Preparing Canada’s AECO Workforce for the AI Era

The workforce discussion surfaced a fundamental tension: AI tools are being designed by people who understand technology, but adopted by people whose daily reality looks nothing like a product demo. For tradespeople and field workers, many of whom are not digitally advanced, tools need to be so simple that adoption feels effortless. The industry has spent decades telling workers not to use mobile phones on site for safety reasons, and is now asking those same workers to engage with AI through those devices. That contradiction has not been reconciled.

The most resonant insight was about framing. Training and adoption efforts succeed when they answer “what’s in it for me” from the worker’s perspective. If a worker perceives that AI threatens their job or primarily benefits someone above them, resistance is rational.

From an academic perspective, the room heard that AI now makes it possible to deliver training through simulation, multilingual interfaces, and adaptive formats that were previously impractical. The technology exists to meet workers where they are. The challenge is institutional willingness and funding to deploy it.

A practical example was shared of a labourer on a high-rise project who was constantly walking between the fifth floor and the site trailer to check drawing information. Over time, with side-by-side support, that worker learned to use the tablet his company had provided and eventually became a champion for technology adoption among his crew. The lesson: adoption happens through accompaniment.

The unionized workforce dimension was also raised. If AI begins to monitor worker activity or performance through computer vision or data tracking, it will intersect with collective bargaining agreements and labour relations in ways the industry has not yet confronted.

The integration discussion converged quickly on a single foundational issue: data interoperability. The ability to develop and deploy AI tools is fundamentally limited by how data is structured, formatted, and shared across organizations. Good tools built on siloed or improperly formatted data cannot deliver on their promise. This was identified as the single biggest blocker to meaningful AI adoption in design and engineering workflows. Realizing the value of AI will require organizations to re-evaluate and redesign their workflows to optimize how data is captured, processed, and used.

A parallel was drawn to the aviation industry, which has spent 35 to 40 years developing robust data-sharing standards across airlines, handlers, and their supply chains. Construction has been earlier to this effort. Without industry-defined standards for the types of data to be shared, the risk is that a small number of large technology platforms end up controlling the ecosystem, with smaller firms and startups unable to participate on equal terms.

The startups in the room pushed on open APIs as the lever. Companies that make their data accessible are seeing more usage, and in some cases, AI systems are recommending tools based on API availability. The larger platforms are beginning to open up, and collective pressure from buyers, even smaller ones, can influence this trajectory.

On the implementation side, the message was clear: without a dedicated person on site walking workers through new tools side by side, no technology adoption is sustainable. Change requires investment in people. And that investment needs to flow all the way down to subcontractors, who are the ones actually performing the work but often have no incentive or requirement to adopt digital tools.

A broader point was made about the digital value chain in construction. The digital component of major project bids has grown significantly over the past decade. Subcontractors will increasingly be expected to contribute data in addition to labour, and to demonstrate data proficiency and security. Those who cannot demonstrate digital capability will find themselves unable to compete, regardless of their trade skills.

The standards discussion was the most candid. The room heard a direct challenge to the structural dynamics of the industry: procurement models that reward lowest price discourage investment in AI capability, data governance, or workforce development. When margins are compressed at every level, the business case for adoption becomes harder to make. The procurement model itself was identified as a barrier.

The room observed that historically, major changes in construction, such as the transformation of safety culture, were driven from within the industry by people with direct operational experience. AI adoption is different. It is being driven from outside the industry by technology companies, and construction’s instinct is to be cautious. Bridging the gap between the technology builders and the people who will use it is a precondition for progress.

From the legal perspective, AI is already showing up in claims and dispute resolution. Predictive tools are being used to forecast arbitration outcomes and analyse judicial tendencies. The legal infrastructure is adapting alongside the operational side of the industry.

A philosophical thread ran through the closing discussion. The promise of AI is the removal of repetitive, data-driven tasks that do not require human judgment. But this framing, while rational, does not account for how people actually experience change. People prefer known, predictable outcomes, even when those outcomes are suboptimal. The room raised the question of whether current investment in AI is aimed at freeing people from drudgery or at reducing headcount. If the industry does not actively choose a path, the default will be shaped by whoever moves first.

The consensus was that progress begins with simplicity. The last major culture shift in construction, around safety, succeeded because it started small and was made accessible. AI adoption needs the same approach: small, demonstrable wins that people can see working before scaling.