Skip to content
Frontiers
Search

Original Research

Frontiers in Sustainability ·

Volume 7 - 2026 | https://doi.org/10.3389/frsus.2026.1965649

Explainable AI and ESG analytics for sustainable investment decision: the roles of ESG forecast credibility and trust in AI-generated information

  • M. Vijayananth1
  • R. Muthu Sundari1
  • N. Saravanabhavan1*corresponding author
  1. Department of Commerce, Vellore Institute of Technology, Vellore, Tamil Nadu, India
Article metrics0 viewsView details
PDFCiteMetrics

Abstract

Artificial intelligence (AI) is transforming sustainable investing, yet limited evidence explains how AI capabilities influence investors stated preferences and willingness toward sustainable investment. This study develops and empirically tests a sequential behavioural framework explaining how AI-related characteristics influence investors stated sustainable investment preferences through perceived ESG forecast credibility and trust in AI-generated ESG information, using survey data from 566 middle-income retail investors across four Indian metropolitan cities. The findings reveal that explainable AI and ESG analytics capability significantly enhance perceived ESG forecast credibility, which sequentially strengthens trust in AI-generated ESG information and is associated with stronger stated preferences and willingness toward sustainable investment. Perceived Greenwashing Risk weakens the positive relationship between Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information, while the interaction between Human-AI Interaction Competency and Trust in AI-generated ESG Information is statistically significant but negligible in practical magnitude (f 2 = 0.012) in relation to Sustainable Investment Decision. The study provides actionable guidance for AI developers, financial institutions, and policymakers to enhance transparency, credibility, investor trust, and responsible AI-enabled sustainable finance.

Keywords

Introduction

In the financial sector, the arrival of Artificial Intelligence (AI) is transforming how investors access, process, and utilize information to drive investment decisions. The developments in machine learning, natural language processing, generative AI and predictive analytics have further enhanced an investor's ability to analyze huge amounts of financial information, improve predictions and make informed investment decisions (Dwivedi et al., 2023; Khan et al., 2025). However, the emphasis on sustainable development has also created a new sphere of investing, known as Environmental, Social, and Governance (ESG) investing, which has included investors looking at companies beyond their financial numbers, and evaluating them on their sustainability, ethical governance, and social responsibility (Berg et al., 2022; Gillan et al., 2021). As a result, AI-driven ESG Analytics has become a valuable decision support tool for investors to provide forward-looking sustainability insights (Del Vitto et al., 2023; Giudici and Wu, 2025; Nurmasari and Bilad, 2026; Ahmad and Ahmed, 2026).

Despite these advances, the use of AI to analyze ESG has introduced challenges around transparency, interpretability, and reliability of AI-generated recommendations. However, many AI systems are considered “black-box” models, which can be hard to understand for investors (Meske et al., 2022; Weber et al., 2024; Khan et al., 2025). Explainable Artificial Intelligence (XAI) aims to make AI systems transparent by providing understandable explanations for how they generate recommendations or decisions (Rai, 2020; Hoffman et al., 2023; Del Vitto et al., 2023; Sullivan and Weger, 2025). But explainability is not enough to drive sustainable investing, if investors believe that AI-generated ESG forecasts are credible, and they perceive AI-generated ESG information as credible and trustworthy. Further, investor confidence in ESG investing remains down as issues such as Perceived Greenwashing Risk, varying ESG rating systems, and uneven ESG disclosure practices persist (Berg et al., 2022; Inderst and Opp, 2025; Zhong et al., 2025).

While research on Perceived Explainable AI, Trust in AI-generated ESG Information, and ESG investing has grown significantly, several theoretical and empirical gaps remain. Previous research on XAI has largely concentrated on healthcare, information systems, and technology adoption, whereas emerging ESG investment research has tended to focus on sustainability disclosure and investor preferences (Meske et al., 2022; Gillan et al., 2021; Giudici and Wu, 2025). Although these studies demonstrate the importance of explainability and ESG information quality, limited attention has been given to how explainability shapes investors stated sustainable investment preferences through credibility and trust mechanisms in AI-assisted ESG investing. Consequently, the cognitive process through which explainable AI and ESG analytics capability influence sustainable investment preferences remains insufficiently understood. Moreover, the moderating roles of Perceived Greenwashing Risk and Human-AI Interaction Competency remain underexplored in AI-driven ESG investing research (Li et al., 2025; Sun and Yin, 2026; Vijayananth and Saravanabhavan, 2026). Addressing these gaps, the present study develops and tests a sequential mechanism linking AI-related characteristics with sustainable investment preferences through credibility and trust, while examining these contextual conditions.

To address these theoretical and empirical gaps, this study develops an integrated behavioral framework that combines Perceived Explainable AI, Perceived ESG Analytics Capability, Perceived ESG Forecast Credibility, and Trust in AI-generated ESG Information to explain Sustainable Investment Decision. The proposed framework is based on the Stimulus – Organism – Response (S–O–R) theory and supported by Signaling Theory and Trust Theory, and analyzes the relationship between the influence of technological characteristics and the investor's stated sustainable investment decision preferences via sequential cognitive evaluations and considers the important boundary conditions.

The study is targeted at middle-income retail investors, one of the fastest-growing groups among the equity market players. They typically have less analytical capacity than institutional investors, tend to use digital investment platforms more and are more susceptible to information asymmetry when assessing information about AI-based ESG. Therefore, it is crucial to understand how this group interprets AI-driven sustainability insights in forming sustainable investment preferences. Recent empirical research also highlights the importance of understanding behavioral responses in applied decision-making contexts (Sundari and Saravanabhavan, 2026).

To address these research gaps, this study investigates the following research questions:

  • RQ1: How do Perceived Explainable AI and Perceived ESG Analytics Capability influence Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information?
  • RQ2: How do Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information influence Sustainable Investment Decision?
  • RQ3: How does Perceived Greenwashing Risk moderate the relationship between Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information?
  • RQ4: How does Human-AI Interaction Competency moderate the relationship between Trust in AI-generated ESG Information and Sustainable Investment Decision?

This study makes three related contributions to the emerging literature on AI-enabled sustainable investing. First, it identifies a sequential cognitive mechanism through which Perceived Explainable AI and Perceived ESG Analytics Capability influence Sustainable Investment Decision through Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information. This highlights credibility as an important cognitive link between AI-related characteristics and investor trust. Second, the study examines Perceived Greenwashing Risk and Human-AI Interaction Competency as contextual conditions that may influence these relationships, thereby identifying circumstances under which credibility and trust translate into sustainable investment preferences. Third, the study provides empirical evidence from middle-income retail investors in India, an emerging-market context that remains comparatively underrepresented in AI-enabled ESG investment research. Collectively, these contributions provide a process-oriented explanation of how AI-generated ESG information may shape sustainable investment preferences through credibility and trust rather than through AI characteristics alone.

The remainder of this manuscript is structured as follows. The following section summarizes the literature and formulates the proposed hypotheses. The next section presents the research methodology, and the empirical findings are presented and interpreted next. The paper then discusses the theoretical and practical implications of the results before concluding with the study's limitations and recommendations for future research.

Review of literature and hypothesis development

Theoretical foundation

This study primarily uses Stimulus–Organism–Response (S–O–R) theory (Mehrabian and Russell, 1974) as the theoretical framework, which helps to understand how external technological stimuli affect internal cognitive evaluation that ultimately leads to responses in behavior. The S–O–R framework is especially well suited to the investment context, where technological cues can shape investors' cognitive evaluations and subsequently influence their stated investment responses (Dwivedi et al., 2023; Huang and Rust, 2021). In this study, the technological stimuli are Perceived Explainable AI (PXAI) and Perceived ESG Analytics Capability (PEAC), while the internal cognitive evaluations through which investors make sense of AI-supported ESG recommendations are Perceived ESG Forecast Credibility (PEFC) and Trust in AI-generated ESG Information (TAEGI). These cognitive evaluations subsequently influence Sustainable Investment Decision (SID), which represents the response component of the S–O–R framework and is operationalized in this study through respondents stated preferences and willingness to incorporate ESG considerations into their investment decisions (Ajzen, 1991).

In addition, to support this behavioral approach, the study also utilizes Signaling Theory (Spence, 1973), which explains how informed parties use observable signals to reduce information asymmetry. In the context of AI-driven ESG investing, AI system developers or providers make observable information about the AI system available through explainability and ESG analytics capability, which serve as informational cues for retail investors who may have limited ability to directly evaluate the underlying AI processes and ESG data. Investors use these cues to assess the credibility of AI-generated ESG forecasts (Connelly et al., 2011; Christensen et al., 2021). Thus, Signaling Theory explains how AI-related characteristics can reduce information uncertainty and inform investors' credibility evaluations, while Perceived Greenwashing Risk may weaken the effectiveness of these signals (Gillan et al., 2021).

The study also adopts the trust theory that suggests reliance on technological systems is influenced by the perceptions of credibility, reliability, and transparency (Mayer et al., 1995; Shin, 2021). Accordingly, credibility and transparency increase investors' willingness to rely on AI-generated ESG information (Mayer et al., 1995; Shin, 2021; Weber et al., 2024).

These theories perform distinct but complementary roles in the proposed framework. S–O–R provides the overall behavioral structure linking AI-related stimuli to investment responses through cognitive evaluations. Signaling Theory explains how explainability and ESG analytics capability provide informational cues for evaluating ESG forecast credibility. Trust Theory explains how favorable credibility evaluations develop into trust and willingness to rely on AI-generated ESG information.

Hypothesis development

AI capabilities and ESG forecast credibility

Perceived Explainable AI (PXAI) refers to investors' perception that AI systems provide transparent, understandable, and interpretable explanations for ESG-based investment recommendations. From a Signaling Theory perspective, explainability provides an observable informational cue through which investors can better understand the reasoning underlying AI-generated ESG recommendations, thereby reducing information uncertainty and supporting their evaluation of forecast credibility (Spence, 1973; Shin, 2021). Investors therefore use AI explanations as cues when evaluating forecast credibility, because the underlying ESG forecasts cannot be directly verified ex ante. As a result, the confidence in AI recommendations is boosted by the use of interpretable AI systems, which improves the credibility of the forecasts (Hoffman et al., 2023).

Likewise, Perceived ESG Analytics Capability (PEAC) is investors' confidence in AI's capability of analyzing intricate ESG data, spotting sustainability risks and opportunities, and delivering precise forecasts. More reliable ESG assessments come from AI systems that have stronger analytical capabilities. From a Signaling Theory perspective, perceived analytical capability provides an informational cue that investors can use to assess the quality and reliability of AI-generated ESG forecasts, as most middle-income retail investors lack the ability to evaluate complex ESG datasets themselves (Akter et al., 2016; Dwivedi et al., 2023; Zhong et al., 2025). Therefore, greater explainability and ESG analytics capability are expected to enhance Perceived ESG Forecast Credibility, leading to the following hypotheses.

  • H1: Perceived Explainable AI influences Perceived ESG Forecast Credibility.
  • H2: Perceived ESG Analytics Capability influences Perceived ESG Forecast Credibility.

Trust in AI-generated ESG information and sustainable investment decision

Perceived ESG Forecast Credibility (PEFC) refers to investors' evaluation of the accuracy, reliability, and believability of AI-generated ESG forecasts. It represents a judgment about the quality and dependability of the information itself, rather than a willingness to rely on that information for decision-making (Spence, 1973; Connelly et al., 2011). Investors face considerable uncertainty regarding firms‘ sustainability performance and therefore rely on the credibility of AI-generated ESG recommendations as a key heuristic in their investment decision-making. Investors are more likely to develop Trust in AI-generated ESG Information (TAEGI) when they perceive the ESG predictions as realistic, evidence-based, and sufficiently reliable for investment decision-making. In line with this perspective, previous studies reveal that when the AI output is transparent and credible, investors are more likely to trust it and depend on it, especially for critically important decision such as investing (Shin, 2021; Hoffman et al., 2023).

Trust in AI-generated ESG Information (TAEGI), in contrast, reflects investors' willingness and confidence to rely on AI-generated ESG information when making investment decisions. Thus, while credibility concerns whether the ESG forecast is perceived as believable and reliable, trust concerns whether investors are willing to accept and rely on that information in their decision-making. Trust Theory suggests that individuals' willingness to accept and rely on a technological system depends on their confidence that the system and its outputs are dependable and suitable for use (Mayer et al., 1995). Accordingly, credibility represents an evaluation of the quality and believability of the information, whereas trust reflects the willingness to rely on that information. In the context of sustainable investing, reliable AI-generated ESG data bolsters investor Trust in AI-generated ESG Information in using ESG data to guide investment decision, strengthening investors' willingness to consider companies that perform well on ESG metrics in their investment decisions (Gillan et al., 2021; Dwivedi et al., 2023; Inderst and Opp, 2025). Consequently, the confidence in AI-generated ESG information will increase, resulting in more sustainable investing decision.

Conceptually, credibility and trust represent related but distinct stages of evaluation. Credibility concerns investors' assessment of whether an AI-generated ESG forecast is believable and reliable, whereas trust reflects their subsequent willingness to rely on that information in decision-making. Accordingly, perceived credibility provides the informational basis upon which trust can develop.

  • H3: Perceived ESG Forecast Credibility influences Trust in AI-generated ESG Information.
  • H4: Trust in AI-generated ESG Information influences Sustainable Investment Decision.

Mediating effects of ESG forecast credibility and trust

According to the Stimulus–Organism–Response (S–O–R) theory, external technological stimuli influence behavioral responses through individuals' internal cognitive evaluations (Mehrabian and Russell, 1974). Perceived ESG Forecast Credibility (PEFC) is a cognitive process by which technological stimuli affect investor Trust in AI-generated ESG Information. Investors are more likely to find ESG forecasts credible as AI explainability and ESG analytics capability improves the transparency and reliability of the forecast (Shin, 2021; Hoffman et al., 2023). Because investors cannot directly observe future ESG performance, forecast credibility reduces information uncertainty and encourages reliance on AI-assisted ESG recommendations, thereby facilitating trust. As a result, these relationships are anticipated to be mediated by forecast credibility.

  • H5: Perceived ESG Forecast Credibility mediates the relationship between Perceived Explainable AI and Trust in AI-generated ESG Information.
  • H6: Perceived ESG Forecast Credibility mediates the relationship between Perceived ESG Analytics Capability and Trust in AI-generated ESG Information.

In addition, Trust Theory suggests that trust is an important psychological process that converts credible information into behavioral choices (Mayer et al., 1995). Trust reduces perceived uncertainty associated with AI-assisted recommendations, increasing investors' willingness to incorporate sustainability information into their investment decision-making. Therefore, successful AI-backed recommendations are more likely to impact the sustainable investment decision (Shin, 2021; Dwivedi et al., 2023). Therefore, it is expected that the relationship between forecast credibility and investment decision would be mediated by trust.

  • H7: Trust in AI-generated ESG Information mediates the relationship between Perceived ESG Forecast Credibility and Sustainable Investment Decision.

These theories suggest that the characteristics of AI affect investment decision through a series of cognitive evaluations. These distinctions support the proposed sequential mechanism. Investors may first evaluate whether AI-generated ESG forecasts are credible before deciding whether the information is sufficiently trustworthy to rely upon in investment decision-making. Thus, credibility represents an evaluative judgment about the information, whereas trust represents the subsequent willingness to rely on that information. This provides the theoretical basis for the proposed sequential mediation. Accordingly, forecast credibility and trust are expected to sequentially mediate these relationships.

  • H8: Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information sequentially mediate the relationship between Perceived Explainable AI and Sustainable Investment Decision.
  • H9: Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information sequentially mediate the relationship between Perceived ESG Analytics Capability and Sustainable Investment Decision.

Moderating effects of Perceived Greenwashing Risk and Human-AI Interaction Competency

Perceived Greenwashing Risk (PGR) represents the investors' concerns that companies might overstate or misrepresent their ESG performance and thus decrease investors' trust in sustainability-related information (Chen and Chang, 2013; Sun and Yin, 2026). According to Signaling Theory, greenwashing weakens the credibility of ESG signals, limiting investors' willingness to Trust in AI-generated ESG Information despite perceiving ESG forecasts as credible (Spence, 1973; de Freitas Netto et al., 2020; Christensen et al., 2021). Despite the seemingly credible nature of ESG forecasts, investors may suspect that the ESG disclosures do not reflect a company's genuine sustainability practices because of the Perceived Greenwashing Risk. Therefore, the relationship between forecast credibility and trust is likely to become weaker when Perceived Greenwashing Risk is high.

  • H10: Perceived Greenwashing Risk moderates the relationship between Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information.

Human-AI Interaction Competency (HAIC) reflects investors' ability to understand and effectively use AI-generated sustainability information. Investors that possess greater Human-AI Interaction Competency can better understand AI explanations, identify high-quality recommendations, and consistently apply their own insight and experience to investment decision in a way that incorporates AI insights. Therefore, investors with a higher level of competency may be better able to translate trust into stronger stated preferences and intentions toward sustainable investment decisions (Longoni et al., 2019; Shin, 2021; Dwivedi et al., 2023). Accordingly, Human-AI Interaction Competency is expected to strengthen this relationship.

  • H11: Human-AI Interaction Competency moderates the relationship between Trust in AI-generated ESG Information and Sustainable Investment Decision.

Proposed conceptual framework

Theoretically, this study builds upon the theories and hypotheses presented in the previous sections to create an integrated conceptual model using Stimulus – Organism – Response (S – O – R) theory with the assistance of Signaling Theory and Trust Theory. Technological stimuli are Perceived Explainable AI (PXAI) and Perceived ESG Analytics Capability (PEAC), while organism within investors‘ internal cognition is Perceived ESG Forecast Credibility (PEFC) and Trust in AI-generated ESG Information (TAEGI). Sustainable Investment Decision (SID) represents the response component and is operationalized as respondents' stated preferences and willingness to incorporate ESG considerations into their investment decisions. Furthermore, Perceived Greenwashing Risk (PGR) and Human-AI Interaction Competency (HAIC) are incorporated as moderators that strengthen or weaken the proposed relationships. The conceptual model is illustrated in Figure 1.

Conceptual framework.
Figure 1. Conceptual framework.

Methodology

Sampling techniques and data collection

The study aimed to identify the target population, which consisted of middle-income retail investors actively investing in the stock market in India. The group was chosen because of the increasing relevance of AI-powered investment platforms and ESG-related digital investment information to retail investors, making this population appropriate for examining perceptions of AI-assisted sustainable investment decision-making (Gillan et al., 2021; Giudici and Wu, 2025). Following the PRICE Survey Report (eople Research on India's Consumer Economy (PRICE), 2022), households with annual incomes ranging from 1.25 lakh to 30 lakh were classified as middle-income households. As this classification is based on household rather than individual income, respondents without regular personal earnings were included provided that their household income met the specified criterion.

The sampling method used in the study was purposive sampling and the respondent-referral method was controlled (Salganik and Heckathorn, 2004; Naderifar et al., 2017). This approach was deemed suitable due to the lack of a comprehensive sampling frame of middle-income retail investors who actively engage in equity investments. Initially, respondents were identified using a mall intercept survey in four major cities: Chennai, Mumbai, Delhi and Kolkata, chosen due to the high density of retail investors, a well-established financial infrastructure, and the availability of digital investment platforms (Bush and Hair Jr, 1985). The eligible respondents were then asked to recommend up to three additional middle-income retail investors, further enlarging the sample and resulting in a lower likelihood of duplication with other respondents while still keeping each participant in the study eligible.

The eligibility criteria for the study were: (i) age of 18 years and above, (ii) middle-income group, according to PRICE Survey 2022, (iii) possessing at least a basic awareness of ESG-related investment concepts, and (iv) investment in equity shares over the last 12 months. The study did not require prior use of a specific AI-based investment or ESG tool as an eligibility criterion; instead, respondents evaluated AI-generated ESG information and related attributes based on their awareness and understanding of AI-assisted investment technologies. To ensure clarity and content validity to the measurement instrument, a pilot study was conducted with 67 retail investors in the city of Chennai prior to the main survey. The pilot responses were excluded from the final analysis.

The primary survey was conducted from January to May, 2026, with a multi-wave and mixed-mode approach to data collection. Printed questionnaires were initially administered through mall intercept surveys, after which referred, respondents were invited to complete the online questionnaire via Google Forms. This strategy expanded the recruitment sources, increased access to the target population, and minimized the possibility of common method bias (Podsakoff et al., 2003). There were 978 questionnaires distributed, with 727 being returned, representing a 74.3% return rate. During screening and data validation, 161 responses were excluded, comprising 73 responses that did not meet the eligibility requirements or failed referral code verification, 54 incomplete questionnaires, and 34 responses showing irregular response patterns or failing data validation. Consequently, 566 questionnaires (77.9% of the returned questionnaires) were retained for the final analysis using SmartPLS 4. A priori power analysis using G*Power 3.1.9.7 indicated that a minimum sample size of 119 respondents was required to detect the hypothesized structural relationship. The final sample of 566 respondents substantially exceeded this requirement, indicating adequate statistical power (Faul et al., 2009; Hair et al., 2019). The corresponding G*Power output is provided in Appendix A (Figure A1).

Operationalization of the study variables

All constructs were assessed with multi-item scales, which were based on previously published studies, as detailed in Table 1 and Appendix A (Table A1). The measurement framework has been developed based on the S–O–R framework which is supported by the Signaling Theory and Trust Theory. All constructs were treated as reflective constructs, meaning that the indicators observed for each construct were seen as reflecting the underlying latent constructs, and therefore were assumed to be highly intercorrelated (Hair et al., 2019; Sarstedt et al., 2021).

Table 1. Operationalization of constructs.

S–O–R componentConstructMeasurement indicatorsSources
StimulusPerceived Explainable AI (PXAI)Perceived transparency and interpretability of AI-generated ESG recommendationsHoffman et al. (2023)
StimulusPerceived ESG Analytics Capability (PEAC)Perceived capability of AI to analyse and evaluate ESG informationAkter et al. (2016); Xiao and Xiao (2025)
OrganismPerceived ESG Forecast Credibility (PEFC)Perceived credibility and reliability of AI-generated ESG forecastsAppelman and Sundar (2016); Shin (2021)
OrganismTrust in AI-generated ESG Information (TAEGI)Trust in AI-generated ESG information for investment decision-makingShin (2021)
ResponseSustainable Investment Decision (SID)Willingness to make investment decision based on ESG considerationsVijayananth and Saravanabhavan (2026)
ModeratorPerceived Greenwashing Risk (PGR)Perceived risk of misleading ESG claimsChen and Chang (2013)
ModeratorHuman-AI Interaction Competency (HAIC)Ability to understand and effectively use AI-generated ESG informationVijayananth and Saravanabhavan (2026)

In the S–O–R framework proposed, Perceived Explainable AI (PXAI) and Perceived ESG Analytics Capability (PEAC) were operationalized as stimulus constructs. The organism construct consisted of Perceived ESG Forecast Credibility (PEFC) and Trust in AI-generated ESG Information (TAEGI); the response construct was Sustainable Investment Decision (SID), operationalized as respondents stated preferences and willingness to incorporate ESG considerations into their investment decisions rather than observed investment behavior. Additionally, two moderating variables were included: Perceived Greenwashing Risk (PGR) and Human-AI Interaction Competency (HAIC). The operationalization of all study constructs is summarized in Table 1.

The constructs were measured using three or four items, depending on the established measurement scale, and were contextualized to the present study. The questionnaire consisted of three parts: (i) informed consent, (ii) demographic information and (iii) measurement items. Respondents rated each item on a 5-point Likert scale ranging from 1 “strongly disagree” to 5 “strongly agree.” To ensure the clarity and content validity of the questionnaire the questionnaire was pre-tested with academic experts and experienced investors before data collection. The full measurement items are included in Appendix A (Table A1).

Common method bias assessment

Common method bias (CMB) was examined using Harman's single-factor test as an initial diagnostic (Podsakoff et al., 2003). The first unrotated factor explained 38.063% of the total variance, which was below the commonly applied 50% threshold. However, this test alone cannot conclusively establish the absence of substantial common method bias. To further minimize the possibility of CMB, several procedural measures were adopted, including anonymous participation, mixed-mode data collection, multi-wave data collection, and questionnaire pre-testing. Taken together, these measures and the diagnostic result suggest that common method bias is unlikely to have substantially influenced the study findings.

Data analysis procedures

The proposed research model was analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM) in SmartPLS 4.1.1.7. For the prediction-oriented research with complex models (direct, mediating, sequential mediating and moderating relationships between latent constructs) PLS-SEM was chosen as the analysis method (Hair et al., 2019; Sarstedt et al., 2021). The analysis was conducted in two steps. Firstly, the measurement model was examined using outer loadings, Cronbach's alpha, composite reliability, average variance extracted (AVE), and discriminant validity assessed through the Fornell–Larcker criterion, HTMT ratio, and HTMT inference (Henseler et al., 2015; Hair et al., 2019).

Then, the structural model was evaluated using collinearity diagnostics, path coefficients, coefficient of determination (R2), effect size (f2) and out-of-sample predictive performance (Q2_predict). Last, the hypothesized relationships (direct, mediational, sequential mediational, and moderating) were explored using the bootstrapping procedure with 5,000 resamples to assess the significance of the structural relationships (Hair et al., 2019; Sarstedt et al., 2021).

Results

The empirical results presented in this chapter are derived from the Partial Least Squares Structural Equation Modeling (PLS-SEM). The results show the respondent profile, the measurement model assessment and the assessment of the structural model, which gives empirical evidence for the research framework proposed. Detailed statistical results are provided in Appendix A.

Respondent profile

The demographic data for the 566 respondents who constituted the sample for the study are shown in Table A2. The sample comprised 56.9% male and 43.1% female respondents. The highest percentage of the respondents were in the age group 26–35 years (32.5%) and 36–45 years (30.0%). With respect to education, 48.4% of the people scored a high level, while 26.1% had a post-graduate. Private sector workers were the largest group (39.6%) followed by business owners and self-employed workers (21.2%). Most respondents reported a monthly personal income between 25,001–50,000 (30.4%) and 50,001–1,00,000 (29.0%). These personal-income figures are reported for descriptive purposes and are distinct from the household-income criterion used to determine eligibility. Regarding investment experience, 33.9% of respondents had 1 to 3 years of investing experience, followed by 30.0% with 4 to 6 years. The respondents were evenly spread in the four major metros of the country (25.4% Chennai, 25.1% Mumbai, 24.7% Delhi, 24.7% Kolkata) thus maintaining a geographical balance.

Measurement model validation

Reliability, convergent validity, discriminant validity, collinearity, and model fit were used to evaluate the measurement model. The detailed measurement model results are provided in Appendix A (Figure A2 and Tables A3–A8).

Reliability and convergent validity assessment

Reliability and convergent validity were assessed using outer loadings, Cronbach's alpha, composite reliability, and average variance extracted (AVE). All the indicator loadings were greater than the recommended value (Hair et al., 2019; Sarstedt et al., 2021) as presented in Appendix A (Table A3), from 0.715 to 0.908. Similarly, Cronbach's alpha for these scales ranged from 0.783 to 0.895 and composite reliability values ranged from 0.860 to 0.927, which were higher than the suggested minimum value of 0.70. In addition, the AVE ranged from 0.607 to 0.762, exceeding the threshold of 0.50 (Hair et al., 2019). The results of the present study validate high internal consistency reliability and convergent validity of all reflective constructs.

Discriminant validity assessment

Discriminant validity was assessed using the Fornell–Larcker criterion and the Heterotrait–Monotrait ratio (HTMT). The square root of the AVE for each construct exceeded its correlations with the other constructs, satisfying the Fornell–Larcker criterion (Fornell and Larcker, 1981), as shown in Appendix A (Table A4). The HTMT values ranged from 0.242 to 0.855 (Table A5), with the highest value of 0.855 remaining below the 0.90 threshold. The 0.90 criterion was adopted as an established threshold for assessing discriminant validity in PLS-SEM, while the more conservative 0.85 criterion was also considered because several HTMT values approached or slightly exceeded this level (Henseler et al., 2015; Hair et al., 2019). Since some HTMT values were close to 0.85, HTMT inference was additionally performed. As shown in Table A6, the 95% bootstrap confidence intervals for all HTMT estimates excluded 1.00, including the intervals for the highest HTMT relationships (0.794–0.911 and 0.797–0.901), providing further evidence of discriminant validity.

Collinearity and model fit assessment

Collinearity was assessed using the variance inflation factor (VIF). As reported in Appendix A (Table A7), all VIF values ranged from 1.147 to 2.153, well below the recommended threshold of 3.30, indicating that multicollinearity was not a concern (Hair et al., 2019). The standardized root mean square residual (SRMR) was used to assess approximate model fit. The SRMR values for the saturated and estimated models were 0.068 and 0.073, respectively (Table A8 in Appendix A), both below the commonly used 0.08 threshold, indicating acceptable model fit. Model fit was interpreted alongside reliability, validity, collinearity, explanatory power, and predictive performance (Hair et al., 2019). Therefore, the model was considered suitable for hypothesis testing.

Structural model validation

Evaluating the structural model was done by analyzing the explanatory power, predictive relevance, direct effects, mediation effects, sequential mediation effects and moderation effects in the proposed research model. The significance of the structural relationships was assessed using the bootstrapping procedure with 5,000 resamples in SmartPLS 4. Results of the detailed structural model are presented in Appendix A (Figures A3–5 and Tables A9–16).

Model explanatory capability and predictive performance

The explanatory power of the endogenous constructs was assessed using the coefficient of determination (R2), following established PLS-SEM evaluation guidelines (Hair et al., 2019). Out-of-sample predictive performance was assessed using the PLSpredict procedure in SmartPLS 4, with Q2_predict calculated as 1 – (SSE_PLS-SEM/SSE_IA) by comparing PLS-SEM prediction errors with the naïve indicator-average (IA) benchmark (Shmueli et al., 2019). Positive Q2_predict values indicate out-of-sample predictive relevance for the corresponding endogenous constructs. As reported in Appendix A (Table A9), Q2_predict values were 0.627 for Perceived ESG Forecast Credibility, 0.545 for Trust in AI-generated ESG Information, and 0.500 for Sustainable Investment Decision, indicating positive out-of-sample predictive relevance. The corresponding RMSE and MAE values are also reported in Table A9. In addition, Table A10 shows that the PLS-SEM prediction errors were lower than the naïve indicator-average (IA) benchmark across all endogenous constructs, supporting the model's predictive performance (Shmueli et al., 2019).

The effect sizes (f2) were also assessed to determine the magnitude of the effects of the predictor constructs on the endogenous constructs (Hair et al., 2019). As reported in Appendix A (Table A11), Perceived Explainable AI (f2 = 0.559) and Perceived ESG Analytics Capability (f2 = 0.435) demonstrated large effect sizes on Perceived ESG Forecast Credibility. The remaining relationships exhibited medium, small, or negligible effect sizes, indicating varying magnitudes of predictor effects across the endogenous constructs.

Direct path analysis

The direct relationships were evaluated using the bootstrapping procedure with 5,000 resamples. The bootstrapping results are presented in Appendix A (Figure A3 and Table A12). All the direct hypothesized relationships were positive and statistically significant (p < 0.001), thus supporting the hypotheses H1–H4. Specifically, Perceived Explainable AI (β = 0.500, t = 13.018) and Perceived ESG Analytics Capability (β = 0.441, t = 13.318) had significant positive effects on Perceived ESG Forecast Credibility. Furthermore, Perceived ESG Forecast Credibility positively influenced Trust in AI-generated ESG Information (β = 0.389, t = 9.712). The path coefficient between Trust in AI-generated ESG Information and Sustainable Investment Decision (β = 0.509, t = 12.935) represented the strongest hypothesized direct effect. Accordingly, all proposed direct hypotheses (H1–H4) were supported. In addition, the non-hypothesized direct effects of Perceived Greenwashing Risk on Trust in AI-generated ESG Information (β = 0.267, t = 6.216, p < 0.001) and Human-AI Interaction Competency on Sustainable Investment Decision (β = 0.350, t = 8.279, p < 0.001) were statistically significant.

Mediation path and sequential mediation path analysis

The mediating effects were assessed using the bootstrapping procedure with 5,000 resamples. The detailed results for the simple and sequential mediation analyses are reported in Appendix A (Tables A13 and A14). All indirect effects were positive and statistically significant (p < 0.001) which supported H5–H7 as seen in Table A13. Specifically, Perceived ESG Forecast Credibility significantly mediated the relationship between Perceived Explainable AI and Trust in AI-generated ESG Information (β = 0.195, t = 6.904), as well as the relationship between Perceived ESG Analytics Capability and Trust in AI-generated ESG Information (β = 0.172, t = 8.712). In addition, the relationship between Perceived ESG Forecast Credibility and Sustainable Investment Decision was supported by Trust in AI-generated ESG Information (β = 0.198, t = 7.075), which was the hypothesis tested in H7. The sequential mediation analysis (Table A14) further indicated that Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information jointly mediated the effects of Perceived Explainable AI (β = 0.099, t = 5.520) and Perceived ESG Analytics Capability (β = 0.087, t = 7.011) on Sustainable Investment Decision, thereby supporting H8 and H9.

Moderation path analysis

The moderating effects were evaluated using the bootstrapping procedure with 5,000 resamples (Hair et al., 2019). The moderation results are summarized in Appendix A (Table A15). The simple-slope results for both moderation effects are reported in Appendix A (Table A16), with the corresponding interaction plots presented in Figures A4 and A5. As presented in the table below, the relationship between Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information was significantly moderated by Perceived Greenwashing Risk (β = −0.170, t = 4.679, p < 0.001) thus supporting H10. The negative interaction coefficient indicates that higher levels of perceived greenwashing risk weaken the positive relationship between perceived ESG forecast credibility and trust in AI-generated ESG information. Moreover, the relationship between Trust in AI-generated ESG Information and Sustainable Investment Decision was moderated by Human-AI Interaction Competency (β = 0.078, t = 2.779, p = 0.005) supporting H11. The inclusion of the HAIC × Trust interaction increased the R2 of Sustainable Investment Decision from 0.573 to 0.578, resulting in a ΔR2 of 0.005 and indicating a limited incremental contribution of the interaction to the explained variance. The interaction effect was statistically significant but negligible in magnitude (f2 = 0.012), indicating limited practical contribution.

Discussion

This chapter provides a critical interpretation of the empirical findings in relation to the proposed theoretical framework and existing literature. The discussion examines the direct, mediating, sequential mediating, and moderating relationships identified in the study and interprets how these relationships help explain the role of AI-related characteristics, ESG forecast credibility, and trust in shaping sustainable investment decision-making. It further discusses the theoretical relevance and practical implications of the findings for understanding AI-enabled ESG investing and the conditions under which AI-generated ESG information may influence investors' sustainable investment preferences.

Findings

Direct effects

The positive relationships between Perceived Explainable AI, Perceived ESG Analytics Capability, and Perceived ESG Forecast Credibility suggest that investors place greater confidence in AI-generated ESG forecasts when the underlying system is perceived as understandable and analytically capable. Explainability may reduce uncertainty by allowing investors to understand the reasoning behind AI-generated forecasts, while analytics capability may increase confidence in the quality and reliability of the information. These complementary mechanisms help explain why technological attributes alone are insufficient; their value depends on how investors interpret the credibility of the resulting ESG information. The findings are also consistent with Signaling Theory (Spence, 1973; Connelly et al., 2011), as explainability and ESG analytics capability provide observable informational cues through which investors can evaluate the credibility of AI-generated ESG forecasts under conditions of information uncertainty. They are also consistent with previous research showing that explainability and advanced ESG analytics can strengthen trust in AI-supported financial decision-making (Shin, 2021; Hoffman et al., 2023; Dwivedi et al., 2023; Ahmad and Ahmed, 2026).

Furthermore, the credibility–trust relationship suggests that investors do not treat AI-generated ESG information as trustworthy simply because it is technologically advanced. Instead, trust appears to develop after investors form a favorable judgement about the credibility, reliability, and believability of the information. This distinction is theoretically important because it positions forecast credibility as an evaluative mechanism that precedes willingness to rely on AI-generated ESG information. The result is consistent with Trust Theory, which emphasizes credibility, reliability, and transparency as foundations of trust (Mayer et al., 1995; Shin, 2021). The present study shows that forecast credibility is an important cognitive mechanism connecting AI-related characteristics with investor Trust in AI-generated ESG Information and highlights the sequential role of credibility between AI-related characteristics and trust.

The positive role of Trust in AI-generated ESG Information suggests that credibility alone may not be sufficient to translate AI-generated ESG information into sustainable investment preferences; investors must also be willing to rely on that information. This finding supports the response component of the S–O–R framework, where favorable internal evaluations are reflected in stated investment preferences and intentions. It also indicates that the behavioral relevance of AI-enabled ESG information depends on investors' willingness to rely on the information rather than merely recognize its credibility.

Mediating and sequential mediating effects

The mediation findings clarify how AI-related characteristics translate into sustainable investment decision through successive psychological evaluations. Rather than operating as direct technological influences alone, Perceived Explainable AI and Perceived ESG Analytics Capability influence sustainable investment decision through investors' evaluation of ESG forecast credibility and their subsequent willingness to rely on AI-generated ESG information. This indicates that credibility and trust represent sequential rather than interchangeable psychological processes. Investors first assess whether AI-generated ESG forecasts appear accurate, reliable, and believable; when this evaluation is favorable, they become more willing to rely on the information in investment decision-making. Thus, Perceived ESG Forecast Credibility functions as an initial informational evaluation, whereas Trust in AI-generated ESG Information represents the subsequent willingness to rely on that information.

This psychological process helps explain the Stimulus–Organism–Response (S–O–R) framework by showing how external technological stimuli influence investors stated investment responses through internal cognitive evaluations. Additionally, the results are consistent with Signaling Theory, which suggests that clear and credible AI-driven ESG predictions can alleviate information asymmetry and enhance investor trust. These findings are broadly consistent with previous studies highlighting the importance of explainability, credibility, and Trust in AI-generated ESG Information assisted financial decision-making (Shin, 2021; Hoffman et al., 2023; Dwivedi et al., 2023).

Further, the results of significant sequential mediation further indicate that the effects of the AI-enabled capabilities on sustainable investment decision are mediated by forecast credibility and trust. The present study reveals that the psychological steps of forecast credibility and trust are consecutive in the chain of translating AI characteristics into sustainable investment decision, and that this differs with previous studies which were focussed on either explainability or trust. In addition to answering the question of whether AI-supported investment technologies can influence investors stated sustainable investment decision preferences, this paper also provides explanations for how these technologies can make a difference in successive cognitive evaluations.

Moderating effect

The moderation analysis highlights the factors that condition investors stated sustainable investment decision preferences regarding the use of AI-generated ESG information. Perceived Greenwashing Risk (PGR) significantly weakened the positive relationship between Perceived ESG Forecast Credibility (PEFC) and Trust in AI-generated ESG Information (TAEGI). This finding suggests that Perceived Greenwashing Risk increases uncertainty regarding the authenticity of ESG information, making investors less willing to rely on AI-generated ESG forecasts even when they appear technically credible. The results are consistent with previous studies showing that Perceived Greenwashing Risk undermines trust and reduces the effectiveness of ESG disclosures in shaping investors' investment-related responses (Chen and Chang, 2013; Delmas and Burbano, 2011; de Freitas Netto et al., 2020; Zhong et al., 2025; Li et al., 2025).

On the other hand, Human-AI Interaction Competency (HAIC) positively moderated the relationship between Trust in AI-generated ESG Information and Sustainable Investment Decision. However, the interaction effect was negligible (f2 = 0.012), indicating limited practical contribution despite statistical significance. The simple-slope analysis showed that the positive relationship increased from β = 0.430 at low HAIC to β = 0.587 at high HAIC. This finding suggests that Human-AI Interaction Competency may strengthen investors' ability to translate trust in AI-generated ESG information into sustainable investment preferences, but its limited effect size indicates that competency alone is unlikely to be a dominant driver of this transition. Thus, the result reveals an important boundary condition: investors' ability to interact with AI may facilitate the translation of trust into investment responses, but other contextual and behavioral factors are likely to remain important. This finding is consistent with recent research emphasizing the importance of AI competencies in strengthening confidence in AI-generated ESG information and improving investment decision-making (Dwivedi et al., 2023; Khan et al., 2025; Sun and Yin, 2026; Inderst and Opp, 2025).

Theoretical implications

The study provides empirical evidence on how AI-related characteristics are associated with sustainable investment preferences through a sequential process involving ESG forecast credibility and trust in AI-generated ESG information, thereby highlighting the importance of cognitive evaluation. The current study, in contrast to previous studies which mainly analyzed AI capabilities or ESG disclosures separately, shows that forecast credibility and Trust in AI-generated ESG Information act sequentially as psychological processes connecting AI characteristics to sustainable investment decision. To explain this process, the study integrates the Stimulus–Organism–Response (S–O–R) theory, Signaling Theory, and Trust Theory. Collectively, these complementary theoretical perspectives provide a multi-perspective explanation of how AI-related characteristics are associated with sustainable investment decision through credibility and trust.

From an S–O–R perspective, the findings provide empirical support for a process in which technological characteristics are associated with investors' responses through internal cognitive evaluations of credibility and trust. In the proposed framework, S–O–R provides the overarching behavioral structure by connecting AI-related technological stimuli with investors' sustainable investment responses through successive cognitive evaluations. This application highlights how credibility and trust operate as internal evaluative processes within an AI-enabled ESG investment context.

From a Signaling Theory perspective, the findings suggest that the informational value of AI-related characteristics depends not only on their presence but also on how investors evaluate the credibility of the resulting ESG information. Explainable AI makes the reasoning behind algorithms more transparent, and ESG analytics capability provides information that investors can use to assess the quality and reliability of AI-generated ESG forecasts. Thus, Signaling Theory provides the informational mechanism for explaining how AI-related characteristics function as cues that inform investors' evaluation of forecast credibility under conditions of information uncertainty.

Finally, from a Trust Theory perspective, the findings position Trust in AI-generated ESG Information as a central psychological mechanism linking AI-enabled capabilities with Sustainable Investment Decision. The significant relationship between Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information indicates that investors' willingness to rely on AI-generated ESG information develops following a favorable evaluation of its credibility. The moderation results further indicate that this reliance mechanism is context dependent: Perceived Greenwashing Risk weakens the credibility–trust relationship, whereas Human-AI Interaction Competency positively moderates the relationship between Trust in AI-generated ESG Information and Sustainable Investment Decision, although the latter interaction has limited practical magnitude. Thus, Trust Theory helps explain how favorable credibility evaluations translate into reliance on AI-generated ESG information and subsequently into sustainable investment responses.

Practical implications

The findings provide practical guidance for strengthening responsible AI-enabled sustainable investing. The sequential mediation results suggest that improving the credibility of AI-generated ESG forecasts should precede efforts to build investor trust, while the moderating results indicate that greenwashing concerns can weaken the credibility–trust relationship. Accordingly, practical actions should focus on transparent AI outputs, reliable ESG data, credible verification mechanisms, and investor capabilities across the AI development, financial services, ESG data, and policy domains.

For AI developers and fintech firms, AI systems should provide understandable explanations for ESG-related recommendations, clearly identify the key factors influencing forecasts, and communicate relevant uncertainty rather than presenting outputs as definitive investment advice. Developers should also provide traceable information on how ESG inputs are processed and periodically validate model outputs to support investor assessment of forecast credibility. These practices can improve the transparency and interpretability of AI-generated ESG information and support more informed investor reliance. For ESG-data providers, the findings highlight the need for reliable and consistent ESG data. Providers should strengthen data validation, document data sources and limitations, and encourage independent verification where feasible. These measures can improve the credibility of AI-generated ESG forecasts and reduce uncertainty.

Transparency, accountability and ethics in sustainable finance continue to be vital for governance frameworks for policy makers. For policymakers and financial regulators, the findings support clearer requirements for transparency and accountability in AI-assisted ESG decision-making. Guidelines on AI-generated ESG disclosures, data verification, and misleading sustainability claims could reduce Perceived Greenwashing Risk and strengthen investor confidence.

For financial institutions and investment advisors, AI-generated ESG information should be incorporated into investment and advisory processes with appropriate human oversight rather than being treated as a standalone decision rule. Institutions could require advisors to review the credibility, source quality, and limitations of AI-generated ESG outputs before presenting them to investors. Investor education should also strengthen Human–AI Interaction Competency by helping investors interpret AI-generated ESG information, recognize uncertainty, and distinguish credible outputs from potentially misleading recommendations.

Limitations and future research

While this study provides further evidence on AI-enabled sustainable investment decision, several limitations should be considered when interpreting the findings. First, the study design is cross sectional, which reduces the scope of drawing causal relationships between variables, as well the ability to examine changes in forecast credibility, investor trust, and stated sustainable investment preferences over time. Future research can employ longitudinal or experimental designs to analyze these relationships over time as AI technologies and changing ESG standards evolve.

Second, the empirical study was conducted on middle-income retail investors of four metros in India. Although this context provides valuable insights into an important investor segment, the findings may not be fully generalizable to rural investors, high-income individuals, institutional investors, or other international markets. Empirical future research needs to confirm the proposed framework in developed and emerging markets, in institutional investor settings as well as under various regulatory and market conditions.

Third, the self-reported nature of the survey may introduce social desirability and common response bias despite the procedural and statistical techniques used. In particular, the Sustainable Investment Decision measure captures respondents stated preferences and willingness rather than observed investment transactions or realized portfolio behavior. Future research could combine perceptual measures with real investment portfolios, real trading data, AI platform use, and behavioral measures to enhance external validity. Future studies may also incorporate additional investor capability (e.g., AI literacy and financial literacy), technological (e.g., algorithmic transparency and perceived AI ethics), and contextual factors (e.g., ESG knowledge and regulatory awareness). Furthermore, other mediating and moderating factors could be explored to better understand whether stated sustainable investment preferences translate into actual sustainable investment behavior under varying market conditions.

Fourth, prior use of specific AI-based investment or ESG tools was not measured. Thus, the findings reflect investors' perceptions of AI-assisted ESG information rather than verified prior AI-tool usage. Future research could include AI-tool use as a screening criterion and compare experienced and inexperienced users of AI-enabled investment or ESG platforms.

Conclusion

This research contributes to the development of knowledge in the field of how AI attributes influence sustainable investment decision making by examining the sequential cognitive process of forecast credibility followed by Trust in AI-generated ESG Information The findings are interpreted through the complementary perspectives of Stimulus–Organism–Response (S–O–R) theory, Signaling Theory, and Trust Theory, which together help explain how AI-related characteristics are associated with sustainable investment decision through credibility and trust.

The findings indicate that both forecast credibility and trust are important psychological processes through which AI-related characteristics are associated with sustainable investment decision-making. Specifically, investors report stronger preferences and willingness to consider sustainable investment options when they perceive AI-generated ESG forecasts as credible and trust the AI-generated ESG information. The results also highlight that the effectiveness of these mechanisms depends on contextual conditions. Perceived Greenwashing Risk weakens the positive relationship between Perceived ESG Forecast Credibility and Trust in AI-generated ESG Information, while Human-AI Interaction Competency positively moderates the relationship between Trust in AI-generated ESG Information and Sustainable Investment Decision. However, the negligible effect size of the HAIC interaction indicates that its practical contribution is limited.

In conclusion, the present research contributes not only to the knowledge on AI characteristics but also indicates that transparency, forecast credibility, and investors‘ interpretation of the AI-generated ESG information have values in the context of sustainable finance. The results offer significant theoretical and practical implications for creating reliable and ethically responsible AI systems for sustainable investing, in order to facilitate the responsible use of AI in sustainable financial decision. The transparency, credibility, and Trust in AI-generated ESG Information will be important for strengthening investors' informed and responsible preferences and intentions toward sustainable investment decisions as AI becomes more involved in supporting sustainable finance.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/frsus.2026.1965649/full#supplementary-material

Supplementary material Supplementary material at the source

Footnotes

Edited by: Simone Leticia Raimundini, State University of Maringá, Brazil

Reviewed by: Ian Staley, eCapital, United States

Dendi Mulyana, Indonesia University of Education, Indonesia

Statements

Conflict of interest

The author(s) declared that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

The author(s) declared that financial support was received for this work and/or its publication. Open access funding was provided by Vellore Institute of Technology.

Data availability

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

Ethics approval

Ethical approval was not required for the studies involving humans because all procedures performed in this study were conducted in accordance with institutional research guidelines and ethical standards applicable to social science research. The study was questionnaire-based and did not involve medical intervention or experimental manipulation. The research involved voluntary participation, anonymous data collection, and no collection of sensitive personal information. Accordingly, formal ethical approval was not required under the applicable institutional requirements. The studies were conducted in accordance with the local legislation and institutional requirements. The participants provided their written informed consent to participate in this study.

Author contributions

MV: Software, Writing – original draft, Resources, Writing – review & editing, Methodology, Formal analysis, Visualization, Data curation, Conceptualization. RM: Formal analysis, Methodology, Visualization, Validation, Data curation, Conceptualization, Writing – original draft, Software, Writing – review & editing, Resources. NS: Software, Methodology, Validation, Writing – review & editing, Conceptualization, Formal analysis, Writing – original draft, Funding acquisition, Supervision.

References

  1. Ahmad I., Ahmed T. (2026). AI-enhanced ESG framework for sustainability: a multi-sectoral analysis through an explainable AI approach. Sustainability, 18: 794. doi: doi:10.3390/su18020794
  2. Ajzen I. (1991). The Theory of Planned Behavior. Organizational Behavior and Human Decision Processes, 50, 179 – 211. doi: doi:10.1016/0749-5978(91)90020-T
  3. Akter S., Wamba S. F., Gunasekaran A., Dubey R., Childe S. J. (2016). How to improve firm performance using big data analytics capability and business strategy alignment? Int. J. Prod. Econ. 182, 113 – 131. doi: doi:10.1016/j.ijpe.2016.08.018
  4. Appelman A., Sundar S. S. (2016). Measuring message credibility: construction and validation of an exclusive scale. J. Mass Commun. Q. 93, 59 – 79. doi: doi:10.1177/1077699015606057
  5. Berg F., Kölbel J. F., Rigobon R. (2022). Aggregate confusion: the divergence of ESG ratings. Rev. Financ. 26, 1315 – 1344. doi: doi:10.1093/rof/rfac033
  6. Bush A. J., Hair Jr J. F. (1985). An assessment of the mall intercept as a data collection method. J. Mark. Res. 22, 158 – 167. doi: doi:10.1177/002224378502200205
  7. Chen Y. S., Chang C. H. (2013). Greenwash and green trust: the mediation effects of green consumer confusion and green perceived risk. J. Bus. Ethics 114, 489 – 500. doi: doi:10.1007/s10551-012-1360-0
  8. Christensen H. B., Hail L., Leuz C. (2021). Mandatory CSR and sustainability reporting: economic analysis and literature review. Rev. Account. Stud. 26, 1176 – 1248. doi: doi:10.1007/s11142-021-09609-5
  9. Connelly B. L., Certo S. T., Ireland R. D., Reutzel C. R. (2011). Signaling theory: a review and assessment. J. Manag. 37, 39 – 67. doi: doi:10.1177/0149206310388419
  10. de Freitas Netto S. V., Sobral M. F. F., Ribeiro A. R. B., Soares G. R. D. L. (2020). Concepts and forms of greenwashing: a systematic review. Environ. Sci. Eur. 32: 19. doi: doi:10.1186/s12302-020-0300-3
  11. Del Vitto A., Marazzina D., Stocco D. (2023). ESG ratings explainability through machine learning techniques. Ann. Oper. Res. 1 – 30. doi: doi:10.1007/s10479-023-05514-z
  12. Delmas M. A., Burbano V. C. (2011). The drivers of greenwashing. Calif. Manag. Rev. 54, 64 – 87. doi: doi:10.1525/cmr.2011.54.1.64
  13. Dwivedi Y. K., Kshetri N., Hughes L., Slade E. L., Jeyaraj A., Kar A. K.. (2023). Opinion paper:“so what if ChatGPT wrote it?” multidisciplinary perspectives on opportunities, challenges and implications of generative conversational AI for research, practice and policy. Int. J. Inf. Manag. 71: 102642. doi: doi:10.1016/j.ijinfomgt.2023.102642
  14. Faul F., Erdfelder E., Buchner A., Lang A. G. (2009). Statistical power analyses using G * Power 3.1: tests for correlation and regression analyses. Behav. Res. Methods. 41, 1149 – 1160. doi: doi:10.3758/BRM.41.4.1149
  15. Fornell C., Larcker D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Mark. Res. 18, 39 – 50. doi: doi:10.1177/002224378101800104
  16. Gillan S. L., Koch A., Starks L. T. (2021). Firms and social responsibility: a review of ESG and CSR research in corporate finance. J. Corp. Financ. 66: 101889. doi: doi:10.1016/j.jcorpfin.2021.101889
  17. Giudici P., Wu L. (2025). Sustainable artificial intelligence in finance: impact of ESG factors. Front. Artif. Intell. 8: 1566197. doi: doi:10.3389/frai.2025.1566197
  18. Hair J. F., Risher J. J., Sarstedt M., Ringle C. M. (2019). When to use and how to report the results of PLS-SEM. Eur. Bus. Rev. 31, 2 – 24. doi: doi:10.1108/EBR-11-2018-0203
  19. Henseler J., Ringle C. M., Sarstedt M. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. J. Acad. Mark. Sci. 43, 115 – 135. doi: doi:10.1007/s11747-014-0403-8
  20. Hoffman R. R., Mueller S. T., Klein G., Litman J. (2023). Measures for explainable AI: explanation goodness, user satisfaction, mental models, curiosity, trust, and human-AI performance. Front. Comput. Sci. 5: 1096257. doi: doi:10.3389/fcomp.2023.1096257
  21. Huang M. H., Rust R. T. (2021). A strategic framework for artificial intelligence in marketing. J. Acad. Mark. Sci. 49, 30 – 50. doi: doi:10.1007/s11747-020-00749-9
  22. Inderst R., Opp M. M. (2025). Sustainable finance versus environmental policy: does greenwashing justify a taxonomy for sustainable investments? J. Financ. Econ. 163: 103954. doi: doi:10.1016/j.jfineco.2024.103954
  23. Khan F. S., Mazhar S. S., Mazhar K., AlSaleh D. A., Mazhar A. (2025). Model-agnostic explainable artificial intelligence methods in finance: a systematic review, recent developments, limitations, challenges and future directions. Artif. Intell. Rev. 58: 232. doi: doi:10.1007/s10462-025-11215-9
  24. Li W., Mao Z., Ren X., Liang J. (2025). Retail investor attention: guardian of corporate ESG integrity or catalyst for greenwashing? Energy Econ., 144: 108361. doi: doi:10.1016/j.eneco.2025.108361
  25. Longoni C., Bonezzi A., Morewedge C. K. (2019). Resistance to medical artificial intelligence. J. Consum. Res. 46, 629 – 650. doi: doi:10.1093/jcr/ucz013
  26. Mayer R. C., Davis J. H., Schoorman F. D. (1995). An integrative model of organizational trust. Acad. Manag. Rev. 20, 709 – 734. doi: doi:10.2307/258792
  27. Mehrabian A., Russell J. A. (1974). An Approach to Environmental Psychology. Cambridge, MA: The MIT Press. p. 266.
  28. Meske C., Bunde E., Schneider J., Gersch M. (2022). Explainable artificial intelligence: objectives, stakeholders, and future research opportunities. Inf. Syst. Manag. 39, 53 – 63. doi: doi:10.1080/10580530.2020.1849465
  29. Naderifar M., Goli H., Ghaljaie F. (2017). Snowball sampling: a purposeful method of sampling in qualitative research. Strides Dev. Med. Educ. 14, 1 – 6. doi: doi:10.5812/sdme.67670
  30. Nurmasari N. D., Bilad M. R. (2026). Artificial intelligence applications for advancing sustainable green finance. Discov. Sustain. 7: 668. doi: doi:10.1007/s43621-026-02980-w
  31. People Research on India's Consumer Economy (PRICE) (2022). The rise of India's middle class: Middle class brochure 2022 (ICE 360° Surveys). Available online at: (Accessed January 10, 2026).
  32. Podsakoff P. M., MacKenzie S. B., Lee J. Y., Podsakoff N. P. (2003). Common method biases in behavioral research: a critical review of the literature and recommended remedies. J. Appl. Psychol. 88: 879. doi: doi:10.1037/0021-9010.88.5.879
  33. Rai A. (2020). Explainable AI: from black box to glass box. J. Acad. Mark. Sci. 48, 137 – 141. doi: doi:10.1007/s11747-019-00710-5
  34. Salganik M. J., Heckathorn D. D. (2004). Sampling and estimation in hidden populations using respondent-driven sampling. Sociol. Methodol. 34, 193 – 240. doi: doi:10.1111/j.0081-1750.2004.00152.x
  35. Sarstedt M., Ringle C. M., Hair J. F. (2021). “Partial least squares structural equation modeling”. In Handbook of Market Research, ed. C. Homburg, M. Klarmann, and A. Vomberg (Cham: Springer International Publishing), 587 – 632. doi: doi:10.1007/978-3-319-57413-4_15
  36. Shin D. (2021). The effects of explainability and causability on perception, trust, and acceptance: implications for explainable AI. Int. J. Hum. Comput. Stud. 146: 102551. doi: doi:10.1016/j.ijhcs.2020.102551
  37. Shmueli G., Sarstedt M., Hair J. F., Cheah J. H., Ting H., Vaithilingam S.. (2019). Predictive model assessment in PLS-SEM: guidelines for using PLSpredict. Eur. J. Mark. 53, 2322 – 2347. doi: doi:10.1108/EJM-02-2019-0189
  38. Spence M. (1973). Job market signaling. Q. J. Econ. 87, 355 – 374. doi: doi:10.2307/1882010
  39. Sullivan V., Weger K. (2025). Transparency and explainability in AI-assisted decision making: effects on trust, perceived reliability, confidence, and ease of understanding. In Proceedings of the Human Factors and Ergonomics Society Annual Meeting (Vol. 69, No. 1) (Sage CA: Los Angeles, CA: SAGE Publications), 1827 – 1833. doi: doi:10.1177/10711813251369473
  40. Sun X., Yin J. (2026). The power of retail investors: the effect of online interactions over environmental issues on corporate greenwashing. J. Bus. Ethics, 1 – 28. doi: doi:10.1007/s10551-026-06258-z
  41. Sundari R. M., Saravanabhavan N. (2026). Taxpayers' perceptions and behavioral responses to GST 2.0 slab rationalization: evidence from MSMEs in the Indian automobile industry. Front. Psychol. 17: 1885139. doi: doi:10.3389/fpsyg.2026.1885139
  42. Vijayananth M., Saravanabhavan N. (2026). Beyond algorithmic trust: human-AI interaction competency strengthens AI-driven financial decision-making and financial resilience. Front. Artif. Intell. 9: 1867266. doi: doi:10.3389/frai.2026.1867266
  43. Weber P., Carl K. V., Hinz O. (2024). Applications of explainable artificial intelligence in finance—a systematic review of finance, information systems, and computer science literature. Manag. Rev. Q. 74, 867 – 907. doi: doi:10.1007/s11301-023-00320-0
  44. Xiao Y., Xiao L. (2025). The impact of artificial intelligence-driven ESG performance on sustainable development of central state-owned enterprises listed companies. Sci. Rep. 15: 8548. doi: doi:10.1038/s41598-025-93694-y
  45. Zhong M., Zhu Y., Li R., Liu Z. Y. R. (2025). ESG greenwashing and retail investor criticisms. Res. Int. Bus. Finance, 79: 103034. doi: doi:10.1016/j.ribaf.2025.103034

Citation

Vijayananth, M., et al. (2026). Explainable AI and ESG analytics for sustainable investment decision: the roles of ESG forecast credibility and trust in AI-generated information. Frontiers in Sustainability, Volume 7. https://doi.org/
Received
Accepted
Published
Volume
7 - 2026

Copyright

Copyright © 2026 Vijayananth, Sundari and Saravanabhavan.

License: https://creativecommons.org/licenses/by/4.0/

*Correspondence: N. Saravanabhavan

Article metrics

Views
0
Downloads
0
Citations
0
Shares
0

Related articles

Systematic ReviewThermochemical and biochemical conversion technologies for the valorization of coffee-industry residues: a systematic reviewMaricielo Yzquierdo-Alejandría, César A. Chomlón-Perales, José A. Vásquez–CoronelOriginal ResearchCorporate social responsibility as a strategic resource: examining the mediating role of corporate reputation in securing competitive advantageHazem Aldabbas, Abdallah M. Elamin, Liza Gernal, Ahmed Z. E. Ahmed, Ruba M. IbnaufOriginal ResearchThe aesthetics of applying 3D printing in the design and manufacture of sustainable bags for womenRania Shawky Ghazy, Afnan Abdallah Al-Amri