AI in construction (AI, artificial construction)

The use of big data and reliance on machine learning capability are becoming standard project control tools. Niru Sundararajah, a data scientist for Aecom’s Cost Intelligence Europe team, shares his perspective on how solid foundations for data management are crucial to successful analytical outcomes using AI in construction

We see many discussions about big data, especially among data scientists, and we’re seeing a shift in that rhetoric towards the untapped potential of Artificial Intelligence (AI) to revolutionise the AEC industry – particularly the construction sector.

AI implementation case studies in this field are more easily found these days touting successes regarding the improvement of cost prediction and risk mitigation – those alluring golden markers of success that are highly convincing – and it’s understandable why the use of AI in construction continues to grow by leaps and bounds.

The lure of AI is justifiable and, within the Aecom Cost Intelligence team, it has improved both the quality and efficiency of our client services. Clients are leading the charge for intelligent data as they see the value proposition of becoming increasingly conscious about the use of their data to enable better prediction and enhance decision-making processes.

Aecom has created AI tools to meet client demands and assist them on their full lifecycle digital journey to ensure their projects work for them to produce accurate cost predictions with less effort compared to current processes.

Despite the accolades and success stories within and beyond our organisation, the potential offered by AI remains largely untapped primarily due to the more traditional challenges surrounding data governance as it relates to where the data resides, the volume of that data and the quality of available data.

The need for data

When using AI applications, we’re relying on a clean dataset of consistent variables in order to infer trends and patterns before formulating predictive results. In order to perform optimally, we require large quantities of that clean or high-quality data. What we mean by that is any deficiencies in either data quality or quantity will prove to be a significant stumbling block for AI in construction.

Data stored outside of a common data environment or scattered across disparate software platforms are significant industry issues to overcome. An overarching data governance programme is the key to unlocking the value of organisational data in order to successfully navigate this hurdle.

What is data governance?

At its core, data governance concerns the right mixture of process, technology and personnel to govern the input, storage and use of data in order to meet and exceed business goals. The Data Governance Institute dives into greater detail, describing data governance as “a system of decision rights and accountabilities for information-related processes, executed according to agreed-upon models which describe who can take what actions with what information, and when, under what circumstances, using what methods.”

Traditionally, data governance frameworks have a wide reach and vary based on client requirements and organisational needs. That said, there are several key aspects that are commonly present across data governance frameworks:

  • Stewardship: Promoting accountability by assigning stewards/custodians to relevant datasets.
  • Accessibility: Facilitating availability of data for relevant stakeholders.
  • Data security: Ensuring sensitivity-based safeguarding measures are implemented on databases.
  • Data quality: Maintaining and monitoring data quality to ensure its suitability for intended applications.
  • Knowledge: Preserving and improving data knowledge within the organisation by ensuring documentation of data systems and related processes are kept up to date.

How is data governance relevant to AI?

While a strong data governance programme establishes baseline approaches to data use, the most critical aspect for the purposes of AI implementation is ensuring high-quality data is available for collection. This is achieved through various means; most critically through its emphasis on data quality.

By defining data quality requirements, an effective data governance framework will promote accountability and encourage initiative among stakeholders to maintain the useable of data both for ease of reporting due to a consistent data set and especially for AI purposes.

Pursuit of a robust data governance programme, like any quality programme, is best shared by all stakeholders – particularly those in leadership positions. The burden of storing and maintaining a clean dataset that is focused on the storage and transfer of data, as well as its consistency, ensures that everyone in the organisation has access to the data they require for each use case and assures continuity in the face of personnel changes.

Who is responsible for data governance?

As we’ve already alluded to, data governance requires collective effort for successful implementation. However, it’s the senior leadership that must firstly buy-in to the criticality of data governance policies and programmes, and drive this message across the organisation – embedding data governance discussions into wider business strategy.

Further, a top-down intervention is pivotal to ensure data governance tasks are prioritised. By promoting data consciousness at the top of the organisation, it will be easier to shape the right data governance culture to produce the metrics and predictive analytics that can mitigate risk and increase profits. By doing this, the benefits offered by AI can be leveraged.

What are the next steps?

At times the route to AI can be a tricky space to navigate due to the complexities surrounding data governance and the wide array of stakeholders who create and use data and data systems across an organisation.

Aecom works with clients across the AEC industry who are looking to improve their data pipeline and systems across diverse portfolio of projects. We work with clients to firstly develop a standardised data quality assessment framework based on recognised industry standards. Secondly, as data quality is a key principle of data governance, we recommend the adoption of company-wide best practices to enable clients to achieve their data governance objectives.

We actively assist clients in reshaping their data processes and systems in order to build the foundations for AI; enabling our clients to begin the journey of fully leveraging the potential AI has to offer – a long journey of transformation but well worth the effort. The hope is that data and AI will not only help us navigate complexity in the future, but it will also help our industry achieve better environmental, social and governance outcomes that allow communities to thrive.



Niru Sundararajah

Data scientist, cost intelligence, Europe


+44 (0)20 7061 7000

Twitter: AECOM

LinkedIn: aecom

YouTube: AECOMTechnologyCorp

Instagram: aecom


Please enter your comment!
Please enter your name here