5 things construction professionals should know about AI in 2026
Are you prepared for what’s next in AECO?
According to the 2026 Bridging the Gap report, 90% of AEC firms are hitting a wall when it comes to deriving real value from artificial intelligence — despite the fact that it's one of the most discussed topics across every market and vertical in the industry. In an episode of Bridging the Gap: The Conversation — 2026 Reality Check, Ian Besford, Global Digital Delivery Leader at Mott MacDonald, shared his perspective on where AI is genuinely useful, where it falls short, and what the industry needs to do differently to close that gap.
1. The time problem isn't really about time
Ask anyone in construction why their team hasn't fully adopted AI tools and the answer is almost always the same: we don't have time. Time to learn the tools, time to experiment, time to figure out where they actually add value on a live project.
Ian Besford has a different read on that answer.
"We all know we work in this world where the pressure's always on to deliver. So people prioritize delivering over learning new ways of doing things. It's very much that tale of the woodcutter who's too busy chopping down trees in the forest to spend his time sharpening his axe." Ian Besford, Global Digital Delivery Leader, Mott MacDonald
The woodcutter keeps chopping because stopping feels like falling behind. But the axe gets duller with every tree, and the work gets harder. The same dynamic plays out across construction teams every day. The pressure to deliver the current project makes learning new ways of working feel like a luxury, even when those new ways would make every subsequent project faster.
The firms making real progress with AI aren't the ones that suddenly found more hours in the day. They're the ones that made a deliberate decision to treat learning as part of delivery, not a break from it.
2. AI's value is in replacement, not addition
The reason most AI pilots stall is the same reason most construction technology pilots stall. The tool gets added on top of an existing workflow rather than replacing part of it. Teams use the AI tool because someone asked them to, continue doing everything else the way they always did, and conclude after a few weeks that the AI didn't save them any time.
It didn't save them time because nothing was removed. Addition never creates efficiency. Substitution does.
"The real issue is how AI can help you by replacing elements of your work rather than adding to it as something else you have to do." Ian Besford, Global Digital Delivery Leader, Mott MacDonald
The construction teams getting genuine value from AI right now are using it to eliminate specific, time-consuming tasks. Summarizing meeting minutes. Producing first drafts of documents. Searching across large volumes of project information. Research and data analysis. These aren't glamorous use cases, but they're real ones — tasks that previously took hours and now take minutes, freeing up the time that the woodcutter never had.
The same principle applies to automation more broadly. Connecting the platforms your team already uses so that data flows automatically — rather than being manually extracted, reformatted, and re-entered — removes a category of work that nobody notices until it's gone. Revizto's integrations and API capabilities work on exactly that basis, connecting coordination data to the wider project environment and eliminating the manual overhead that sits between systems.
3. ChatGPT is not the ceiling
For most construction professionals, the mental model of what AI can do was formed somewhere between their first ChatGPT conversation and their most recent Copilot experience. That's a reasonable starting point, but it's a dramatically limited view of where the technology is heading.
"You can do things today which you couldn't even imagine 18 months ago. Most people have experienced ChatGPT or Copilot over the last year or so, and that's kind of their view of what AI can do." Ian Besford, Global Digital Delivery Leader, Mott MacDonald
The gap between where most people's AI expectations sit and where the technology actually is — let alone where it's heading — is significant. In software engineering, bespoke AI tools are already making developers two to three times more productive than they were before. That level of productivity gain hasn't hit the engineering and construction space yet, but Ian's view is that it's coming.
The firms that will be best positioned when it arrives are the ones that have been experimenting with the current generation of tools, building the habits and the workflows that will absorb the next wave of capability rather than scrambling to catch up with it.
For now, the practical opportunity is clear: use AI for research, data analysis, document drafting, and information retrieval. Get comfortable with what it can do today, and stay curious about what it will be able to do next.
4. Human accountability cannot be outsourced to AI
One of the most important questions to come out of the Bridging the Gap: The Conversation — 2026 Reality Check webinar came from the audience: when you connect AI to BIM, whose responsibility is it to know good data from bad data?
Ian's answer:
"When it comes to the use of AI, everything we have to do still has to be signed off at the end of the day by a human, and there has to be an accountable person. It can't just happen in a black box where you don't see anything that's going on, you just see the output of it. You have to understand what it's doing and what the process is doing." Ian Besford, Global Digital Delivery Leader, Mott MacDonald
This matters particularly in construction, where the output of design and coordination decisions gets built. If it goes wrong, people can get hurt. The accountability cannot disappear into an algorithm. It has to sit with a person who understands what the AI did, why it did it, and whether the output can be trusted.
The practical implication is that AI in construction workflows needs to be implemented with the same quality assurance logic applied to any other design input. You wouldn't put a graduate with no experience on the most complex structural design without oversight. The same principle applies to AI generated outputs. Human review is not optional.
FAQ
Selon le Rapport 2026 Bridging the Gap, 90 % des entreprises du secteur AEC se heurtent à un mur lorsqu'il s'agit de tirer une réelle valeur de l'IA. La raison la plus courante est que les outils d'IA sont ajoutés aux flux de travail existants plutôt que de remplacer une partie de ceux-ci, ce qui crée une charge de travail supplémentaire au lieu de la réduire. L'adoption durable de l'IA se produit lorsque les équipes identifient les tâches spécifiques que la technologie peut éliminer, et non pas seulement augmenter.
Les applications les plus pratiques à l'heure actuelle concernent des tâches spécifiques : résumer des comptes rendus de réunion, produire des premières versions de documents, rechercher des informations dans de vastes volumes de données de projet et analyser des données. Ces applications ont fait leurs preuves, permettent de réduire un travail réellement chronophage et créent les habitudes nécessaires à une adoption plus poussée de l'IA à mesure que la technologie progresse.
La responsabilité humaine ne peut pas être déléguée à l'IA dans les processus de construction. Chaque production générée par l'IA qui influence une décision de conception ou de coordination doit être examinée et validée par une personne qui comprend ce que l'IA a fait et si l'on peut faire confiance à cette production. La responsabilité incombe à l'humain, et non à l'algorithme, quelle que soit la manière dont l'IA a été utilisée dans le processus.
Privilégiez le remplacement plutôt que l'ajout. Identifiez les tâches les plus chronophages et répétitives dans vos flux de travail actuels et évaluez si l'IA peut les éliminer plutôt que de simplement s'y ajouter. Développez des habitudes avec les outils actuels plutôt que d'attendre des capacités plus avancées. Et créez un environnement où l'expérimentation est encouragée — les entreprises qui développent des capacités d'IA dès maintenant seront mieux positionnées lorsque la prochaine génération d'outils arrivera.
Pas encore, et pas sans une supervision humaine significative. Les outils d'IA progressent rapidement et rendent déjà les ingénieurs logiciels deux à trois fois plus productifs dans certains contextes. Ce niveau de capacité n'a pas encore atteint la conception et la coordination de la construction à grande échelle. La tendance est claire, mais le calendrier est incertain. Les entreprises devraient expérimenter et développer leurs capacités dès maintenant, plutôt que d'attendre que la technologie mûrisse avant de l'adopter.
