Conversational chatbot ChatGPT has been used for everything from composing music to writing student essays. Could it have a role in construction? CM asked a panel of experts to pose questions to ChatGPT and test out its capabilities. Read on to find out how the chatbot responded, and what our panel thought of its answers.
Charles Tincknell, efficiencies and systems lead, Willmott Dixon Interiors.
CT: Tell me how the construction industry could benefit from ChatGPT in 100 words.
ChatGPT is a variant of the popular language model GPT, designed to be able to perform tasks through natural language conversations. In the construction industry, ChatGPT could be used to help automate various processes, such as answering frequently asked questions, scheduling appointments or tasks, and providing real-time updates on project progress. Additionally, ChatGPT could be used to assist in the design process by generating detailed specifications and plans based on high-level input from architects and engineers. Overall, ChatGPT has the potential to greatly improve efficiency and accuracy in the construction industry by enabling the automation of tasks and providing accurate and detailed information.
CT: I think the real advance that ChatGPT offers is as a first draft generator for virtually any output that requires a written response. The answers are OK to good, to make them great will be down to the user and the quality of the follow-up interactions with the tool.
ChatGPT is also learning so the answers and outputs will also improve. However, the huge leap ChatGPT brings is that it is incredibly easy to use and this means the power of AI is now really available to all. Best of all, it’s currently free to use, so the only barrier to entry is your own time and inclination.
Sam Stacey, CEO, Stacx International, and senior adviser, Construction Leadership Council.
SS: What is pre-manufactured value in construction?
Pre-manufactured value in construction refers to the value that is added to a product before it is shipped to a job site for final construction or installation. This value includes the cost of materials, labour, and overhead associated with manufacturing or fabricating the product in a factory or other controlled environment. Pre-manufactured value can be a significant component of the overall cost of a construction project, especially when using prefabricated or modular components. By manufacturing these components off-site, it is often possible to reduce labour costs, improve quality control, and reduce the overall construction schedule.
SS: The purpose of building information modelling (BIM) is to provide the right information to the right people at the right time. A useful tool would be a digital assistant that could provide instant answers to questions as they arise – in other words, a chatbot.
Recent work by ChatGPT has made some progress, providing sensible responses to generic queries. Where the real value will come in is complex project-specific queries. These might need real-time answers such as to ‘When will panels XYZ arrive on site?’ Or geometry-related queries such as ‘Show me the setting-out for the light switches in room ABC.’ Use of this sort of AI will require processing of information from common data environments and from dynamic sources such as sensors.
The user interfaces would ideally be more than just text – for example, augmented reality. Through the Transforming Construction Challenge, we sponsored some impressive work led by University of West of England, Conversational-BIM, which addressed some of these demands.
Emma Hooper, head of R&D at Bond Bryan Digital, and Government and Industry Interoperability Group (GIIG) workstream member.
EH: What is IFC in construction?
In construction, IFC (International Finance Corporation) is a set of standards and guidelines for the design, construction, and operation of buildings and infrastructure projects. The IFC standards are used to ensure that projects are designed, built, and operated in a way that is sustainable, environmentally responsible, and economically feasible. The IFC standards cover a wide range of topics, including site planning and design, materials and resources, indoor environmental quality, and energy and water efficiency. These standards are often used in conjunction with other industry standards and guidelines, such as LEED (Leadership in Energy and Environmental Design) and BREEAM (Building Research Establishment Environmental Assessment Method).
EH: The definition provided by ChatGPT is not accurate. IFC stands for Industry Foundation Classes and is governed by the single standard, ISO 16739. It is an off-the-shelf standardised data model and schema specification providing a digital representation of a building or infrastructure facility.
Chatbots work from spotting patterns in information and piecing it together, a bit like how the human brain does. Unlike humans they cannot think laterally, and it can only be as good as the information it has access to.
Like everything in digital construction, we must get the underlying information organised, which will help accelerate accuracy. Data standards, including IFC, are imperative to provide a starting point for a machine to make sense of what it is being asked. AI is not magic; it is all based on logic. If we skip the information part and move on to the next buzz technology – as we often do – and start believing what we read from chatbots without questioning it then we’re going to find ourselves in a dark place.
Neil Thompson, director, digital integration & delivery, Atkins
NT: What is the relationship between data and information in the context of digital twins?
In the context of digital twins, data is the raw material that is used to create the digital twin, while information is the processed and organized form of that data. A digital twin is a virtual representation of a physical object or system, and it is created by collecting and analyzing data about the object or system. This data can come from a variety of sources, such as sensors, monitoring systems, and other data-gathering devices. The data is then processed and organized to create the digital twin, which provides information about the object or system that it represents. The information contained in the digital twin can be used to understand the object or system in greater detail, predict its behaviour, and optimize its performance.
NT: I wanted to see if we can get a meaningful answer that captures the essence of data and its difference to information. It’s given a soundbite, but not something I think will hold water in technical communities. Although, if this bot can reference its sources, I would be happier.
I feel a sense of irony, asking a machine learning system to provide a commentary on the difference between data and information, as it processed data to give me information on my questions.
In summary, I think this helps the layperson and can aid in marketing. Interestingly I can see a path to how one could manipulate the system to provide answers to specific questions! Which is a worry if it is not going to tell you its data sources.
Professor Jacqui Glass, vice dean research for The Bartlett Faculty of the Built Environment, University College London
JG: What are the benefits of self-compacting concrete?
Self-compacting concrete (SCC) is a type of concrete that has a high level of workability and is able to flow and compact under its own weight. It does not require mechanical vibration to achieve full consolidation. There are several benefits to using self-compacting concrete:
1. Improved productivity: Because SCC does not require vibration to consolidate, it can be placed and finished more quickly than conventional concrete, leading to increased productivity on the construction site.
2. Enhanced durability: SCC has a denser structure than conventional concrete, which can improve its durability and resistance to chemical attacks and freeze-thaw cycling.
3. Better surface finish: SCC can produce a smoother and more uniform surface finish than conventional concrete, as it is able to fill and consolidate around complex reinforcement and formwork without leaving voids or honeycombing.
4. Increased design flexibility: SCC can be used in a variety of applications and has the ability to flow into and around complex and irregularly shaped formwork, allowing for increased design flexibility.
5. Improved energy efficiency: The production of SCC requires less energy than the production of conventional concrete, as it does not require vibration for consolidation. This can lead to reduced greenhouse gas emissions and improved energy efficiency.
JG: It was my first time using ChatGPT and I was impressed. Its answers were not 100% perfect, but the speed, succinctness, broad-brush coherence, and match to what I was expecting were genuinely remarkable.
My first queries featured straightforward terms like project management which elicited some reasonable, although rather unsophisticated, responses.
My final choice was a well-established technical term (self-compacting concrete), and I think this demonstrates well the kind of answer it will provide to a basic question. Here’s the ‘pinch of salt’ moment though: ChatGPT answers the question, but it won’t ‘school’ you. If you look at point five, you’ll see that it only notes energy savings through less vibration and does not suggest that you might like to think about the carbon consequences of SCC’s higher cementitious content.
The next time I need a speedy take on a subject, I’ll be happy to use AI to supplement my usual go-to wikis. ChatGPT is not quite capable of taking my job, yet AI tech like this is causing huge ripples in education right now, as we try to come to terms with both its potential to support learning and its unfortunate deployment as a weapon of mass plagiarism.