Embracing Generative Risk Mitigation in Construction

By Georgia Stillwell
January 25, 2024

Project delays have long plagued the construction industry, with risk often identified as the primary culprit. However, finding effective solutions to mitigate risk on complex projects has remained daunting. Traditional methods for simulating risk primarily focus on extending project timelines, overlooking the diverse range of opportunities available for risk mitigation. With the construction industry’s digital transformation, generative methodologies have emerged to handle complex decision-making in uncertain situations. This article aims to shed light on the limitations of existing risk modeling and introduce a novel approach known as generative risk mitigation to enhance decision-making under deep uncertainty.

According to McKinsey, 98% of megaprojects experience cost overruns exceeding 30%. Project delays have become so pervasive that the industry has grown accustomed to them. For example, in 2022, the UK government issued ‘The Green Book,’ which requires contingency funds in projects, such as a 44% contingency budget for standard civil projects. This implies that for a $100 million project, you should allocate $144 million to manage expected risks. There is no denying significant academic literature on the root cause of these delays: it is ‘risk,’ and there is an entire industry based on it.

Conversations with project directors and risk experts reveal the same issue, different project. And that issue is that we cannot easily forecast risk, qualify the impacts or fully understand the opportunities that exist to mitigate risks and make timely decisions. A method that will finally help us overcome this has emerged within the industry.


Before delving into the solution, examining the current landscape of risk modeling in construction is crucial. Commonly recognized concepts in this field include risk registers, quantitative schedule risk analysis, Monte Carlo simulation, S-curves and evaluations of project completion timing. These traditional methods have faced criticism owing to their inherent limitations, a viewpoint shared by experienced project experts.

One widely used method mentioned above, QSRA, models risks and uncertainty onto an existing schedule, simulating multiple times and summarizing those results into a cumulative probability distribution curve, which indicates the likelihood of finishing your project on time. Regardless of where you get your risk and uncertainty information, whether it originates from reference class forecasting, benchmarking data, extensive datasets or your in-house project team, the method of simulating them onto a schedule where the logic is all hard logic is unrealistic.

Hard logic implies that all links made in a schedule are obeyed as hard rules and never broken. This is different from soft or preferential logic, which allows a degree of freedom on what activities should proceed next.

Consider the following example: a construction site divided into four zones—north, east, south and west. A single crew is responsible for laying concrete in all four zones, and the order in which they tackle these zones is not prescribed—soft logic. However, as soon as links are drawn between the zones, forcing the crew to go north to east, etc., the logic is now hard and unalterable.

All schedules going into a QSRA are 100% hard logic and, therefore, fixed in sequence. In practice, this means when we add a risk and model it on a typical schedule, all links between activities must be maintained, effectively pushing out the entire schedule end date.

This approach is unrealistic. Why? Project teams do not stand by idly when risks materialize. Instead, project directors leverage their expertise and instincts to make prompt decisions on recovering from a delay. These decisions take into account the variables in play and rely on experience—additional excavation equipment, overtime work for crews or a complete resequencing of project scope. It is a tough, time-sensitive decision, and the project’s success counts on it being right. Instincts also play a big part, which is why project directors are hired based on their experience. But experience can also include biases—recency bias, optimism bias and the creation of a narrative fallacy, to name a few.


The Project Management Journal article "What Are the Causes and Cures of Poor Megaproject Performance? A Systematic Literature Review," which reviewed over 6,007 titles and abstracts and 86 full papers, highlights that there is "little or no explanation of how performance may be improved by making decisions to address unforeseen events and circumstances when a megaproject is underway.” Today, we can change this. And the crux of it is generative scheduling, which means identifying only hard logic, leaving all soft/preferential logic out of the schedule, and allowing AI to generate the soft logic that best satisfies the project constraints. That means not just pushing out a sequence driven by fixed logic like a typical QSRA when simulating risks, but rather exploring new logic sequences and considering all parameterized variables: crews, equipment, production rates, working hours, etc. This reflects the reality of a project. If the risk occurs, the project director considers resequencing and which variables to dial up and down.

With generative scheduling, instead of laboring over that decision, those scenarios can be run in minutes, exploring which levers are worth pulling, making decisions the very next day. Construction professionals can supplement their experience and make rapid, data-driven decisions. This is a breakthrough as, for the first time, the risk and mitigation plans modeled produce outcomes that reflect reality, as opposed to being frozen in time while a risk materializes and carrying on with the original plan. Unsurprisingly, many schedules are thrown out as soon as something goes awry because they are not flexible enough, the logic is hard and the inputs are static.

Generative risk mitigation is an innovative approach to risk modeling that harnesses AI and generative scheduling simulation. It is predicted to revolutionize how owners and contractors handle project risk by replicating real-time decision-making and considering a broader range of variables. Similar to navigation software recalculating routes when you deviate off course, generative risk mitigation—or indeed generative project recovery—determines how to meet the original project end date and identifies necessary adjustments in the face of risks and uncertainties.

Embracing the philosophy of proactive risk management, generative risk mitigation heralds a new era of construction project management. It confronts risks head-on, acknowledging that their nature or timing cannot always be predicted. Generative risk mitigation empowers the construction industry to navigate the unpredictable with agility and data-driven precision, marking a significant step towards ensuring better project success.

by Georgia Stillwell
ALICE Technologies

Related stories

Calculating the True Cost of Underinvesting in Construction Health and Safety
By James Alexander
Research from the Construction Safety Research Alliance aims to set a global standard for safety, efficiency and finance at construction companies.
Mitigating Four Major Risk Factors in the Construction Industry
By Julia Holden Davis
ESG, inflation, the supply chain and weather pose major risks to the construction industry. Knowing them is the key to understanding and mitigating them.
Surety Trends to Keep an Eye on in the Construction Industry
By Oliver Craig
Even a market as sure as surety bonding is expecting trends and uncertainties in 2024. From interest rates and credit availability to labor and insurance, challenges and opportunities abound.

Follow us

Subscribe to Our Newsletter

Stay in the know with the latest industry news, technology and our weekly features. Get early access to any CE events and webinars.