THANK YOU FOR SUBSCRIBING
AI and Machine Learning's Role in Enhancing the PropTech Landscape
William Reddaway, Group Head of Innovation, J. Murphy & Sons Limited
As an industry that likes the comfort of tried and tested, we forget that we have innovated over the years, as well as having adopted and adapted new technologies over the decades. One could call it evolution, but we are doing ourselves a disservice—we have developed new materials, applied new engineering processes, improved efficiencies whilst reducing waste and improving safety—themes that are still very relevant today. Although our industry seldom makes original mistakes, we do learn, and we are able to build incredible structures, safely dig massive tunnels under cities among a myriad of obstacles with millimetre accuracy. Do we innovate? Of course, we do—we just do not talk about it or realise we are doing it!
However, as an industry, our maturity is still either very siloed or behind when introducing these type of technologies to the industry. To say that we are tech-shy is a misnomer. As an industry, we have changed significantly over the years, and continue to do so. Most civil engineering firms are now well entrenched in digital technology.
To name some examples:
• Laser scanning including LIDAR
• Using mobile computing for red-lining, snags, and reporting are becoming increasingly normal
• The introduction and adoption of BIM and its standards
• Digital authoring tools, now with the added opportunities that UAVs and satellite data can bring (photogrammetry and point clouds
Collecting and sorting data is all very well and good, but data is only useful if it can be used
• Robotics and machinery are allowing for increased efficiency or lowering personnel risk.
However, this new data revolution being an intangible element for many raises a lot of questions fuelled by trepidation. Often I hear the following: ‘Artificial intelligence/ big data/analytics so what?’ ‘What is in it for me?’ ‘It is not applicable to us.’
Collecting and sorting data is all very well and good, but data is only useful if it can be used. The challenge for my industry is unlocking but also understanding the potential and the benefit that data can provide us. Many AI developers promise a plethora of data and insights. We already have a plethora of data, insights are only useful if they add value.
I think the question we should be asking ourselves is: we know we have data, but what can we do with the data to make it more valuable than it currently is? What is the data telling us that we do not already know or make sense of? Furthermore, there is a significant amount of supervised learning that needs to be done to ensure the insights actually offer value, how can we learn to trust the outputs?
Lastly, the economics of going ‘all in’ with AI and machine learning are significant: investment in software, sensors or data acquisition devices, data warehousing, costs of integrating or structuring all the data together then processing it, requires careful consideration in an industry with small profit margins (scraping past fractions under 1 percent). Ultimately, to do this requires a significant outlay without knowledge about AI and machine learning capability, which is a risk.
Our demands are relatively simple. Make our data provide more value than it currently does. We deliver complex multi-disciplinary and million pound projects successfully (and generally on time and budget), our safety records are constantly improving (and market leading). What can the data we have offer to us that we do not already know? With the decades of experience we have within our businesses, this is where the challenge lies. That said, we still have a way to go to fully establish lessons learned and spotting trends and repeated mistakes.
What will our businesses need to invest in to be able to gather new data to provide insights we did not know we needed? Is the reward worth the risk of investment?
Merely exploring this is a very bold first step for our industry. We would have to give an external party access to all our project data (be it good or bad). They would have to give our project or programme planning a shakedown with several terabytes of data. The outputs may well be positive, or uncover some uncomfortable truths, or provide insights that we thought to be good that are actually the polar opposite.
If this revolution succeeds, the information that AI, sensors, or Big Data could provide some spectacular business intelligence and insights that were never known, creating a paradigm shift in behaviours in how we deliver projects as well as enhance our learning and knowledge, indeed, in how we run our businesses. If this revolution fails, just like the dreadful incident of Elaine Herzberg’s death owing to the trial of an autonomous car, it will set our industry back several years entrenching further scepticism.