You’re the CEO of a construction company who has decided to embrace the power of artificial intelligence (AI). You’ve learned that AI can play an integral role in improving productivity, fostering innovation, and leading to a stronger bottom line. 

The time has come to act. You’re ready to go. Not so fast.

It’s true that the construction industry has begun to realize the transformative potential of AI. The market forecasts bear this out: According to Adroit Market Research, the global market for AI in construction was estimated in 2022 to be worth USD $1.3 billion, and by 2030, it is anticipated to grow to USD $13.5 billion, with an eye-popping Compound Annual Growth Rate (CAGR) of 36.03 percent. 

However, integrating AI is not as simple as plugging in a new software system. It requires careful preparation, robust infrastructure, and most critically clean data. Companies eager to jump into the AI space must understand that it is not a magic solution but a tool that needs a solid foundation to function effectively.

Many companies express interest in adopting AI, expecting immediate benefits. However, successful AI integration requires a proactive approach. Before diving into AI, organizations must establish clear objectives. What problems do they want AI to solve? What kind of useful information do they hope to extract? 

Identify what you want to do or accomplish with AI. Then work backwards to lay a technological foundation which will accept the AI tools that can meet your objectives. 

ESTABLISH OBJECTIVES

The first step is to determine the objectives for AI implementation. Are current processes capturing the necessary information? Often, companies will find they have substantial data, but it’s not adequate or appropriately organized for coordination with AI tools. This is a common scenario where data quality and completeness become significant hurdles.

There’s a saying in the world of artificial intelligence: “AI is the only place where ‘B’ comes before ‘A.’” Before you can begin integrating “artificial” intelligence, companies need to focus “business” intelligence. This involves getting an analyst to scrutinize existing data to solve your target problems manually. If a human analyst can derive valuable insights from the data, it indicates that the data is robust enough for AI. This step serves as a proof of concept, demonstrating that the data can indeed solve the intended problems.

Conversely, if a business analyst can’t actually look at your information, build some sort of mental model, and extract answers to pertinent questions from that data, then there’s little chance that AI, despite all the hype, will be able to produce any meaningful results. It is critical to understand that AI solutions scale well in terms of cost but are almost never as accurate as a detailed human analyst.

A great deal of the data that companies have accrued might have been collected for a specific use case that doesn’t lend itself to the use of AI. It might have even been stored with no real purpose in mind.

DATA CLEANING

A critical aspect of preparing for AI is data cleaning. Industry figures suggest that a whopping 80 percent of accumulated data is “dirty data,” unclean and, consequently, unusable. This necessitates a thorough review and cleaning process, ensuring that only high-quality data is fed into AI systems. 

AI models suffer from GIGO or “Garbage In Garbage Out,” hence the criticality of the data-cleaning process. Additionally, companies need to put significant effort into refining their data collection and storage processes to maintain high data quality.

For companies dealing with planning data, having a central repository is vital. Tools like Asta Vision enable centralized data storage and establish workflows that ensure data consistency. This eliminates the common issue of multiple, fragmented versions of data files, ensuring that the most accurate and final versions are the ones used for analysis.

Once the data is clean, the next step is to enrich it with additional layers of information. For example, in construction planning, adding metadata can help provide deeper insights. This iterative process involves regular reviews and enhancements, gradually building a robust dataset that AI can leverage effectively.

BUSINESS INTELLIGENCE AND BIM

Companies that have already invested in Business Intelligence and BIM (Building Information Modeling) will enjoy a marked advantage. These tools provide a solid foundation for AI by ensuring data is significantly richer and considerably easier to make machine readable. A thoughtful approach to “business intelligence” means that specific metrics are collected to derive actionable insights.

These tools will also form the foundation of an effective AI roadmap, which requires collaboration between software providers and clients. In industries like construction, where projects vary widely in scope and regulatory requirements, AI solutions need to be adaptable. This adaptability is achieved through enriched data and continuous feedback from real-world applications.

One of the primary applications of AI in construction is risk assessment and project planning. AI can use historical data to quantify risks, such as underestimating activity durations or identifying potential delays. This proactive risk management enables companies to address issues early, improving project outcomes and client communication.

FUTURE IN CONSTRUCTION

In the construction industry, AI can be a game-changer, reducing the risk of errors, enhancing project accuracy, and improving bidding processes. Companies that adopt AI tools will have a major advantage over those that do not.

The integration of AI in construction is akin to the evolution that took place from performing manual computations to the use of calculators. It’s a subjective perspective, but it is no exaggeration to say that the leap from traditional construction planning methods to AI-driven processes is even more dramatic and transformative. 

TIMEFRAME

AI represents a significant opportunity to enhance efficiency and accuracy, but none of this is going to be quick. The timeline for AI integration varies based on the organization’s size and readiness. For larger organizations, the process can take up to a year or more, whereas smaller companies might complete it within a quarter. This timeline includes setting up the necessary infrastructure, cleaning data, and iterating processes to ensure everything is in place for AI deployment.

At the end of this extensive process, you’ll have a robust data set that is going to build over time. That’s when you’ll realize it was worth it. 

about the author

Daniel Hewson is the data capability manager for Elecosoft. He has a strong background in mathematics, computer science, and engineering, with a focus on machine learning and how to apply it to real-world processes, including construction. For more, visit www.elecosoft.com.