General Purpose Technologies and the Construction Industry
The modern construction industry had its roots in the take-off of industrialization in the early nineteenth century, and there was a comparable period of rapid, disruptive technological development not unlike the present one in the late nineteenth century.
Between 1860 and 1900 building and construction was restructured as an industry by the rise of large, international contractors, and project management and delivery was reorganized around steam powered machinery and equipment. Major projects like the Suez Canal, railways, tunnels and the new factories for mass production were typically built by new, global European contractors employing workers from around the world on their projects. These projects also required a new organizational form that integrated components, systems and processes.
In materials, the disruptive new technologies of steel, glass and concrete, which came together in the last decades of the century, led to fundamental changes in both processes and products, along Peters (1996) three dimensions of industry development: industrialization, mechanization and organization. Over a hundred years later construction is a mature system, based around standards and professional roles, with a high degree of technological lock in due to the age of the system. The ‘embeddedness’ of the construction technological system is found across the various combinations of the complex array of professional institutes and organizations, trade and industry associations, government regulations and licensing, standards and codes, and insurance and finance providers and regulators.
The impacts of new technology on a mature technological system like the construction industry are often thought to be gradual, changing industry practice over time without significantly affecting industry structure or dynamics. This was the case for twentieth century General Purpose Technologies like electricity, computers and the internal combustion engine. These became universal without significantly restructuring and reorganising construction in the way steam powered mechanization did, because they essentially upgraded existing capabilities. At the starting point for a cycle of development is a new GPT, then industries and products evolve and develop as the underlying knowledge base and technological capabilities increase and become more complex.. If, after a period of development, this GPT gives a technological shock to an existing system of production, it leads to a transition period where the firms involved have to adjust to a new business environment, which in turn leads to a restructuring and consolidation of the industry. This is what happened to construction in the second half of the nineteenth century, with iron-framed and steel-reinforced concrete buildings the industry had to not only master the use of these new materials, but also develop the processes and project management skills the new technology required. With electricity, computers and the internal combustion engine in the twentieth century, the construction industry adopted these new GPTs and used them to improve efficiency, but they did not require a major change in the form of industry organization that had emerged during industrialization and mechanization in the nineteenth century.
How and why a new technology spreads through the economy and society is determined by many factors, however studies of historical cases such as tractors, electricity, TV and phones have given good examples of technology diffusion and its dynamics. A GPT takes time to diffuse through the economy because parallel changes in forms of organization, methods of production and patterns of consumption are required, and these are not decisions firms and households make quickly or easily. Studies on the introduction of new technologies found it takes 15 to 30 years for a new technology to reach 90 percent of its potential market, for example electrification in the US, which took 30 years from 1900 because of the fundamental changes industry and households needed to make to take advantage of electrical power. Another example is how the tractor displaced horses and mules in US agriculture between 1910 and 1960. Horses and mules declined from about 26 million in 1920 to about 3 million by 1960, while the number of tractors rose from zero to 4.5 million by 1960. One reason for the slow spread of tractors was the incremental innovation needed to increase their reliability. A second was an increase in farm wages after 1940. The relative price of labour and mechanization has been found to be the most significant factor in technological innovation, diffusion and automation of work.
How firms utilise technological capabilities differentiates them within a diverse, location-based technological system. It is widely recognised there are differences between industries in the way that technology is adopted, adapted and applied, but differences within industries generally get less attention. The technology adoption literature discusses rank effects, which are the different individual characteristics of firms such as their size, and how they affect the rate and extent of adoption of new technologies, and the effects of competitive dynamics, which is how the adoption of new technology by one company in an industry influences the adoption of technology by other companies in that industry. For building and construction this is significant, not only because of the number of small and medium size firms, but because of the size and reach of the major firms. A global contractor might have over 50,000 employees, suppliers of basic materials and sophisticated components are large multinational or multilocational industrial firms, many of these firms are publicly listed, and so on. These firms have the management and financial resources required to invest in twenty-first century technology, if and when they decide to do so. The issue may be the ability of incumbent firms to capture knowledge externalities, adopt new technologies, and adapt to the impacts of emerging technologies and their requirements.
Importantly, there is a class of more nimble, faster growing firms that have been identified as technology leaders, some of which are incumbents but often are not. Andrews et al. (2015) called these ‘frontier firms’, or firms pushing at the technological frontier through experimentation and development. Frontier firms bring with them radical new production technologies that rely in various ways on smart machines, like the three studied by Hall et al. (2019) and firms like Katerra, Esko, FBR and Daqri (from Table 3). Those firms are new entrants, but incumbents are also on the frontier. Examples are Trimble and Autodesk, Skanska embedding wireless sensors in buildings, Arup’s data collection systems and Atkins water infrastructure design system.
The technological frontier
The construction technological system is wide and diverse, and the various parts of the digital construction technological system are in various stages of development (Gruska et al. 2017). There are many possible futures that could unfold over the next few decades, recent industry scenarios for AI include Agarwal et al. (2016), WEF/BCG (2017) and Quezada et al. (2016), but there is little probability of some breakthrough technology that leads to some different, new industry. Instead, development of AI and associated digital fabrication and production technologies will more likely reshape the existing industry, led by fundamental changes in demand (the function, type and number of buildings), design (the opportunities new materials offer), and delivery (through project management). The fourth industrial revolution has already affected demand for structures like renewable energy sources and buildings like data centres, warehouses and retail, ‘dark’ kitchens and supermarkets for online delivery services. Some of these buildings and structures already use forms of applied AI in their management and operation.
At the end of the second decade of the twenty-first century, automation technology is at the point where intelligent machines are moving from operating comfortably in controlled environments, in manufacturing or social media, to unpredictable environments, like driving a car or truck. In many cases, like remotely controlled and autonomous trucks and trains on mining sites, the operations are run as a partnership between humans and machines, or as Brynjolfsson and McAfee (2014) put it “running with the machines not against them”. These innovations might reasonably be expected to affect site processes and project organization, as concrete and steam power did in the past. Table 1 has examples of where the technological frontier is in 2020 for plant and equipment, also for construction materials, as an indication of the range and extent of this wave of innovations. Missing from these lists is smart contracts using blockchain.
Invention and innovation based around BIM, digital twins, digital fabrication and advanced manufacturing technology, is starting to fundamentally affect the production system through economies of scale. Over time this will alter the balance between on-site and off-site production of building modules and components, and how they are handled, assembled and integrated. The combination of BIM and digital fabrication could be transformational if it allows on-site production of building components, fundamentally altering the economies of scale in the industry. Mass production will always have a role, but market niches currently occupied by some manufacturing firms may disappear, replaced by new production technologies based on digital fabrication and online design databases. Adding new materials to the fabrication palette through molecular design and engineering may be significant, or other new materials, or upgraded versions of existing structural materials. Combining robotic and automated machinery with digital fabrication and standardized parts opens up many possibilities. Exoskeletons combine human skill with machine strength.
While firms involved in construction of the built environment are facing technological advances that will affect many aspects of the technological system, this is a process that happens over years and decades. Lipsey et al. (2005: p. 211) found “the gestation period of individual GPTs does not seem to have shortened much since the industrial revolution” and it takes 50 years between invention of a GPT and its use becoming widespread (their examples were discovering the double-helix and biotechnology, the dynamo and electricity, and the first electronic computers in the 1940s). For the tractor and electrification cases used above, starting from the date of invention of the internal combustion engine and dynamo would add around three decades to those timelines.
In fact, how long a transition to a new technological system built on automation and digital fabrication coordinated by AI takes is unknown. While machines can replicate individual tasks, integrating different capabilities into solutions where everything works together is another matter. Combining a range of technologies is needed for workplace automation, but solving specific problems involves specific and organizational technical challenges, and once the technical feasibility has been resolved and the technologies become commercially available it can take many years before they are adopted. Importantly, this suggests there will be many new jobs in construction over coming years, for project information managers, BIM supervisors, integration specialists and other fourth industrial revolution roles. Because these jobs will be primarily on new projects, they will not quickly replace the many existing jobs in the industry required to maintain the built environment. McKinsey (MGI 2017) sees construction as an industry where AI does not significantly reduce the number of jobs. In their paper on ‘Construction 4.0’ de Soto et al. (2019) conclude: “there will be a time in which conventional construction and robotic technologies will coexist, leading to a higher job variability and new roles.”
Nevertheless, the technological frontier is moving again, and new construction projects will generally utilise the most cost-effective technology. Current AI technology provides services such as GPS navigation and trip planning, spam filters, language recognition and translation, credit checks and fraud alerts, book and music recommendations, and energy management systems. It is being used in law, transport, education, healthcare and security, and for engineering, economic and scientific modelling. Advanced manufacturing is almost entirely automated. As expected with a new GPT, there are many new applications under development (Mitchell 2019).
The next level of AI envisages those capabilities extended in the near future to a group of intelligent machines that have been individually trained to collect and manage data from the stages of a construction project, and that outsourced business processes can provide such data for intelligent machines, supervised by users and helping them manage complicated processes. An AI acting as an overall project data manager could integrate the data from many sources to continually update a project’s schedule, work plan and cost estimates, matching progress and performance to iterate those plans for the project’s managers. This AI assists users’ decision-making by generating and evaluating options. Such a system would be operated by a voice activated interface, with the progress updates included and access to expert systems for specialist areas provided. It would generate design options and provide full visualisation of a shared BIM model linked to the schedule and site work plan. There would be real-time supply chain data on fabrication and logistics through cloud-based platforms. The AI can iterate the schedule and cost plans for a project, based on that data, allowing the project management team to match performance with plans, in real-time, for every aspect of a project. The data required for the coordination and management role of intelligent machines can come from widespread use of standardized, outsourced cloud-based business processes. That data then becomes a series of training sets needed for deep learning, the current level of AI technology.
References Agarwal, R., Chandrasekaran, S. and Sridhar, S. 2016. Imagining Construction’s Digital Future, McKinsey & Co Andrews, D., C. Criscuolo and P. N. Gal, 2015. Frontier Firms, Technology Diffusion and Public Policy: Micro Evidence from OECD Countries, OECD Productivity Working Papers, 2015-02, OECD Publishing, Paris. Brynjolfsson, E. and McAfee, A. 2014. The Second Machine Age: Work, Progress and Prosperity in a Time of Brilliant Technologies, New York: W. W. Norton & Co. de Soto, B. J., Agustí-Juan, I., Joss, S. and Hunhevicz, J. 2019. Implications of Construction 4.0 to the workforce and organizational structures, International Journal of Construction Management, DOI: 10.1080/15623599.2019.1616414 Gruszka, A., Jupp, J. R. and de Valence, G. 2017. Digital Foundations: How Technology is Transforming Australia's Construction Sector, Sydney: Startup Australia. Lipsey, R. G., Carlaw, K. I. and Bekar, C. T. 2005. Economic Transformations: General Purpose Technologies and Long-term Economic Growth, Oxford: Oxford University Press. MGI, 2017. A Future that Works, McKinsey Global Institute. Mitchell, M. 2019. Artificial Intelligence: A Guide for Thinking Humans, New York: Farrar, Straus, and Giroux. Peters, T. F. 1996. Building the Nineteenth Century, Cambridge, Mass. MIT Press. Quezada, G., Bratanova, A., Boughen, N. and Hajowicz, S. 2016. Farsight for Construction: Exploratory scenarios for Queensland’s construction industry to 2036, CSIRO, Australia. WEF/BCG, 2017. Future Scenarios and Implications for the Construction Industry, World Economic Forum and the Boston Consulting Group, Geneva.