Posted on September 04, 2019

Construction sites are collecting more and more data from wearables, telematics, GPS, drones, smartphones, BIM, and the list goes on. By using this data effectively and aggregating it, better strategic decisions can be formed. Big data is the lifeblood of machine learning, which will help the industry increase productivity, reduce risks and improve safety. 

Sounds great, right? However, big data - so-named because it is significant in volume, velocity and variety - poses issues for the industry, especially around maximising its usefulness and privacy. 

As part of our digital construction series, this blog will look at the challenges - and opportunities - of data privacy.




Last year, privacy concerns dominated the press in light of the Cambridge Analytica scandal. This British political consulting firm mined data through a personality test circulated via Facebook. The personal information of up to 87m users was shared illegally and then allegedly harvested for political campaigning. Subsequently, the US Federal Trade Commission has fined the social channel $5bn for this data privacy breach. As we write this blog, a documentary all about it, “The Great Hack” has just been released on Netflix:



This scandal signalled a stark reminder for social users that their data had market value and that they should protect it more carefully - or have it protected as a right. There was a backlash against the social media behemoth with 26% of people in the USA deleting their Facebook app last year

Skip a few months later to May 2018. Collectively we received a flurry of almost identical emails from businesses requesting that we “click here” or that we would be banished from their contact lists forever. These communications were in response to the EU General Data Protection Regulation (GDPR), which embodied a massive shakeup of the data privacy laws, to reflect technological advancements. 

Broadly the GDPR legislation means that

  • People have the right to be forgotten and have their data removed.
  • Businesses need to obtain an explicit opt-in to email and marketing communications.
  • There are limits placed on what data is collected in the first place.
  • Personal data must be secure and protected. 

GDPR is European legislation, but there are worldwide shifts to create similar privacy regulations in response to data breaches and cases like Cambridge Analytica. 




Analysts indicate that the update of AI in construction will be incremental steps, and it will undoubtedly impact on the sector as in other areas of society (McKinsey, 2018). Big data is essential to push forward developments in machine learning, which can bring considerable benefits to the industry. However, how does privacy impact on AI and construction?

Much of the big data that would be used by construction is not personal data. For example, connected IoT could collect data on weather conditions, temperature, or humidity. However, there will be personal information in big data analytics too. Wearables come to mind here. Businesses will need to show the case for tracking employees through this technology and assess reasonable time limits for keeping the data stored. 

It can be challenging to apply personal data protection laws in the big data context, especially where it involves using techniques made possible by AI. The ICO Big Data Paper 2017 states:

“These implications arise not only from the volume of the data but from the ways in which it is generated, the propensity to find new uses for it, the complexity of the processing and the possibility of unexpected consequences for individuals”.

AI data image

Big data analytics differs from traditional processing in these ways: 

  • It uses complex algorithms for processing data, which usually involves a discovery phase to find relevant correlations so that the algorithms can be created.
  • The way that the algorithms work lacks transparency - a “black box” effect.
  • Often all the data is collected rather than a randomised sample. 
  • If a third party has collected the data, it may be used for a different purpose than for which it was collected.
  • With IoT data, they have collected data through behaviours rather than direct data provided by the individual. 




There are concerns that the GDPR could limit the development of AI, but there are several ways to manage big data analytics in terms of privacy. These include:



This can help businesses use big data analytics while protecting personal privacy. Where anonymisation is not possible, pseudonymization should be considered - but those individuals should not be able to be identified.



As a reaction to the GDPR legislation, it is crucial to be transparent about what information you are collecting and what you will do with it. Such an approach makes sense to rebuild trust with your internal and external audiences around respecting privacy and developing long-term relationships. Privacy notices can be (surprisingly) engaging and a way to further explain your brand personality. A video could be a great way to show users what you are doing.


A privacy risk assessment: 

Regular privacy risk assessments are needed to mitigate risks before you process personal data. Many organisations may feel that GDPR is done and dusted, but it is an ongoing process, and embedding data privacy in big data analytics should be done at an organisational level - from marketing and IT to the boardroom.

AI data chip.


Explainable AI:

We’re sure you will have heard of cases where the algorithm showed racist or sexist bias. Bias can occur at any level of machine learning, and it is difficult to correct this. The data used could be unrepresentative or reflect existing prejudices (MIT).

Data privacy laws can help society to think about the ethics of big data usage and create explainable AI to reduce the impact of bias of future projects. Algorithm auditing is a new process to add to the list, could prevent us collectively from continually repeating the same bias. 


Real-time analytics:

Real-time analytics takes data that has just been collected and puts it to immediate use and analysis. There are systems which can help to filter, mask or encrypt data in real-time and detect any areas which would not comply. With collected data being processed immediately, there is no need for keeping that data around for an extended period in line with GDPR.




As we’ve discussed, machine learning or AI needs lots of accurate training data. BIM integration steps up here. For us, BIM usage is about more than software or modelling; it has the potential to bring digital into the whole build and specification process collaboratively. 

Building information modelling becomes a magnificent data pool of information. When harnessed correctly, construction industry professionals and their customers can analyse and benefit from the insights generated from the data with the help of AI and machine learning systems. 

The amount of digital data collected on a construction site is growing exponentially. IoT sensors can track fuel consumption and activity rates to monitor cost and carbon footprint, and geolocation can help to track when repairs or spare parts are needed. We can add weather, temperature, traffic, and business data to help optimise the schedule and manage the project, helping to bring the building wok in on time. 

This data needs to be structured so that it can be feed into machine learning, robotics and enable offsite construction. Using BIM, buildings can now not just be designed with a digital twin but also constructed. This twin helps to train the AI generating data from which it can continue to learn and predict issues before they happen to help improve design and maintenance. It results in a database of structured information that can interrogate all the projects essentials such costs, carbon footprint, energy use, waste or downtime. This data can then be analysed collaboratively across teams and feeds into machine learning to improve the construction of buildings, the performance of a building over its lifetime, demolition and future design.


To give you an idea of the power of BIM integration and AI, at Lobster Pictures, we have millions of high resolution fixed images of constructions sites. The data includes weather, light and a whole host of meta-data. The advantages of harnessing this data in terms of safety improvements, building efficiencies, time and cost-savings are many-fold. 

We believe that in the areas of digital construction that we can make better progress by being open and collaborative. That’s why we have a fully open API that we share with other platform providers on projects across the globe. 

Over a year on from the introduction of GDPR, security and personal privacy should be prioritised and kept at board level. It’s not in question that AI and machine learning will transform the construction sector. But it is important to bring consumers with us and take an open approach to reap these benefits.