Lersak Phothong
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Lersak Phothong has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Anupong Sukprasert
Mahasarakham Business School, Mahasarakham University, Mahasarakham 44150, Thailand
9 shared publications - Prapassorn Wareesri
Mahasarakham Business School, Mahasarakham University, Maha Sarakham, Thailand
3 shared publications - Mehtabhorn Obthong
Mahasarakham Business School, Mahasarakham University, Mahasarakham 44150, Thailand
2 shared publications Anupong Sukprasert
1 shared publication- Nantana Ngamtampong
Mahasarakham Business School, Mahasarakham University, Maha Sarakham, Thailand
1 shared publication - Nantana Ngamtampong
Mahasarakham Business School, Mahasarakham University, Mahasarakham 44150, Thailand
1 shared publication
Latest Publications
- Open Access
When Predictive Ranking Is Not Enough: Warning Governance and Economic Value in Bitcoin Tail-Risk Analytics
This study forecasts unusually large daily Bitcoin price movements and examines whether the resulting warnings improve investment decisions. We compare simple volatility signals, a benchmark based on the previous day’s return magnitude, logistic regression, and random forests in chronological tests covering 2340 days. Large movements include gains and losses whose absolute daily log returns exceed the 95th percentile of the preceding training sample. The evaluation contains 51 such events. The selected volatility signal issues 77 warnings and captures nine events. Better predictive ranking does not consistently produce better warnings. Additional comparisons use model tuning, common nominal warning budgets, and a separate analysis of large losses. Investment evaluation accounts for trading costs, cash returns, execution delays, and investor preferences. Moving fully to cash after the selected loss warnings can reduce investor utility relative to buying and holding Bitcoin. In contrast, adjusting exposure using volatility over the preceding 20 days improves utility at the highest tested risk aversion, despite lowering total return. These historical comparisons are exploratory; uncertainty estimates condition on recorded forecasts and are not adjusted for multiple comparisons. The findings support separate evaluation of predictive ranking, warning governance, and economic value.
Open AccessWhen Predictive Ranking Is Not Enough: Warning Governance and Economic Value in Bitcoin Tail-Risk Analytics
This study forecasts unusually large daily Bitcoin price movements and examines whether the resulting warnings improve investment decisions. We compare simple volatility signals, a benchmark based on the previous day’s return magnitude, logistic regression, and random forests in chronological tests covering 2340 days. Large movements include gains and losses whose absolute daily log returns exceed the 95th percentile of the preceding training sample. The evaluation contains 51 such events. The selected volatility signal issues 77 warnings and captures nine events. Better predictive ranking does not consistently produce better warnings. Additional comparisons use model tuning, common nominal warning budgets, and a separate analysis of large losses. Investment evaluation accounts for trading costs, cash returns, execution delays, and investor preferences. Moving fully to cash after the selected loss warnings can reduce investor utility relative to buying and holding Bitcoin. In contrast, adjusting exposure using volatility over the preceding 20 days improves utility at the highest tested risk aversion, despite lowering total return. These historical comparisons are exploratory; uncertainty estimates condition on recorded forecasts and are not adjusted for multiple comparisons. The findings support separate evaluation of predictive ranking, warning governance, and economic value.
- Open Access
- 8 November 2026
- Published by MDPI AG in Informatics
- Vol. 13 (11) , 189
Public Negotiation of AI-Mediated Cognition in Reddit Discourse
We examine how acceptable AI assistance is related to continuing human responsibility in public discussion. Our analysis covers 208,252 comments from selected Reddit communities using nonexclusive lexical indicators, BERTopic on a fixed 50,000-comment subset, contextual interpretation, engagement measures and monthly distributions. AI-assisted learning and broader educational discussion formed the largest non-outlier clusters. Illustrative arguments made assistance conditional on retained understanding, source checking and accountable professional judgment; other selected topics supplied educational, institutional or human capacity context. The six largest primary clusters recurred across 13 observed months, while frequency, scores and extended comments captured different forms of prominence. A coverage-oriented 120-case label fit assessment yielded coder agreement of 50/60 for indicators and 60/60 for topics, including one shared topic label rejection. Nine sensitivity reconstructions produced 14–35 non-outlier topics, with variation in cluster boundaries and the separation of smaller themes. The study contributes to social informatics by showing how cognitive responsibility is articulated through relations among AI assistance, human competence, verification and institutional expectations within a purposively sampled platform environment.
- 8 November 2026
- Published by MDPI AG in Informatics
- Vol. 13 (11) , 189
Open AccessPublic Negotiation of AI-Mediated Cognition in Reddit Discourse
We examine how acceptable AI assistance is related to continuing human responsibility in public discussion. Our analysis covers 208,252 comments from selected Reddit communities using nonexclusive lexical indicators, BERTopic on a fixed 50,000-comment subset, contextual interpretation, engagement measures and monthly distributions. AI-assisted learning and broader educational discussion formed the largest non-outlier clusters. Illustrative arguments made assistance conditional on retained understanding, source checking and accountable professional judgment; other selected topics supplied educational, institutional or human capacity context. The six largest primary clusters recurred across 13 observed months, while frequency, scores and extended comments captured different forms of prominence. A coverage-oriented 120-case label fit assessment yielded coder agreement of 50/60 for indicators and 60/60 for topics, including one shared topic label rejection. Nine sensitivity reconstructions produced 14–35 non-outlier topics, with variation in cluster boundaries and the separation of smaller themes. The study contributes to social informatics by showing how cognitive responsibility is articulated through relations among AI assistance, human competence, verification and institutional expectations within a purposively sampled platform environment.
- 26 October 2026
- Published by Elsevier in Research in International Business and Finance
- Vol. 92
Machine learning-based hedge ratios for cryptocurrency futures under regime-dependent tail risk
- 26 October 2026
- Published by Elsevier in Research in International Business and Finance
- Vol. 92
Machine learning-based hedge ratios for cryptocurrency futures under regime-dependent tail risk
- Open Access
- 25 October 2026
- Published by MDPI AG in Sustainability
- Vol. 18 (21) , 10831
Responsible GenAI Augmentation and Cognitive Agency in Higher Education
Responsible generative artificial intelligence (GenAI) integration in higher education requires attention to useful assistance and students’ continuing cognitive agency. We examined how learner capability, cognitive offloading and institutional support relate to intention to use a university GenAI learning service, alongside reported reduced independent thinking. A cross-sectional survey of 593 students at one Thai public university was analyzed using ordinal structural equation modeling with WLSMV estimation. AI self-efficacy was positively associated with information utilization, utilization with perceived work efficiency, and efficiency with service intention. The serial indirect association was positive, while adjusted direct self-efficacy associations with efficiency and intention were inconclusive. Offloading was positively associated with both efficiency and reduced independent thinking. Its positive indirect association with intention through efficiency coexisted with a negative adjusted direct association. Institutional support was positively associated with intention. The prespecified TRUST comparison and additional covariates, screening and complete-case checks, retained the main association pattern. However, replacing utilization with information analysis reduced the analogous serial association, and reversing the utilization–efficiency link gave better descriptive fit. The findings therefore concern concurrent perceptions and reported practices, without establishing serial ordering or causal mediation. Independent learning and actual service engagement remain essential criteria for assessing educational value.
- 25 October 2026
- Published by MDPI AG in Sustainability
- Vol. 18 (21) , 10831
Open AccessResponsible GenAI Augmentation and Cognitive Agency in Higher Education
Responsible generative artificial intelligence (GenAI) integration in higher education requires attention to useful assistance and students’ continuing cognitive agency. We examined how learner capability, cognitive offloading and institutional support relate to intention to use a university GenAI learning service, alongside reported reduced independent thinking. A cross-sectional survey of 593 students at one Thai public university was analyzed using ordinal structural equation modeling with WLSMV estimation. AI self-efficacy was positively associated with information utilization, utilization with perceived work efficiency, and efficiency with service intention. The serial indirect association was positive, while adjusted direct self-efficacy associations with efficiency and intention were inconclusive. Offloading was positively associated with both efficiency and reduced independent thinking. Its positive indirect association with intention through efficiency coexisted with a negative adjusted direct association. Institutional support was positively associated with intention. The prespecified TRUST comparison and additional covariates, screening and complete-case checks, retained the main association pattern. However, replacing utilization with information analysis reduced the analogous serial association, and reversing the utilization–efficiency link gave better descriptive fit. The findings therefore concern concurrent perceptions and reported practices, without establishing serial ordering or causal mediation. Independent learning and actual service engagement remain essential criteria for assessing educational value.
- Open Access
Walk-Forward Evaluation of Early-Warning Models for Large Bitcoin Movements Under Fixed and Train-Only Event Definitions
Evaluations of financial early-warning models can appear stronger when temporal separation is incomplete or when the event definition uses information unavailable at the forecast origin. This study examines one-day-ahead risk ranking of large absolute Bitcoin returns using 334 daily observations from 31 August 2024 to 30 July 2025, five expanding-window folds, lagged predictors, training-fold preprocessing, and train-only classification-threshold selection. The original q85 and q90 cutoffs, estimated from the full study-period return distribution, are retained as retrospective fixed-label benchmarks and are complemented by fold-specific train-only cutoffs. Under fixed q85 labels, class-weighted Logistic Regression with technical and market-activity variables achieved pooled ROC-AUC = 0.723 (moving-block 95% CI [0.556, 0.822]) and Average Precision = 0.338 ([0.103, 0.565]); Random Forest yielded 0.739 and 0.352. Pooled MCC was 0.207, but the primary model issued no positive warnings in three of the five fixed-q85 test folds. Under train-only q85 labels, ROC-AUC was 0.674 and Average Precision was 0.174, but the retained MCC threshold-selection rule produced no true positives. Conventional GARCH(1,1), EWMA, and 14-day historical-volatility scores did not exceed the primary model on both ranking metrics. Adding the complete six-variable sentiment and attention block lowered point estimates, although one-variable ablations and a matched-dimension control do not isolate sentiment content from dimensionality and small-sample overfitting. The results provide limited, sample-specific evidence for risk ranking, not evidence of deployment readiness, economic profitability, or cross-asset generalizability.
Open AccessWalk-Forward Evaluation of Early-Warning Models for Large Bitcoin Movements Under Fixed and Train-Only Event Definitions
Evaluations of financial early-warning models can appear stronger when temporal separation is incomplete or when the event definition uses information unavailable at the forecast origin. This study examines one-day-ahead risk ranking of large absolute Bitcoin returns using 334 daily observations from 31 August 2024 to 30 July 2025, five expanding-window folds, lagged predictors, training-fold preprocessing, and train-only classification-threshold selection. The original q85 and q90 cutoffs, estimated from the full study-period return distribution, are retained as retrospective fixed-label benchmarks and are complemented by fold-specific train-only cutoffs. Under fixed q85 labels, class-weighted Logistic Regression with technical and market-activity variables achieved pooled ROC-AUC = 0.723 (moving-block 95% CI [0.556, 0.822]) and Average Precision = 0.338 ([0.103, 0.565]); Random Forest yielded 0.739 and 0.352. Pooled MCC was 0.207, but the primary model issued no positive warnings in three of the five fixed-q85 test folds. Under train-only q85 labels, ROC-AUC was 0.674 and Average Precision was 0.174, but the retained MCC threshold-selection rule produced no true positives. Conventional GARCH(1,1), EWMA, and 14-day historical-volatility scores did not exceed the primary model on both ranking metrics. Adding the complete six-variable sentiment and attention block lowered point estimates, although one-variable ablations and a matched-dimension control do not isolate sentiment content from dimensionality and small-sample overfitting. The results provide limited, sample-specific evidence for risk ranking, not evidence of deployment readiness, economic profitability, or cross-asset generalizability.
- Open Access
- 15 June 2026
- Published by Elsevier in SSRN Electronic Journal
Machine Learning-Based Hedge Ratios for Cryptocurrency Futures under Regime-Dependent Tail Risk
This study examines whether machine learning models generate more resilient hedge-ratio estimates than traditional econometric models under regime-dependent tail risk in cryptocurrency futures markets. Using daily BTC futures data from 2019 to 2025 and ETH futures data from 2021 to 2025, we compare Ordinary Least Squares (OLS), Dynamic Conditional Correlation GARCH (DCC‑GARCH), and Random Forest (RF) against a naïve benchmark across four stress regimes — the COVID-19 crash, the TerraLuna collapse, the FTX bankruptcy, and the Federal Reserve tightening cycle. Three findings emerge. First, Random Forest exhibits economically meaningful resilience advantages for BTC during stress periods — hedging effectiveness declines by 10.4 percentage points versus 20.7 for DCC‑GARCH — though this difference does not reach statistical significance given the limited stress subsample. Second, for ETH, dynamic models generally outperform OLS in variance-based and CVaR-based measures, though OLS achieves the highest full-sample VaR reduction. Third, CVaR reduction declines sharply during stress periods across all models; CER analysis confirms that the economic utility of futures hedging is severely limited for ETH investors during extreme dislocations. The findings underscore the importance of regime-sensitive, asset-specific hedging strategies and demonstrate that machine learning can enhance hedging resilience under certain stress conditions, with direct implications for institutional risk management in volatile cryptocurrency markets.
- 15 June 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessMachine Learning-Based Hedge Ratios for Cryptocurrency Futures under Regime-Dependent Tail Risk
This study examines whether machine learning models generate more resilient hedge-ratio estimates than traditional econometric models under regime-dependent tail risk in cryptocurrency futures markets. Using daily BTC futures data from 2019 to 2025 and ETH futures data from 2021 to 2025, we compare Ordinary Least Squares (OLS), Dynamic Conditional Correlation GARCH (DCC‑GARCH), and Random Forest (RF) against a naïve benchmark across four stress regimes — the COVID-19 crash, the TerraLuna collapse, the FTX bankruptcy, and the Federal Reserve tightening cycle. Three findings emerge. First, Random Forest exhibits economically meaningful resilience advantages for BTC during stress periods — hedging effectiveness declines by 10.4 percentage points versus 20.7 for DCC‑GARCH — though this difference does not reach statistical significance given the limited stress subsample. Second, for ETH, dynamic models generally outperform OLS in variance-based and CVaR-based measures, though OLS achieves the highest full-sample VaR reduction. Third, CVaR reduction declines sharply during stress periods across all models; CER analysis confirms that the economic utility of futures hedging is severely limited for ETH investors during extreme dislocations. The findings underscore the importance of regime-sensitive, asset-specific hedging strategies and demonstrate that machine learning can enhance hedging resilience under certain stress conditions, with direct implications for institutional risk management in volatile cryptocurrency markets.
- 11 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. PP, 1-14
Reddit as a Semantic Sensor: Forecasting Forex Movements via Emotionally Weighted Signals in Computational Social Systems
This study presents an empirical framework for forecasting directional movements of the EUR/USD foreign exchange market using sentiment signals extracted from Reddit. Sentiment is classified with FinBERT (a financial-domain adaptation of Bidirectional Encoder Representations from Transformers, BERT) and transformed through an engagement-weighted formulation to capture social amplification effects, with a two-day behavioral lag applied to align sentiment with market responses. Feature engineering incorporates linguistic attributes, interaction terms, and rolling sentiment averages. Four machine learning models—Logistic Regression, Random Forest, XGBoost (XGB), and Neural Networks—are evaluated using walk-forward time-aware cross-validation. Results indicate that XGB achieves the highest average predictive performance, with accuracy around 0.78 and consistently high AUC under walk-forward evaluation, relative to standard price-based and naive baseline models. Model interpretability is examined using SHapley Additive Explanations, while a posthoc semantic analysis employing BERTopic, WordClouds, and uniform manifold approximation and projection (UMAP) shows that correct predictions are associated with coherent and recurring discourse patterns related to market conditions and sentiment expression. Overall, the findings demonstrate that engagement-weighted Reddit sentiment provides a robust and interpretable signal for short-horizon Forex forecasting.
- 11 May 2026
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Computational Social Systems
- Vol. PP, 1-14
Reddit as a Semantic Sensor: Forecasting Forex Movements via Emotionally Weighted Signals in Computational Social Systems
This study presents an empirical framework for forecasting directional movements of the EUR/USD foreign exchange market using sentiment signals extracted from Reddit. Sentiment is classified with FinBERT (a financial-domain adaptation of Bidirectional Encoder Representations from Transformers, BERT) and transformed through an engagement-weighted formulation to capture social amplification effects, with a two-day behavioral lag applied to align sentiment with market responses. Feature engineering incorporates linguistic attributes, interaction terms, and rolling sentiment averages. Four machine learning models—Logistic Regression, Random Forest, XGBoost (XGB), and Neural Networks—are evaluated using walk-forward time-aware cross-validation. Results indicate that XGB achieves the highest average predictive performance, with accuracy around 0.78 and consistently high AUC under walk-forward evaluation, relative to standard price-based and naive baseline models. Model interpretability is examined using SHapley Additive Explanations, while a posthoc semantic analysis employing BERTopic, WordClouds, and uniform manifold approximation and projection (UMAP) shows that correct predictions are associated with coherent and recurring discourse patterns related to market conditions and sentiment expression. Overall, the findings demonstrate that engagement-weighted Reddit sentiment provides a robust and interpretable signal for short-horizon Forex forecasting.
- Open Access
- 18 March 2026
- Published by MDPI AG in Information
- Vol. 17 (3) , 295
Seeing the Message but Not the Machine: Digital Skepticism and AI Discernment in Online Information Environments
Artificial intelligence (AI) increasingly mediates how information is generated, ranked, and circulated in digital environments. However, it remains unclear under what conditions users explicitly articulate recognition of AI involvement in routine news-related discourse. This study examines how digital skepticism and AI-related discernment are expressed in naturally occurring social media discussions. Using an exploratory observational design, 6065 user-generated comments from 305 news-related Reddit threads were analyzed through a rule-based framework distinguishing general skepticism, structural suspicion, and explicit AI-related discernment. Within the sampled corpus, generalized digital skepticism is proportionally more visible than explicit attribution to AI-generated or synthetically produced content. Explicit AI-related attribution is unevenly distributed across discourse contexts, appearing more frequently in technology-oriented communities and remaining limited in mainstream news-related discussions. Differences across score-based visibility contexts do not correspond to a consistently higher representation of explicit AI attribution. These findings indicate a distributional difference between generalized skepticism and publicly articulated recognition of AI mediation. Rather than measuring levels of awareness, the results illuminate the contextual and linguistic conditions under which AI involvement becomes explicitly named in public interaction. By focusing on observable discourse rather than self-reported attitudes, the study provides a corpus-bound account of when AI mediation becomes discursively articulated in algorithmically mediated environments.
- 18 March 2026
- Published by MDPI AG in Information
- Vol. 17 (3) , 295
Open AccessSeeing the Message but Not the Machine: Digital Skepticism and AI Discernment in Online Information Environments
Artificial intelligence (AI) increasingly mediates how information is generated, ranked, and circulated in digital environments. However, it remains unclear under what conditions users explicitly articulate recognition of AI involvement in routine news-related discourse. This study examines how digital skepticism and AI-related discernment are expressed in naturally occurring social media discussions. Using an exploratory observational design, 6065 user-generated comments from 305 news-related Reddit threads were analyzed through a rule-based framework distinguishing general skepticism, structural suspicion, and explicit AI-related discernment. Within the sampled corpus, generalized digital skepticism is proportionally more visible than explicit attribution to AI-generated or synthetically produced content. Explicit AI-related attribution is unevenly distributed across discourse contexts, appearing more frequently in technology-oriented communities and remaining limited in mainstream news-related discussions. Differences across score-based visibility contexts do not correspond to a consistently higher representation of explicit AI attribution. These findings indicate a distributional difference between generalized skepticism and publicly articulated recognition of AI mediation. Rather than measuring levels of awareness, the results illuminate the contextual and linguistic conditions under which AI involvement becomes explicitly named in public interaction. By focusing on observable discourse rather than self-reported attitudes, the study provides a corpus-bound account of when AI mediation becomes discursively articulated in algorithmically mediated environments.
- Top Cited PapersOpen Access
- 23 January 2026
- Published by MDPI AG in Forecasting
- Vol. 8 (1) , 10
An Explainable Voting Ensemble Framework for Early-Warning Forecasting of Corporate Financial Distress
Accurate early-warning forecasting of corporate financial distress remains a critical challenge due to nonlinear financial relationships, severe data imbalance, and the high operational costs of false alarms in risk-monitoring systems. This study proposes an explainable voting ensemble framework for early-warning forecasting of corporate financial distress using lagged accounting-based financial information. The proposed framework integrates heterogeneous base learners, including Decision Tree, Neural Network, and k-Nearest Neighbors models, and is evaluated using financial statement data from 752 publicly listed firms in Thailand, comprising sixteen financial ratios across six dimensions: liquidity, operating efficiency, debt management, profitability, earnings quality, and solvency. To ensure robustness under imbalanced and rare-event conditions, the study employs feature selection, data normalization, stratified cross-validation, resampling techniques, and repeated validation procedures. Empirical results demonstrate that the proposed Voting Ensemble delivers a precision-oriented and decision-relevant forecasting profile, outperforming classical classifiers and maintaining greater early-warning reliability when benchmarked against advanced tree-based ensemble models. Probability-based evaluation further confirms the robustness and calibration stability of the proposed framework under repeated cross-validation. By adopting a forward-looking, early-warning perspective and integrating ensemble learning with explainable machine learning principles, this study offers a transparent and scalable approach to financial distress forecasting. The findings offer practical implications for auditors, investors, and regulators seeking reliable early-warning tools for corporate risk assessment, particularly in emerging market environments characterized by data imbalance and heightened uncertainty.
- 23 January 2026
- Published by MDPI AG in Forecasting
- Vol. 8 (1) , 10
Top Cited PapersOpen AccessAn Explainable Voting Ensemble Framework for Early-Warning Forecasting of Corporate Financial Distress
Accurate early-warning forecasting of corporate financial distress remains a critical challenge due to nonlinear financial relationships, severe data imbalance, and the high operational costs of false alarms in risk-monitoring systems. This study proposes an explainable voting ensemble framework for early-warning forecasting of corporate financial distress using lagged accounting-based financial information. The proposed framework integrates heterogeneous base learners, including Decision Tree, Neural Network, and k-Nearest Neighbors models, and is evaluated using financial statement data from 752 publicly listed firms in Thailand, comprising sixteen financial ratios across six dimensions: liquidity, operating efficiency, debt management, profitability, earnings quality, and solvency. To ensure robustness under imbalanced and rare-event conditions, the study employs feature selection, data normalization, stratified cross-validation, resampling techniques, and repeated validation procedures. Empirical results demonstrate that the proposed Voting Ensemble delivers a precision-oriented and decision-relevant forecasting profile, outperforming classical classifiers and maintaining greater early-warning reliability when benchmarked against advanced tree-based ensemble models. Probability-based evaluation further confirms the robustness and calibration stability of the proposed framework under repeated cross-validation. By adopting a forward-looking, early-warning perspective and integrating ensemble learning with explainable machine learning principles, this study offers a transparent and scalable approach to financial distress forecasting. The findings offer practical implications for auditors, investors, and regulators seeking reliable early-warning tools for corporate risk assessment, particularly in emerging market environments characterized by data imbalance and heightened uncertainty.
- Top Cited PapersOpen Access
- 21 January 2026
- Published by MDPI AG in Sustainability
- Vol. 18 (2) , 1109
Artificial Intelligence Governance in Smart Cities: A Causal Model of Citizen Sustainability Co-Creation Through Acceptance, Trust, and Adaptability
Urban sustainability has become a defining governance challenge as smart cities increasingly integrate artificial intelligence (AI) into public service delivery and decision-making. While AI-enabled systems promise efficiency and responsiveness, growing concerns regarding trust, legitimacy, and citizen engagement suggest that technological adoption alone does not guarantee sustainable urban outcomes. Existing studies have largely emphasized technological performance or individual adoption, paying limited attention to the governance mechanisms through which AI acceptance translates into sustainability co-creation. To address this gap, this study develops and empirically examines the AI–Urban Citizen Sustainability Co-Creation Framework (AI–CSCF) within the context of smart cities in Thailand. A quantitative survey was conducted with 1002 citizens across three smart city settings, and structural equation modeling (SEM) was employed to examine the relationships among AI acceptance, trust in AI, citizen adaptability, and sustainability co-creation. The results indicate that AI acceptance functions as a foundational condition shaping trust in AI and citizen adaptability, through which its influence on sustainability co-creation is indirectly transmitted. Trust in AI emerges as a key mediating mechanism linking AI-enabled governance to participatory sustainability outcomes. These findings underscore the importance of human-centered and trustworthy AI governance that strengthens citizen trust, enhances adaptive capacities, and positions citizens as active co-creators of sustainable urban development aligned with SDG 11.
- 21 January 2026
- Published by MDPI AG in Sustainability
- Vol. 18 (2) , 1109
Top Cited PapersOpen AccessArtificial Intelligence Governance in Smart Cities: A Causal Model of Citizen Sustainability Co-Creation Through Acceptance, Trust, and Adaptability
Urban sustainability has become a defining governance challenge as smart cities increasingly integrate artificial intelligence (AI) into public service delivery and decision-making. While AI-enabled systems promise efficiency and responsiveness, growing concerns regarding trust, legitimacy, and citizen engagement suggest that technological adoption alone does not guarantee sustainable urban outcomes. Existing studies have largely emphasized technological performance or individual adoption, paying limited attention to the governance mechanisms through which AI acceptance translates into sustainability co-creation. To address this gap, this study develops and empirically examines the AI–Urban Citizen Sustainability Co-Creation Framework (AI–CSCF) within the context of smart cities in Thailand. A quantitative survey was conducted with 1002 citizens across three smart city settings, and structural equation modeling (SEM) was employed to examine the relationships among AI acceptance, trust in AI, citizen adaptability, and sustainability co-creation. The results indicate that AI acceptance functions as a foundational condition shaping trust in AI and citizen adaptability, through which its influence on sustainability co-creation is indirectly transmitted. Trust in AI emerges as a key mediating mechanism linking AI-enabled governance to participatory sustainability outcomes. These findings underscore the importance of human-centered and trustworthy AI governance that strengthens citizen trust, enhances adaptive capacities, and positions citizens as active co-creators of sustainable urban development aligned with SDG 11.