AI

AI as a discipline studies models that predict, classify or propose based on patterns in data, including the question of transparency, bias and the limits of generalisation. The field is generic — it does not itself produce FM applications, but the mathematical and methodological foundation on which every AI application, including in FM, rests.

FM is offered AI functionality in almost every new system, from CAFM to BMS, often without clarity on which model sits behind it and what data it was trained on. The discipline gives FM the capacity to question an AI claim rather than accept it.

• What data was this model trained on, and is that data representative of our situation?
• What happens when the model is wrong — is there a human checking the output before it leads to action?

• The academic field of machine learning and AI ethics; the EU AI Act as a recent regulatory framework (jurisdiction-dependent, verify status per application).

• An AI system that makes decisions for people rather than supporting them may fall into a higher risk category under the EU AI Act, with corresponding obligations.

• Multiple Part 3 services with AI-supported functionality (CAFM, BMS, service desk)

• AI-in-FM (a Part 3 capability): the applied skill of selecting, implementing and controlling AI functionality in facility systems. This knowledge domain supplies the underlying methodology and limitations that capability must reckon with; it is not itself an application.

• Demand organisation

  The client role must determine what level of AI-use risk is acceptable, before a supplier delivers it as a standard feature.

• EU AI Act (Regulation 2024/1689) — risk-based classification; verify status and applicability per use case.

The standards section arrives in Part 7.

• 'AI' used as a label for rule-based automation (if-then logic) without any learning model.

The EU AI Act is recent and compliance practice for building-related AI applications (for instance camera surveillance with facial recognition) is still very much developing; classifying concrete FM applications into risk categories is not yet an established practice.

• EU AI Act, Regulation (EU) 2024/1689.