Construction based AI: Is it a pipe dream?

I recently read a great article by McKinsey on Artificial intelligence: Construction technology’s next frontier. Its a good read about construction based AI and I would commend it to you.

Are you ready for AI?

A lot of organisations that I speak to report that they aren’t ready for an AI future within a project context because they are still embedding the fundamentals of project management; “Our priority is to develop a deep foundation of project management skills” rather than develop the foundation for an AI based future. Its a position that I can empathise with.

But AI is marching ahead at a considerable pace and when the technology takes off within the construction sector it is likely to transform the competitive landscape. AI isn’t a simple plug in solution. It needs data…lots of data. Data needs to be collected, governed (data quality, version control and all the essential boring stuff), retained and made available to those who need it. For most organisations, this is a significant challenge because project data is developed specifically for the project and may be difficult to find when the project has finished; or some data may simply not be available. By way of example consider the project risk register. How many projects version control their risk registers so that analysts can understand the trajectory of project risk over its lifecycle, or better still, how the risk influenced schedule and cost variance. It requires a set of interconnected data; something that a lot of projects simply don’t have.

There is a perception that this is all too difficult and costly. But with foresight, how difficult is it to save version controlled snapshots of the risk register, project costs and schedules, linking the data together within a data model? Its an inexpensive investment that begins to build a foundation for what will follow.

There is also a perception that its a huge leap from today into a new AI world. But such huge leaps will rarely be successful for a whole host of reasons, with cultural change being at the forefront. Success will only come from a series of small stepping stones; agile based implementation experiments that build capability incrementally, aligned to business goals and use cases. Such stepping stones can be as simple as building out a connected dataset and using the data to derive insights that were previously hidden. Techniques such as regression and clustering can then be used to move up the maturity ladder from diagnostics (what happened) through to predictive analytics, and then machine learning and AI to deliver the final optimisation step of prescriptive analytics.

Example use cases

The McKinsey paper describes a number of use cases, but we could go so much further. Let me run through a few examples:

  • Scheduling. Construction projects all have their own schedules. The schedules evolve over time. Techniques already exist to extract and compare the work breakdown structure to provide comparables and benchmarks, either at a planning module level or across the entire plan. By understanding schedule and cost variance across the aggregated dataset, it is also possible to understand where the greatest risk to delivery and margin exist. This technology is already being applied in other sectors and will be applied to construction within the next 12-24 months. The challenge for industry is how they wish to engage with it.
  • Risk management. By aggregating risk data we begin to understand which risks are difficult to treat, which management actions work or don’t, the level of risk management budget required and insights to help compensate for optimism bias. Again, the technology exists. Its a matter of gaining access to data.
  • Benefits management. Gaining an understanding of which benefits are intrinsically difficult to deliver, which are oversold, the actions which have the greatest influence on successful benefits realisation, how benefits relate to each other and are interdependent. This could help to transform how benefits management is delivered within the P3M profession.

This is just a small insight into the potential use cases and when the capabilities mature they will be transformational. The most enlightening insights will come from the integration of all of this data, connecting risks, lessons learned, schedules, costs and project insights. In my view it will deliver the largest advancement that our profession has ever seen and it will emerge at an incredible pace within a period of years.

If you would like to be involved in helping to shape this future then please sign up to the London Projects Data and Analytics Meetup where we are building a community to help to drive this agenda. Wouldn’t it be cool to be part of it?