Why AI may be better understood as the next layer of computing in AEC
Every major technology shift invites comparison, especially in an industry like architecture, engineering, and construction, where change often arrives through tools, standards, processes, and deliverables that reshape how project teams work together. When something new starts gaining momentum, we naturally try to understand it by comparing it to something we have already experienced. In AEC, that often means comparing new technologies to CAD, BIM, cloud collaboration, digital twins, or other shifts that changed how we design, document, coordinate, and deliver projects.
That instinct is understandable, and in many cases it can be helpful. Comparisons give us language. They give us a starting point. They allow us to say, “This feels like that,” and then begin making sense of what is changing. But comparisons can also become limiting when they place a new technology into the wrong category. That is why I struggle a bit when I hear AI described as “the next BIM.”
I understand why the comparison is being made. BIM was one of the most significant changes our industry has experienced. It was never just about replacing lines with models. At its best, BIM changed how teams thought about information, coordination, standards, responsibility, project execution, and downstream value. It affected how we create, manage, exchange, and rely on building information throughout the life of a project.
But that is also why I do not think AI and BIM are the same kind of thing.
BIM is a process and methodology supported by technology. AI is becoming something broader than a process, a platform, or a single category of software. AI is increasingly becoming a capability that gets embedded into the tools, systems, devices, and workflows we already use. It is not simply another thing we will “do” alongside CAD, BIM, coordination, project management, or documentation. More and more, it is becoming part of the environment in which those activities happen.
AI is becoming something broader than a process, a platform, or a single category of software.
That distinction matters because if we think of AI as “the next BIM,” we may unintentionally make the conversation too narrow. We may focus too quickly on questions like which AI tool to buy, who owns the AI workflow, what the AI deliverable is, or how to write an AI execution plan. Those questions may have value in the right context, but they do not fully describe what is happening. The larger issue is not whether AI becomes another project delivery method. The larger issue is whether AI becomes part of the normal fabric of how information is created, searched, summarized, validated, exchanged, and acted upon.
A better analogy, at least in my view, may be the computer itself.
When I was younger, personal computers were not embedded into everything the way they are today. A computer was a specific thing, often in a specific place, used for specific tasks. You sat down at the computer to do computer work. Over time, that changed so completely that we almost stopped noticing it. Today, computing is in our phones, watches, cars, cameras, televisions, appliances, thermostats, jobsite equipment, building systems, and nearly every platform we use at work. We do not usually stop and say we are “using a computer” every time we check a notification on a watch, open a model on a laptop, follow directions in a vehicle, adjust a smart thermostat, or send a message from a phone. Computing moved from being a separate activity to becoming part of everyday life.
AI is moving in a similar direction.
Right now, many people still experience AI as a destination. They open ChatGPT, Microsoft Copilot, Gemini, Claude, or another AI-enabled tool and intentionally ask it to do something. They may use it to summarize notes, generate a first draft, brainstorm ideas, explain a concept, or create a starting point for a task. That is a meaningful stage of adoption, but I do not think it is the final form of how most people will experience AI.
Over time, AI will likely become less visible as a separate destination and more common as an embedded capability. It will appear inside email, search, meeting platforms, project management tools, design authoring applications, model coordination software, document management systems, estimating platforms, knowledge bases, support portals, and facilities management systems. In some cases, users may know they are using AI because the feature is clearly labeled. In other cases, they may simply experience a tool that is better at helping them find, organize, compare, summarize, or act on information.
That is where the comparison to BIM starts to break down. BIM is something a project team intentionally implements and manages. AI will increasingly appear in many of the systems that project teams already depend on. It may support BIM workflows, but it is not limited to BIM. It may help with model review, but it may also help with meeting documentation, specification analysis, issue tracking, knowledge retrieval, proposal development, training, project controls, internal support, and client communication. AI is not one workflow replacing another workflow. It is a capability that can be layered across many workflows.
For AEC firms, this should change how we think about strategy. If AI were simply another application, the path forward would be relatively straightforward. Evaluate the available tools, select the best option, train the users, and measure the results. That kind of approach still has a place, especially for targeted use cases, but it is not enough by itself. AI adoption touches more than software selection because AI depends heavily on the quality, structure, accessibility, and reliability of the information it is asked to work with.
This is where our experience with BIM still has something important to teach us. BIM showed the industry that technology alone does not create better outcomes. A model without standards, clear expectations, defined responsibilities, and quality control can create as much confusion as value. The same will be true with AI. A firm with inconsistent data, scattered standards, outdated project information, unclear permissions, and undocumented workflows should not expect AI to magically create clarity. In many cases, AI may simply make existing confusion faster, easier to package, and harder to notice.
That does not mean firms should wait until everything is perfect before exploring AI. Waiting for perfect data, perfect standards, and perfect governance would likely mean waiting forever. But it does mean the most successful firms may not be the ones chasing every new AI demo. They may be the ones that understand their own workflows well enough to know where AI can safely help, where it should be limited, and where professional judgment must remain firmly in control. In practical terms, AI readiness in AEC is not only about teaching people how to write better prompts. It includes understanding what information the firm trusts, where that information lives, who owns it, who has permission to access it, how current it is, and whether it is appropriate to use in a given context. It also includes training people to question AI-generated output rather than treating it as automatically reliable. The more embedded AI becomes, the more important it will be for professionals to understand both its usefulness and its limitations.
That point is especially important in our industry because AEC work is tied to real projects, real budgets, real schedules, real risks, and real communities. AI can help professionals move faster, but speed does not guarantee quality. AI can summarize a meeting, but someone still needs to confirm whether the summary reflects the actual decisions made. AI can generate a first draft, but someone still needs to evaluate whether the content is accurate, appropriate, and complete. AI can help identify patterns in project data, but someone still needs to determine whether those patterns matter. AI can assist with model review or document analysis, but it cannot carry professional responsibility for the final decision.
This is why I believe AI should be viewed as an assistant to professional judgment, not a replacement for it. The value of AI is not that it removes the need for expertise. In many ways, it increases the need for expertise because someone has to know whether the output is useful, misleading, incomplete, or simply wrong. As AI becomes more common, the people who understand the work deeply may become even more valuable, not less, because they will be better equipped to guide, evaluate, and apply AI-assisted results responsibly.
AI should be viewed as an assistant to professional judgment, not a replacement for it.
The challenge for firms is to shift from a mindset of simple adoption to one of readiness. Adoption asks, “What AI tools should we use?” Readiness asks a broader question: “How should our people, data, standards, workflows, and governance evolve as AI becomes part of the way work gets done?” That second question is more difficult, but it is also more important.
For some firms, the best starting point may be low-risk productivity use cases such as meeting summaries, internal knowledge search, training support, or first-draft content creation. For others, the opportunity may be in project data, document review, model quality checks, or workflow automation. The right starting point will depend on the firm’s goals, risk tolerance, data maturity, and existing technology environment. What matters is that experimentation happens with intention rather than hype, and that teams learn from practical use cases before scaling too quickly.
The goal should not be to use AI simply because it is new. The goal should be better work. Better access to information. Better use of people’s time. Better coordination. Better consistency. Better support for the professionals responsible for delivering projects. When AI is evaluated through that lens, it becomes easier to separate meaningful value from noise.
So, while I understand why people compare AI to BIM, I think the more useful way to frame AI is as the next layer of computing. BIM changed how we create and manage building information. AI is becoming a capability that will increasingly sit across many kinds of information, many kinds of tools, and many kinds of work.
That shift will not eliminate the need for standards. It will make standards more important. It will not eliminate the need for process. It will reveal where processes are unclear. It will not eliminate the need for expertise. It will require expertise to be applied in new ways.
The question for AEC firms is not simply whether AI is coming. It is already here, and it is becoming more embedded every day. The better question is whether we are preparing for AI as a separate tool we occasionally use, or as a capability that will increasingly become part of the working environment itself.
AI is where BIM was 20 years ago. Twenty years ago, firms didn't need "BIM," they needed the right technology, workflows, standards, and expertise to successfully implement it. Today, organizations don't simply need AI. They need to identify where it creates business value, integrate it into their workflows, and equip their people to use it effectively. That's where ARKANCE can help.