“I’m Sorry, Dave”: Should AI Really Control the Heating?

The promise and pitfalls of AI in building energy management

In the cult sci-fi film 2001: A Space Odyssey, astronaut Dave is famously locked out of his spacecraft by HAL 9000, the ship’s AI, with the calm but chilling words: “I’m sorry, Dave. I’m afraid I can’t do that.” The line has become a cultural reference point for the moment when machines begin making decisions humans can’t override.

While we’re far from HAL-level artificial intelligence in our buildings (thankfully), the rise of AI in energy management and building control systems does raise serious questions, particularly about reliability, trust, and the real value of advanced technology in our built environment.

Let’s explore where AI fits in, and where it might not.

.

The Modern Building Paradox: Smart but Not Trained

Most new commercial buildings come equipped with fairly advanced (albeit non-AI) building management systems. These are capable of controlling lighting, heating, and ventilation across zones, adjusting setpoints based on occupancy or time of day, and even optimizing energy use across seasons.

The problem is rarely the system. It’s often the users.

Staff often aren’t trained to use the system properly. As a result, it gets overridden, ignored, or left to operate on outdated schedules. When comfort levels drop or bills rise, the blame is placed on the system, rather than the lack of engagement with it.

In these cases, AI isn’t the missing piece, it’s proper commissioning, training, and engagement.

.

The Legacy Building Reality

In older buildings, or those that have been extended, subdivided, or repurposed over the years, it’s a different story.

The control system may not have been updated to reflect the current layout or usage. Thermostats may now sit in closed-off corners. Sensors may have been painted over or covered by filing cabinets. Light switches may control the wrong zone, or worse, require entering another room entirely.

Here, AI could be genuinely transformative.

.

What AI Could Add

A true AI-enabled control system would go beyond pre-set rules and static zoning. It would learn how people use the space. It would notice that the meeting room is empty on Fridays, that the open-plan area fills up at 8am in winter but not until 9am in summer, that some rooms are too warm during afternoon sun.

It could offer feedback, “Zone 3 has seen a 15% increase in after-hours usage this month”, and adjust heating and lighting schedules accordingly. It could react to weather forecasts, occupancy patterns, and even work shift rosters.

For poorly performing buildings, the leap from broken control to AI-enhanced automation would be huge.

.

The Devil in the Interface

But there’s a catch.

For a building to have a truly AI-enabled system, its existing control infrastructure must connect to an AI engine, usually cloud-based and hosted by a third-party provider. That means data flows off-site. It means internet connections, APIs, interfaces, and cybersecurity. It may mean giving external systems a degree of control over building operations.

For many organisations—especially those in finance, healthcare, or government, this introduces concerns. Even for others, there’s the question of reliability and support. What happens if the AI says, “I’m sorry, Dave. I’m afraid I can’t do that,” and there’s no manual override?

.

AI vs. “Just Working”

There’s a lesson here.

The difference between a good, properly commissioned and maintained control system and an AI-enabled one may not be large. The difference between a badly configured or out-of-date system and one that simply works properly is vast.

Before investing in AI, it’s worth asking:

• Has the current system been commissioned properly?

• Do staff know how to use it?

• Does the system reflect how the building is actually used today?

In other words: don’t chase AI before making sure your control system is fit for purpose. AI can optimise a good system, but it can’t fix one that was never set up right in the first place.

.

Final Thoughts

AI in building energy management is exciting. It promises intelligent control, predictive adjustments, and long-term savings. But it’s not magic. It builds on the foundations of working infrastructure and good design.

If your building is whispering, “I’m sorry, Dave,” maybe it’s not AI you need, just someone to retrain the system, not the staff.

.

While You Are Here…

Why not use one or all of Bryce Energy Services free online calculators.

Carbon Footprint — Solar Panels — Electric Chargers — ISO 50001 Gap Analysis

Or demonstrate your business is environmentally and socially responsible, with the CSR Accreditation.

Try our ‘new’ New Vehicle Carbon Footprint Calculator now.