How predictive AI will curb construction risk and improve decision-making

Construction projects are highly complex and interconnected undertakings, and the potential for inefficiency and risk – which inevitably lead to project costs and delays – can grow exponentially along with the size of the project.

Many engineering and construction (E&C) organizations have accelerated their automation efforts as they feel the squeeze of growing risk, strained supply chains, and narrowing margins. Traditionally organizations have focused on gaining operational improvements by using technology to refine processes and procedures, but the data accrued from digitization can sometimes be an afterthought.

What if you could improve your chances of delivering a project on time and on budget by utilizing the growing volume of data that you previously only archived for future reference? Artificial intelligence (AI) holds enormous potential to help E&C organizations optimize their decision-making, and to drive project success by proactively unlocking new predictive insights from project data.

Turning data into insights

Vast amounts of data are being generated by the construction industry as digitization is embraced, and with it there is a significant opportunity for teams to learn from and use this data to create better estimates, plan smarter, and avoid – or at least mitigate – potential risks.

Historical data provides a critical starting point for organizations to provide deeper analysis of their business. By looking at existing project data, E&C professionals can answer questions such as:

What is the amount of time it takes to complete a process?
Which subcontractors are the best and worst performing?
Which activities have typically been delayed in the past?

Once you've been able to define baselines, benchmarks, standards and key performance indicators (KPIs) from your data, you can now apply these insights to present conditions. For example, in the case of estimating, you can discern:

Are our estimates accurate and have we accounted for historical delays?
How are we performing compared to historical benchmarks and organizational baselines?
Did we select the best partners for the job based on previous project performance?
Should we change the requirements or frequency of our reporting to ensure we avoid "surprises"?

The ability to gain insights from historical data and apply them to current projects is key to providing a basis to prevent mistakes from being repeated and ensure there is a focus on driving continuous improvements.

Looking ahead with predictive AI

To date, business intelligence (BI) technologies have generally only provided a backward-looking view into project data. While these insights are valuable, new developments in AI have unlocked a new level of project intelligence based on real-time data, enabling predictive insights to turn into better outcomes.

This transformative change in data science for the E&C industry yields a dynamic view into such variables as:

The factors which might delay a project
The probability of delay on a project
Amount of predicted delay
Likelihood (and severity) of a cost overrun
Hidden risks around safety, design, rework, and litigation

These AI technologies are powering active intelligence, helping organizations learn from the past while continually assessing the present. This enables organizations to regularly monitor developments and adjust plans using up-to-date predictive insights. Such a system is always current and learns from its re-trainable machine learning models, growing smarter and improving accuracy over time.

Predictive insights at work

Active intelligence yields predictive insights that add value to nearly every aspect of construction project management, including critical areas such as schedule, cost/budget, quality, safety, risk, and collaboration.

Schedule – Project schedules, which seem to change daily, are the lifeblood for project managers. Predictive intelligence can improve schedule accuracy for planned and in-progress projects by identifying which are most likely to have delays, the probability and extent of potential schedule delays, and which activities are most likely to create delays.

Active intelligence leverages rich internal and external data sources to improve prediction accuracy and precision. Internal data can include past schedules, subcontractor delays, rework histories, missed delivery dates, request-for-information (RFI) bottlenecks, actual versus estimated project duration, work breakdown structures, and more. When combining these along with external data, such as weather forecasts/history, supply chain disruptions, and workforce disruptions, it is possible to create a more accurate schedule and manage them proactively to reduce risk.
Cost/budget – Countless variables come into play projecting budgets, and there’s no shortage of historical data active intelligence can leverage to build better budgets and spot potential issues early on.

Active intelligence can leverage past cost breakdown structures and actual costs and budget, along with change request history, subcontractor performance, geographic considerations, and project types to build better budgets from the start. Organizations can proactively monitor projects and manage costs due to change requests and other variations, with the goal of fewer surprises and reduced risks.
Risk – Project risk can take on many different forms in construction, including added costs, compliance violations, litigation, and safety. Active intelligence has the potential to mitigate risk on all of these fronts.

Using natural language processing capabilities, active intelligence can analyze sentiment and detect early signs of conflict between two parties, such as a general contractor and subcontractor. After factoring in volume of correspondence on specific issues, as well as the history of work delays and outcomes with specific subcontractors, AI can provide an effective early warning system that surfaces potential issues long before they boil over.

Firms can also leverage active intelligence to manage their project portfolio better, with insight into which projects are likely to be most profitable and successful based on the risk associated with them.
Quality – Active intelligence can improve quality of work, ranging from workmanship to material integrity, by leveraging their data on materials providers, subcontractor performance, and site performance to stay ahead of potential risk. For example, natural language processing can detect potential quality issues by mining data as diverse as correspondence, punch lists, inspection reports, change requests, and work orders.

The Internet of Things (IoT) sensors and the rich data they can provide are additionally set to play a growing role with how active intelligence can improve quality of work.
Safety – The health and safety of everyone on a project is always top priority. The emerging signs that can alert organizations of potential safety issues are all too obvious after an accident occurs.

Predictive intelligence can help to detect potential issues by analyzing images from the jobsite, data from sensors, safety reports, correspondence, training logs, past incidents, and more. This can provide organizations with an early warning of potential emerging safety issues so that they can address them proactively.
Collaboration – Project managers depend on collaboration for project success—and this starts with an open and transparent supply chain. Active intelligence combined with graph analytics offer the potential to extend and strengthen collaboration. By visualizing the complete supply chain in a single image with an overlay of various risks factors, active intelligence can help identify problems before they happen.

Using concepts borrowed from social network analysis, it becomes easy to identify information bottlenecks, inter-dependencies between partners, and easily make sense of the vast amounts of information typically shared between the different parties.

There's a lot that can be learned from the past; but the real benefit of active intelligence is in its ability to predict the future and make the decisions needed to change the course of a project before it's too late. The challenge has been identifying the right models for the construction industry and enabling organizations to connect their data from all aspects of a project: schedule, budget, resources, safety, quality, and risk.

AI and machine learning are turning the tide in the construction industry, putting active intelligence within our reach. Today, with AI, we can look ahead and improve the decision making of tomorrow.