late morning, so it automatically increases ventilation in those rooms earlier in the morning to prevent CO2 from getting too high once the rooms are fully occupied.
In the end, these are all great visions for how AI fits into the future of IAQ optimization. But, before we can take the next step, we have to focus on getting the right building blocks in place. That starts with measuring air quality on a continuous basis, pairing accurate and consistent sensor data with relevant, high-quality metadata (sensor location, mechanical equipment connections, etc.) to, ultimately, set the stage for AI to deliver real value.
