Clear, high-quality data builds trust in AI decision-support systems, and that trust helps drive organizations' intentions to adopt them; those intentions correlate with higher self-reported decision-making efficiency.

Decision-Making in Complex Systems Using AI-Based Decision Support: The Role of Trust, Transparency, and Data Quality

G. Bondac, S. Stanescu, C. Ionescu, Anisoara Duica, Marilena Carmen Uzlău · Fetched April 11, 2026 · Electronics

semantic_scholar correlational low evidence 7/10 relevance DOI Source PDF

In a survey of 324 Romanian organizational respondents, data transparency and quality strongly predict trust in AI decision-support systems, and trust—together with perceived usefulness and ease of use—positively predicts adoption intentions, which are associated with higher self-reported decision-making efficiency.

In the context of accelerated digital transformation, organizations increasingly operate as complex systems in which strategic decision-making is challenged by uncertainty, data heterogeneity, and bounded rationality. The integration of artificial intelligence (AI) into organizational processes is therefore redefining how decisions are supported and enacted. This study develops and validates an integrated conceptual model that explains how trust in AI-based decision support systems (AI-DSSs), data transparency and quality, perceived usefulness, and ease of use influence decision-making efficiency and the intention to adopt AI-DSS in complex organizational contexts. The empirical analysis is based on a questionnaire survey administered to 324 respondents from Romanian organizations operating in IT, services, industry, and public administration. Data were analyzed using partial least squares structural equation modeling (PLS-SEM) implemented in SmartPLS 4. The results show that data transparency and quality strongly enhance trust in AI-DSS (β = 0.784, p < 0.001). Trust positively influences both perceived usefulness (β = 0.229, p < 0.01) and perceived ease of use (β = 0.482, p < 0.001), confirming its role as a key psychological enabler of favorable technology perceptions. Furthermore, perceived ease of use significantly affects perceived usefulness (β = 0.597, p < 0.001). Regarding adoption-related attitudes, perceived usefulness (β = 0.352, p < 0.001), trust (β = 0.311, p < 0.001), and perceived ease of use (β = 0.135, p < 0.05) exert significant positive effects on the intention to adopt AI-DSS, which in turn demonstrates a strong association with decision-making efficiency (β = 0.544, p < 0.001). By extending traditional technology acceptance models (TAM) with AI-specific dimensions—namely transparency, data quality, and trust—this study contributes to the literature on decision-making in complex systems and offers practical insights for organizations seeking to improve decision effectiveness through AI-based support.

Summary

Main Finding

Extending the Technology Acceptance Model with AI-specific dimensions (data transparency, data quality, and trust) explains adoption of AI-based decision support systems (AI-DSS) in complex organizational settings. Data transparency and quality strongly increase trust in AI-DSS, which in turn raises perceived ease of use and perceived usefulness. These perceptions and trust drive intention to adopt AI-DSS, and adoption intention strongly predicts improvements in decision-making efficiency.

Key Points

Sample: 324 respondents from Romanian organizations across IT, services, industry, and public administration.
Analytical method: Partial least squares structural equation modeling (PLS-SEM) using SmartPLS 4.
Strongest path:
- Data transparency & quality → Trust (β = 0.784, p < 0.001).
Trust effects:
- Trust → Perceived ease of use (β = 0.482, p < 0.001).
- Trust → Perceived usefulness (β = 0.229, p < 0.01).
Perceived ease of use → Perceived usefulness (β = 0.597, p < 0.001).
Adoption intention drivers:
- Perceived usefulness → Intention to adopt (β = 0.352, p < 0.001).
- Trust → Intention to adopt (β = 0.311, p < 0.001).
- Perceived ease of use → Intention to adopt (β = 0.135, p < 0.05).
Outcome link:
- Intention to adopt → Decision-making efficiency (β = 0.544, p < 0.001).
Contribution: Validates that AI-specific constructs (transparency, data quality, trust) are crucial complements to TAM in explaining AI-DSS adoption and downstream efficiency gains.

Data & Methods

Design: Cross-sectional questionnaire survey.
Respondents: 324 organizational participants from Romania (sectors: IT, services, industry, public administration).
Measurement: Constructs for data transparency, data quality, trust, perceived usefulness, perceived ease of use, intention to adopt AI-DSS, and decision-making efficiency.
Analysis: PLS-SEM (SmartPLS 4) to estimate path coefficients and significance.
Notes on limits of methods:
- Self-reported, cross-sectional data limit causal claims.
- Context-specific (Romanian organizations); generalizability should be tested in other regions/sectors.
- Paper does not report (here) R² or model fit indices—interpretation should consider potential omitted-variable bias.

Implications for AI Economics

Investment priorities: Improving data quality and transparency yields high returns via trust formation, which cascades into higher perceived usefulness and adoption—suggesting organizations should allocate resources to data governance and explainability features alongside algorithm development.
Adoption economics: Trust acts as a psychological complement to technical usability; investments in UX and trust-building (explainability, auditability, certification) can increase adoption and thus realize efficiency gains faster.
Productivity and ROI: The strong link from adoption intention to decision-making efficiency (β = 0.544) implies that managerial focus on adoption barriers can produce measurable organizational productivity benefits; economic evaluations of AI projects should include these mediating factors.
Policy and market design: Regulators and standards bodies that improve transparency and data quality (through guidelines, certifications, or required disclosures) can lower frictions in AI adoption and accelerate organizational digitization.
Strategic fit in complex systems: In environments of bounded rationality and heterogeneous data, trust-enabled AI-DSS can serve as a coordination mechanism—affecting diffusion dynamics, complementarities between human capital and AI tools, and the distribution of gains from AI-enabled decisions.
Future research directions for AI economics: quantify cost-effectiveness of data governance vs. model improvements, measure long-term impacts of adoption on firm performance, and model adoption dynamics under heterogeneous trust/skill distributions.

Assessment

Paper Typecorrelational Evidence Strengthlow — The study is a cross-sectional, self-reported survey analyzed with PLS-SEM and does not use experimental or quasi-experimental variation to identify causal effects; relationships are associative and potentially affected by common-method bias, reverse causality, and omitted variables despite statistically significant path coefficients. Methods Rigormedium — The paper applies established structural equation modeling (PLS-SEM) to test an extended TAM framework and reports standardized coefficients and significance levels, and the sample size (N=324) is adequate for PLS; however, rigor is limited by reliance on cross-sectional self-reports, likely convenience sampling, limited discussion of measurement validation/robustness checks and common-method bias, and lack of external validation or longitudinal data. SampleQuestionnaire survey of 324 respondents from Romanian organizations operating in IT, services, industry, and public administration (roles not fully specified); cross-sectional self-reported measures of trust, data transparency and quality, perceived usefulness/ease of use, adoption intention, and decision-making efficiency. Themesadoption human_ai_collab productivity GeneralizabilitySingle-country sample (Romania) — cultural and institutional contexts may limit applicability to other countries, Likely non-probability/convenience sampling and unspecified role distribution — sample may not represent broader managerial populations, Cross-sectional self-reports may not reflect actual adoption behavior or objective decision-making performance, Types of AI-DSS and organizational size/sector heterogeneity not fully specified, limiting transferability across technologies and firms

Claims (9)

Claim	Direction	Confidence	Outcome	Details
The empirical analysis is based on a questionnaire survey administered to 324 respondents from Romanian organizations operating in IT, services, industry, and public administration. Other	null_result	high	sample description / data source	n=324 0.5
Data were analyzed using partial least squares structural equation modeling (PLS-SEM) implemented in SmartPLS 4. Other	null_result	high	analysis method	n=324 0.5
Data transparency and quality strongly enhance trust in AI-based decision support systems (AI-DSS) (β = 0.784, p < 0.001). Other	positive	high	trust in AI-based decision support systems	n=324 β = 0.784, p < 0.001 0.3
Trust positively influences perceived usefulness of AI-DSS (β = 0.229, p < 0.01). Adoption Rate	positive	high	perceived usefulness of AI-DSS	n=324 β = 0.229, p < 0.01 0.3
Trust positively influences perceived ease of use of AI-DSS (β = 0.482, p < 0.001). Adoption Rate	positive	high	perceived ease of use of AI-DSS	n=324 β = 0.482, p < 0.001 0.3
Perceived ease of use significantly affects perceived usefulness (β = 0.597, p < 0.001). Adoption Rate	positive	high	perceived usefulness of AI-DSS	n=324 β = 0.597, p < 0.001 0.3
Perceived usefulness (β = 0.352, p < 0.001), trust (β = 0.311, p < 0.001), and perceived ease of use (β = 0.135, p < 0.05) exert significant positive effects on the intention to adopt AI-DSS. Adoption Rate	positive	high	intention to adopt AI-DSS	n=324 β = 0.352 (perceived usefulness); β = 0.311 (trust); β = 0.135 (perceived ease of use) 0.3
Intention to adopt AI-DSS demonstrates a strong association with decision-making efficiency (β = 0.544, p < 0.001). Decision Quality	positive	high	decision-making efficiency	n=324 β = 0.544, p < 0.001 0.3
By extending traditional technology acceptance models (TAM) with AI-specific dimensions—namely transparency, data quality, and trust—this study contributes to the literature on decision-making in complex systems and offers practical insights for organizations seeking to improve decision effectiveness through AI-based support. Other	positive	high	conceptual/methodological contribution and practical insights	n=324 0.15