Since the mid nineties productivity in manufacturing has more than doubled while over the same period in the construction sector it has barely improved.

One of the main reasons attributed to this is that construction is one of the slowest sectors in the world to adopt new technologies. But machine learning and artificial intelligence (AI) is gradually becoming recognised as a potential game changer in construction efficiency.

The construction sector has become very good at creating data – lots of it. The availability of so much data presents challenges to those (humans) trying to analyse its meaning, sometimes slowing critical decision making or smothering the obvious simple answers. But machines love data and the more of it there is, the better that they can learn from it.

There is a huge potential for the application of AI in construction. Because AI can analyse data that overwhelms humans it can alert the critical issues that need a project manager’s attention, saving the hours it would normally take to make the analysis.

We are already seeing the use of AI in BIM modelling as new technology creates modelling systems that learn from each design iteration to make sure that mechanical, electrical and plumbing plans don’t clash with building architecture and actually create the optimum solution.

One AI company has now developed robots that can physically scan construction sites and feed the data back to a neural network that identifies progress on all aspects of the project.  Self-driving machinery will become more prevalent on sites for repetitive work and many construction companies already use AI to manage resources more effectively.

A recent McKinsey report said that real time analysis of data could improve productivity in construction by up to 50%. Imagine the value of that to HS2 or Crossrail.

While some fear significant job losses to AI, in reality it’s unlikely to replace the human workforce but there is absolutely no doubt that in the construction sector it will change business models, increase efficiency and reduce expensive errors.