INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 942
OPPORTUNITIES AND ADOPTION CHALLENGES OF ARTIFICIAL
INTELLIGENCE IN THE CONSTRUCTION INDUSTRY
Igamova Shaxinya Zikrilloyevna
Associate professor ,PhD Asian International University
Abstract:
Over the past decade, while artificial intelligence (AI) has rapidly transformed
numerous industries, the construction sector has been slow to adopt these advancements.
However, the rise of sophisticated large language models (LLMs) such as OpenAI’s GPT,
Google’s PaLM, and Meta’s Llama has demonstrated significant potential, sparking
widespread global interest. Despite this surge, there remains a lack of research specifically
examining the opportunities and challenges of integrating Generative AI (GenAI) within the
construction industry, resulting in a crucial knowledge gap for both researchers and
practitioners. Addressing this gap is essential to effectively leveraging GenAI during its early
adoption phase in construction. Given GenAI’s remarkable ability to generate human-like
content by learning from existing data, this study explores two key questions: What does the
future hold for GenAI in the construction industry? What are the potential opportunities and
challenges associated with its implementation? To answer these questions, the study conducts
a literature review, assesses industry perspectives using programming-based word cloud and
frequency analysis, and incorporates the authors’ insights. Additionally, the paper proposes a
conceptual framework for GenAI implementation, offers practical recommendations,
highlights future research directions, and establishes a foundational knowledge base to
support further exploration of GenAI in construction, architecture, and engineering
disciplines.
Keywords
: generative AI; implementation framework; construction; AEC; GPT; LLM;
PaLM; Llama; fine-tuning.
Introduction
Over the past four decades, machine learning (ML), particularly deep learning based on
artificial neural networks, has advanced significantly, driving transformations across various
industries. Within the construction sector—an industry often lagging in efficiency and
productivity—ML has been instrumental in automating processes. However, widespread
adoption faces challenges, including issues related to data quality management and the lack
of clear guidelines for integrating domain expertise with data-driven insights. These
challenges manifest in three key areas: (1) the disparity between a feature-rich dataset and a
limited number of samples, (2) the trade-off between model accuracy and broad applicability,
and (3) the difficulty of aligning machine learning outputs with industry-specific knowledge.
For example, a construction company may possess extensive data on project features but
only a limited number of actual projects, making it difficult to develop a precise cost
prediction model. Similarly, an organization seeking to forecast project completion times
must balance model accuracy with its ability to generalize across diverse projects.
Additionally, a safety manager using ML to predict fall risks may find that traditional models
fail to account for human factors and unforeseen conditions. These limitations highlight the
INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 943
constraints of conventional machine learning in addressing complex construction industry
challenges. The rapid evolution of artificial intelligence (AI)—which enables machines to
replicate human-like reasoning and decision-making—has led to the emergence of advanced
large language models (LLMs) such as OpenAI’s GPT, Google’s PaLM, and Meta’s Llama.
Generative AI (GenAI), a subset of deep learning, utilizes neural networks and various
learning methods (supervised, unsupervised, and semi-supervised) to generate new content,
including text, images, and audio. LLMs train on extensive datasets, developing statistical
models that allow them to generate coherent outputs based on input prompts. At their core,
transformer-based architectures power GenAI models, processing inputs through encoders
and generating outputs via decoders. GenAI can be classified into four major types: text-to-
text, text-to-image, text-to-video/3D, and text-to-task. Text-to-text models generate text-
based outputs, while text-to-image models create visuals from textual descriptions, often
utilizing diffusion techniques. Text-to-video and text-to-3D models extend these capabilities
to multimedia generation. Meanwhile, text-to-task models perform functions such as
answering questions, retrieving information, making predictions, and executing complex
tasks. Large-scale, pre-trained GenAI models like GPT are designed for adaptability and can
be fine-tuned for various applications, including sentiment analysis, object recognition,
instruction following, and more. In recent decades, AI and ML research in construction has
explored applications such as safety management, cost prediction, schedule optimization,
progress monitoring, quality control, supply chain and logistics management, risk mitigation,
dispute resolution, waste management, sustainability assessment, visualization, and
infrastructure inspection. Furthermore, integrating AI with Building Information Modeling
(BIM) has enhanced information extraction, workflow efficiency, and project management.
The incorporation of robotics and AI has also led to improvements in construction quality,
safety, and project timelines, while mitigating labor shortages. Despite these advancements,
research on the specific applications, opportunities, and adoption barriers of GenAI in
construction remains limited. This gap is likely due to the technology’s recent emergence,
resulting in a slower pace of adoption compared to other industries that have already begun
leveraging its benefits. Given this context, this study aims to answer two key research
questions: (1) What are the current perspectives, evidence, and challenges surrounding
GenAI adoption in construction? and (2) What are the most critical research directions for
future investigations into GenAI’s role in construction? The paper is structured as follows:
Section 2 outlines the research methodology. Section 3 explores different GenAI models and
their relevance to construction. Section 4 synthesizes existing insights, identifies potential
application areas, and presents a conceptual implementation framework. Section 5 discusses
challenges, ranging from technical limitations to industry-specific barriers. Section 6 offers
implementation recommendations and highlights priority research questions. Finally,
concludes with key findings and future considerations for advancing GenAI in construction.
Results and methodology.
To achieve our research objectives—identifying the opportunities and challenges of
Generative AI (GenAI) in construction, developing a conceptual implementation framework,
and outlining future research directions—we followed the research framework illustrated in
Figure 1. Given the limited existing literature on GenAI in construction, we adopted a non-
systematic review approach.
INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 944
Our initial literature search used keywords such as
“Generative AI AND Construction”
,
“Generative AI”
, and
“Large Language Models AND Construction”
in academic
databases like Scopus and Google Scholar. To further expand our sources, we applied the
snowball method, examining key articles and tracing their references and citations to uncover
additional relevant studies.
Figure 1. Research framework
In recent years, researchers have been refining generative AI (GenAI) models to address
industry-specific challenges. The choice of a GenAI model depends on the task at hand, as
different models excel in different applications. There are five major types of GenAI models,
as shown in Figure 2, and ongoing advancements continue to shape the field. Researchers
actively explore novel architectures and methodologies to enhance model capabilities.
The five major types of GenAI models include:
Generative Adversarial Networks (GANs):
Commonly used for image generation,
GANs create realistic images by training a generator and discriminator in a competitive
setting.
Variational Autoencoders (VAEs):
Typically applied in text generation, VAEs learn
the underlying data distribution, enabling the production of grammatically coherent and
meaningful text samples.
INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 945
Autoregressive Models:
These models generate text token by token, conditioning on
previous tokens to ensure coherence and fluency. They are widely used in natural language
processing tasks.
Diffusion Models:
Used primarily in image synthesis, these models begin with noise
and gradually refine images by reversing a diffusion process, leading to high-quality outputs.
Flow-Based Models:
These models transform data into a latent representation,
allowing for creative and diverse content generation in both image and text domains.
Each of these models has unique strengths and limitations. In the following subsections,
we will explore their architectures, operational mechanisms, and constraints. Additionally, we
will examine their relevance to the construction industry, identify existing use cases, and
summarize their key advantages and disadvantages.
Figure 2. GenAI Models.
Recent studies leveraging Large Language Models (LLMs) to solve construction-related
challenges highlight the long-term opportunities of generative AI (GenAI) in the industry.
Some key developments include:BIM-GPT Integration: In 2023, Zheng and Fischer
introduced a BIM-GPT framework that enables LLMs to retrieve, summarize, and answer
questions from Building Information Modeling (BIM) databases. This addresses challenges
in automating complex information extraction from rich BIM models. Despite BIM's growing
adoption, issues like interoperability and standardization remain. Using LLMs to extract and
standardize BIM data can streamline workflows.Automated Construction Scheduling: Prieto
et al. (2023) tested ChatGPT for generating coherent construction schedules, improving
activity sequencing and scope alignment.Safety and Risk Classification: Hasan et al.
developed a BERT-based model to classify injury narratives, identifying risks and hazards in
construction sites. This technique has also been used for detecting contractual risk clauses in
INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 946
construction specifications.Document Analysis and Synthesis: While construction projects
generate extensive documentation (e.g., contracts, reports, and drawings), traditional systems
struggle to utilize this data effectively. GenAI models like ChatGPT and Bard can synthesize
construction documents, answering queries and extracting key information from these data
sources.
The use of robotic systems in construction requires efficient sequence planning.
Traditional mathematical and machine-learning methods have struggled to adapt to dynamic
construction environments. RoboGPT, a model leveraging ChatGPT, was introduced to
enhance robotic assembly sequence planning. However, challenges remain in translating
human language into robot-interpretable instructions.Studies suggest that current LLMs still
face limitations in numerical and physical reasoning, necessitating further improvements for
fully autonomous robotic construction planning.
AI Policy and Regulation
:
The CREATE AI Act (2024) and the National Artificial Intelligence Research Resource
(NAIRR) indicate growing governmental support for AI in construction. NAIRR aims to
expand access to AI resources, fostering innovation in sectors like architecture, engineering,
and construction.
LLMs are being developed under both open-source and closed-source approaches, each
with distinct implications: Promote transparency by providing access to source code, training
data, and model parameters.Enable collaboration, allowing researchers and developers to
enhance models for specific construction applications.Require significant infrastructure and
hosting costs for deployment.Closed-Source LLMs:Are proprietary models restricted to
license holders, limiting external development opportunities.Offer reliable cloud-based
deployment with dedicated resources, ensuring uptime and scalability.Provide better data
privacy by restricting access to training data and algorithms.As GenAI adoption in
construction continues to grow, balancing the trade-offs between model scale, cost,
accessibility, and transparency will be critical for optimizing its impact in the industry.
Conclusion.
The construction industry, historically slow to adopt digital transformation, is now
experiencing a paradigm shift with the emergence of Generative AI (GenAI). This study
explored the current applications, challenges, and future opportunities of GenAI in
construction, demonstrating its potential to enhance productivity, streamline processes, and
improve decision-making.Key advancements, such as BIM-GPT for data extraction, AI-
driven scheduling, risk classification, and robotics integration, highlight the transformative
role of GenAI in addressing industry-specific challenges. However, limitations persist,
including issues related to data availability, model accuracy, domain knowledge integration,
and real-world adaptability. The balance between open-source and closed-source LLMs also
presents strategic trade-offs for organizations seeking to optimize cost, performance, and
transparency.Despite these challenges, government initiatives like the CREATE AI Act and
NAIRR indicate a growing commitment to fostering AI development across sectors,
including construction. To fully leverage GenAI, further research is needed to refine LLM-
INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 947
based automation, improve interoperability in BIM workflows, and enhance AI-driven
decision-making frameworks. Ultimately, GenAI represents a promising frontier for the
construction industry, offering solutions to long-standing inefficiencies. By addressing
existing barriers and investing in innovation, the industry can harness AI’s full potential,
driving increased efficiency, safety, and sustainability in the built environment.
References:
1. Bamgbade, J.; Nawi, M.; Kamaruddeen, A.; Adeleke, A.; Salimon, M. Building
sustainability in the construction industry through f irm capabilities, technology and
business innovativeness: Empirical evidence from Malaysia. Int. J. Constr. Manag. 2019,
23, 1–6. [CrossRef]
2. Zhang, S.; Li, Z.; Ning, X.; Li, L. Gauging the impacts of urbanization on CO2
emissions from the construction industry: Evidence from China. J. Environ. Manag. 2021,
288, 112440. [CrossRef]
3. Mahbub, P.; Goonetilleke, A.; Ayoko, G.; Egodawatta, P.; Yigitcanlar, T. Analysis of
build-up of heavy metals and volatile organics on urban roads in Gold Coast, Australia.
Water Sci. Technol. 2011, 63, 2077–2085. [CrossRef]
4. Javed, A.; Pan, W.; Chen, L.; Zhan, W. A systemic exploration of drivers for and
constraints on construction productivity enhancement. Built Environ. Proj. Asset Manag.
2018, 8, 234–238. [CrossRef]
5. The
Next
Normal
in
Construction.
Available
online:
https://www.mckinsey.com/~{}/media/McKinsey/Industries/Capital%
20Projects%20and%20Infrastructure/Our%20Insights/The%20next%20normal%20in%2
0construction/The-next-normal-in construction.pdf (accessed on 5 May 2021)
6. Young, D.; Panthi, K.; Noor, O. Challenges involved in adopting BIM on the
construction jobsite. Built Environ. 2021, 3, 302–310
7. Adwan, E.; Al-Soufi, A. A review of ICT technology in construction. Int. J. Manag. Inf.
Technol. 2016, 8, 1–10.
8. Yun, J.J.; Lee, D.; Ahn, H.; Park, K.; Yigitcanlar, T. Not deep learning but autonomous
learning of open innovation for sustainable artificial intelligence. Sustainability 2016, 8,
797. [CrossRef]
9. Yigitcanlar, T.; Cugurullo, F. The sustainability of artificial intelligence: An urbanistic
viewpoint from the lens of smart and sustainable cities. Sustainability 2020, 12, 8548.
[CrossRef].
10. Yigitcanlar, T. Greening the artificial intelligence for a sustainable planet: An editorial
commentary. Sustainability 2021,
11. Chien, C.F.; Dauzère-Pérès, S.; Huh, W.T.; Jang, Y.J.; Morrison, J.R. Artificial
intelligence in manufacturing and logistics systems: Algorithms, applications, and case
studies. Residential 2020, 58, 2730–2731. [CrossRef]
12. Grabowska, S.; Saniuk, S. Business models in the industry 4.0 environment: Results of
Web of Science bibliometric analysis. J. Open Innov. Technol. Market Complex. 2022, 8,
19. [CrossRef]
13. Xin, X.; Tu, Y.; Stojanovic, V.; Wang, H.; Shi, K.; He, S.; Pan, T. Online reinforcement
learning multiplayer non-zero sum games of continuous-time Markov jump linear
systems. Appl. Math. Comput. 2022, 412, 126537.
INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 948
14. Qudratova, G. M. (2025). TEXNOLOGIK PARKLARNING MINTAQA
INNOVATSION RIVOJLANISHINI TA'MINLASHDAGI AHAMIYATI. YANGI O
‘ZBEKISTON, YANGI TADQIQOTLAR JURNALI, 2(8), 170-178.
15.
Sodiqova, N. (2025). IQTISODIYOT FANLARINI OʻQITISHDA TALABALAR
TEXNIK TAFAKKURINI RIVOJLANTIRISHNING AMALDAGI HOLATI VA
TAKOMILLASHTIRISH YOʻLLARI. " ПЕДАГОГИЧЕСКАЯ АКМЕОЛОГИЯ"
международный научно-методический журнал, 2(19).
16.
Bahodirovich, K. B. (2025, April). STRUCTURE OF THE CASH FLOWS
STATEMENT. In CONFERENCE OF MODERN SCIENCE & PEDAGOGY (Vol. 1,
No. 1, pp. 325-330).
17.
Алимова, Ш. А. (2025). УСТОЙЧИВЫЕ ЦЕПОЧКИ ПОСТАВОК: ОТ ТРЕНДА К
НЕОБХОДИМОСТИ
РАСШИРЕННАЯ
ВЕРСИЯ.
Modern
Science
and
Research, 4(5), 76-81.
18.
Toshov, M. H. (2025). SАNОАT KОRXОNАLАRIDА MEHNАTGА HАQ TО'LАSH
TIZIMINI BОSHQАRISH. Modern Science and Research, 4(4).
19.
Azimov, B. (2025). METHODS AND MODELS FOR ASSESSING THE SOCIO-
ECONOMIC
EFFICIENCY
OF
REGIONAL
INNOVATION
INFRASTRUCTURE. International Journal of Artificial Intelligence, 1(3), 685-691.
20.
Ikromov, E. I., & Safarova, J. (2025). O’ZBEKISTONDA YASHIL
TADBIRKORLIKNI HUDUDLARDA RIVOJLANTIRISHI ISTIQBOLLARI. Modern
Science and Research, 4(4), 421-428.
21.
Raxmonqulova, N. O. (2025). DEVELOPMENT OF THE DIGITAL ECONOMY ON A
GLOBAL SCALE AND THE EXPERIENCE OF COUNTRIES. SHOKH LIBRARY.
22.
Shadiyev, A. X. (2025). MINTAQANING IJTIMOIY-IQTISODIY RIVOJLANISHINI
BOSHQARISH MEXANIZMINI TAKOMILLASHTIRISH. STUDYING THE
PROGRESS OF SCIENCE AND ITS SHORTCOMINGS, 1(7), 145-150.
23.
Naimova, N. (2025). THE IMPACT OF GLOBALIZATION ON MODERN
ECONOMIC
PROFESSIONS. Journal of Multidisciplinary Sciences
and
Innovations, 1(2), 153-155.
24.
Bazarova, M. (2025). FEATURES OF ASSESSING THE EFFECTIVENESS OF
INNOVATION RISK MANAGEMENT OF AN EDUCATIONAL ORGANIZATION
IN THE PROCESS OF DIGITAL TRANSFORMATION OF ACTIVITIES. Journal of
Multidisciplinary Sciences and Innovations, 1(2), 161-164.
25.
Jumayeva, Z. (2025). THE ROLE OF MICROECONOMIC ANALYSIS IN
ENHANCING ECONOMIC EFFICIENCY THROUGH MARKET EQUILIBRIUM
ANALYSIS. International Journal of Artificial Intelligence, 1(3), 634-637.
26.
Bobojonova, M. (2025). GREEN ENTREPRENEURSHIP IN UZBEKISTAN AND ITS
OPPORTUNITIES. International Journal of Artificial Intelligence, 1(3), 592-595.
27.
Jumayeva, Z. (2025). THE FORMATION OF THE GREEN ECONOMY CONCEPT,
STAGES OF DEVELOPMENT AND ITS RELEVANCE. International Journal of
Artificial Intelligence, 1(3), 262-266.
28.
Ibragimov, A. (2025). TAX POLICY AND IMPACT ON ECONOMIC
DEVELOPMENT. International Journal of Artificial Intelligence, 1(3), 259-261.
29.
Djurayeva, M. (2025). ISSUES OF SMALL BUSINESS AND PRIVATE
ENTREPRENEURSHIP DEVELOPMENT. International Journal of Artificial
Intelligence, 1(3), 596-598.
INTERNATIONAL JOURNAL OF ARTIFICIAL INTELLIGENCE
ISSN: 2692-5206, Impact Factor: 12,23
American Academic publishers, volume 05, issue 05,2025
Journal:
https://www.academicpublishers.org/journals/index.php/ijai
page 949
30.
Umarova, H. (2025). RIVOJLANGAN MAMLAKATLARDA KORXONA
RISKLARINI BOSHQARISH VA BAHOLASH AMALIYOTI TAHLILI. Modern
Science and Research, 4(5), 158-161.
31.
Aslanova,
D.
(2025).
CHALLENGES
OF
IMPLEMENTING
MODERN
MANAGEMENT PRINCIPLES IN THE TOURISM INDUSTRY. Journal of
Multidisciplinary Sciences and Innovations, 1(2), 119-121.
32.
Rajabova, D. (2025). SPECIFIC FEATURES AND FACTORS OF SUSTAINABLE
DEVELOPMENT OF THE INNOVATIVE ENVIRONMENT IN INDUSTRIAL
ENTERPRISES. Journal of Applied Science and Social Science, 1(2), 474-479.
33.
Игамова,
Ш.
З.
(2024).
МЕТOДИЧЕCКИЕ
РЕКOМЕНДАЦИИ
ПO
ФOРМИРOВАНИЮ OРГАНИЗАЦИOННO-ЭКOНOМИЧЕCКOГO МЕХАНИЗМА
OБЕCПЕЧЕНИЯ
ЭФФЕКТИВНOCТИ
ИННOВАЦИOННOГO
развития
ПРЕДПРИЯТИЙ CТРOИТЕЛЬНЫХ МАТЕРИАЛOВ. Gospodarka i Innowacje., 43,
335-340.
34.
Akramova, O. (2025). FOREIGN COUNTRIES IN EXPERIENCE INVESTMENT
ATTRACTIVENESS INCREASE MECHANISMS AND UZBEKISTAN IN
PRACTICE USE OPPORTUNITIES. Journal of Multidisciplinary Sciences and
Innovations, 1(1), 395-398.
35.
Jumayev, B. (2025). BIG DATA: CUSTOMER CREDIT ANALYSIS USING DIGITAL
BANKING DATABASE. International Journal of Artificial Intelligence, 1(2), 1056-
1059.
36.
Gafarova, D. (2025). INNOVATION POLICY OF THE REPUBLIC OF UZBEKISTAN:
ACHIEVEMENTS AND PROSPECTS. Journal of Multidisciplinary Sciences and
Innovations, 1(2), 165-167.
