This codebook provides a structured framework for analyzing the six Gulf Cooperation Council (GCC) National AI Strategies (NASs) with a focus on capacity-building references. It links policy priorities explicitly to the constructs used in our empirical model, Technical Infrastructure (TI), Organizational Capability (OC), and Governance (GOV), and supports alignment with survey items Q2–Q12.The codebook combines qualitative categorization and quantitative weighting. Each NAS passage is coded into a primary category (TI, OC, or GOV) using well-defined sub-codes, while secondary tags capture intended outcomes (AIO: service efficiency, citizen trust, performance) and contextual dimensions (e.g., sector, localization, cultural factors). This ensures that both policy intentions and implementation pathways are captured consistently across countries.To assess emphasis on specific AI capabilities (for example, NLP, robotics, cybersecurity), the codebook applies term-salience analysis (TF–IDF) within the coded passages. This allows us to differentiate between universally mentioned enablers, such as AI ethics or cybersecurity, and more specialized but heavily emphasized capabilities, such as deep learning or robotics, that feature in only some national strategies.The design emphasizes reproducibility and rigor: coders follow clear inclusion/exclusion rules, use EN/AR synonym lexicons, and apply double-coding with inter-coder reliability checks (target κ ≥ 0.79). Outputs include normalized construct-level signals per country and comparative TF–IDF tables, which are then triangulated with survey and PLS-SEM findings.In sum, the codebook acts as a bridge between policy texts and empirical analysis, enabling transparent, comparable, and replicable measurement of AI adoption enablers across the GCC.

The AI Adoption Survey was designed to empirically validate the constructs derived from the GCC National AI Strategies and the literature review. It targeted mid- to senior-level government employees across the six GCC member states who are directly involved in digital transformation, AI strategy, or IT/innovation functions.Survey Design and StructureThe questionnaire included 12 core items (Q1–Q12).Q1 captured demographic and categorical information (country, role, sector).Q2–Q4 measured Technical Infrastructure (TI), focusing on the availability of ICT, data, and cloud resources to support AI.Q5–Q7 measured Organizational Readiness (OR), including workforce readiness, leadership support, and institutional processes for AI integration.Q8–Q9 measured Governance Environment (GE), assessing regulatory clarity, ethical oversight, and legal frameworks surrounding AI.Q10–Q12 measured AI Outcomes (AIO), including perceived impact on service delivery, efficiency, and citizen satisfaction.All items were measured using a 5-point Likert scale (1 = strongly disagree, 5 = strongly agree).Sampling and Data CollectionThe survey was distributed online between September 2024 and February 2025. Invitations were sent to approximately 400 eligible participants, with 203 valid responses recorded, representing all six GCC countries. Respondents were purposively sampled to ensure coverage of ministries, agencies, and professional associations engaged in AI deployment.Response CharacteristicsThe final dataset provides a balanced representation across GCC states, with respondent distribution ranging from 15–17% per country. All respondents held positions in AI, IT, or digital transformation roles, ensuring relevance to the constructs under study. A non-response bias test (early vs. late respondents) indicated no statistically significant differences, supporting the validity of the sample.Link to Model DevelopmentResponses to Q2–Q9 were used to construct the latent variables Technical Infrastructure, Organizational Readines, and Governance Environment, while Q10–Q12 were modeled as AI Outcomes in the PLS-SEM. These survey data provided the empirical basis for validating the GCC-specific AI Adoption Index proposed in this study.

The SmartPLS Analysis Report documents the Partial Least Squares Structural Equation Modeling (PLS-SEM) procedures used to validate the GCC AI Adoption Index. It provides a transparent, step-by-step account of how measurement and structural models were assessed, ensuring that the study’s findings are reproducible, interpretable, and robust.Key ContentsMeasurement Model ChecksReliability metrics (Cronbach’s α, Composite Reliability, AVE).Outer loadings/weights for reflective and formative constructs.Discriminant validity (Fornell–Larcker criterion, HTMT ratios).Structural Model ResultsPath coefficients (β values) with bootstrapped t- and p-values.R² and adjusted R² values showing explanatory power.Effect sizes (f²) and predictive relevance (Q²).Model fit indices (SRMR, NFI) for overall adequacy.Construct RelationshipsTechnical Infrastructure, Organizational Capability, and Governance are modeled as predictors of AI Outcomes.Results highlight the dominant role of Technical Infrastructure (β ≈ 0.657) and significant contribution of Governance (β ≈ 0.206), while Organizational Capability shows negligible influence (β ≈ 0.016).Index DerivationThe report also includes the path-based weights used to compute the GCC AI Adoption Index, rescaled to a 0–100 range.PurposeThis report serves as the technical backbone of the study, allowing other researchers to audit, replicate, or extend the model. By providing full outputs (tables, figures, and algorithmic settings), it strengthens the study’s credibility and supports transparent, evidence-based policy design for AI adoption in the GCC.

The SmartPLS Analysis Report documents the Partial Least Squares Structural Equation Modeling (PLS-SEM) procedures used to validate the GCC AI Adoption Index. It provides a transparent, step-by-step account of how measurement and structural models were assessed, ensuring that the study’s findings are reproducible, interpretable, and robust. Key Contents Measurement Model Checks Reliability metrics (Cronbach’s α, Composite Reliability, AVE). Outer loadings/weights for reflective and formative constructs. Discriminant validity (Fornell–Larcker criterion, HTMT ratios). Structural Model Results Path coefficients (β values) with bootstrapped t- and p-values. R² and adjusted R² values showing explanatory power. Effect sizes (f²) and predictive relevance (Q²). Model fit indices (SRMR, NFI) for overall adequacy. Construct Relationships Technical Infrastructure, Organizational Readiness, and Governance Environment are modeled as predictors of AI Outcomes. Results highlight the dominant role of Technical Infrastructure (β ≈ 0.657) and significant contribution of Governance Enviroment (β ≈ 0.206), while Organizational Readiness shows negligible influence (β ≈ 0.016). Index Derivation The report also includes the path-based weights used to compute the GCC AI Adoption Index, rescaled to a 0–100 range. Purpose This report serves as the technical backbone of the study, allowing other researchers to audit, replicate, or extend the model. By providing full outputs (tables, figures, and algorithmic settings), it strengthens the study’s credibility and supports transparent, evidence-based policy design for AI adoption in the GCC.

This codebook provides a structured framework for analyzing the six Gulf Cooperation Council (GCC) National AI Strategies (NASs) with a focus on capacity-building references. It links policy priorities explicitly to the constructs used in our empirical model, Technical Infrastructure (TI), Organizational Capability (OR), and Governance Environment (GE), and supports alignment with survey items Q2–Q12. The codebook combines qualitative categorization and quantitative weighting. Each NAS passage is coded into a primary category (TI, OR, or GE) using well-defined sub-codes, while secondary tags capture intended outcomes (AIO: service efficiency, citizen trust, performance) and contextual dimensions (for example, sector, localization, cultural factors). This ensures that both policy intentions and implementation pathways are captured consistently across countries. To assess emphasis on specific AI capabilities (for example, NLP, robotics, cybersecurity), the codebook applies term-salience analysis (TF–IDF) within the coded passages. This allows us to differentiate between universally mentioned enablers, such as AI ethics or cybersecurity, and more specialized but heavily emphasized capabilities, such as deep learning or robotics, that feature in only some national strategies. The design emphasizes reproducibility and rigor: coders follow clear inclusion/exclusion rules, use EN/AR synonym lexicons, and apply double-coding with inter-coder reliability checks (target κ ≥ 0.79). Outputs include normalized construct-level signals per country and comparative TF–IDF tables, which are then triangulated with survey and PLS-SEM findings. In sum, the codebook acts as a bridge between policy texts and empirical analysis, enabling transparent, comparable, and replicable measurement of AI adoption enablers across the GCC.

The SmartPLS Analysis Report documents the Partial Least Squares Structural Equation Modeling (PLS-SEM) procedures used to validate the GCC AI Adoption Index. It provides a transparent, step-by-step account of how measurement and structural models were assessed, ensuring that the study’s findings are reproducible, interpretable, and robust. Key Contents Measurement Model Checks Reliability metrics (Cronbach’s α, Composite Reliability, AVE). Outer loadings/weights for reflective and formative constructs. Discriminant validity (Fornell–Larcker criterion, HTMT ratios). Structural Model Results Path coefficients (β values) with bootstrapped t- and p-values. R² and adjusted R² values showing explanatory power. Effect sizes (f²) and predictive relevance (Q²). Model fit indices (SRMR, NFI) for overall adequacy. Construct Relationships Technical Infrastructure, Organizational Readiness, and Governance Environment are modeled as predictors of AI Outcomes. Results highlight the dominant role of Technical Infrastructure (β ≈ 0.657) and significant contribution of Governance Enviroment (β ≈ 0.206), while Organizational Readiness shows negligible influence (β ≈ 0.016). Index Derivation The report also includes the path-based weights used to compute the GCC AI Adoption Index, rescaled to a 0–100 range. Purpose This report serves as the technical backbone of the study, allowing other researchers to audit, replicate, or extend the model. By providing full outputs (tables, figures, and algorithmic settings), it strengthens the study’s credibility and supports transparent, evidence-based policy design for AI adoption in the GCC.

Replication package (v6.5) for the study “AI-Enabled Transport for Kuwait’s 2035 Roadmap: Bridging the Tech-Policy Gap vis-à-vis the UAE and Singapore”. This dataset contains the public data derivatives, analysis code, and documentation required to reproduce the study’s results.Documentation included: (1) Online Codebook: authoritative data dictionary and coding guide covering case/source attributes; a nine-family thematic code hierarchy (policy, finance, data, safety, implementation, AI initiative types, outcomes, capacity, transferability); and Technology–Policy Readiness Matrix (TPRM) scoring rules for TRL and PRL, gap banding, and confidence. It also specifies inclusion/exclusion criteria, coder workflow, reliability targets, and provides machine-readable templates/exports. (2) Online Appendix: extended methods (TPRM ladders; scoring basis, evidence rules, and adjudication; intercoder reliability), the TF-IDF specification for National AI Strategy salience, and the screening protocol for the 2014–2025 documentary corpus. It includes printable templates/SOPs (TPRM scoring sheet; corridor KPI pack; AV sandbox safety reporting; Transferability Box; Scenario Mapping; minimal Transport Data & API circular), the full corpus list (Kuwait 16; UAE 12; Singapore 11), reproducibility and access notes (with restricted human-subjects materials), and supplementary scenario figures (E1–E5) with scripts to regenerate the panels.What the files cover: Kuwait, UAE (RTA), and Singapore (LTA); TRL/PRL scoring with adjudication; TF-IDF counts and scripts; intercoder kappa (κ) inputs/logs; figure scripts/outputs; and machine-readable codebook exports.Access and ethics: Public files include data derivatives, code, and documentation. Human-subjects derivatives (for example, de-identified excerpts) are provided under Restricted access with Terms of Access aligned to IRB Protocol #420 25.Licensing: Documentation (Appendix, Codebook): CC BY 4.0. Code: MIT/Apache 2.0 (see LICENSE in /code). Restricted files are governed by their Terms of Access.Study cutoff: 31 December 2025.