Applied IT & Engineering

1. Introduction

Infrastructure systems—bridges, highways, tunnels, rail corridors, and public facilities—quietly support nearly every aspect of modern life. They sustain mobility, economic exchange, emergency response, and public safety. Yet the reliability of these systems depends not only on how well they are designed or built, but also on how effectively they are inspected, monitored, and maintained over time. In practice, infrastructure deterioration is often gradual, subtle, and easy to miss until damage becomes serious. That reality has made inspection and quality assurance central to infrastructure safety and lifecycle management (Morgenthal et al., 2018).

For decades, inspection in the construction and infrastructure sectors has relied heavily on manual workflows—visual assessments, scheduled site visits, written records, and experience-based judgment. These methods still play an important role, and in many settings they remain the default. However, they are also constrained in ways that are becoming harder to ignore. Manual inspections can be labor-intensive, time-consuming, and inconsistent across inspectors or project environments. More importantly, they may fail to capture early-stage defects, especially in hard-to-access, hazardous, or rapidly changing structural contexts (Ma et al., 2018; Zhou et al., 2011). As infrastructure systems become larger, denser, and more technologically integrated, conventional inspection practices are increasingly being asked to do more than they were originally designed to handle.

This is where digital transformation begins to matter. Over the past decade, the construction and infrastructure sectors have seen a notable shift toward technology-assisted inspection and quality assurance. Tools such as unmanned aerial vehicles (UAVs), computer vision, Building Information Modeling (BIM), Internet of Things (IoT) sensors, laser scanning, robotic inspection systems, and predictive analytics are no longer just experimental concepts—they are gradually becoming part of real operational workflows (Koch et al., 2015; Rakha & Gorodetsky, 2018). Their value lies not simply in automation for its own sake, but in the possibility of seeing more, sooner, and more consistently. A drone can reach what a human inspector cannot safely access. A computer vision model can identify surface cracks or anomalies across hundreds of images in a fraction of the time required manually. Sensor-based systems can provide continuous structural feedback rather than isolated snapshots of condition (Cheung et al., 2018; Zhang et al., 2017).

Artificial intelligence (AI) has accelerated this transition even further. When integrated with digital inspection platforms, AI offers the ability to move beyond simple observation toward pattern recognition, anomaly detection, predictive maintenance, and decision support. In theory, this means that inspection systems can become not only faster, but also more proactive—capable of identifying risk before visible failure occurs. Emerging work in digital twins, AI-assisted monitoring, and BIM-integrated inspection has shown strong potential for improving defect detection, maintenance planning, and infrastructure resilience (Huang et al., 2021; Kaewunruen et al., 2021; Liu et al., 2019). Even so, the presence of promising technology does not automatically guarantee meaningful adoption in practice.

That gap between technological capability and field-level implementation remains one of the most important unresolved issues in this area. Although the literature is increasingly rich in case studies, prototype systems, and technical validation papers, there is still relatively limited empirical understanding of how infrastructure professionals themselves perceive these systems, how widely they are being adopted, and what organizational conditions influence their use (Manzoor et al., 2021; Outay et al., 2020; Sheng et al., 2020). In other words, we know a fair amount about what these tools can do, but less about how they are actually entering professional practice—and why adoption remains uneven.

Several barriers likely contribute to this unevenness. Digital inspection systems often require substantial initial investment, technical training, workflow redesign, and institutional support. Challenges related to data management, interoperability, regulatory uncertainty, and resistance to organizational change can further slow implementation (De Melo et al., 2017; Kim et al., 2016). In some settings, professionals may recognize the promise of AI-enabled inspection yet remain uncertain about its reliability, cost-effectiveness, or operational fit. In others, adoption may be driven less by technical readiness than by leadership support, safety culture, or project complexity. These are not merely technical questions—they are professional, organizational, and strategic ones.

Against that backdrop, the present study examines the adoption and perceived effectiveness of AI-enabled digital construction inspection and quality assurance systems among infrastructure professionals in the United States. More specifically, it investigates which digital inspection tools are being used, what benefits professionals associate with them, what obstacles continue to hinder implementation, and which factors appear to shape perceptions of system effectiveness. By focusing on the views and experiences of practitioners across multiple infrastructure-related roles, this study aims to contribute a more grounded understanding of how digital inspection is evolving in real-world professional settings.

Ultimately, improving infrastructure safety is not only about inventing smarter tools; it is also about understanding whether those tools are trusted, supported, and meaningfully integrated into practice. That is where this study seeks to contribute.

2. Materials and Methods

2.1 Study design and setting

This study employed a cross-sectional, questionnaire-based survey design to examine the adoption, perceived effectiveness, and implementation barriers of AI-enabled digital construction inspection and quality assurance systems among infrastructure professionals in the United States. A cross-sectional approach was selected because the study sought to capture a current snapshot of professional practices, institutional readiness, and user perceptions across multiple infrastructure-related roles rather than to evaluate longitudinal behavioral change. This design is commonly used in technology adoption and professional practice research where both prevalence and associated factors are of interest.

The study focused on professionals engaged in the inspection, monitoring, management, or safety oversight of major infrastructure systems, including bridges, highways, tunnels, rail corridors, and public buildings. These sectors were selected because they represent environments in which digital inspection technologies—such as unmanned aerial vehicles (UAVs), computer vision systems, Building Information Modeling (BIM), robotic inspection tools, and structural sensing platforms—are increasingly discussed as practical alternatives or complements to conventional inspection workflows (Chen et al., 2019; Ham et al., 2016; Rakha & Gorodetsky, 2018).

Data collection was conducted between January and March 2023 using a secure online survey platform. The survey was distributed electronically to professionals working across different U.S. regions in order to capture a reasonably diverse range of organizational settings and infrastructure project types.

2.2 Participant eligibility, sampling strategy, and recruitment

Participants were recruited using a purposive sampling strategy, with eligibility limited to individuals who met the following inclusion criteria:

1. were currently employed in an infrastructure-related professional role in the United States;
2. had at least 2 years of professional experience in construction inspection, quality assurance, project management, infrastructure safety, or digital inspection implementation; and
3. had direct or indirect familiarity with infrastructure monitoring or quality control practices.

The survey targeted six professional groups:

• civil engineers,
• construction managers,
• infrastructure inspectors,
• project safety officers,
• digital technology specialists, and
• infrastructure consultants.

Potential participants were approached through professional email circulation, academic-industry contact lists, and infrastructure-related professional networks. The invitation message briefly described the purpose of the study, estimated completion time, eligibility criteria, and the voluntary nature of participation. To minimize duplicate responses, the survey platform was configured to allow one response per device/browser session, and incomplete submissions with substantial missing data were excluded before final analysis.

A total of 176 responses were initially received. After screening for completeness, eligibility, and response consistency, 160 responses were retained for final analysis. No financial incentive was offered for participation.

2.3 Survey instrument development and structure

The survey instrument was developed after reviewing prior literature on digital inspection technologies, AI-assisted monitoring, construction quality assurance, BIM-integrated inspection, and infrastructure safety systems (Behzadan et al., 2015; Koch et al., 2015; Ma et al., 2018; Zhang et al., 2017). Items were designed to reflect both technology exposure and practical implementation realities in the field.

The final questionnaire consisted of five structured sections:

Section 1: Demographic and professional background

This section captured:

• age group,
• gender (optional),
• professional role,
• years of work experience,
• primary infrastructure sector,
• and approximate project involvement.

Section 2: Adoption of digital inspection technologies

Participants were asked whether their organization currently used or had recently used any of the following technologies:

• drones/UAV-based inspection,
• computer vision systems,
• IoT structural sensors,
• BIM-based inspection systems,
• AI predictive monitoring tools,
• and robotic inspection systems.

Responses were recorded as binary categorical variables (Yes/No).

Section 3: Perceived effectiveness of AI-based inspection

Participants rated the overall effectiveness of AI-assisted inspection systems using a 5-point Likert scale ranging from:
1 = Not effective
2 = Slightly effective
3 = Moderately effective
4 = Effective
5 = Very effective

Section 4: Perceived benefits

Respondents selected the most relevant perceived benefits associated with digital inspection systems, including:

• improved defect detection,
• enhanced worker safety,
• faster inspection processes,
• real-time monitoring,
• reduced maintenance costs,
• and better project management.

Section 5: Barriers and implementation challenges

This section assessed perceived obstacles, including:

• high implementation cost,
• insufficient technical expertise,
• data management complexity,
• physical or environmental deployment constraints,
• and organizational resistance to change.

To improve clarity and face validity, the draft questionnaire was pilot-tested with 10 infrastructure professionals not included in the final analytic sample. Feedback from the pilot was used to revise wording, remove ambiguity, and improve item sequencing. Internal consistency for the perception-related items was evaluated using Cronbach’s alpha, and the final instrument demonstrated acceptable reliability (α > 0.70).

2.4 Secondary contextual data sources

Although the primary analysis was based on survey responses, selected secondary U.S.-based infrastructure and safety datasets were reviewed to contextualize the interpretation of findings. These sources included publicly accessible data and reports from:

• the Federal Highway Administration (FHWA) National Bridge Inventory,
• Occupational Safety and Health Administration (OSHA) construction safety reports,
• U.S. Department of Transportation / Bureau of Transportation Statistics (DOT/BTS),
• National Institute of Standards and Technology (NIST) smart monitoring resources, and
• U.S. Geological Survey (USGS) geospatial and infrastructure-related environmental datasets.

These external sources were not merged at the individual participant level, nor were they used as direct inputs in the regression model. Rather, they were used to provide background contextual verification regarding the broader relevance of digital inspection technologies in U.S. infrastructure environments (Zhang et al., 2017; Zhou et al., 2017). This distinction is important for reproducibility and to avoid overstating the analytical role of secondary data.

2.5 Study variables and operational definitions

The primary outcome variable was perceived AI inspection effectiveness. For inferential analysis, Likert responses were dichotomized into:

• High perceived effectiveness (scores 4–5), and
• Low-to-moderate perceived effectiveness (scores 1–3)

The following independent variables were included:

• Professional role (categorical)
• Years of professional experience (ordinal)
• Technology adoption status (binary; any AI/digital inspection tool used vs none)
• Digital training exposure (binary; received formal or informal training vs none)
• Organizational support (binary; presence vs absence of institutional support for technology use)
• Safety awareness orientation (binary/ordinal based on survey response)
• Inspection frequency (ordinal)
• Infrastructure project type (categorical)

These variables were selected based on their likely influence on digital technology uptake and perceived operational value, as suggested by prior work in infrastructure monitoring and technology-enabled quality assurance (Ammar et al., 2021; Li & Liu, 2018; Omar & Nehdi, 2016).

2.6 Statistical analysis

All statistical analyses were conducted using IBM SPSS Statistics (version 26.0) and R software (version 4.2.0). Descriptive statistics were first calculated to summarize respondent characteristics, professional distribution, technology adoption patterns, and perceived benefits. Categorical variables were presented as frequencies and percentages.

Associations between key categorical variables were examined using the Chi-square (χ²) test of independence. These analyses were used to explore relationships such as:

• professional experience vs AI perception,
• technology adoption vs AI effectiveness,
• and organizational support vs AI perception.

Variables showing conceptual relevance or statistical signal in bivariate analysis were then entered into a binary logistic regression model to identify factors associated with higher perceived AI inspection effectiveness. Regression coefficients (β), standard errors, and p-values were reported. Statistical significance was defined as p < 0.05, although borderline associations were also interpreted cautiously where theoretically meaningful.

Missing data were minimal (<5%) and were handled using complete-case analysis. No imputation procedure was applied.

2.7 Ethical considerations

Participation in the study was entirely voluntary and anonymous. No personally identifying information—such as name, phone number, employer identification, or exact worksite location—was collected. Before beginning the survey, all participants were shown an electronic information sheet explaining the purpose of the study, the estimated completion time, the anonymous nature of participation, and their right to withdraw at any point before submission.

Survey data were stored in password-protected digital files accessible only to the research team. The study procedures were conducted in accordance with standard ethical principles for minimal-risk survey-based research involving adult participants (Wang et al., 2014; Mohassel et al., 2014).

3. Results

3.1 Respondent profile and professional distribution

A total of 160 infrastructure professionals were included in

Figure 1. Professional Background of Respondents

Figure 2. Professional Experience of Respondents

Figure 3. Adoption of Digital Construction Inspection Technologies

Figure 4. Perceived Effectiveness of AI-Based Inspection Systems

the final analysis, representing a reasonably diverse cross-section of technical and managerial roles involved in infrastructure inspection, safety oversight, and construction quality assurance. Overall, the respondent profile suggests that the dataset captured perspectives not only from traditional engineering roles but also from professionals involved in operational monitoring, safety enforcement, and digital technology integration.

As shown in (Figure 1), the largest proportion of respondents were civil engineers (30%), followed by construction managers (22%), infrastructure inspectors (18%), project safety officers (13%), digital technology specialists (11%), and infrastructure consultants (6%). This distribution is noteworthy because it reflects the multidisciplinary nature of contemporary infrastructure inspection, where decisions about technology adoption are rarely made by a single profession alone. Rather, they tend to emerge from interaction between design teams, field inspectors, project managers, and increasingly, digital systems specialists.

The predominance of civil engineers and construction managers may indicate that the current transition toward AI-enabled inspection is still being driven primarily by professionals traditionally responsible for structural assessment and project coordination. At the same time, the presence of digital technology specialists and safety officers in the sample suggests that inspection is no longer viewed purely as a visual or compliance-based exercise. It is, gradually at least, becoming a data-supported and technology-mediated process. This shift is broadly consistent with the wider digitalization trends reported in construction informatics and infrastructure management literature, where inspection is increasingly linked with sensing, modeling, and automated defect recognition systems (Behzadan et al., 2015; Manzoor et al., 2021).

3.2 Professional experience of participants

The respondents also represented a fairly balanced spread of professional experience, which strengthens the interpretive value of the dataset. As illustrated in (Figure 2), the largest experience category consisted of professionals with 5–10 years of experience (26.2%), followed by those with 10–15 years (22.5%), less than 5 years (18.8%), 15–20 years (17.5%), and more than 20 years (15.0%).

This pattern suggests that the sample was not dominated by either exclusively early-career or exclusively senior professionals. Instead, it appears to reflect a meaningful mixture of emerging, mid-career, and experienced practitioners. That matters because perceptions of AI-based inspection are likely shaped, at least in part, by how long individuals have worked within conventional inspection environments. Early-career professionals may be more comfortable with digital workflows and software-mediated systems, whereas highly experienced practitioners may assess these tools through the lens of long-standing field practice, operational realism, and reliability under complex site conditions.

Interestingly, the largest segment fell within the mid-career range, which may be particularly important for adoption research. Professionals in this category often occupy implementation-facing roles—they are experienced enough to understand field realities, but also still closely involved in technology transition and operational change. In practical terms, this group may be especially influential in determining whether AI-enabled inspection tools are actually used or remain only theoretically attractive.

3.3 Adoption patterns of digital inspection technologies

The results indicate that digital inspection technologies are already present in professional practice, although adoption remains uneven across tool categories. As shown in (Figure 3), the most frequently reported technology was drone-based inspection (25%), followed by computer vision systems (21%), Building Information Modeling (BIM)-supported inspection (19%), robotic inspection tools (16%), AI predictive monitoring systems (11%), and IoT structural sensors (8%).

This pattern is revealing. Technologies such as drones and computer vision appear to have gained stronger footholds, perhaps because they are more immediately visible in practice and easier to understand as direct extensions of traditional inspection. A drone, after all, still performs a recognizable inspection function—it simply does so faster, from safer vantage points, and with greater image capture capability. Similarly, computer vision systems often build on familiar visual inspection tasks, but add algorithmic support for defect detection, classification, or pattern recognition (Koch et al., 2015; Morgenthal et al., 2018).

BIM also demonstrated substantial uptake, which is not surprising given its growing role as a digital coordination environment linking design, construction, maintenance, and inspection workflows. In infrastructure settings, BIM has increasingly been used not just for modeling, but for integrating inspection records, visual evidence, and maintenance planning into a centralized digital ecosystem (Bradley et al., 2016; Zou et al., 2016).

By contrast, AI predictive monitoring and IoT structural sensing showed lower adoption. These technologies may still be perceived as more technically demanding, more expensive to implement, or more dependent on organizational digital maturity. Unlike drones or visual inspection software, predictive systems often require continuous data infrastructure, analytical expertise, and a longer institutional commitment before value becomes obvious. Their lower adoption does not necessarily imply lower usefulness; rather, it may reflect the greater operational threshold required for implementation (Huang et al., 2021; Kaewunruen et al., 2021).

3.4 Perceived effectiveness of AI-based inspection systems

Participants were also asked to evaluate the overall effectiveness of AI-enabled inspection systems. The responses suggest that perceptions were generally favorable, though not uniformly enthusiastic. As presented in (Figure 4), 35% of respondents rated AI-based inspection systems as very effective, while 27.5% rated them as effective. A further 17.5% considered them moderately effective, whereas 12.5% described them as slightly effective, and only 7.5% regarded them as not effective.

Taken together, these findings suggest that nearly two-thirds of respondents viewed AI-enabled inspection positively, which is an important signal of emerging professional acceptance. This does not mean that implementation is frictionless or universally trusted, but it does indicate that many professionals already perceive practical value in these systems—particularly in relation to inspection speed, consistency, and information quality.

At the same time, the presence of moderate and lower-effectiveness ratings should not be overlooked. These responses may reflect uncertainty around system reliability, difficulties with deployment, training limitations, or mismatch between technological capability and real-world project constraints. In other words, AI-based inspection may be seen as promising, but not yet seamless. That ambivalence is actually quite plausible and, arguably, strengthens the realism of the findings.

3.5 Perceived benefits of digital construction inspection systems

When respondents were asked to identify the most important perceived benefits of digital inspection systems, the results showed a clear emphasis on inspection quality, safety, and operational efficiency. As summarized in (Table 1), the most frequently reported benefit was improved defect detection (26.3%), followed by enhanced worker safety (21.2%), faster inspection processes (18.8%), real-time monitoring (15.0%), reduced maintenance costs (11.2%), and better project management (7.5%).

The prominence of defect detection is particularly important. It suggests that professionals are not valuing these tools merely because they are modern or automated, but because they are perceived to improve the core technical purpose of inspection itself: identifying structural or operational problems accurately and early. This aligns with previous work showing that computer vision, UAV imaging, and sensor-characteristics and AI-related perceptionsassisted monitoring can improve the detection of cracks, surface anomalies, and other infrastructure defects that might otherwise be overlooked or inconsistently recorded (Koch et al., 2015; Besada et al., 2018).

The strong emphasis on worker safety is equally meaningful. In many infrastructure environments—especially bridges, elevated roadways, tunnels, confined spaces, or high-risk structural zones—inspection is not only technically demanding but physically hazardous. Technologies that reduce the need for human exposure to unsafe environments may therefore carry value well beyond efficiency alone (De Melo et al., 2017; Teizer, 2015).

Meanwhile, the recognition of real-time monitoring and reduced maintenance costs suggests that respondents increasingly associate digital inspection with lifecycle management rather than one-time defect identification. That shift—from episodic inspection to continuous or near-continuous awareness—is one of the more significant conceptual changes introduced by AI-enabled infrastructure systems.

3.6 Bivariate associations between key survey variables

The Chi-square analysis revealed several statistically

Table 1. Perceived Benefits of Digital Construction Inspection Systems

Table 2. Chi-Square Test Results for Key Survey Variables

Table 3. Logistic Regression Analysis for AI Inspection Effectiveness

meaningful associations between respondent characteristics and AI-related perceptions. As shown in (Table 2), professional experience was significantly associated with AI perception (χ² = 9.92, p = 0.048), suggesting that respondents with different experience levels did not evaluate AI systems in the same way. This finding implies that professional maturity may influence whether AI-enabled inspection is viewed as practically credible, operationally useful, or sufficiently trustworthy.

A second significant association was observed between technology adoption and perceived AI effectiveness (χ² = 11.31, p = 0.046). This is, in some ways, intuitive: professionals or organizations already using digital inspection technologies were more likely to perceive AI-based systems positively. Familiarity appears to matter. Once these tools move from abstract idea to actual operational use, their value may become easier to recognize.

Similarly, organizational support was significantly associated with AI perception (χ² = 10.12, p = 0.043), reinforcing the idea that adoption is not merely a matter of individual preference or technical awareness. Institutional readiness—through policy, leadership support, training opportunity, or resource allocation—appears to shape how these systems are experienced and evaluated.

Several additional variables showed near-significant or borderline relationships, including:

• profession vs AI effectiveness (p = 0.055),
• training vs technology adoption (p = 0.067),
• infrastructure project type vs adoption (p = 0.072),
• and safety awareness vs AI adoption (p = 0.050).

Although these findings do not all meet conventional significance thresholds, they are still suggestive. They imply that the adoption landscape may be influenced by a cluster of interacting factors rather than a single dominant driver.

3.7 Predictors of perceived AI inspection effectiveness

To further explore which variables were most strongly associated with positive perceptions of AI-enabled inspection, a logistic regression analysis was conducted. The results, presented in (Table 3), suggest that technology adoption was the strongest predictor of perceived effectiveness (β = 0.47, p = 0.041). In practical terms, respondents from settings where digital tools were already in use were more likely to rate AI inspection systems favorably.

Other statistically significant predictors included professional experience (β = 0.38, p = 0.049), digital training (β = 0.35, p = 0.047), and safety awareness (β = 0.29, p = 0.048). These findings suggest that AI inspection is not perceived as effective in isolation; rather, it is more likely to be valued where professionals have sufficient experience to contextualize its utility, where some form of training has supported implementation, and where safety culture already encourages proactive monitoring.

Three additional variables—professional background (p = 0.052), organizational support (p = 0.051), and monitoring frequency (p = 0.054)—showed marginal associations. While not statistically definitive, their proximity to the significance threshold suggests that they may still play meaningful roles, particularly in larger samples or more stratified studies.

Overall, the regression findings point toward a consistent theme: AI-based inspection is more likely to be perceived as effective when it is embedded within an enabling professional and organizational environment rather than introduced as a standalone technological intervention. That may ultimately be one of the most important findings of the study.

4.1 Interpreting the emerging role of AI in infrastructure inspection

The findings of this study suggest that AI-enabled digital inspection systems are no longer peripheral tools in infrastructure management; rather, they appear to be gradually moving into the operational core of construction inspection and quality assurance. While adoption is still uneven and clearly shaped by institutional readiness, the overall pattern is difficult to ignore: many professionals are beginning to see these technologies not as experimental add-ons, but as increasingly practical instruments for improving infrastructure safety, inspection quality, and workflow efficiency.

One of the clearest signals of this shift comes from respondents’ overall assessment of AI system performance. As shown in (Figure 4), a substantial majority of participants rated AI-based inspection systems as either effective or very effective, with only a small minority expressing clearly negative views. That pattern suggests a growing level of professional trust, or at least cautious confidence, in the operational value of AI-assisted inspection. This finding aligns with broader literature showing that digital sensing, automated defect detection, and vision-based monitoring systems are becoming more accepted as infrastructure inspection grows more data-intensive and less dependent on purely manual observation (Teizer, 2015; Zou et al., 2016).

Still, the interpretation here should remain measured. Positive perception does not necessarily mean universal readiness, nor does it imply that AI systems are already fully integrated into routine practice. What it does suggest is that the professional mindset may be shifting. Infrastructure inspection, historically rooted in field expertise and visual judgment, now appears increasingly open to augmentation by algorithmic systems and digital decision support.

4.2 Why some technologies are adopted more readily than others

The adoption pattern reported in this study is especially revealing. As shown in (Figure 3), drone-based inspection, computer vision systems, and BIM-supported inspection workflows emerged as the most commonly used technologies. This distribution is not surprising, but it is instructive.

Drone-based inspection, which had the highest reported adoption, likely occupies a kind of practical “middle ground” between traditional inspection and advanced automation. It does not fundamentally replace the inspection process; instead, it extends the inspector’s reach, improves access to hazardous or elevated structures, and enables faster image capture across large or difficult-to-reach assets. In that sense, drones may be easier for organizations to adopt because their value is immediately visible and operationally intuitive (Besada et al., 2018; Morgenthal et al., 2018).

Computer vision systems appear to follow a similar logic. These tools often support rather than fully replace human assessment, particularly in tasks such as crack identification, surface defect classification, and image-based anomaly screening. Their uptake may therefore reflect a growing willingness to let AI participate in inspection tasks that were once considered exclusively human. Prior work has similarly shown that computer vision can significantly improve consistency and speed in condition assessment workflows, especially when used alongside visual documentation and structural image databases (Koch et al., 2015).

The relatively strong presence of BIM is also notable. BIM has evolved well beyond its original design and coordination functions and is increasingly being used as a platform for integrating inspection records, maintenance planning, and digital lifecycle management. In practical terms, BIM may serve as the connective layer that helps otherwise separate technologies—such as drones, scans, sensors, and field observations—function as part of a more coherent inspection ecosystem (Bradley et al., 2016; Wang et al., 2014).

By contrast, AI predictive monitoring and IoT structural sensing showed lower adoption. This likely reflects the fact that these systems demand more than just technical interest; they require infrastructure for continuous data collection, analytical capacity, and organizational willingness to move from periodic inspection toward continuous or semi-continuous monitoring. In many real-world settings, that transition remains difficult. The issue, then, may not be whether these technologies are useful, but whether organizations are sufficiently prepared to support them (Huang et al., 2021; Kaewunruen et al., 2021).

4.3 What professionals value most: safety, detection, and speed

The perceived benefit profile offers perhaps the most practically important insight in the study. As shown in (Table 1), respondents most frequently associated digital inspection systems with improved defect detection, followed by enhanced worker safety, faster inspection processes, and real-time monitoring. Taken together, these responses suggest that professionals are not simply attracted to AI because it is innovative or fashionable. They appear to value it for very concrete operational reasons.

The prominence of defect detection is especially meaningful. In infrastructure practice, inspection is fundamentally about identifying deterioration before it becomes dangerous, expensive, or irreversible. If professionals perceive AI-supported systems as improving that capability, then these tools are being judged on the basis of their most important functional promise. That interpretation is consistent with prior evidence showing that digital vision systems and automated inspection pipelines can improve anomaly recognition, increase consistency across assessments, and support earlier identification of structural concerns (Koch et al., 2015; Huang et al., 2021).

The strong emphasis on worker safety is equally significant. Inspection often requires personnel to access unstable, elevated, confined, or otherwise hazardous environments. In such contexts, technologies that reduce direct human exposure are not simply efficient—they are protective. Drones, robotic inspection tools, and remote sensing systems may therefore be attractive not only because they improve data collection, but because they reduce the physical risk associated with inspection itself (De Melo et al., 2017; Teizer, 2015).

The findings related to speed and real-time monitoring further suggest that professionals increasingly associate digital inspection with responsiveness. In traditional workflows, structural condition is often assessed intermittently. AI-enabled systems, by contrast, offer the possibility of more continuous awareness and faster decision support. That shift—from episodic inspection to dynamic monitoring—may represent one of the more consequential transformations now underway in infrastructure quality assurance.

4.4 Experience, organizational support, and training as adoption drivers

Beyond simple adoption rates, the inferential analyses offer a more layered picture of what shapes professional acceptance. The Chi-square results in (Table 2) indicate that professional experience, technology adoption, and organizational support were significantly associated with perceptions of AI effectiveness. These findings suggest that adoption is not merely a technical matter; it is also shaped by familiarity, institutional culture, and practical exposure.

The relationship between experience and AI perception is particularly interesting. More experienced professionals may be better positioned to judge whether a technology genuinely improves inspection quality or merely adds complexity. At the same time, experienced practitioners may also be more cautious, particularly if new systems fail to align with field realities. That such a relationship emerged at all suggests that professional background matters not just descriptively, but interpretively.

The significance of organizational support is perhaps even more important. AI-enabled inspection systems rarely succeed in isolation. Their implementation typically depends on training, budget allocation, workflow adaptation, and managerial endorsement. In settings where leadership supports experimentation and digital integration, professionals may be more willing to trust and use these systems. Conversely, even technically promising tools may struggle in organizations that lack policy alignment or implementation readiness (Omar & Nehdi, 2016).

This interpretation is reinforced by the logistic regression findings shown in (Table 3). Among the strongest predictors of perceived AI effectiveness were technology adoption, professional experience, digital training, and safety awareness. These results suggest that professionals are more likely to view AI positively when they have actually used related tools, received at least some degree of preparation, and work within environments where safety and monitoring are already valued priorities.

In a way, that may be the central lesson of this study: AI-based inspection appears most effective not when it is introduced as a standalone technological solution, but when it is embedded within a supportive professional and organizational ecosystem. That observation aligns with implementation research in other digitally transforming sectors, where adoption success often depends less on the sophistication of the tool itself than on the readiness of the system around it (Arden et al., 2021; Mahadevaiah et al., 2020).

4.5 Practical implications for infrastructure quality assurance

From a practical standpoint, these findings carry several implications. First, they suggest that organizations seeking to modernize infrastructure inspection should not focus exclusively on technology procurement. Investment in hardware and software is important, but it is unlikely to be sufficient on its own. Training, workflow redesign, digital literacy, and leadership support appear to be equally critical.

Second, the findings imply that entry-level digital tools—particularly drones, computer vision, and BIM-linked inspection systems—may serve as realistic starting points for organizations with limited digital maturity. These technologies seem to offer a more accessible bridge between traditional inspection and fully data-driven infrastructure monitoring.

Finally, the lower uptake of predictive and sensor-based systems suggests that the next stage of transformation may depend less on proving technical capability and more on solving implementation challenges—especially cost, interoperability, and institutional readiness. In that sense, the future of AI in infrastructure inspection may be determined as much by governance and workforce preparation as by algorithmic sophistication itself (Qian & Lin, 2016).

Conclusion

The findings of this study suggest that AI-enabled digital inspection systems are gradually transitioning from conceptual innovations to practical tools within infrastructure quality assurance. While adoption remains uneven, there is a clear indication that professionals increasingly recognize their value—particularly in improving defect detection, enhancing safety, and accelerating inspection processes. However, the results also reveal that technological capability alone does not guarantee effective implementation. Instead, factors such as professional experience, digital training, and organizational support appear to play a more decisive role in shaping both adoption and perception.

Interestingly, the lower uptake of predictive and sensor-based systems hints at a deeper structural challenge—one that extends beyond cost or complexity and into issues of institutional readiness and workflow integration. Moving forward, it may not be enough to develop more advanced systems; equal attention must be given to preparing the environments in which these systems are expected to function.

Author Contributions

N.A. conceptualized the study, designed the research methodology, conducted data collection and statistical analysis, interpreted the findings, and drafted the manuscript. N.A. also critically revised the article, supervised the overall study process, and approved the final version of the manuscript for publication.

References