Artificial intelligence is increasingly positioned as a solution for accelerating software development across the digital construction sector. From construction management platforms to BIM-enabled asset systems, development teams are under pressure to modernise software quickly. Evidence suggests AI can help. A McKinsey & Company report found that developers generated code up to 45% faster using generative AI. However, speed alone does not guarantee resilience, security, or long-term value.
Edward Anderson Jr., professor of information, risk, and operations management at Texas McCombs, warns that AI can create significant problems when applied to legacy systems. These systems underpin much of the built environment, including infrastructure scheduling, enterprise asset management, and construction finance platforms. Many were developed decades ago and contain shortcuts and undocumented fixes that increase technical debt.
Technical debt represents the cumulative cost of poor or outdated coding practices. The Consortium for Information & Software Quality estimates this burden costs US companies $1.5 trillion annually through reduced productivity and cybercrime. In construction and infrastructure, the consequences can be operationally severe. In 2022, Southwest Airlines’ 20-year-old scheduling system failed, cancelling nearly 17,000 flights. Similar failures in construction software could disrupt project delivery, payments, or compliance.
Anderson explains that AI tools learn from existing codebases, including flawed ones. When used to patch or extend outdated construction software, AI may reproduce poor practices at scale. As a result, “it tends to create more technical debt per line of code than trained, experienced human software engineers would.”
Research conducted by Anderson with Geoffrey Parker of Dartmouth College and Burcu Tan of the University of New Mexico highlights several measures organisations should adopt when deploying AI-assisted software development in legacy environments.
Technical debt
First, technical debt must become an explicit engineering priority. In fast-moving construction technology markets, debt is often deferred in favour of new features. Anderson argues that organisations should integrate debt reduction into everyday workflows, particularly when AI is involved. “If you’re going to use AI, and there’s a chance that you could be increasing the rate of technical debt generation, you’re going to have to allocate more time to doing the retirement.”
Second, clear guidelines for AI-assisted coding are essential. While many boards have high-level AI policies, day-to-day protocols remain inconsistent. Development teams should document when AI is used, for which tasks, and why. Human oversight remains critical. “You really want to make sure you’ve got somebody who has a lot of training in software engineering and experience to catch the AI when it’s making mistakes.”
Finally, training and knowledge transfer are central to managing risk. As experienced developers retire, junior staff may rely heavily on AI tools, particularly within complex legacy construction systems. Anderson stresses that mentoring must be formalised, ensuring senior engineers guide responsible AI use alongside code review. “Let me be clear: I think AI is a productivity booster. It’s just that you have to use it thoughtfully, and give software engineers the time to do that.”
Read next: Building an industrial AI operating system – Build in Digital
Are you a building professional? Sign up for a FREE MEMBERSHIP to upload news stories, post job vacancies, and connect with colleagues on our secure social feed.
