Table 9 Future research agenda Theme... | Emerald Publishing

Table 9

Future research agenda

Theme	Examples of AI context	Future research agenda	Proposed methodology	Proposed variables across all themes
AI as a strategic infrastructure for digital transformation in banking and financial services	Use of AI in auditing and accounting information systems; adoption of AI in banking services across different cultures	How does AI function as a multi-layered infrastructure across different geographies (e.g. developed vs emerging markets) and organizational types (e.g. local vs global banks)? How is risk aversion toward AI implementation different in various organization types? How do implementation costs, limited executive sponsorship and/or skepticism toward opaque algorithmic decisions influence AI adoption in financial institutions? How do different stages of development across data, technology and governance shape firm performance? How does AI infrastructure interact with institutional strategies, internal policies and regulatory environments? Does AI infrastructure improve efficiency at the cost of perceived fairness? How does AI-driven lending affect investor-side decisions (not just borrowers)?	Qualitative: Grounded theory analysis, multiple case study analysis Quantitative: Multilevel modeling, mediation/moderation analysis, event study methodology	Antecedents Psychological traits: regret aversion coefficient, rarer probability Firm-level: innovation intensity, marketing orientation, firm performance, ROA, ROI Top Management Team (TMT)- level: Power/presence/tenure, TMT diversity, credit repayment behavior AI literacy, TMT risk orientation Marketing-level: advertising intensity, innovation intensity, marketing orientation Bias: sampling bias Creditworthiness indicators: debt-paying ability, growth ability Culture: social norms, openness to new ideas, trust in others, fear of being cheated, attitude toward science, value placed on independence, Guanxi Personal attributes: race, income, education, age, location, experience Outcomes Fairness perception Investment behavior: likelihood of investment, propensity to invest, willingness to fund, likelihood of investment Borrower repayment behavior: probability of mortgage prepayment, credit repayment behavior TMT level: AI literacy, TMT risk orientation Firm-level: strategic orientation, innovation orientation Market-level: market share, brand equity, switching behavior, returns, volatility Firm-level: firm performance, strategic orientation, innovation orientation, corporate social responsibility Technology adoption behavior: resistance to AI adoption, attitude toward fintech chatbots, willingness to switch to fintech chatbots Model performance: brier score, AUC classification quality, AUC model performance, prediction accuracy of repayment behavior Loan performance: loan default, default status, default rate, project failure likelihood Moderators Personal attributes: gender, experience, age, race, income, region/location, education Bias Decision outcomes: decision outcome: acceptance rate, classification task complexity AI perceived use: AI's perceived usefulness, AI's ease of use, model updates AI system attributes: Chain-of-Thought (CoT) prompt, class imbalance, model updates Mediators Bias: perceived bias in facial expressions Trust: reliability trust, security trust, capability trust
Human–algorithm interaction and the duality of algorithmic opacity: Toward bias mitigation and behavioral correction	Use of AI to mitigate cognitive biases in financial planning industry; use of AI to mitigate bias in credit-risk analysis in financial institutions	How do cognitive, emotional and behavioral dynamics shape human-algorithm collaboration? How do trust, overreliance and interpretive reasoning evolve in adaptive AI environments? How does behavioral correction mechanisms, such as feedback loops, adaptive nudges and cognitive calibration systems, enhance human–algorithm interactions and systemic stability? How can integrating behavioral economics with machine learning research reveal how biases evolve across time? How does organizational learning process prevent amplification of biases? How can conditioning LM models to reflect sociocultural traits provide insights into different patterns of attitudes across groups? How do neurophysiological measures of attention and confidence enhance understanding of how users determine trust? How do varying levels of algorithmic opacity and explainability affect trust calibration? How does algorithmic opacity influence investor willingness to fund?	Qualitative: Ethnographic analysis, qualitative interview analysis Quantitative: Experiments, mediation/moderation analysis, SEM
Fairness frameworks and responsible AI governance: Trust, ethics and systemic stability	Use of AI to assess and monitor the trustworthiness of AI applications in finance; use of AI in financial decision-making and fairness perception	How can the studies of fairness be extended from technical to socio-organizational domains? How do fairness-by-design principles interact with firm culture, regulatory mandates and consumer perceptions? How do different regulatory regimes influence the adoption of responsible AI? How do different institutional architectures foster or constrain responsible AI adoption? How can mixed method approaches bridge quantitative fairness metrics with qualitative evaluations of legitimacy and trust? Examine how governance policies influence algorithmic fairness across banks How does fairness regulation influence loan default outcomes? Does perceived fairness affect portfolio allocation decisions?	Qualitative: Case study analysis, qualitative interview analysis Quantitative: Multilevel modeling, SEM, event study methodology
Behavioral finance in the age of AI and LM	Estimating individual bid responses in banking using causal machine learning; use of AI in automated pricing for credit terms	Are certain machine learning architectures inherently more bias-resistant than others? How does reliance on AI-based forecasting change the mental workload of financial decision-makers? How do human behaviors and biases influence AI model training? How do AI models reshape future human judgments? How is the loop created between human and algorithmic learning? Does AI adoption change risk-taking in financial markets?	Qualitative: Qualitative interview analysis, narrative analysis Quantitative: Experiments, mediation/moderation analysis, multilevel modeling

Theme	Examples of AI context	Future research agenda	Proposed methodology	Proposed variables across all themes
AI as a strategic infrastructure for digital transformation in banking and financial services	Use of AI in auditing and accounting information systems; adoption of AI in banking services across different cultures	How does AI function as a multi-layered infrastructure across different geographies (e.g. developed vs emerging markets) and organizational types (e.g. local vs global banks)? How is risk aversion toward AI implementation different in various organization types? How do implementation costs, limited executive sponsorship and/or skepticism toward opaque algorithmic decisions influence AI adoption in financial institutions? How do different stages of development across data, technology and governance shape firm performance? How does AI infrastructure interact with institutional strategies, internal policies and regulatory environments? Does AI infrastructure improve efficiency at the cost of perceived fairness? How does AI-driven lending affect investor-side decisions (not just borrowers)?	Qualitative: Grounded theory analysis, multiple case study analysis Quantitative: Multilevel modeling, mediation/moderation analysis, event study methodology	Antecedents Psychological traits: regret aversion coefficient, rarer probability Firm-level: innovation intensity, marketing orientation, firm performance, ROA, ROI Top Management Team (TMT)- level: Power/presence/tenure, TMT diversity, credit repayment behavior AI literacy, TMT risk orientation Marketing-level: advertising intensity, innovation intensity, marketing orientation Bias: sampling bias Creditworthiness indicators: debt-paying ability, growth ability Culture: social norms, openness to new ideas, trust in others, fear of being cheated, attitude toward science, value placed on independence, Guanxi Personal attributes: race, income, education, age, location, experience Outcomes Fairness perception Investment behavior: likelihood of investment, propensity to invest, willingness to fund, likelihood of investment Borrower repayment behavior: probability of mortgage prepayment, credit repayment behavior TMT level: AI literacy, TMT risk orientation Firm-level: strategic orientation, innovation orientation Market-level: market share, brand equity, switching behavior, returns, volatility Firm-level: firm performance, strategic orientation, innovation orientation, corporate social responsibility Technology adoption behavior: resistance to AI adoption, attitude toward fintech chatbots, willingness to switch to fintech chatbots Model performance: brier score, AUC classification quality, AUC model performance, prediction accuracy of repayment behavior Loan performance: loan default, default status, default rate, project failure likelihood Moderators Personal attributes: gender, experience, age, race, income, region/location, education Bias Decision outcomes: decision outcome: acceptance rate, classification task complexity AI perceived use: AI's perceived usefulness, AI's ease of use, model updates AI system attributes: Chain-of-Thought (CoT) prompt, class imbalance, model updates Mediators Bias: perceived bias in facial expressions Trust: reliability trust, security trust, capability trust
Human–algorithm interaction and the duality of algorithmic opacity: Toward bias mitigation and behavioral correction	Use of AI to mitigate cognitive biases in financial planning industry; use of AI to mitigate bias in credit-risk analysis in financial institutions	How do cognitive, emotional and behavioral dynamics shape human-algorithm collaboration? How do trust, overreliance and interpretive reasoning evolve in adaptive AI environments? How does behavioral correction mechanisms, such as feedback loops, adaptive nudges and cognitive calibration systems, enhance human–algorithm interactions and systemic stability? How can integrating behavioral economics with machine learning research reveal how biases evolve across time? How does organizational learning process prevent amplification of biases? How can conditioning LM models to reflect sociocultural traits provide insights into different patterns of attitudes across groups? How do neurophysiological measures of attention and confidence enhance understanding of how users determine trust? How do varying levels of algorithmic opacity and explainability affect trust calibration? How does algorithmic opacity influence investor willingness to fund?	Qualitative: Ethnographic analysis, qualitative interview analysis Quantitative: Experiments, mediation/moderation analysis, SEM
Fairness frameworks and responsible AI governance: Trust, ethics and systemic stability	Use of AI to assess and monitor the trustworthiness of AI applications in finance; use of AI in financial decision-making and fairness perception	How can the studies of fairness be extended from technical to socio-organizational domains? How do fairness-by-design principles interact with firm culture, regulatory mandates and consumer perceptions? How do different regulatory regimes influence the adoption of responsible AI? How do different institutional architectures foster or constrain responsible AI adoption? How can mixed method approaches bridge quantitative fairness metrics with qualitative evaluations of legitimacy and trust? Examine how governance policies influence algorithmic fairness across banks How does fairness regulation influence loan default outcomes? Does perceived fairness affect portfolio allocation decisions?	Qualitative: Case study analysis, qualitative interview analysis Quantitative: Multilevel modeling, SEM, event study methodology
Behavioral finance in the age of AI and LM	Estimating individual bid responses in banking using causal machine learning; use of AI in automated pricing for credit terms	Are certain machine learning architectures inherently more bias-resistant than others? How does reliance on AI-based forecasting change the mental workload of financial decision-makers? How do human behaviors and biases influence AI model training? How do AI models reshape future human judgments? How is the loop created between human and algorithmic learning? Does AI adoption change risk-taking in financial markets?	Qualitative: Qualitative interview analysis, narrative analysis Quantitative: Experiments, mediation/moderation analysis, multilevel modeling

Source(s): Created by authors