Construction Productivity: Why it is Falling and why AI is the key

Automated decision systems (aka AI) are rapidly transforming the way industries work. Over the last few months, we’ve posted some of our key theses in fields of mining, manufacturing, and warehousing & logistics.

Despite the talk of “Software is eating the world,” the reality is that some industries have barely changed in the last 50 years. Nowhere is this more evident than in the world of construction — the only industry where productivity has consistently dropped for the last 20 years and is still lower today than it was in 1964. This graphic by Plangrid outlines how far we’ve come.

This post is the first of a series of blog posts from our team that outlines why we believe that, finally, we will begin seeing wholesale change.

To properly understand what opportunities exist for AI to transform the world of construction, we need to examine some of the causes of low productivity to-date:

Realistic Depiction of a Construction Site

Inefficiency due to misaligned incentives: In a normal factory setting, workers are hired to complete the necessary work. However, because of the “contractor” model (a result of labor-laws and worker-compensation exemption), owners don’t hire workers to complete work that needs to be done. Instead, for most commercial construction projects, there are usually at least four parties (and often up to hundreds) that have different goals:

• The owners (or developers): The owners (or developers): Generally, want the fastest work done for the lowest price. (Seems reasonable!)
• The financing banks: Want accountability for where their money is being spent, with as little risk as possible. Minimal risk helps to ensure on-time repayment.
• The various consultants: I.e. architects (a person who designs, builds and sometimes oversees construction), Mechanical, electrical, and plumbing engineers, (MEPs plan, design, and manage their designated areas of expertise) structural engineers (consult and advise contractors and architects throughout the building process on possible structural issues), etc., put their reputations on the line (literally by “signing / stamping” their work). As a result, they are primarily concerned with the status of the end-product. This includes things such as safety, quality, design aesthetic, etc. (note, cost of construction is generally not a significant factor)
• The general contractor: The general contractor has the responsibility to provide all labor, material, and equipment ( including tools and vehicles) necessary for a job to be completed. They are mainly concerned with 1) winning the bid to secure the job, and 2) maximizing their profits while 3) maintaining a quality standard to avoid getting fired from the job.
• The sub-contractors: A sub-contractor is given a certain amount of existing work to complete through a contract with the general contractor. The sub-contractors are concerned with completing their contracts well enough to pass inspection while trying to maximize profit (billed work minus actual input to the project) and worker safety. Sub-contractors also purchase materials for a job and are concerned with getting paid on-time.

As Charlie Munger said: “show me the incentive, and I can show you the outcome.”The overall efficiency here is significantly lower when compared to a company where incentives are relatively simple. Within a simple incentive system the company has a goal they are striving towards. Workers who help the company achieve this goal, receive rewards). This lowered efficiency leads to seldom kept budgets and timelines.

Inefficiency because of non-standard jobs: Construction is similar to the movie industry. For each new project, a new director, producer, and new set design, as well as new actors must be pulled together to complete a complicated task. Once the project is complete, they go their separate ways. Unlike factories that build best-practices, tracking, and trusting relationships, construction projects generally do not have the time to work better together on the “next” project. Instead, the role of the General Contractor is to use a series of essential tools to conduct the necessary work. For example, the idea of paper drawings, or PDF versions are nothing more than a 2-dimensional representation of an intended end-result, generally without any details of the methodology, an order of work, or feasibility of the work. It is not unusual for a subcontractor to receive a set of drawings, and need significant time re-drawing that work into separate and easy to complete necessary tasks.

Inefficiency due to the complexity of coordination: In a typical commercial project, it is common to have dozens of subcontractors and hundreds of workers collaborating to build a project. That this arrangement works at all is a marvel of human ingenuity, but there are still many lost opportunities to increase productivity. For example, drywall installation typically is dependent on the completion of framing work and electrical / plumbing. After the installation of drywall, the electrical team has to return to the site and complete the installation of switches, fixtures, etc. Since the electricians and the drywall installers are different subcontractors, their work happens in a sequence. First, the framer works, then the electrician, then the drywall installer, finally the electrician returns. By coordinating these tasks, much of the work could happen in parallel, with each group working on one side of the job site while others prepared for them.

As a result of the sequential nature of the work, project planners often resort to *massive* Gantt charts to order work. These charts include tightly scheduled times for project completion. Unfortunately, since one part or another is often behind schedule, the cascading effects are never accounted for quickly enough, and all parts of the project end up delayed.

Inefficiency due to lack of precision: On the manufacturing floor, each step of the process is confirmed and reconfirmed. Precision tolerances are much lower than on a construction site. This lower tolerance is often chalked up to the “uncontrolled environment” of a construction site, however, with modern tools, it has gradually become possible to quantify the job site with precision, and create tools that auto-correct and bring much higher accuracy to the work product.

Why Now?

Although AI can not solve all of these problems, it will undeniably play a huge role because of a few key recent developments:

• Because of new low-power low-cost connected sensors, computer vision, and other tools, we can now effectively digitize the physical world: creating a high-fidelity digital representation is critical for the second step.
• The ability of computers to understand a complex mix of factors, and use it to make decisions has increased dramatically because of developments in deep learning. Rather than simple rule-based systems, computers can approach or surpass human decision-makers by relying on past experiences to make decisions.
• Automation & action: by using technologies like projection mapping, robotics, and wearable devices, computers can implement plans autonomously or with the assistance of human workers. Companies like Shaper Tools are already enabling non-experts to “magically” do the work of trained woodworkers, and this trend is sure to continue.

There are also significant changes in the industry afoot that are not reliant on AI, but will nonetheless transform the construction of buildings, and as a result, require developers, engineers, architects, and contractors to adapt their approaches. These include developments such as modular construction (prefabricated parts), the discovery of new materials and techniques (3D printed lattices, new glass facades, along with others.)

In our next few posts, we’ll outline new business models in the world of construction, challenges with selling into the construction industry, a startup landscape, as well as a series of interviews with experts in the field.

We look forward to your feedback!