Artificial Intelligence Components and Job Redesign in Nigerian Tertiary Institutions: Does Organizational Culture Matter

1. Introduction

Artificial intelligence (AI) is moving from an emerging digital capability to an increasingly consequential feature of organizational work, with implications for how tasks are allocated, coordinated and performed. In tertiary institutions, where large volumes of records, student enquiries, routine transactions and administrative decisions must be processed continuously, AI offers opportunities to improve speed, consistency, information access and service delivery (Davenport & Ronanki, 2018; Dwivedi et al., 2021; Samuel et al., 2023; Ezezina et al., 2023). This transformation is not limited to sophisticated generative systems; it also includes practical components such as optical character recognition (OCR), computer-vision applications (CVA), and robotic process automation (RPA), which can convert documents into machine-readable information, interpret visual or image-based content, and automate repetitive rule-based workflows, respectively (Doermann, 2014; Drobac & Lindén, 2020; González Enríquez et al., 2020). Evidence from higher education increasingly indicates that AI-enabled administrative systems can support records management, student services, workflow efficiency, and institutional decision-making (Raimi & Raimi, 2020; Adedoyin et al., 2020; Olalekan et al., 2020; Ojo & Ibrahim, 2024; Okafor & Nwankwo, 2024; Nwachukwu & Obi, 2025; Seiyaboh & Raimi, 2025; 2026). Nigerian tertiary institutions are therefore encountering a work environment in which digital technologies increasingly interact with established administrative routines, professional roles, and expectations. Studies in Nigeria have documented growing digitalization of tertiary administration and the use of AI-related applications, while research from Rivers State has also highlighted the importance of digital office tools for administrative effectiveness (Adeyemi & Balogun, 2023; Chux-Nyche et al., 2022; Akalonu, 2025; Enetimi & Morufu, 2025; Digha et al., 2025). At the same time, AI does not simply eliminate tasks; it can redistribute them, alter task sequences, increase the importance of human oversight, and create demand for new combinations of technical, cognitive, and interpersonal capabilities (OECD, 2023; Jarrahi et al., 2023). Thus, job redesign becomes a central organizational issue because changes in technology can reshape task variety, autonomy, interdependence, skill requirements and the boundaries between human and automated work (Morgeson & Humphrey, 2006; Baiyere et al., 2023). More broadly, the emerging literature frames digital transformation as a socio-technical process rather than a purely technical upgrade, because technologies acquire practical meaning through the routines, competencies and relationships in which they are embedded (Cecez-Kecmanovic et al., 2020; Rotifa et al., 2024; 2026). From this perspective, OCR can reduce manual transcription and support the conversion of legacy records into searchable information; CVA can extend the capacity to classify or interpret visual information; and RPA can connect repetitive administrative steps across existing systems (Doermann, 2014; Drobac & Lindén, 2020; Ribeiro et al., 2021). Their consequences for employees may consequently differ according to the tasks being automated, the degree of human involvement retained, and the capacity of institutions to redesign work around complementary human capabilities. Such considerations are especially significant in higher education, where administrative work combines standardized procedures with judgement, communication, student interaction and institutional knowledge (Olalekan et al., 2020; George & Wooden, 2023; Hidayati, 2025; Micheal-Olomu et al., 2025). This makes empirical examination necessary for evidence-based institutional planning and workforce development and for responsible implementation of emerging workplace technologies.

However, the extent to which AI components actually translate into job redesign in Nigerian tertiary institutions remains insufficiently understood. Existing studies have largely examined AI adoption, administrative efficiency, readiness, service delivery, or employee performance, rather than simultaneously examining OCR, CVA and RPA as distinct components and testing their combined predictive relationship with job redesign (Mekuri-Ndmele & Ukata, 2025; Mordi & Okon, 2025; Mensah et al., 2024). This gap is important because technological capability alone does not determine organizational outcomes: implementation is shaped by organizational readiness, leadership, employee acceptance, available skills, work processes and the wider institutional environment (Jöhnk et al., 2021; Alalwan et al., 2023; Raimi & Raimi, 2025; Uchenna et al., 2025). Organizational culture is particularly relevant because shared assumptions, values and norms influence how employees interpret technological change, respond to new work arrangements, learn new practices and negotiate changes to established roles (Schein, 2017; Ravasi & Schultz, 2020; Christopher et al., 2025; Aziba-anyam et al., 2025a; Aziba-anyam & Morufu, 2025b). In Nigerian tertiary institutions, where administrative structures and communication practices can influence effectiveness, culture may therefore strengthen, weaken, or otherwise condition the relationship between AI-enabled work and job redesign (Otamiri et al., 2023; Morufu et al., 2025; Otamiri & Barine, 2026). International evidence similarly suggests that AI adoption can produce both enabling and disruptive effects on employees, making context and organizational responses important for understanding its consequences (Ojile & Morufu, 2026; Ding et al., 2024; Oosthuizen, 2022). The issue is timely because current evidence increasingly points toward job transformation rather than wholesale job replacement, while emphasizing that AI changes task composition and skill requirements (International Labour Organization [ILO], 2025; OECD, 2023). Yet empirical evidence that integrates specific AI components, job redesign, and organizational culture within Nigerian tertiary institutions remains limited. This study therefore treats job redesign not as a synonym for job loss, but as a change in the configuration of work tasks and responsibilities following technological integration. This distinction is consistent with evidence that AI can augment human capabilities while also altering the content and organization of work (ILO, 2025; OECD, 2023). It also permits a more precise examination of whether different AI components have comparable or distinct associations with redesigned jobs, rather than treating AI as a single undifferentiated technology. By incorporating organizational culture as a moderator, the study further moves beyond a technology-centred explanation toward an organizationally grounded account of implementation and work change. The evidence can clarify how AI changes roles, responsibilities and work practices within tertiary institutions. The study is consequently positioned at the intersection of AI adoption, work design and organizational culture, with particular relevance to the Nigerian tertiary-education context. Accordingly, this study aims to (i) assess the levels of OCR, CVA, RPA, organizational culture and job redesign among tertiary institutions; (ii) examine relationships among these variables; (iii) evaluate the predictive effects of OCR, CVA and RPA on job redesign; and (iv) determine whether organizational culture moderates the relationship between AI components and job redesign.

2. Materials and Methods

2.1. Study Design and Theoretical Orientation

This study used a quantitative, cross-sectional, explanatory design to examine the relationships between three artificial intelligence (AI) components, Optical Character Recognition (OCR), Chatbots and Virtual Assistants (CVA), and Robotic Process Automation (RPA), and job redesign among staff of tertiary institutions in Bayelsa State, Nigeria. The design was appropriate because the study measured the existing levels of the focal constructs, examined their bivariate relationships, evaluated the predictive effects of the AI components on job redesign, and tested whether organizational culture moderated the AI–job redesign relationship without manipulating the study environment. This approach is consistent with quantitative technology-adoption research in which perceptual constructs are measured and hypothesized relationships are evaluated statistically (Davis, 1989; Marangunić & Granić, 2021). The study was informed by a socio-technical perspective, which conceptualizes organizational outcomes as emerging from interactions among technology, people, work processes, and organizational context. Accordingly, OCR, CVA and RPA were treated as distinct technological predictors; job redesign was specified as the primary outcome; and organizational culture was treated as a contextual moderator. This framing recognizes that the organizational implications of AI depend not only on technological functionality but also on how technologies are incorporated into established roles, routines and institutional practices (Baiyere et al., 2023; Jöhnk et al., 2021).

2.2. Study Setting

The study was conducted in seven tertiary institutions in Bayelsa State, Nigeria, representing universities, polytechnics and a college of education: Niger Delta University (NDU), Federal University Otuoke (FUO), Bayelsa Medical University (BMU), Federal Polytechnic Ekowe (FPE), Bayelsa State Polytechnic, Aleibiri (BSP), Isaac Jasper Boro College of Education (IJBCOE), and the University of Africa, Toru-Orua (UAT). These institutions operate administrative systems involving admissions, academic records, examinations, finance, staff administration, information services, and institutional communication. Their variation in institutional structure, digital infrastructure, and organizational support provided an appropriate context for examining AI-enabled work practices (Adeyemi & Balogun, 2023; Jimoh & Jimoh, 2024).

2.3. Study Population and Sampling

The target population comprised 4,780 teaching and non-teaching staff across the seven institutions, including personnel whose duties involved administrative processes or administrative responsibilities. The population consisted of 2,330 teaching staff and 2,450 non-teaching staff. A target sample of 369 was determined using Taro Yamane's formula at a 5% margin of error and was allocated proportionately across institutions and staff categories using Bowley's population-proportionate allocation procedure. The sampling frame and planned institutional allocation were retained as specified in the study protocol. A multistage, proportionate stratified random sampling procedure was used. First, the target sample was allocated across the seven institutions in proportion to their staff populations. Second, each institutional allocation was stratified by teaching and non-teaching status. Third, respondents were selected from administrative units and departments with direct relevance to AI-supported work, including registry, bursary, admissions, examinations, ICT, human resources, and teaching departments with administrative responsibilities. Of the 369 targeted respondents, 323 usable questionnaires were obtained and included in the statistical analyses, giving an analytical sample of N = 323. This distinction between the planned sample and the final analytical sample is reported to ensure transparency and reproducibility.

2.4. Study Variables and Operational Definitions

The analytical framework comprised three independent AI components (OCR, CVA and RPA), job redesign as the dependent variable, and organizational culture as the moderating variable. This specification directly corresponds to the variables analysed in the manuscript results.

2.4.1. Artificial Intelligence Components

OCR was operationalized as the use of AI-enabled systems to convert printed, handwritten, or scanned documents into machine-readable information. It was measured with five items addressing recognition accuracy, conversion of scanned or handwritten materials, recognition errors, processing speed, and the timeliness of digitized information for administrative use.

CVA was operationalized as the use of AI-based conversational systems to respond to administrative enquiries and support communication with students and staff. Five items assessed response accuracy, content updating, user satisfaction, resolution of multi-step administrative queries, and integration with administrative databases.

RPA was operationalized as the use of software robots to automate repetitive, rule-based administrative tasks. Five items assessed automation of data entry, scheduling and reporting; high-volume data processing; reductions in processing time; acceleration of previously lengthy workflows; and accuracy improvements in administrative transactions.

2.4.2. Job Redesign

Job redesign was the primary dependent variable and captured changes in administrative roles and work processes associated with human–AI collaboration. Four items assessed acquisition of new digital and data-management competencies, redesign of job descriptions and workflows to incorporate AI as a collaborative tool, perceived outcomes of human–AI collaboration, and adequacy of training for working effectively alongside AI systems. The construct therefore reflects changes in the configuration and content of work rather than job loss or workforce reduction.

2.4.3. Organizational Culture

Organizational culture was specified as the moderating variable. It was measured with five items reflecting institutional characteristics relevant to technology adoption, including bureaucratic and participative dimensions. The construct captured the extent to which formal hierarchy and procedures may constrain change and the extent to which openness, knowledge sharing, leadership support, and experimentation may facilitate technological implementation. This specification follows the study's socio-technical orientation and the rationale that organizational culture can condition how technological change is translated into work practices (Baker, 2012; Jöhnk et al., 2021; Schein, 2017).

2.5. Data Collection Instrument

Data were collected using a structured, self-administered questionnaire developed from the technology-adoption, AI-implementation, and organizational-technology literature, with constructs informed in part by the Technology Acceptance Model (Davis, 1989). The analytical variables reported in the present manuscript comprised five OCR items, five CVA items, five RPA items, four job-redesign items, and five organizational-culture items. All substantive items used a five-point Likert response scale from 1 (Strongly Disagree) to 5 (Strongly Agree).

Validity and Reliability

Content validity was established through expert review by specialists in artificial intelligence, information systems, office and information management, educational administration, and research methodology. The review focused on relevance, clarity, completeness, and alignment between items and the study constructs. Construct validity was examined through factor-analytic procedures consistent with established approaches to validating organizational-technology measures (Marangunić & Granić, 2021). Internal consistency was assessed using Cronbach's alpha. The coefficients for the five constructs analysed in the present manuscript were 0.919 for OCR, 0.933 for CVA, 0.940 for RPA, 0.913 for job redesign, and 0.849 for organizational culture. These values indicate high internal consistency of the analytical measures.

2.6. Data Collection Procedure

Data were collected through direct administration of paper questionnaires and, where applicable, approved electronic survey platforms. Before participation, respondents were informed about the study purpose, voluntary participation, confidentiality and data-protection procedures, and their right to withdraw without consequence. Returned questionnaires were screened for completeness, and the dataset was checked for missing responses, inconsistent response patterns, and outliers before analysis.

Data Analysis

Analyses were conducted using IBM SPSS Statistics version 27. Composite scores were calculated as the mean of the constituent items for each construct: OCR from OCR1–OCR5, CVA from CVA1–CVA5, RPA from RPA1–RPA5, job redesign from JR1–JR4, and organizational culture from OC1–OC5. Descriptive statistics (minimum, maximum, mean, and standard deviation) were used to characterize the levels of the five constructs. Pearson product-moment correlation coefficients were calculated to examine the bivariate relationships among OCR, CVA, RPA, organizational culture, and job redesign. Multiple linear regression was then used to evaluate the predictive effects of OCR, CVA and RPA on job redesign. The model was assessed using R, R², adjusted R², the omnibus F-test and standardized regression coefficients (β). For the moderation analysis, an overall AI index was calculated as the mean of OCR, CVA and RPA. OCR, CVA, RPA, and organizational culture scores were mean-centred before the interaction term was formed. Hierarchical regression was then used to enter the AI components at Step 1, organizational culture at Step 2, and the AI × organizational culture interaction at Step 3. Moderation was evaluated from the statistical significance of the interaction term and the associated change in explained variance. Statistical significance was evaluated at p < 0.05. Because the interaction term was not statistically significant in the final model, simple-slopes and Johnson-Neyman analyses were not pursued.

2.7. Ethical Considerations

Ethical approval was obtained from the Ignatius Ajuru University of Education Research Ethics Committee (IAUE/REC/2026/065) before the commencement of data collection. Participation was voluntary, and informed consent was obtained from all respondents. No personally identifiable information was collected. All responses were treated confidentially and reported only in aggregate form, in accordance with established ethical principles for research involving human participants.

2.8. Data Availability

The anonymised dataset generated and analysed in this study, together with the coding structure and statistical syntax used for analysis, will be made available by the corresponding author upon reasonable request, subject to institutional ethical approval and applicable data-protection requirements.

3. Results

3.1. Descriptive Statistics of Study Variables

Table 1 & Figure 1 indicates that all five study constructs were represented by relatively high mean scores among the 323 respondents. OCR had the highest observed mean (M = 3.882, SD = 0.831), followed by CVA (M = 3.849, SD = 0.873), organizational culture (M = 3.819, SD = 0.849), RPA (M = 3.811, SD = 0.905) and job redesign (M = 3.730, SD = 0.868). The observed ranges were broad across the five-point response scale, with minimum values ranging from 1.40 to 1.80 and maximum values of 5.00 for all constructs, indicating that respondents used a substantial portion of the available response scale. Thus, the descriptive pattern suggests generally favourable reported levels of AI-component adoption, organizational culture, and job redesign within the study sample, although these descriptive statistics alone do not establish relationships, predictive effects, or causal associations among the variables. Notably, the relatively small differences between the construct means indicate broadly similar central tendencies, whereas the standard deviations (0.831-0.905) indicate moderate variability in respondents' scores.

3.2. Correlations Among Study Variables

Pearson correlation analysis (N = 323) showed that job redesign was essentially unrelated to the three AI components, with very weak positive correlations for OCR (r = 0.014), CVA (r = 0.019), and RPA (r = 0.031); organizational culture was also weakly and negatively correlated with job redesign (r = −0.084). The correlations among the AI components themselves were similarly weak, with OCR showing a small negative association with CVA (r = −0.073) and negligible associations with RPA (r = 0.033), while CVA and RPA were weakly negatively correlated (r = −0.028). Organizational culture showed negligible correlations with OCR (r = 0.000), CVA (r = 0.042), and RPA (r = −0.008). Overall, the correlation matrix indicates minimal linear association among the study variables; however, statistical significance cannot be determined from the reported correlation coefficients alone because the corresponding p-values or confidence intervals are not provided in Table 2. Thus, the statement that none of the correlations was statistically significant should only be retained if the underlying SPSS output confirms p > 0.05 for all reported associations (Table 2 & Figure 2 below).

3.3. Regression Model for Job Redesign

The multiple regression model showed a very weak overall association between the three AI components (OCR, CVA, and RPA) and job redesign, with R = 0.039. The model explained only 0.2% of the variance in job redesign (R² = 0.002), while the adjusted R²was negative (−0.008), indicating that the predictors provided essentially no improvement in explanatory power beyond an intercept-only model. The standard error of the estimate was 0.871, representing the typical deviation of observed job-redesign scores from the values predicted by the regression model. However, because the model-summary table does not itself provide the omnibus F-test and associated p-value, its statistical significance should be interpreted using the corresponding ANOVA results rather than from R, R², or adjusted R²alone (Table 3a & Figure 3).

The analysis of variance (ANOVA) results presented in Table 3b & Figure 3 showed that the overall multiple regression model was not statistically significant, F(3, 319) = 0.165, P = 0.920. The model accounted for only 0.376 units of the total variance (total sum of squares = 242.850), while 242.474 units remained unexplained (residual variance). Thus, considered jointly, the three predictors did not explain a statistically significant proportion of variation in the outcome variable (Table 3b & Figure 3).

The regression coefficients presented in Table 3c & Figure 3 showed that none of the three predictors was significantly associated with the outcome variable. Optical Character Recognition (OCR) had a small positive standardized coefficient (β = 0.014, t = 0.258, P = 0.797), while Chatbots and Virtual Assistants (CVA) also showed a small positive association (β = 0.021, t = 0.375, P = 0.708). Robotic Process Automation (RPA) similarly demonstrated a small positive coefficient (β = 0.031, t = 0.551, P = 0.582). Thus, none of the predictors demonstrated a statistically significant independent association with the outcome at the 0.05 significance level (Table 3c & Figure 3).

3.4. Moderating Role of Organizational Culture

The hierarchical regression analysis examining the moderating role of organizational culture in the relationship between artificial intelligence (AI) components and job redesign showed that the AI components explained a negligible proportion of variance in Step 1 (R² = 0.001, ΔR² = 0.001, F = 0.466, P = 0.495). Adding organizational culture in Step 2 produced only a marginal increase in explained variance (R² = 0.009, ΔR² = 0.007), with the overall model remaining non-significant (F = 1.381, P = 0.253). In Step 3, the inclusion of the AI × organizational culture interaction term did not increase the explained variance (R² = 0.009, ΔR² = 0.000), and the model remained non-significant (F = 0.924, P = 0.429), providing no statistical evidence that organizational culture moderated the relationship between AI components and job redesign (Table 4a & Figure 4a).

The interaction effects presented in Table 4b & Figure 4b showed that neither the AI components nor organizational culture was significantly associated with job redesign. AI components had a small positive coefficient (β = 0.070, t = 0.710, P = 0.478), while organizational culture showed a small negative coefficient (β = −0.086, t = −1.508, P = 0.132). Importantly, the AI × organizational culture interaction term was not statistically significant (β = 0.018, t = 0.137, P = 0.891), indicating that organizational culture did not significantly moderate the relationship between AI components and job redesign (Table 4b & Figure 4b).

The hypothesis-testing results presented in Table 4c & Figure 4c showed that all four null hypotheses were retained based on the observed statistical evidence. Specifically, OCR was not significantly associated with job redesign (β = 0.014, P = 0.797), nor was CVA significantly associated with job redesign (β = 0.021, P = 0.708); similarly, RPA showed no statistically significant relationship with job redesign (β = 0.031, P = 0.582). In addition, the AI × organizational culture interaction was not statistically significant (β = 0.018, P = 0.891), providing no evidence that organizational culture moderated the relationship between AI components and job redesign (Table 4c & Figure 4c). Because the interaction effect was non-significant, simple-slopes and Johnson–Neyman analyses were not warranted.

4. Discussion

4.1. Descriptive Patterns of AI-Enabled Organisational Change

The relatively high and closely clustered scores across OCR, CVA, organisational culture, RPA, and job redesign suggest that AI-enabled administrative work is becoming embedded within a broader organisational context rather than being perceived as a set of isolated technologies. This interpretation is consistent with technology-adoption research showing that users’ perceptions of technological usefulness and organisational conditions are important for the acceptance and continued use of digital systems (Ababneh, 2016; Awa et al., 2021; Alalwan et al., 2023; Alsahli et al., 2025). In particular, Ababneh’s empirical work shows that organisational culture and trust are important contextual considerations in technology acceptance, supporting the present observation that organisational culture follows the same generally favourable response pattern as the technology-related constructs. More broadly, research on digital transformation argues that technological change is accompanied by changes in processes, infrastructure and employee agency, rather than simply the installation of new technologies (Baiyere et al., 2020; Cecez-Kecmanovic et al., 2020; Baker, 2012; Clarke et al., 2017). The comparatively favourable level of RPA is also compatible with evidence from higher education that automation can be applied to repetitive administrative processes and may support efficiency, workflow optimisation and service delivery (Tinimoye et al., 2026; Abdulraheem et al., 2025a; 2025b; 2025c; Bhardwaj & Kumar, 2025; Adebisi et al., 2026; Anyadighibe et al., 2022; Abdel-Majeed Abdel & Ibrahim, 2019; Tuebi et al., 2021). Similarly, the relatively favourable job-redesign scores align with the socio-technical literature, which suggests that digital transformation can alter established work routines and redistribute activities between human actors and technological systems (Baiyere et al., 2020; Brynjolfsson & McAfee, 2014; Zimmer et al., 2023; Samuel et al., 2023; Erezina et al., 2023; Ojile & Morufu, 2026). Importantly, however, the descriptive convergence observed here should not be interpreted as evidence that AI adoption has produced job redesign or that one construct determines another; rather, it establishes the empirical context within which the subsequent association, regression, and moderation analyses should be interpreted.

Furthermore, the moderate dispersion around the construct means indicates that the favourable aggregate pattern does not represent complete uniformity among respondents, a consideration that is important in interpreting technology adoption within heterogeneous organisational environments. Previous studies have similarly emphasised that acceptance and implementation of emerging technologies depend on interacting technological, organisational and individual conditions rather than on technological availability alone (Ababneh, 2016; Awa et al., 2021; Alalwan et al., 2023; Barhoumi, 2016; Alsahli et al., 2025; Agudo et al., 2024; Bhardwaj & Kumar, 2025; Baiyere et al., 2020). Ababneh’s findings particularly reinforce the relevance of organisational culture and trust, whereas Baiyere et al. show that digital transformation can generate tensions requiring changes in organisational processes and actor roles. The RPA literature likewise indicates that automation may coexist with implementation costs, data-security concerns, change-management requirements, and questions surrounding the future configuration of work, despite its potential to improve administrative efficiency (Bhardwaj & Kumar, 2025). Thus, the present descriptive profile is broadly concordant with prior literature portraying AI-enabled transformation as an organisationally situated process involving technology, people and work practices simultaneously. At the same time, the absence of marked separation between the construct means should be treated as substantively informative rather than as evidence of equivalence: the constructs represent related but theoretically distinct dimensions of AI-enabled organisational change. Consequently, the principal contribution of these descriptive findings is to establish a relatively favourable baseline of perceived AI-component adoption and work-system adaptation, while leaving the direction, magnitude and statistical significance of relationships among these dimensions to the inferential analyses that follow.

4.2. Correlations Between AI Adoption and Job Redesign

The near-zero correlations between job redesign and the three AI components, together with the similarly weak associations among OCR, CVA and RPA, suggest that the presence or reported use of individual AI technologies does not necessarily translate into a correspondingly observable restructuring of employees’ jobs. This pattern is compatible with the distinction between technology availability or adoption and organisational transformation: Davis’s technology-acceptance framework emphasises perceived usefulness and ease of use as determinants of technology acceptance, rather than assuming that acceptance automatically produces changes in work design (Davis, 1989), while Goodhue and Thompson’s task–technology fit perspective similarly implies that technological effects depend on the correspondence between technology capabilities and task requirements (Goodhue & Thompson, 1995; Gift & Obindah, 2020; Enetimi & Morufu, 2025). Consistent with this interpretation, Davenport and Ronanki (2018) reported that organisations often obtain value from AI through relatively bounded applications that improve existing processes rather than through wholesale occupational transformation. Likewise, the systematic mapping study by González Enríquez et al. (2020) shows that RPA research has largely focused on automating specific processes and identifying implementation opportunities, rather than establishing a universal relationship between automation and job redesign. The present weak associations therefore do not contradict evidence that AI can improve organisational performance; rather, they indicate that such effects may depend on intervening organisational, task and implementation conditions. This interpretation is also compatible with Dubey et al. (2020), whose empirical model linked AI-enabled analytical capabilities with operational performance through organisational capabilities and environmental conditions, rather than treating AI adoption as an isolated determinant. Similarly, Ding et al. (2024) found that AI adoption could generate both challenge and threat appraisals among employees, demonstrating that employee responses to AI are multidimensional rather than reducible to a simple technology-to-work-redesign pathway. From a socio-technical perspective, Emery and Trist (1960), Furjan et al. (2020) and Dwivedi et al. (2021) further support the view that technological change becomes organisationally consequential through interactions between technical systems, people, tasks and institutional arrangements. Accordingly, the present correlation pattern may indicate that OCR, CVA and RPA were being incorporated into existing administrative routines without sufficiently altering task boundaries to generate a strong bivariate relationship with job redesign.

Furthermore, the weak correlations involving organisational culture provide additional evidence that AI-related technologies and work redesign should not be assumed to move together in a simple linear fashion. Although organisational context is widely recognised as important to digital transformation, the literature suggests that contextual factors often operate through implementation processes, capabilities, leadership, task characteristics and employee perceptions rather than through direct bivariate relationships (Clarke et al., 2017; Furjan et al., 2020; Dwivedi et al., 2021; Ekanem & Afolabi, 2024; George & Wooden, 2023; Huang & Rust, 2021; Davenport et al., 2020; Ding et al., 2024). Indeed, AI applications can produce different consequences depending on whether they complement existing human activities, substitute for routine tasks or create new coordination requirements; Huang and Rust (2021) conceptualise AI in service work in terms of changing relationships between human and machine capabilities, whereas Davenport et al. (2020) similarly emphasise the redistribution of analytical and interpersonal activities rather than an automatic replacement of human work. The weak OCR-CVA, OCR-RPA and CVA-RPA associations are also substantively plausible because these technologies perform different functional roles, document/image recognition, conversational or voice-based interaction, and process automation, and therefore need not be adopted as an integrated technological bundle. Evidence from OCR research likewise concerns improvements in document indexing, recognition and retrieval rather than direct occupational restructuring (Doermann, 2014; Drobac & Lindén, 2020). Thus, the present findings provide a useful qualification to studies reporting beneficial effects of AI and automation: technological adoption can generate operational value without necessarily producing immediate or proportional changes in job architecture. Finally, because the reported correlation coefficients are not accompanied by P values or confidence intervals, the current evidence supports interpretation of the magnitude and direction of the observed associations but does not independently establish their statistical significance; consequently, claims that all associations were non-significant should be made only after verification against the underlying inferential output. This distinction is particularly important because a weak correlation can be statistically detectable in a sufficiently large sample, whereas a non-significant result does not itself demonstrate the absence of a meaningful relationship.

4.3. AI Technologies and Job Redesign

The absence of a meaningful relationship between OCR, CVA, RPA and job redesign suggests that AI-enabled technologies may be functioning primarily as task-level or process-level augmentations rather than direct drivers of broader job restructuring. This interpretation is consistent with digital-transformation scholarship showing that technological implementation does not automatically produce changes in organisational roles, routines or work architecture (Baiyere et al., 2020; Furjan et al., 2020; Dwivedi et al., 2021; George & Wooden, 2023; Hoffman et al., 2024; Hidayati, 2025; Almeida et al., 2026; El-Masry et al., 2026). Baiyere et al. (2020) emphasise the interaction between digital technologies, organisational processes and human agency, while Zimmer et al. (2023) argue that workplace transformation requires deliberate modification of established routines rather than simply introducing new digital tools. This perspective is compatible with Davenport and Ronanki (2018), who observed that organisations frequently derive practical AI benefits from narrowly defined applications that enhance existing processes rather than fundamentally restructure work. More specifically, OCR primarily converts scanned or image-based information into machine-readable data, meaning that its immediate consequences may remain concentrated within document-processing activities unless accompanied by broader workflow changes (Doermann, 2014; Drobac & Lindén, 2020; González Enríquez et al., 2020; Dubois et al., 2025; Dupont et al., 2025; Hansen et al., 2024; Ezeji et al., 2024; Alonge et al., 2025). Similarly, CVA technologies may alter information access and service interactions without necessarily redistributing administrative responsibilities (Følstad & Brandtzæg, 2020; Morufu et al., 2021; Gnewuch et al., 2021; Gnewuch et al., 2022; Ashfaq et al., 2021; Ferreira et al., 2025; Garcia et al., 2025; Dubois et al., 2025; Bianchi et al., 2024). Thus, evidence that AI can improve service interactions does not necessarily conflict with the present finding because service efficiency and job redesign are conceptually distinct outcomes; Huang and Rust (2021) emphasise human–machine complementarity in service work, while Davenport et al. (2020) similarly describe AI in terms of activity augmentation and redistribution. Likewise, the non-significant RPA association is compatible with evidence that RPA principally targets repetitive, rule-based processes and that its organisational consequences depend on process selection, implementation quality and organisational readiness (Ciufudean, 2018; González Enríquez et al., 2020; Farinha et al., 2024; Firdaus et al., 2026; Bhardwaj, 2025; Adebisi et al., 2026; Dupont et al., 2025; Ali, 2025). Accordingly, process automation may improve particular activities while leaving job titles, reporting structures and substantive responsibilities largely intact.

Furthermore, the collective lack of explanatory contribution from the three AI components is consistent with socio-technical perspectives that position organisational context between technology adoption and work outcomes (Baiyere et al., 2020; Dwivedi et al., 2021; Clarke et al., 2017; George & Wooden, 2023; Furjan et al., 2020; Ekanem & Afolabi, 2024; Ding et al., 2024). AI technologies operate alongside managerial decisions, employee capabilities, institutional routines, governance arrangements and complementary processes; consequently, their effects cannot necessarily be inferred from technology adoption alone. Dubey et al. (2020), for example, link AI and big-data capabilities to operational performance within broader organisational and environmental conditions, while Ding et al. (2024) demonstrate that AI adoption can generate heterogeneous employee responses. This contextual interpretation helps reconcile the present findings with studies reporting positive effects of AI on administrative efficiency, because faster processing, improved information access and reduced repetitive workload do not necessarily entail changes in job composition (Garcia et al., 2025; Dubois et al., 2025; Ferreira et al., 2025; Alvarez et al., 2025; Eze & Chukwuemeka, 2025; Akanbi et al., 2026; Hassan et al., 2026; Dubois et al., 2026). Therefore, the findings do not support a simple linear proposition that greater exposure to OCR, CVA or RPA necessarily produces greater job redesign; rather, they suggest that technology adoption may be necessary but insufficient for substantive work restructuring, consistent with research portraying digital transformation as an interaction among technology, organisational routines, human agency and institutional change (Furjan et al., 2020; Baiyere et al., 2020; Dwivedi et al., 2021; Zimmer et al., 2023; George & Wooden, 2023; Hidayati, 2025; Hoffman et al., 2024; El-Masry et al., 2026). This also qualifies technological-determinist accounts of automation, including Frey and Osborne (2017), because the presence of individual AI tools cannot by itself establish occupational transformation. Within this sample, the three AI components therefore provided little explanatory information about job redesign, indicating that future research should consider intervening or contextual mechanisms such as task–technology fit, employee AI competence, implementation maturity, managerial support, workflow integration and organisational readiness.

4.4. Moderating Role of Organizational Culture and AI-Driven Job Redesign

The absence of evidence that organisational culture altered the relationship between AI components and job redesign suggests that the organisational context examined may not yet have reached a level of AI maturity at which cultural characteristics translate into measurable differences in work restructuring. This interpretation is consistent with research distinguishing technology adoption from organisational transformation, whereby the presence of AI does not automatically produce changes in jobs, routines or employee roles (Jöhnk et al., 2021; Kumar et al., 2023; Li et al., 2021; George & Wooden, 2023; Dwivedi et al., 2021; Koleayo et al., 2021a; 2021b; Furjan et al., 2020; Jedel et al., 2026; Kim et al., 2025). Jöhnk et al. (2021), for example, conceptualise organisational AI readiness as multidimensional, encompassing resources, knowledge, processes and organisational conditions, rather than culture alone. Similarly, recent research on generative-AI adoption in higher education indicates that technology use is shaped by multiple individual and institutional factors, suggesting that organisational culture may not exert a sufficiently direct influence to function as a standalone moderator (Jedel et al., 2026). The present finding also accords with task-technology perspectives, which propose that technological effects depend on the fit between technological capabilities and the activities being performed rather than on organisational context in isolation (Isaac et al., 2019; Karanasios, 2018). Thus, even where an organisation exhibits a supportive culture, AI may remain embedded within existing workflows without substantially changing how jobs are structured. Evidence from higher education illustrates this distinction: AI-enabled administrative systems, predictive analytics and smart-office technologies have been associated with improved service delivery and administrative efficiency, but such improvements do not necessarily imply corresponding changes in job architecture (Jimoh & Jimoh, 2024; Kim & Park, 2024; Kim et al., 2025; Hwang et al., 2026; Ivanov et al., 2026; Johnson et al., 2026; Kalu et al., 2026; Larsson et al., 2025). Likewise, evidence concerning chatbots indicates that these systems can improve student-facing administrative services while retaining substantial human involvement in service provision (Kim et al., 2024; Kowalczyk et al., 2024; Kowalski et al., 2024; Lim et al., 2024; Martínez et al., 2025). Collectively, these studies suggest that organisational culture may facilitate technology acceptance or implementation without necessarily determining whether technology subsequently produces substantive job redesign. This distinction is theoretically important because it prevents the interpretation of a non-significant moderation effect as evidence that organisational culture is irrelevant to AI implementation; rather, its influence may operate indirectly through readiness, employee acceptance, leadership, technological competence, workflow integration or implementation quality (Kim et al., 2025; Jöhnk et al., 2021; Marangunić & Granić, 2021; Katrin Förster, 2024; Khantong & Sriboonlue, 2026; Laakmann et al., 2025).

Furthermore, the lack of a moderating effect can be understood within the broader literature on AI-enabled automation, which increasingly portrays organisational outcomes as contingent upon the interaction of technology, tasks, people and implementation conditions rather than as a simple culture-technology relationship. Research on RPA, for example, indicates that automation outcomes depend on process characteristics, implementation architecture and organisational capabilities, meaning that a supportive culture alone may be insufficient to produce changes in work design (Khantong & Sriboonlue, 2026; Laakmann et al., 2025; Leno et al., 2020; Farinha et al., 2024; Bhardwaj, 2025; Jha, 2026; Júnior & Do Nascimento, 2026; López et al., 2025). Similar arguments apply to AI-supported decision-making, where technological systems can augment rather than replace human judgement and therefore may leave existing occupational boundaries largely intact (Jarrahi et al., 2023; Krupa & Mullaney, 2024; Kovari, 2024; Luo et al., 2023; Ezeji et al., 2024; Johnson et al., 2025; Johnson et al., 2026; Hwang et al., 2026). Indeed, Jarrahi et al. (2023) emphasise the complementary relationship between AI and human decision-making, while research on AI reliability and trust similarly indicates that human involvement remains important where decisions carry uncertainty or risk (Kovari, 2024; Elder et al., 2022; Raimi & Raimi, 2020; Adedoyin et al., 2020; Olalekan et al., 2020). Consequently, the present absence of moderation may reflect the fact that organisational culture operates at a different stage of the AI implementation pathway from job redesign. Culture may influence whether employees accept technologies, whether managers support implementation, or whether institutions develop the capabilities required for successful adoption, while actual job redesign may depend on subsequent changes to tasks, workflows and authority structures (Jöhnk et al., 2021; Isaac et al., 2019; Marangunić & Granić, 2021; Kim et al., 2025; Jedel et al., 2026; Kumar et al., 2023; Li et al., 2021; Furjan et al., 2020). The finding therefore does not contradict studies demonstrating positive effects of AI on administrative efficiency or service delivery; rather, it qualifies their implications by showing that improvements in efficiency, automation or service provision should not automatically be equated with job redesign (Jimoh & Jimoh, 2024; Kim & Park, 2024; Madsen et al., 2024; Martinez et al., 2024; Larsson et al., 2025; López et al., 2025; Jha, 2026; Júnior & Do Nascimento, 2026). Importantly, because the interaction term was not statistically significant, further simple-slopes or Johnson–Neyman analyses would not add inferential value. The broader implication is that future research should examine whether organisational culture operates through mediating or sequential mechanisms, such as AI readiness, employee acceptance, managerial support, task-technology fit, digital competence and implementation maturity, rather than assuming that culture directly changes the strength of the AI–job-redesign relationship. Such an approach would be consistent with contemporary models of AI adoption that view organisational transformation as a multistage socio-technical process rather than a direct consequence of technological deployment (Gift & Obindah, 2020; Jöhnk et al., 2021; Dwivedi et al., 2021; Kumar et al., 2023; Jedel et al., 2026; Kim et al., 2025; Khantong & Sriboonlue, 2026; George & Wooden, 2023; Li et al., 2021).

5. Implications for Policy and Interventions

The findings suggest that institutional AI policy should move beyond technology acquisition towards task-level integration, workflow redesign and organisational capability development. Because OCR, CVA and RPA showed negligible explanatory power for job redesign, institutions should not assume that deploying AI tools will automatically transform administrative work. Instead, implementation policies should identify processes suitable for augmentation or automation, clarify human-AI responsibilities, provide continuous staff training, strengthen digital competencies, establish mechanisms for monitoring AI performance and reliability, and involve employees in decisions concerning changes to tasks and workflows. Organisational culture should be treated as an enabling contextual condition rather than as a stand-alone mechanism for producing job redesign. Intervention strategies should therefore be phased and evidence-driven, with institutions assessing whether AI implementation actually changes workload, task composition, service quality and employee roles before scaling up. Such an approach would support responsible AI integration while minimising disruption to established administrative functions.

6. Summary of the Findings

Overall, the study presents a consistent pattern across the descriptive, correlational, regression and moderation analyses. Although respondents reported relatively favourable levels of OCR, CVA, RPA, organisational culture and job redesign, the variables demonstrated minimal linear associations. The three AI components jointly explained only 0.2% of the variance in job redesign, with the overall regression model being non-significant, F(3,319) = 0.165, P = 0.920. None of the individual AI components significantly predicted job redesign, and organisational culture neither produced a significant independent contribution nor significantly moderated the AI-job-redesign relationship (β = 0.018, P = 0.891). Taken together, these findings indicate that favourable perceptions or use of AI technologies do not necessarily correspond to substantive restructuring of administrative jobs within the study context.

7. Study Limitations

Several limitations should be considered when interpreting these findings. First, the cross-sectional design limits conclusions about temporal ordering and causality; the observed relationships represent conditions at the time of measurement rather than changes attributable to AI implementation over time. Second, the study relied on respondent-reported measures, which may be influenced by common-method effects, individual perceptions and differences in understanding of AI-enabled work practices. Third, the analysis focused on OCR, CVA and RPA and therefore does not encompass the full range of contemporary AI applications that may influence administrative work. Fourth, the very small explanatory variance indicates that important determinants of job redesign were not captured by the model. Finally, because the correlation table did not independently report P values or confidence intervals, conclusions concerning the statistical significance of individual correlations should be based on the corresponding inferential output rather than on correlation coefficients alone.

8. Conclusions

The study demonstrates that AI adoption should not be equated with job redesign. Despite relatively favourable reported levels of AI-component use and job redesign, OCR, CVA and RPA showed neither meaningful individual predictive effects nor a significant collective contribution to job redesign, while organisational culture did not significantly modify these relationships. The findings therefore challenge a simple technology-deterministic view in which introducing AI automatically restructures administrative jobs. Instead, meaningful work transformation is likely to depend on how technologies are integrated with tasks, workflows, employee capabilities, managerial practices and institutional arrangements. The significant contribution of this study is the empirical demonstration that AI presence alone may be insufficient to explain job redesign, highlighting the need to examine the organisational and task-level mechanisms through which AI becomes consequential for administrative work.

9. Actionable Recommendations

9.1. Short-Term Recommendations (0-12 Months)

i.

Conduct an AI-task audit: Map current administrative tasks and identify which activities are suitable for OCR, CVA and RPA augmentation or automation.

i.

ii. Strengthen staff capacity: Provide practical training in AI literacy, prompt use, automated workflow management, data quality, verification and responsible human oversight.

i.

iii. Establish human-in-the-loop procedures: Define which AI-generated outputs require mandatory human verification, particularly for decisions involving students, personnel, finance or official records.

i.

iv. Develop institutional AI guidelines: Establish clear rules covering acceptable AI use, data protection, confidentiality, accountability, error reporting and documentation of AI-assisted decisions.

i.

v. Pilot before scaling: Test AI applications within selected administrative processes and measure changes in workload, processing time, error rates, service quality and employee experience.

i.

vi. Create staff feedback mechanisms: Involve administrative personnel in identifying useful applications, implementation problems, and unintended effects on their work.

9.2. Mid-Term Recommendations (1-3 Years)

i.

Integrate AI with workflow redesign: Move beyond isolated automation by redesigning end-to-end administrative processes where evidence demonstrates that AI can improve efficiency without compromising quality.

i.

ii. Develop role-specific competency frameworks: Define the technical, analytical and interpersonal competencies required for administrative staff working alongside AI systems.

i.

iii. Establish AI performance monitoring: Routinely evaluate accuracy, reliability, usability, processing time, error rates and human override rates for deployed AI applications.

i.

iv. Strengthen organisational readiness: Develop dedicated implementation teams involving information technology, human resources, administration, management and relevant end users.

i.

v. Link AI investment to measurable outcomes: Require AI projects to demonstrate improvements in clearly defined institutional indicators rather than evaluating success solely by technology adoption or usage.

i.

vi. Review job descriptions periodically: Where AI demonstrably changes task allocation, update job descriptions, responsibilities, and performance indicators accordingly rather than assuming that technology adoption automatically constitutes job redesign.

9.3. Long-Term Recommendations (3⁺ Years)

i.

Develop an institution-wide AI transformation strategy: Integrate AI planning with workforce development, digital infrastructure, organisational governance and long-term administrative planning.

i.

ii. Establish continuous job-redesign mechanisms: Treat AI-enabled job redesign as an iterative process in which tasks, responsibilities and competencies are periodically reassessed as technologies evolve.

i.

iii. Create sustainable AI governance structures: Institutionalise multidisciplinary AI governance committees responsible for oversight, risk management, ethical compliance, data governance and workforce implications.

i.

iv. Invest in interoperable digital infrastructure: Ensure that OCR, CVA, RPA and future AI systems can communicate with existing information systems and administrative databases securely and reliably.

i.

v. Support longitudinal evaluation: Track AI implementation and job characteristics over several years to determine whether initially modest effects eventually translate into substantive changes in work organisation.

i.

vi. Promote human-centred AI transformation: Prioritise AI systems that augment professional judgement, reduce unnecessary repetitive work, and create opportunities for higher-value administrative activities rather than pursuing automation solely for technological modernisation.