Transforming E-Commerce with AI: Navigating Innovation, Personalization, and Ethical Challenges

1. Introduction

Artificial Intelligence (AI) has become a primary agent of change in the contemporary e-commerce landscape. The expansion of online retail and the rapid adoption of AI tools have reconfigured both firm capability and consumer experiences. Worldwide online retail sales reached roughly USD 6 trillion in 2024, reflecting the scale of the digital marketplace and the economic importance of platform-mediated commerce [1]. In parallel, the commercial ecosystem for AI-enabled e-commerce solutions has shown marked growth, as industry analyses estimate the AI-in-e-commerce market rose from about USD 8 billion in 2024 to more than USD 9 billion in 2025, with longer-term forecasts indicating continued acceleration [2]. This combination of market scale and technological progress has produced substantial opportunities for innovation while also raising complex questions about personalization, governance, and social impact.

It is against this backdrop that the 27th International Conference on Electronic Commerce (ICEC 2025) convened researchers and practitioners to explore cutting-edge developments. This Special Issue of the Journal of Theoretical and Applied Electronic Commerce Research (JTAER), titled “Transforming E-Commerce with AI: Navigating Innovation, Personalization, and Ethical Challenges,” presents extended versions of five selected high-quality papers from ICEC 2025. Rather than summarizing each contribution in isolation, this Editorial synthesizes insights and trends across these studies, organizing the discussion around three key dimensions: Innovation, Personalization, and Ethical Challenges. Together, these articles illuminate how AI is reshaping e-commerce and what this implies for future research and practice.

2. Innovation in AI-Driven E-Commerce

AI’s contribution to e-commerce innovation is multifaceted. First, firms that build integrated digital capabilities, including encompassing data collection, algorithmic analytics, and cross-functional coordination, appear to realize broader organizational benefits. Empirical work on digital marketing capability, using evidence from listed manufacturing firms, indicates that investments in digital and data capabilities confer advantages beyond marketing: they support operational improvements and foster R&D outcomes, thereby demonstrating spillovers that strengthen firm competitiveness. These results emphasize that AI-enabled capabilities should be considered organization-wide strategic assets rather than isolated marketing tools. This socio-technical view is reinforced by recent empirical assessments of AI adoption in health and services, which show that organizational routines, workforce readiness, and cross-functional coordination shape whether AI investments translate into improved service outcomes and operational [3].

Second, AI is enabling new kinds of customer-facing products and services that materially reduce information asymmetries in traditionally opaque domains. The AI “Magic Mirror,” developed for medical aesthetic services, is one such example: by generating individualized visual previews, the system reduces uncertainty and increases consumers’ purchase intentions for credence and experiential services where ex ante evaluation is difficult. Field evidence indicates that the effect is strongest for highly customized or unfamiliar procedures, suggesting that AI can be particularly transformative in settings where conventional product representations fall short in conveying likely outcomes. More broadly, augmentative visualization and virtual-try-on tools are beginning to approximate the experiential affordances of in-person commerce and to close the information gap inherent in online purchase decisions.

Third, innovation extends to pricing and monetization of AI services themselves. As advanced AI capabilities such as large language models (LLMs) are commercialized, firms must decide how to package and price those offerings. Theoretical modeling of competitive pricing for LLM services highlights how consumers’ psychological reactions to payment formats, for example, the salience of small recurrent micro-payments in pay-per-use models, can alter perceived utility and hence influence equilibrium pricing strategies. When users are frequent consumers or particularly sensitive to recurring transaction salience, subscriptions become relatively more attractive; in other circumstances, usage-based pricing may persist. These findings underscore that monetization strategies are an intrinsic element of innovation design for AI services and must be informed by behavioral consideration, as well as technological capability.

Fourth, organizational and service-delivery innovations enabled by platform technologies show that human coordination and algorithmic support are complementary. In domains that require domain expertise and trust, such as online healthcare, platforms are experimenting with team-based service models, where multiple experts jointly address a case and provide a consolidated recommendation. Empirical analysis shows that such team consultations increase patient adoption of advice relative to single-doctor interactions, particularly when teams include a visible, reputable lead and a complementary mix of specialties. Moreover, evidence from online healthcare platforms indicates that patients’ adoption of generative AI services is primarily driven by perceived usefulness and social influence, with trust in the platform playing a central mediating role, while ease of use exerts a comparatively limited effect on adoption intentions. These findings point to a broader lesson: technological innovations often achieve their fullest value when coupled with organizational arrangements that facilitate knowledge sharing and collective accountability.

Taken together, the studies in this section enlarge our understanding of what innovation with AI entails: it is not merely algorithm performance but a combination of capability building, new user-facing tools, careful monetization design, and service-delivery models that bring human expertise and automation into productive alignment.

3. Personalization and Customer-Centric AI

Personalization sits at the heart of AI’s promise for e-commerce. Modern machine learning systems enable tailoring at multiple levels, i.e., personalized recommendations, individualized visualizations, conversational interactions, and dynamic offers, all of which aim to make digital shopping more relevant, efficient, and engaging.

Interactive visualization (e.g., virtual try-on or the AI Magic Mirror) exemplifies personalization that directly influences decision quality by presenting outcomes that are specific to the individual consumer. Studies show that individualized visual previews increase perceived value and purchase intentions by decreasing subjective uncertainty and improving outcome salience; the benefit is particularly pronounced for services that are high in heterogeneity and low in prior familiarity. Such applications illustrate the capacity of AI to make intangible services more tangible and to shift the locus of trust toward algorithmically supported representation.

Conversational and generative AI agents further extend personalization into dialogic interactions. Research on generative AI assistants in healthcare contexts finds that adoption intentions are driven primarily by perceived usefulness and social endorsement, the degree to which peers or professionals vouch for the tool, rather than by ease of use alone. This suggests modern users often regard interface usability as a baseline expectation and instead evaluate AI systems chiefly on the relevance and trustworthiness of the content they provide. It also underscores heterogeneity across user segments: individuals with higher domain literacy scrutinize output quality more intensely and are less influenced by social cues. Consequently, personalization strategies must adapt both content and interaction style to user characteristics, rather than applying uniform settings.

Beyond the user–system interface, personalization must be embedded into organizational practices. Firms that integrate personalization into their digital marketing and product development processes can extract cross-departmental value: personalized insights inform assortment, promotions, and inventory decisions, thereby contributing to firm performance beyond immediate sales metrics. Effective personalization therefore entails data governance, analytical capability, and operational alignment.

Evidence from large-scale platform data indicates that physician engagement on online healthcare platforms is strongly shaped by economic incentives and their perceived linkage to offline income opportunities. Empirical findings show that introductory incentives can effectively activate physician participation by signaling future earning potential and reputational benefits, while excessive or poorly aligned incentives may crowd out intrinsic motivation and reduce sustained engagement. These results highlight that platform design choices regarding incentive structures function as a form of governance, shaping professional behavior, effort allocation, and the long-term quality of service provision [4].

Complementary evidence from information systems research indicates that advanced personalization mechanisms, particularly those enabled by generative AI and intelligent automation, can transform how platforms mediate interactions, not merely by tailoring content but by shaping the quality of user–system engagement over time. Specifically, dynamic, context-aware responses and hyper-personalized interaction pathways foster a sense of relevance and reciprocity that supports users’ continued engagement. These findings suggest that effective personalization strategies should extend beyond static feedback to include adaptive, evolving interaction structures that align with users’ expectations and sustained value perceptions [5].

Recent comprehensive reviews in 2025 highlight that fairness and diversity are central limits of standard personalization pipelines; they recommend integrating multi-stakeholder fairness metrics and exposure-diversity objectives into algorithm design and evaluation to mitigate popularity and provider-side biases [6]. In practice, personalization strategies should therefore integrate algorithmic precision, interaction design, and affective support while accounting for user heterogeneity to avoid one-size-fits-all outcomes and to preserve fairness and trust.

4. Ethical Challenges and Considerations

The deployment of AI at scale raises critical ethical, social, and policy issues that require explicit attention in both research and practice. The papers in this Special Issue engage these themes in various ways, and together they point to several cross-cutting concerns.

Personalization and automated decision-making depend on large datasets, often including sensitive information. In healthcare and other sensitive contexts, users’ concerns about data handling, storage, and potential misuse are prominent and can impede adoption. Research underscores that effective deployment requires not only technical solutions, such as federated learning or differential privacy, but also clear institutional frameworks and compliance with jurisdictional data-protection standards. Platforms should incorporate privacy by design and be transparent about data practices to maintain user trust. Systematic reviews from 2025 conclude that federated learning and related privacy-enhancing architectures present practical pathways for reconciling personalization with data protection in regulated settings while highlighting technical challenges (heterogeneity, communication cost, robust aggregation) that must be addressed for real-world deployment [7].

While personalization can reduce uncertainty and improve match quality, certain applications may have unintended psychological consequences. For example, prototype visualizations that render idealized outcomes for aesthetic procedures can inadvertently amplify body image concerns or create pressure to pursue treatments. The literature cautions designers to avoid manipulative presentation and to provide calibrated, evidence-based previews alongside educational resources and opt-out mechanisms. More generally, recommendation systems and dynamic pricing must be assessed for potential exploitative patterns, addictive behaviors, or reinforcement of harmful norms.

Personalized pricing or micro-segmentation raises questions of equity. Recent studies and policy analyses show that algorithmic and dynamic pricing can produce opaque price dispersion and potential consumer harm in practice, motivating proposals for disclosure rules, monitoring tools, and targeted regulatory interventions to preserve consumer welfare [8]. Theoretical work on AI service pricing shows how different monetization models distribute value across user groups; empirical attention to distributional consequences and potential discrimination is therefore essential. Transparency, in the form of explainability, disclosures about automation, and user control options, is a necessary complement to fairness evaluations. Where algorithmic outputs materially affect consumer outcomes, auditability and accountability mechanisms are warranted.

Evidence from online healthcare platforms suggests that human-in-the-loop arrangements, such as a team-based review of AI-generated advice, can strengthen reliability and credibility. Platforms operating in high-stakes domains should consider hybrid governance models that combine algorithmic assistance with professional supervision, thereby preserving human accountability while harnessing AI efficiencies. Such designs can mediate liability concerns and support reliable escalations when automated outputs are uncertain.

Regulatory frameworks often follow rapid technological change, and recent analyses of the EU digital legislative architecture highlight complex overlaps and gaps, for example, between the AI Act, GDPR, and other digital laws, that complicate coherent governance of cross-border AI-enabled services [9].

At the same time, cross-border data flows, both personal and non-personal, are subject to an increasingly fragmented patchwork of national measures; mapping exercises show that this fragmentation raises compliance costs, impedes interoperability, and can materially constrain the international deployment of data-intensive AI services [10].

Moreover, focused research on algorithmic pricing and market oversight demonstrates that automated pricing systems can produce novel consumer-protection and competition risks in practice, which strengthens the case for targeted regulatory instruments such as enhanced disclosure, market monitoring, and interoperable oversight mechanisms rather than ad hoc unilateral rules [8].

From a policy perspective, these findings imply that regulators and standard setters should treat platform design choices, i.e., team composition rules, contribution visibility, incentive framing, and accountability allocation, as governance instruments and include them explicitly in oversight conversations about service quality and market integrity [11].

Addressing these ethical challenges requires interdisciplinary collaboration that blends technical safeguards, organizational governance, and policy frameworks. Only by aligning these domains can AI-driven personalization and innovation proceed in a manner consistent with public values.

5. Conclusions and Future Outlook

Taken together, the contributions in this Special Issue demonstrate that AI is reshaping e-commerce through intertwined processes of innovation, personalization, and ethical negotiation. The studies provide evidence that AI-enabled tools and capabilities can enhance firm performance, improve user experience, and enable the development of new service models while also highlighting the importance of aligning technological advances with human values and expectations.

Looking ahead, continued advances in generative AI, multimodal systems, and immersive technologies are likely to further expand the scope of AI-driven e-commerce. At the same time, further research is needed to deepen our understanding of organizational transformation, user heterogeneity, and governance mechanisms. The balance between personalization and privacy, in particular, will remain a central challenge.

From an industry perspective, the findings suggest that firms should pursue AI adoption to promote not only efficiency but also strategic capability while simultaneously investing in ethical design and transparent governance. For the research community, this Special Issue underscores the value of interdisciplinary inquiry that integrates technical, behavioral, and ethical perspectives. By advancing such integrated approaches, scholars and practitioners can help ensure that the ongoing transformation of e-commerce through AI contributes to sustainable and inclusive digital ecosystems.