Artificial Intelligence and the Reconfiguration of Emotional Well-Being (2020–2025): A Critical Reflection

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript presents a timely and critical reflection on a topic of significant relevance: the psychosocial impact of Artificial Intelligence (AI) on emotional well-being. The interdisciplinary approach and the proposed conceptual model are valuable contributions. However, the article is marred by significant methodological weaknesses and a lack of analytical depth, necessitating a major revision to achieve the expected rigor and scholarly contribution.

• The title and abstract should more accurately reflect the nature of the study as a critical review or reflective article, rather than an empirical research paper.

• While the introduction effectively establishes the topic's relevance, the temporal delimitation "2020–2025" appears artificial for a manuscript being written in 2024/2025. This choice requires justification or rephrasing.

• A "knowledge gap" is correctly identified, but it could be articulated with greater precision. Specifically, which conceptual frameworks are missing? How is the existing literature insufficient for explaining the "constitutive tensions" the author mentions?

• The methodology section is the most substantial weakness. It describes a "critical analysis" of 40 articles but fails to provide the search strategy. It is essential to detail:

  • The specific keywords and Boolean operators used.

  • The exact databases and the filtering criteria applied (e.g., by date and field).

  • A flow diagram (or an equivalent narrative description) of the article selection process, demonstrating how the inclusion/exclusion criteria were applied. This is a standard practice for reviews, including critical ones.

• A three-phase process (exploratory reading, thematic clustering, interpretive synthesis) is mentioned, but the description is vague. It is necessary to clarify:

  • How the "preliminary categories" were defined.

  • Providing greater detail on the "interpretive synthesis" process would strengthen the credibility of the analysis.

• The analysis, while well-written, tends to be overly descriptive and enumerative. It cites numerous authors but lacks a deeper layer of critical synthesis and confrontation of ideas.

• The proposed conceptual model (Section 2.4) is a positive contribution, but its presentation is insufficient. It is strongly recommended to include a figure or diagram that visually illustrates the three levels (technological–structural, psychosocial–relational, ethical–existential) and their interrelationships. This would enhance its clarity and utility.

• The limitations section is adequate, but the future research lines are somewhat generic. They could be made more specific and directly linked to the article's concrete findings.

The manuscript addresses a highly contemporary and important topic from a valuable perspective, and the proposed model of AI-mediated emotional well-being is a potentially significant contribution to the field. However, in its current state, I cannot recommend it for publication due to critical deficiencies, primarily in the transparency and robustness of the methodology and the depth of the critical analysis and synthesis.

Author Response

Please review the attached PDF document.

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The article offers a very interesting critical review of a timely and important topic. The paper is generally well and clearly written. However, I would like to make some suggestions to increase the quality of this paper.

1. A socio-material or a sociotechnical approach would be very suitable as a theoretical approach of this study. A relationship between human and AI tools is technology-mediated and they constantly shape and re-shape each other. AI tools have an impact on human's practices, behavior, and choices.
2. Abbreviations, such as JCR and SJR, are recommended to open upon first mention.
3. Methods section: I am interested in how the 40 articles in the review were selected. Please, elaborate the description of the methodology, i.e. the concrete steps that were taken to compile the articles. In fact, which articles were included in the review? For the sake of transparency, it would be good if the authors can included a list of the articles either as a part of the text (perhaps as a table?) or as an appendix. Also, in which scientific fields were the studies conducted and where? Further, the analysis was guided by critical analysis, but how, that could be refined.
4. On page 4, the authors write "Artificial intelligence (AI) has become a key mediator of contemporary emotional life." Is this a bit too strong statement? Rather, it is one of a key mediators of contemporary emotional life.
5. The conceptual model of AI mediating emotional well-being is interesting and I suggest the authors to make an illustrative figure of it to clarify the text.

The authors have thoroughly considered the theoretical ja practical implications, as well as limitations of the study and future research.  Overall, I recommend publishing with revisions.

Author Response

Please review the attached PDF document.

Author Response File: Author Response.pdf

Reviewer 3 Report

Comments and Suggestions for Authors

To systematically improve the identified weaknesses, the paper should be further developed along six interconnected lines. First, the objectives should be reformulated into two to four precise research questions that operationalize the overall aim, for example focusing on specific effects of AI on emotional expression, self-regulation, trust, and dehumanization in clearly defined contexts such as work, education, or healthcare. These questions should explicitly specify target populations, settings (clinical vs. non-clinical), and types of AI systems (chatbots, conversational assistants, recommendation systems), so that the scope and boundaries of the review become transparent and empirically tractable.

Second, the theoretical grounding needs to be made explicit and differentiated. The paper should identify and briefly elaborate core theoretical frameworks—such as human-centered AI, trust in automation, theories of dehumanization, mediated care and support, and organizational stress and coping models—and clearly link them to the formulated research questions. In addition, a concise “state of the art” section should position the article among existing reviews and conceptual papers on AI in mental health and organizational psychology, explicitly stating which gap it addresses, for instance by integrating organizational and mental-health perspectives or by focusing on paradoxes of support versus dependence. This will clarify whether the article extends, challenges, or synthesizes prior work, and avoid the impression of a merely generic overview.

Third, the review methodology should be made more systematic and transparent. The authors can draw on established reporting guidance (e.g. PRISMA-adequate logic) to document search strategies, databases, time frames, inclusion and exclusion criteria, screening processes, data extraction procedures, and any quality appraisal tools used. Even if the review is not fully systematic, a flow-style description of how the 40 articles were identified and selected would significantly increase credibility and reproducibility. Furthermore, the notion of “conceptual triangulation” should be specified: the paper should explain which sources and theoretical lenses were combined, how conflicting findings were handled, and in what way triangulation helped to reduce bias and enhance the robustness of interpretations.

Fourth, the presentation of results should move from a condensed paradox to a structured, differentiated synthesis. The findings can be organized along a clear conceptual framework with separate subsections, for example: access and effectiveness of emotional support, forms of self-regulation mediated by AI, dynamics of trust and distrust, and mechanisms and indicators of dehumanization and technological dependence. Within each subsection, the paper should illustrate patterns with multiple study examples, indicating types of designs, samples, and contexts, and should systematically qualify the strength and consistency of evidence—highlighting where effects are robust, where they are mixed, and where the evidence is thin or methodologically weak. Tables or matrices summarizing the main characteristics and key findings of the 40 studies would make the argument more transparent and allow readers to critically assess the underlying evidence.

Fifth, the implications need to be translated into concrete, audience-specific guidance and clearly anchored in the empirical patterns. For researchers, the paper should articulate targeted research gaps—such as the need for longitudinal studies on AI dependence, experimental work on trust calibration, or mixed-methods studies on subjective experiences of dehumanization—and link these directly to identified limitations in the current evidence. For practitioners, designers, and policymakers, the article should derive practical design and governance principles for human-centered AI in mental-health and organizational contexts, for example guidelines on conversational style and empathy, transparency and explainability, escalation paths to human professionals, monitoring protocols, and ethical oversight structures. Each recommendation should be explicitly connected to the corresponding findings, and it should be clearly indicated where guidance is strongly supported by evidence and where it remains primarily normative.

Finally, the claimed contribution and originality should be articulated in a precise and modest but convincing way. Rather than stating that the paper “redefines” the role of AI in organizational psychology and mental health in abstract terms, the authors should specify what is new: for example, the proposal of a specific conceptual model or typology of AI-mediated psychological support that systematically accounts for the observed paradox between expanded support and risks of dehumanization; the integration of previously separate organizational and clinical literatures into a unified framework; or the development of a set of design and governance principles derived from a cross-contextual synthesis of studies. These elements should be presented as a coherent contribution that extends existing theories and reviews, allowing readers to see exactly how the article advances the field and making its added value traceable and assessable.

Author Response

Please review the attached PDF document.

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I have carefully reviewed the revised version of the manuscript and appreciate the evident effort made to address the observations raised in the previous round. The paper exhibits substantive improvements in methodological, conceptual, and expository dimensions.

The abstract now explicitly states that the study is a critical analytical review, which enhances conceptual coherence and provides greater transparency for the reader.

The introduction incorporates a solid justification grounded in the post-pandemic context, the rise of generative AI, and the increased volume of scientific production during this period.

The revision also includes keywords, Boolean operators, and filtering criteria.

All observations raised in the previous round were satisfactorily addressed. For further improvement, I recommend the following:

1. Ensure consistent use of terminology, particularly regarding: generative AI/artificial intelligence; emotional well-being/socioemotional well-being; sociotechnical technologies/sociotechnical systems.

With these minor adjustments, the manuscript will achieve a presentation more closely aligned with the standards of scientific publishing.

Author Response

Please see the attachment

Author Response File: Author Response.pdf