THE IMPACT OF ARTIFICIAL INTELLIGENCE ON

CONSTRUCTION COSTING PRACTICE

Kudirat Ayinla1, Abdullahi Saka3 Rafiu Seidu2, and Upeksha Madanayake4
1&2
School of the Built Environment and Architecture, London South Bank University, Borough Road,
London, SE1 0AA, UK
3
School of Built Environment, Engineering and Computing, Leeds Beckett University, Leeds, LS2 8AG, UK
4
School of Construction Management and Engineering, University of Reading, Chancellors Way,
Reading, RG6 6AG, UK

Cost estimation is a crucial process in the construction sector as the efficiency of the
overall project cost serves as one metric in determining project success. Prevailing
traditional approach suffers from human subjectivity and bias which affect accuracy.
With the development and adoption of Artificial Intelligence (AI) such as the use of
machine learning (ML) and deep learning (DL) algorithms, the construction industry
is experiencing brisk technological change and new ways of working, particularly in
terms of cost predictions and estimations. However, the application of AI is still in its
infancy and the industry still prioritises traditional cost modelling approaches in
determining early estimates. This research explores the application of the various ML
methods for costing and assesses their usage and application in the costing practice
via an exploratory critical review. Findings indicate that ML algorithms would
improve the accuracy and efficiency of costing practice but cannot replace the
professionals and data availability.

Keywords: AI; artificial neural network; cost estimating; machine learning

INTRODUCTION
Every construction project is faced with a series of risks and uncertainties at the early
design stages and throughout the project's life cycle. One of the major contributors to
this challenge is cost estimation inaccuracies (Alqahtani and Whyte 2016), whereby
lack of accurate cost data leads to failure to realise set project objectives. Therefore,
cost is a major criterion in decision-making throughout the life cycle of a project
(Juszczyk 2017, Elhegazy et al., 2022). Cost estimation is argued as one of the most
important preliminary processes in a construction project (Elfaki et al., 2014) and it
involves the prediction of the cost required to perform the work within project the
scope. At the early stage of a project, where the scope of the project is uncertain and
involves lots of ambiguities, it is particularly challenging to obtain input data for cost
estimation. Therefore, the impact of overestimating or underestimating may lead to
resource allocation challenges and cost overrun (Hashemi et al., 2020). The accuracy
and comprehensiveness of cost estimation at this stage are perceived as delicate issues
that can be easily influenced by various parameters. Meanwhile, this crucial task of

1
ayinlak@lsbu.ac.uk

Ayinla, K, Saka, A, Seidu, R and Madanayake, U (2023) The Impact of Artificial Intelligence
on Construction Costing Practice In: Tutesigensi, A and Neilson, C J (Eds) Proceedings of the
39th Annual ARCOM Conference, 4-6 September 2023, University of Leeds, Leeds, UK,
Association of Researchers in Construction Management, 65-74.
Ayinla, Saka, Seidu, and Madanayake

predicting cost and generating accurate estimates has in the past depended on the
expertise of human professionals, despite the notion that an expert is prone to
subjectivity and unconscious bias which can influence the result (Elfaki et al., 2014,
Elmousalami 2021). Traditionally, this process depends on the know-how of an
estimator and assumptions are made based on experience and comparison (Bodendorf
et al., 2021). Therefore, it is argued that the level of accuracy required in the
estimating process is largely impossible to achieve manually (Elfaki et al., 2014).
The importance of the use of intelligent techniques in dealing with the challenges
faced with cost estimating in the construction sector has been overemphasised since
intelligent costing methods have the potential to significantly reduce effort and time
(Bodendorf et al., 2021). The implementation of machine learning (ML), a subgroup
of artificial intelligence (AI) in the construction industry is transforming project
delivery and redefining tasks executed by construction professionals and has the
potential to shape construction delivery processes (Xu et al., 2021). ML focuses on
mimicking human intelligence and is generally described as the ability of a computer
to learn without being explicitly programmed because of extracting patterns from
historical data (Akinosho et al., 2020). This aspect of AI investigates the work and
composition of algorithms which can take advantage of and create assumptions about
data thus enabling computers to make decisions, recognize speech, and visualize in 3D
(Ford 2015). There has been a growing interest in ML research particularly deep
learning (DL), a branch of ML, due to its capabilities in automating construction
processes and improving productivity and performance (Xu et al., 2021), with some
studies focused on how construction processes can benefit from digitisation and AI
(Adesi et al., 2018).
While machine and deep learning are arguably in their infancy in the construction
sector (Hong et al., 2020, Xu et al., 2021, Ang 2022), they present opportunities in
addressing challenges with early cost prediction (Ang 2022). AI techniques are now
being considered as a key solution in handling the ambiguity and challenges with cost
estimation in construction projects (Elhegazy et al., 2022). Therefore, this study
explores the application of ML in relation to the practice of construction cost
estimating throughout a project lifecycle and critically analyses the impact of ML
techniques on practices of cost estimating and the challenges and opportunities
presented. The outcome of this study will provide necessary information on the
current use of various AI techniques and their application in costing practice in the
construction sector and provide information on strategic future directions on these
practices and how to harness emerging opportunities that AI presents.
METHOD
This study employs an exploratory research approach through systematic
identification of publications on the theme of Artificial Intelligence techniques for
costing practice in the construction industry. Search queries on AI techniques
('Machine learning', OR 'Artificial intelligence' OR 'Artificial Neural Network'),
costing practice ('Costing' OR 'Estimating' OR 'Cost modelling'), and construction
industry ('Construction Industry' OR 'AEC industry' OR Architecture Construction
and Engineering*') were developed and searched on Scopus, Web of Science and
Google scholar. This is because these databases house relevant publications and have
been employed in similar studies. Inclusion criteria include publications in English
language and are limited to ‘Construction Building Technology’ and 'Engineering
civil' categories without year limitations and document type restrictions. The final

66
 Impact of AI on Construction Costing Practice

publications were critically reviewed to identify and aggregate key findings and lesson
learnt from these studies which forms the basis of the output of this research.
LITERATURE REVIEW
Evolution of Construction Cost Modelling and Estimating Practice
There are various methods and models for generating the cost of a product. The
suitability of each model is often dependent on the type of project, the information
required for completing the cost estimate, and the field of application. Particularly in
the construction field, the use of these cost models could be a question of what phase
the project is and what data is available at that point in time (Günaydin and Doǧan
2004). Niazi et al., 2006) gave a comprehensive classification system for construction
cost modelling methods using a 2-category classification namely the quantitative and
qualitative approaches (Figure 1). This classification simply groups cost modelling
techniques based on the level of information required to generate an estimate. For
instance, the parametric and analytical methods which require some form of
computational analysis and requires a lower level of granularity to derive the cost falls
into the quantitative category, while the more subjective method such as intuitive and
analogical methods are grouped under the qualitative approach. The basis of use for
these categories of cost modelling methods will be reviewed next.

Figure 1: Classification of cost modelling techniques

Progression of the cost modelling practices.
The prediction of project cost received great interest during the early 1970s to the late
1980s in the construction industry because of the need for more accurate estimation
due to the capital value of construction projects and the level of uncertainties involved
throughout a project life cycle. Although few models were already in existence before
this time, the early cost models were criticised for being less value-driven because of
their failure to account for future uncertainties in construction and their inability to
generate reliable cost estimates (Khosrowshahi and Kakat 1996, Yaman and Tas
2007). Accordingly, the historical development of cost-estimating techniques/tools
have progressed from three stages in the construction industry: first, second and third
generations, however given the recent technology drive in the construction sector, a
fourth category is now necessary. As outlined in Table 1, the first-generation
techniques are based majorly on building functional cost analysis approaches and

67
Ayinla, Saka, Seidu, and Madanayake

consist of resource-based costing (RBC), also known as elemental-based costing

(EBC) and activity-based costing (ABC) methods. The RBC method is the most
widely used method in the construction sector and forms the basis of construction cost
estimating guides for professionals as seen in the RICS new rules of measurements
(RICS 2021). The ABC approach, on the other hand, presents an advancement to the
RBC method in terms of its ability to accurately trace the cost/unit of products, thus
improving the accuracy of the cost data. However, both these first-generation models
(RBC and ABC) have limitations in the aspect of cost prediction (Khosrowshahi and
Kakat 1996). This led to the development of other tools to fill this gap in the early
70s.
The parametric method is based on statistical approaches such as regression analysis
(linear regression and multiple regression models) involving the use of historical data
to predict cost. These methods became popular in the late 1970s and are described as
second-generation methods. This method is considered powerful at the feasibility
stage of a project due to its ease, and speed of application (Chou 2011). However, the
accuracy of the method when relationships are non-linear has been an issue as they are
augured to undermine the role of many variable cost drivers that influence a project
cost. Third-generation cost modelling tools used for construction projects experienced
usage at the beginning of the 1980s. These techniques are based on simulations and
risk models such as the Monte Carlo simulation and are initiated as the field of project
management continues to expand in the sector. The Monte Carlo simulation method
significantly reduces the risk associated with cost estimates using a range estimating
approach where estimators determine the minimum, most likely, and maximum
possible cost (Chou et al., 2009) and the probability of exceeding the ‘most likely’
estimate.
More novel approaches have since arrived after the simulation method because of the
application of artificial intelligence and its implementation into cost modelling.
Artificial intelligence (AI) approaches such as artificial neural networks (ANN),
expert systems (ES) and case-based reasoning (CBR) models have been investigated
since the late 1980s (Kim and Shim 2014) and classified as the fourth-generation
models. These methods imitate the human brain function by learning from previous
experience to predict cost and produce very accurate results as well as reducing the
time for estimation. The recent technological advancement in the construction sector
over the last decades has propagated the use and implementation of Building
Information Modelling (BIM), which can be regarded as the fifth-generation
approach. However, this method uses the first-generation RBC/EBC techniques and
differs in that the quantity take-offs and estimation are now automated and supported
by BIM applications (Wu et al., 2014). Measurements are automatically taken off
from digital models of a building, which traditionally has been very time-consuming.
However, there is still a limitation to this method as BIM-based cost estimations are
generated only based on the information in the BIM model while disregarding
possible other external factors (Abioye et al., 2021). More recently, there is ongoing
research on the integration of the combination of AI and BIM to create more accurate
estimates through the application of AI-based prediction models in the estimation and
predictions of construction costs (Ang 2022). Therefore, the next section will focus
on the progress with the use of AI-based approaches in cost-estimating practice and
the benefits and challenges they present.

68
 Impact of AI on Construction Costing Practice

Table 1: Progressive development of Cost Estimating Practices

Impact of ML Techniques on Costing Practice

To develop machines that can simulate human cognitive mechanisms (machines),
different AI methods have been developed. AI systems utilise different inter-
connected sensors for data collection through a process known as data fusion to
integrate and detect possible inferences and characterisations from the data (Pan and
Zhang 2021). Machine learning (ML) is concerned with the design of computer
programs and algorithms that are capable of cognitive skills and capable of reaching
decisions which are traditionally regarded as human skills (Abioye et al., 2021, Xu et
al., 2021, Oluleye et al., 2023). ML has 3 categories (i) supervised learning (ii)
unsupervised learning (iii) reinforcement. Other studies have also used classifications
based on (i) shallow learning which uses a single layer of neural network nodes (ii)
deep learning which uses multiple layers to process large amounts of training data (Xu

69
Ayinla, Saka, Seidu, and Madanayake

et al., 2021). Deep learning (DL) being the current state-of-the-art in ML has been
proven to provide more accurate predictions than conventional ML techniques.
There has been a growing interest in research on ML in the AEC industry which is
partly because these technologies can play an important role in processing the 'huge'
amount of data being generated and used by construction professionals during a
project lifecycle and can be attributed to their potential impact on the cloud-based
computing technologies used within the industry. The idea of AI technologies
implementation in the AEC industry is to improve the industry's productivity and
efficiency to support the complexities in factors such as varying roles, uncertainty in
environmental hazards and others (Pan and Zhang 2021, Oluleye et al., 2023). It is
expected that the development of ML and DL will reshape the whole costing practice,
however, the use of human-based methods is still common in the construction costing
practice. ML techniques could be employed to solve classification or regression
problems. Costing practice that deals with regression and ML algorithms such as
ANN, Logistic Regression, Support Vector Regression (SVR), and Deep neural
networks (DNN) can be leveraged. The following section presents a brief discussion
of the algorithms and studies that have employed them for costing practice in the
extant literature.
Artificial Neural Network (ANN)
Artificial neural network (ANN) is one of the many algorithms of machine learning
which models biological learning processes by computers (Hashemi et al., 2020), and
is commonly used for cost prediction/forecasting. ANN has been applied to mimic the
human system of information processing and can predict the cost of relevant
construction tasks. Its application in the construction sector is well documented in
existing literature. For instance, Aibinu et al., (2011) predicted the cost of pre-tender
cost estimates using ANN algorithm. Similarly, Bala et al., 2014) employed ANN for
predicting the cost of institutional buildings. The method has also been used for life
cycle cost modelling for buildings (Alqahtani and Whyte 2016) and for predicting the
cost of composite flooring systems for multistorey buildings (Elhegazy et al., 2022).
The main limitation of ANN like any other ML algorithm is the need for adequate cost
data from past projects to prevent overfitting of the ANN models (Elmousalami 2021).
Furthermore, the 'black box' nature of ANN models makes it difficult for the
stakeholders to understand how the predictions were made.
Case-Based Reasoning (CBR)
CBR is a technique that makes use of the information contained in past cases (i.e.,
previous projects) to generate cost estimates. This is a data mining technique which
tends to remember information and uses the solution implemented for similar projects
in solving new problems (Ji et al., 2011). The best matching example like the project
at hand is determined to cost of the new project (Niazi et al., 2006). CBR has
experienced wide application in construction costing practice such as for modelling
the cost of new construction projects (An et al., 2007, Ji et al., 2011, Ahn et al., 2020).
Information on previous projects is usually stored in a database and the characteristics
that match the specification of the new project (based on percentage similarity score)
while taking note of changes in systems. There have been well-documented efforts in
improving the result of CBR (Ahn et al., 2020). However, there are still challenges
with the attribution of weight values (Ji et al., 2011).

70
 Impact of AI on Construction Costing Practice

Regression Algorithm (R)

There are different regression algorithms that could be employed for costing practice
in the construction industry such as single learners and ensembles. Single-learner
algorithms include linear regression, multiple linear regression, polynomial
regression, decision tree regression and support vector regression. On the other hand,
ensemble algorithms include random forest regression, gradient boosting regression
and Bayesian regression. Ensemble algorithms do outperform single learners and
single algorithm such as Support Vector Regression (SVR) has an advantage in self-
learning and high performance in generalisation (Kim et al., 2013). The limitations of
these algorithms are linearity assumption, overfitting, underfitting, sensitivity to
outliers, extrapolation, multicollinearity, and difficulty in handling categorical
variables (Géron 2022).
Deep Neural Networks (DNNs)
Deep neural network (DNN) provides more depth to a standard neural network and
are typically trained to model complex non-linear relationships through their ability to
extract unique features (Akinosho et al., 2020) Diverse DNN algorithms can be
leveraged for costing practice in the construction industry. For instance,
Convolutional Neural Network (CNN) can be employed for automated quantity
estimation from images and blueprints, Recurrent Neural Networks (RNN) can be
employed for time series analysis in construction cost forecasting, Generative
Adversarial Networks (GAN) can generate synthetic cost data based on past historical
data to improve accuracy; and lastly, transformer networks like ChatGPT which are
large language model can be prompt with textual specification and project description
to automate cost estimation (Saka et al., 2023). However, the use of DNNs in costing
practice is limited because of the data and expertise needed in deploying these models
for costing. One instance found in published journals is Wang et al., (2022)where it
was used to determine the effects of economic factors on construction costs for public
school projects. This technique, therefore, needs more practical applications in
constriction costing practices.
CONCLUSIONS
The practice of construction cost estimating continues to be a major area of interest to
construction researchers due to the importance of accurate cost estimation in decision-
making for construction projects. Application of AI such as machine learning
algorithms hold immeasurable benefits for costing practice because of their predictive
strengths, capability to infer complex relationships and accuracy. However, the
findings of this study revealed that the use of ML algorithms for costing practice is
limited and majorly in the early estimating stage. There is limited deployment of
these algorithms in tendering and construction phase of the project because of the
unavailability of data to train the model and 'black box' nature of AI which affects
trust.
There is a need for quality and structured data to improve the accuracy of the ML
models and adequate size to prevent overfitting and underfitting of the models. The
study highlighted that although the current applications are limited, there are
opportunities for deploying ANN, regression models, and DNNs for costing practice
in the construction industry. These ML algorithms would improve the accuracy and
efficiency of the costing and estimating practice but cannot replace the professionals.
Domain knowledge is required in fitting of the model - determining the right
predictors and understanding the data - explaining and deploying the models in

71
Ayinla, Saka, Seidu, and Madanayake

practice. As such, it is important for professionals to leverage these algorithms and

not see them as competitors and usurpers.
Lastly, this study is a critical review on the impact of AI on costing practice and is
limited by the size of publications employed for the review. Further study would
employ a systematic review to evaluate the impact of ML on costing practice whilst
highlighting the challenges, and opportunities and providing a case validation.
REFERENCES
Abioye, S O, Oyedele, L O, Akanbi, L, Ajayi, A, Davila Delgado, J M, Bilal, M, Akinade, O
O and Ahmed, A (2021) Artificial intelligence in the construction industry: A review
of present status, opportunities and future challenges, Journal of Building
Engineering, 44, 103299.
Adesi, M, Murphy, R and Kehily, D (2018) The role of digitisation in the strategic planning
process of Irish quantity surveying practices. In: Gorse, C and Neilson, C J
(Eds.), Proceedings 34th Annual ARCOM Conference, 3-5 September 2018, Queen’s
University, Belfast, UK. Association of Researchers in Construction Management,
250–259.
Ahn, S J, Han, S U and Al-Hussein, M (2020) Improvement of transportation cost estimation
for prefabricated construction using geo-fence-based large-scale GPS data feature
extraction and support vector regression, Advanced Engineering Informatics, 43,
101012.
Aibinu, A, Dassanayake, D and Chau, T (2011) Use of artificial intelligence to predict the
accuracy of pre-tender building cost estimate, In: Management and Innovation for a
Sustainable Built Environment, Amsterdam: CIB.
Akinosho, T D, Oyedele, L O, Bilal, M, Ajayi, A O, Delgado, M D, Akinade, O O and
Ahmed, A A (2020) Deep learning in the construction industry: A review of present
status and future innovations, Journal of Building Engineering, 32, 101827.
Akintoye, A and Fitzgerald, E (2000) A survey of current cost estimating practices in the UK,
Construction Management and Economics, 18(2), 161-172.
Alqahtani, A and Whyte, A (2016) Estimation of life-cycle costs of buildings: regression vs
artificial neural network, Built Environment Project and Asset Management, 6(1), 30-
43.
An, S H, Kim, G H and Kang, K I (2007) A case-based reasoning cost estimating model using
experience by analytic hierarchy process, Building and Environment, 42(7), 2573-
2579.
Ang, Y Q (2022) Data science and machine learning for quantity surveying: A brief
perspective, In: Proceedings of the PAQS Congress, Singapore: Singapore Institute of
Surveyors and Valuers, 5-16.
Ayachit, A C, Attarde, P M and Kulkarni, S (2014) Activity based costing in construction
projects, International Journal of Advanced Engineering Research and Studies, 4(1),
9-14.
Bala, K, Bustani, S A and Waziri, B S (2014) A computer-based cost prediction model for
institutional building projects in Nigeria an artificial neural network approach,
Journal of Engineering, Design and Technology, 12(4), 518-529.
Bodendorf, F, Merkl, P and Franke, J (2021) Intelligent cost estimation by machine learning
in supply management: A structured literature review, Computers and Industrial
Engineering, 160, 107601.

72
 Impact of AI on Construction Costing Practice

Chou, J S (2011) Cost simulation in an item-based project involving construction engineering

and management, International Journal of Project Management, 29(6), 706-717.
Chou, J S, Yang, I T and Chong, W K (2009) Probabilistic simulation for developing
likelihood distribution of engineering project cost, Automation in Construction, 18(5),
570-577.
Elfaki, A O, Alatawi, S and Abushandi, E (2014) Using intelligent techniques in construction
project cost estimation: 10-Year survey, Advances in Civil Engineering, 107926.
Elhegazy, H, Chakraborty, D, Elzarka, H, Ebid, A M, Mahdi, I M, Aboul Haggag, S Y and
Abdel Rashid, I (2022) artificial intelligence for developing accurate preliminary cost
estimates for composite flooring systems of multi-storey buildings, Journal of Asian
Architecture and Building Engineering, 21(1), 120-132.
Elmousalami, H H (2021) Comparison of artificial intelligence techniques for project
conceptual cost prediction: A case study and comparative analysis, IEEE
Transactions on Engineering Management, 68(1), 183-196.
Géron, A (2022) Hands-on Machine Learning with Scikit-Learn, Keras and Tensorflow 3rd
Edition, New York: O’Reilly Media, Inc.
Günaydin, H M and Doǧan, S Z (2004) A neural network approach for early cost estimation
of structural systems of buildings, International Journal of Project Management,
22(7), 595-602.
Hashemi, S T, Ebadati, O M and Kaur, H (2020) Cost estimation and prediction in
construction projects: A systematic review on machine learning techniques, SN
Applied Sciences 2, 1703.
Hong, T, Wang, Z, Luo, X and Zhang, W (2020) State-of-the-art on research and applications
of machine learning in the building life cycle, Energy and Buildings, 212, 109831.
Ji, S H, Park, M and Lee, H S (2011) Cost estimation model for building projects using case-
based reasoning, Canadian Journal of Civil Engineering, 38(5), 570-581.
Juszczyk, M (2017) The challenges of nonparametric cost estimation of construction work
with the use of artificial intelligence tools, Procedia Engineering, 196, 415-422.
Khosrowshahi, U K F and Kakat, A P (1996) Estimation of Project Total Cost and Duration
for Housing Projects in the UK, Building and Environment, 31(4), 373-383.
Niazi, A, Dai, J S, Balabani, S and Seneviratne, L (2006) Product cost estimation: Technique
classification and methodology review, Journal of Manufacturing Science and
Engineering, 128(2).
Oluleye, B I, Chan, D W M and Antwi-Afari, P (2023) Adopting Artificial Intelligence for
enhancing the implementation of systemic circularity in the construction industry: A
critical review, Sustainable Production and Consumption, 35, 509-524.
Pan, Y and Zhang, L (2021) A BIM-data mining integrated digital twin framework for
advanced project management, Automation in Construction, 124, 103564.
RICS (2021) New Rules of Measurement: Detailed Measurement for Building Works 2nd
Edition, London: Royal Institution of Chartered Surveyors (RICS).
Saka, A B, Oyedele, L O, Akanbi, L A, Ganiyu, S A, Chan, D W M and Bello, S A (2023)
Conversational artificial intelligence in the AEC industry: A review of present status,
challenges and opportunities, Advanced Engineering Informatics, 55, 101869.
Wang, R, Asghari, V, Cheung, C M, Hsu, S C and Lee, C J (2022) Assessing effects of
economic factors on construction cost estimation using deep neural networks,
Automation in Construction, 134, 104080.

73
Ayinla, Saka, Seidu, and Madanayake

Wu, S, Wood, G and Ginige, K (2014) A technical review of BIM based cost estimating in
UK quantity surveying practice, standards and tools, Journal of Information
Technology in Construction, 19, 534-563.
Xu, Y, Zhou, Y, Sekula, P and Ding, L (2021) Machine learning in construction: From
shallow to deep learning, Developments in the Built Environment, 6, 100045.
Yaman, H and Tas, E (2007) A building cost estimation model based on functional elements,
ITU Journal of Faculty of Architecture, 4(1), 73-87.

74