Educating for Complexity: A Learning Architecture for Systems Thinking in Professional Education and Generative AI Governance

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

First, I would like to recognize the work that has gone into this manuscript. The topic is clearly relevant and timely, especially given the growing interest in systems thinking, professional education, and the governance challenges associated with generative artificial intelligence. The paper sets out to build a conceptual framework that connects established theoretical traditions with implications for curriculum design, assessment, and future research. This is an ambitious and worthwhile goal. The manuscript reflects careful engagement with both foundational texts in systems thinking and more recent discussions on generative AI and educational change.

At the same time, I believe the article would benefit from substantial revision before it is ready for publication. The issue is not the importance of the topic, but rather the level of methodological clarity needed to support the scope of the claims being made. In particular, the research design underlying the development of the conceptual framework requires further elaboration.

Although the manuscript identifies itself as a conceptual article and briefly outlines its methodological approach, the explanation remains somewhat limited. The authors indicate that they conducted a structured conceptual synthesis across several bodies of literature, but it is not entirely clear how this synthesis was carried out in practice. How were the sources identified and selected? What criteria guided decisions about inclusion or emphasis? Even if the review was problem-driven rather than exhaustive, readers still need enough information to understand how the different strands of literature were brought together and why certain perspectives were prioritized. At present, the relevant domains are described, but the analytical process connecting them is not fully transparent.

A related concern arises in the move from literature synthesis to the proposed learning architecture. The model organized around four iterative practices is promising and potentially valuable. However, the manuscript occasionally shifts quite quickly from theoretical discussion to normative recommendations. Curricular implications, assessment strategies, and governance considerations are introduced as logical extensions of the model, yet the intermediate reasoning is sometimes compressed. Making the progression from theory to model, and from model to institutional implications, more explicit would strengthen the overall argument and make it more persuasive.

There is also a question of tone in certain sections. In places—particularly when outlining propositions or governance strategies—the language suggests a level of certainty that may exceed what a conceptual paper can reasonably claim. This does not undermine the proposal itself, but it would help to distinguish more clearly between elements grounded directly in existing scholarship, interpretive conclusions drawn by the authors, and genuinely novel components of the framework. Greater precision here would enhance the article’s intellectual transparency.

The methodological section could be deepened further by clarifying how the key categories and practices were developed. The manuscript refers to a multi-step process, but the steps are described at a fairly general level. For instance, was there a systematic comparison of concepts across traditions? Some form of thematic mapping? An analysis of overlapping constructs? Even in conceptual research, readers benefit from understanding how a framework takes shape and why it should be considered robust rather than illustrative.

With regard to the integration of systems thinking and generative AI—arguably the core contribution of the article—the balance sometimes feels uneven. The discussion of systems thinking in professional education is detailed and well grounded. By contrast, generative AI occasionally appears as an added layer rather than as a fully integrated element of the model. The paper would be stronger if it demonstrated more concretely how each of the four practices interacts with the specific affordances and limitations of generative AI. Brief examples illustrating how AI tools might influence system mapping, boundary critique, feedback analysis, or intervention design would make the framework more tangible.

The engagement with the literature is generally solid and appropriately combines classical and contemporary sources. However, the manuscript could engage more critically with the limits of its own proposal. The framework is presented as broadly applicable across professional education contexts, yet the practical conditions required for implementation are not examined in depth. Attention to disciplinary, institutional, or cultural differences would add nuance and avoid the impression of universal transferability.

In terms of structure, the manuscript is overall well organized, and the tables are helpful in clarifying several components of the framework. That said, some sections become conceptually dense, with multiple propositions and implications presented in close sequence. While this reflects the richness of the model, clearer differentiation between core elements and supporting elaborations would improve readability. In a few cases, consolidating material might actually make the central contribution more visible.

The language is generally clear and academically appropriate. Nonetheless, certain passages—particularly in the methodological and governance sections—contain long and syntactically complex sentences that may hinder readability. A careful stylistic revision aimed at concision and flow would enhance the accessibility of what is already a sophisticated argument.

Overall, the manuscript addresses an important and intellectually promising topic. It brings together significant strands of scholarship and proposes a framework with potential relevance for curriculum design and institutional strategy. To reach that potential, however, the paper would benefit from a more detailed account of its methodological foundations, a clearer separation between synthesis and original contribution, stronger integration of generative AI within the core model, and a slightly more measured articulation of its normative claims. Addressing these issues would substantially strengthen the clarity and credibility of the work.

Comments on the Quality of English Language

The manuscript is written in generally clear and understandable academic English, and the main ideas can be followed without major difficulty. The terminology used is appropriate for the fields of systems thinking, professional education, and artificial intelligence in education. Overall, the language does not represent a major barrier to understanding the argument.

However, the text would benefit from a careful stylistic revision. In several sections, particularly in the methodological discussion and in the parts addressing governance and the research agenda, sentences become quite long and syntactically dense. This occasionally affects readability and makes some passages more difficult to follow than necessary. There are also instances of lexical repetition and some segments where the argument could be expressed more concisely.

A moderate language edit by a proficient academic English speaker or a professional editing service would help improve clarity, flow, and precision. Streamlining certain sentences and slightly reducing structural complexity would make the manuscript easier to read and would strengthen the presentation of what is otherwise a conceptually rich and ambitious contribution.

Author Response

Comment 1: First, I would like to recognize the work that has gone into this manuscript. The topic is clearly relevant and timely, especially given the growing interest in systems thinking, professional education, and the governance challenges associated with generative artificial intelligence. The paper sets out to build a conceptual framework that connects established theoretical traditions with implications for curriculum design, assessment, and future research. This is an ambitious and worthwhile goal. The manuscript reflects careful engagement with both foundational texts in systems thinking and more recent discussions on generative AI and educational change.

At the same time, I believe the article would benefit from substantial revision before it is ready for publication. The issue is not the importance of the topic, but rather the level of methodological clarity needed to support the scope of the claims being made. In particular, the research design underlying the development of the conceptual framework requires further elaboration.

Although the manuscript identifies itself as a conceptual article and briefly outlines its methodological approach, the explanation remains somewhat limited. The authors indicate that they conducted a structured conceptual synthesis across several bodies of literature, but it is not entirely clear how this synthesis was carried out in practice. How were the sources identified and selected? What criteria guided decisions about inclusion or emphasis? Even if the review was problem-driven rather than exhaustive, readers still need enough information to understand how the different strands of literature were brought together and why certain perspectives were prioritized. At present, the relevant domains are described, but the analytical process connecting them is not fully transparent.

Response 1: We thank the reviewer for this thoughtful comment and apologise for the confusion that the earlier version may have created. We agree that, even in a conceptual paper, readers need a clearer explanation of how the relevant literature was identified, prioritised, and brought together in support of framework construction.

In response, we revised the Methodological Approach section to clarify that the manuscript adopts a conceptual, integrative theory-building approach rather than a systematic review. We now specify that relevant literature was identified primarily through Google Scholar and selected purposively according to conceptual relevance, theoretical centrality, and usefulness for linking systems concepts to curriculum, assessment, and governance. We also make the methodological boundaries of the synthesis more explicit by stating that it is selective rather than exhaustive and oriented toward framework construction rather than corpus mapping.

In addition, we clarified the analytic process through which the different strands of scholarship were connected. The revised section now explains that systems traditions were prioritised for conceptual centrality in defining the architecture itself, while the literatures on professional education, assessment, and GenAI were used to translate that architecture into curricular, evaluative, and governance implications. It also explains that framework development proceeded iteratively through comparative conceptual mapping, moving from design-problem identification to concept comparison and grouping, and then to curricular, assessment, and governance derivation.

We believe these changes address the reviewer’s concern by making the research design, selection logic, and analytic integration of the synthesis more transparent.

 

Comment 2: A related concern arises in the move from literature synthesis to the proposed learning architecture. The model organized around four iterative practices is promising and potentially valuable. However, the manuscript occasionally shifts quite quickly from theoretical discussion to normative recommendations. Curricular implications, assessment strategies, and governance considerations are introduced as logical extensions of the model, yet the intermediate reasoning is sometimes compressed. Making the progression from theory to model, and from model to institutional implications, more explicit would strengthen the overall argument and make it more persuasive.

Response 2: We thank the reviewer for this thoughtful observation. We agree that the manuscript needed to make the progression from theory to model, and from model to curricular, assessment, and governance implications, more explicit. In the earlier version, these extensions were conceptually present, but some of the intermediate reasoning was stated too compactly.

In response, we revised the manuscript to clarify the derivational logic linking levels of the argument. The revised text now makes clearer that the systems traditions reviewed in Section 3 provide the conceptual foundations for the four-practice architecture in Section 4, and that this architecture is then translated into professional capabilities, artefacts, and forms of judgement. We also added stronger transitional framing to show that the later sections derive curriculum design, assessment, and GenAI-governance implications from the architecture itself.

More specifically, the revised manuscript now clarifies that curriculum follows from the need to sequence and revisit the four practices over time, assessment follows from the need to evaluate the artefacts and decisions through which those practices become visible, and governance follows from the fact that GenAI can shape each practice and must therefore be aligned with the pedagogical aims of the framework. We believe these revisions make the progression of the argument more explicit and strengthen the persuasiveness of the overall model.

 

Comment 3: There is also a question of tone in certain sections. In places—particularly when outlining propositions or governance strategies—the language suggests a level of certainty that may exceed what a conceptual paper can reasonably claim. This does not undermine the proposal itself, but it would help to distinguish more clearly between elements grounded directly in existing scholarship, interpretive conclusions drawn by the authors, and genuinely novel components of the framework. Greater precision here would enhance the article’s intellectual transparency.

Response 3: We thank the reviewer for this important observation. We agree that a conceptual paper must be especially careful in distinguishing between claims grounded directly in prior scholarship, interpretive conclusions derived from the framework, and genuinely novel components proposed by the authors. In the previous version, some parts of the manuscript, particularly the propositions and aspects of the governance discussion, used language that could be read as more certain or prescriptive than was warranted for a conceptual contribution.

In response, we revised these sections to improve intellectual transparency and to calibrate the tone more carefully. The revised manuscript now clarifies more explicitly that the propositions are conceptually derived and intended as hypotheses for future testing rather than as established empirical claims. We also moderated modal language where appropriate by replacing stronger formulations such as “will” or “predicts” with more qualified formulations such as “may,” “is likely to,” or “suggests.”

In addition, we revised the governance section so that the proposed mechanisms are framed more clearly as analytically grounded design implications. Where these mechanisms are directly supported by existing scholarship, this is now signalled more explicitly; where they represent interpretive extrapolations from the framework, they are presented as proposed strategies requiring further empirical validation. We believe these revisions make the manuscript’s evidentiary status, interpretive reasoning, and novel contribution more transparent.

 

Comment 4: The methodological section could be deepened further by clarifying how the key categories and practices were developed. The manuscript refers to a multi-step process, but the steps are described at a fairly general level. For instance, was there a systematic comparison of concepts across traditions? Some form of thematic mapping? An analysis of overlapping constructs? Even in conceptual research, readers benefit from understanding how a framework takes shape and why it should be considered robust rather than illustrative.

Response 4: We thank the reviewer for this valuable observation and apologise for the confusion that the earlier version may have created. We agree that, even in conceptual research, readers benefit from a clearer explanation of how the framework took shape and why it should be regarded as analytically robust rather than merely illustrative. In the previous version, the manuscript referred to an iterative multi-step process, but the analytic operations within that process were described too generally.

In response, we revised the Methodological Approach section to provide a more explicit account of how the key categories and practices were developed. The revised text now clarifies that framework development proceeded through an iterative process of comparative conceptual mapping. More specifically, we explain that the analysis first identified recurrent design problems in systems thinking education, then compared core concepts across the selected systems traditions in terms of their educational and professional relevance, and subsequently grouped overlapping and complementary constructs according to the kind of learning work they implied. This process led to the formulation of the four iterative practices that structure the proposed architecture.

We also clarify that the emerging architecture was then checked for cross-level coherence by examining whether the four practices could be translated consistently into professional capabilities, curricular sequences, artefacts, assessment criteria, and GenAI-governance implications. In this way, the revised section makes clearer not only how the framework was constructed, but also why it is presented as a robust integrative model rather than as a purely illustrative arrangement of ideas.

 

Comment 5: With regard to the integration of systems thinking and generative AI—arguably the core contribution of the article—the balance sometimes feels uneven. The discussion of systems thinking in professional education is detailed and well grounded. By contrast, generative AI occasionally appears as an added layer rather than as a fully integrated element of the model. The paper would be stronger if it demonstrated more concretely how each of the four practices interacts with the specific affordances and limitations of generative AI. Brief examples illustrating how AI tools might influence system mapping, boundary critique, feedback analysis, or intervention design would make the framework more tangible.

Response 5: We thank the reviewer for this important observation. We agree that the manuscript is stronger when the integration of systems thinking and GenAI is shown not only at the level of conceptual structure, but also through more concrete illustrations of how GenAI may interact with each of the four practices in the proposed architecture.

In response, we strengthened this integration at both the visual and textual levels. First, Figure 1 was revised so that GenAI is now represented explicitly as a cross-cutting support/risk/governance layer within the four-practice architecture, rather than as an external or loosely attached component. This visual revision was accompanied by corresponding changes in Section 4.2 to ensure closer alignment between the figure and the textual argument.

Second, we revised Sections 7.1–7.3 and Table 6 so that GenAI is connected more directly to each of the four practices—sensemaking and boundary setting, co-modelling and causal representation, intervention reasoning, and meta-learning—rather than discussed as a parallel AI-related layer. Third, we added a brief illustrative example showing how GenAI may shape stakeholder framing, causal-loop critique, scenario comparison, and reflective revision within an authentic learning task.

We believe these revisions address the reviewer’s concern by making the GenAI component more fully embedded within the proposed model and by strengthening the coherence between Figure 1, Section 4.2, and Section 7.

 

Comment 6: The engagement with the literature is generally solid and appropriately combines classical and contemporary sources. However, the manuscript could engage more critically with the limits of its own proposal. The framework is presented as broadly applicable across professional education contexts, yet the practical conditions required for implementation are not examined in depth. Attention to disciplinary, institutional, or cultural differences would add nuance and avoid the impression of universal transferability.

Response 6: We thank the reviewer for this important observation. We agree that the framework should not be read as uniformly transferable across all professional education settings, and that the manuscript needed to engage more explicitly with the practical and contextual conditions that may shape its implementation.

In response, we revised the manuscript to clarify more directly the limits and boundary conditions of the proposal. The revised text now states more explicitly that the architecture is intended to be transferable at the level of design logic rather than as a uniform implementation template. We also clarify that its enactment will depend on domain-sensitive and context-sensitive factors, including disciplinary epistemologies, programme structures, institutional resources, stakeholder access, and local cultures of assessment and participation.

In addition, we strengthened the manuscript’s treatment of contextual variation, especially in the future research agenda and, more briefly, in the concluding discussion of limitations. The revised version now places greater emphasis on cross-professional comparative research, equity and inclusion questions, and the institutional conditions under which governance practices may be sustained over time. It also clarifies that the framework is transferable at the level of design logic, but requires domain-sensitive adaptation and validation in local curricular contexts.

 

Comment 7: In terms of structure, the manuscript is overall well organized, and the tables are helpful in clarifying several components of the framework. That said, some sections become conceptually dense, with multiple propositions and implications presented in close sequence. While this reflects the richness of the model, clearer differentiation between core elements and supporting elaborations would improve readability. In a few cases, consolidating material might actually make the central contribution more visible.

Response 7: We thank the reviewer for this helpful observation. We agree that, although the manuscript is generally well organized, some sections became conceptually dense because core elements and supporting elaborations were presented in close sequence. In the earlier version, this sometimes made it harder to distinguish the central contribution of the paper from material intended as illustrative, methodological, or agenda-setting support.

In response, we revised the manuscript to clarify this internal hierarchy more explicitly. We strengthened the distinction between core elements and supporting elaborations, particularly in the sections on assessment and empirical development. More specifically, the revised text now signals more clearly that the core assessment contribution lies in the performance criteria and evidence blueprint, while the rubric template and links to existing instruments are presented as supporting extensions.

We also revised the final section so that the propositions remain the core empirical extension of the framework, while the methodological alignments and future research directions are more clearly consolidated and framed as supporting elaborations. We believe these revisions improve readability while preserving the integrative richness of the model.

 

Comment 8: The language is generally clear and academically appropriate. Nonetheless, certain passages—particularly in the methodological and governance sections—contain long and syntactically complex sentences that may hinder readability. A careful stylistic revision aimed at concision and flow would enhance the accessibility of what is already a sophisticated argument.

Response 8: We thank the reviewer for this helpful observation. We agree that, although the manuscript’s argument is conceptually structured, some passages—especially in the methodological and governance sections—were expressed in sentences that were longer and syntactically denser than necessary. In the earlier version, this may have reduced readability and made some parts of the argument harder to follow.

In response, we undertook a careful stylistic revision of these sections with particular attention to concision, sentence flow, and syntactic clarity. More specifically, we shortened several long sentences, reduced clause stacking, and redistributed dense formulations across shorter units in order to make the logic of the argument easier to follow without altering its substance.

We believe these stylistic changes enhance the accessibility of the manuscript while preserving the sophistication of the underlying argument.

 

Comment 9: Overall, the manuscript addresses an important and intellectually promising topic. It brings together significant strands of scholarship and proposes a framework with potential relevance for curriculum design and institutional strategy. To reach that potential, however, the paper would benefit from a more detailed account of its methodological foundations, a clearer separation between synthesis and original contribution, stronger integration of generative AI within the core model, and a slightly more measured articulation of its normative claims. Addressing these issues would substantially strengthen the clarity and credibility of the work.

Response 9: We thank the reviewer for this thoughtful and constructive overall assessment. We appreciate the recognition of the manuscript’s topic as important and intellectually promising, and we agree that the paper was strengthened by greater clarity in its methodological foundations, its contribution claim, the integration of GenAI into the core model, and the calibration of its normative language.

In response, we revised the manuscript along these lines. First, we expanded the Methodological Approach section to clarify more explicitly how the conceptual synthesis was conducted, how relevant literature was identified and selected, how the different strands of scholarship were prioritised, and how the framework was developed through iterative comparative conceptual mapping. Second, we sharpened the distinction between synthesis and original contribution by clarifying the specific scholarly gap addressed by the paper and by foregrounding the four-practice architecture as its central conceptual contribution, while framing curriculum, assessment, governance, and empirical development as translational extensions of that core model.

Third, we strengthened the integration of GenAI within the architecture itself through revisions to Figure 1, Section 4.2, Sections 7.1–7.3, and Table 6, so that GenAI is now represented and discussed as a cross-cutting support/risk/governance layer operating across the four practices rather than as a loosely attached add-on. Finally, we revised the tone of the manuscript where appropriate, especially in the propositions and governance discussion, to distinguish more clearly between literature-grounded claims, interpretive inferences, and novel framework components.

We believe these revisions directly address the concerns summarised by the reviewer and substantially improve the clarity, coherence, and credibility of the manuscript.

 

Comment 10: The manuscript is written in generally clear and understandable academic English, and the main ideas can be followed without major difficulty. The terminology used is appropriate for the fields of systems thinking, professional education, and artificial intelligence in education. Overall, the language does not represent a major barrier to understanding the argument.

However, the text would benefit from a careful stylistic revision. In several sections, particularly in the methodological discussion and in the parts addressing governance and the research agenda, sentences become quite long and syntactically dense. This occasionally affects readability and makes some passages more difficult to follow than necessary. There are also instances of lexical repetition and some segments where the argument could be expressed more concisely.

A moderate language edit by a proficient academic English speaker or a professional editing service would help improve clarity, flow, and precision. Streamlining certain sentences and slightly reducing structural complexity would make the manuscript easier to read and would strengthen the presentation of what is otherwise a conceptually rich and ambitious contribution.

Response 10: We thank the reviewer for this careful and constructive assessment of the manuscript’s language and presentation. We appreciate the recognition that the paper is generally clear and that its terminology is appropriate to the intersecting fields of systems thinking, professional education, and artificial intelligence in education. We also agree that some sections—particularly the methodological discussion, the governance section, and parts of the research agenda—contained sentences that were longer and syntactically denser than necessary.

In response, we undertook a focused stylistic revision of the manuscript aimed at improving clarity, flow, and concision. More specifically, we shortened a number of long sentences, reduced clause stacking, improved transitions, and simplified phrasing in places where the argument had become overly compressed. We also revised passages where lexical repetition was noticeable and streamlined some formulations so that the progression of the argument is easier to follow.

These revisions were applied especially to the Methodological Approach section, the discussion of GenAI governance, and the final section on propositions and future research. Our aim was to improve readability and precision without reducing the conceptual richness of the manuscript.

[All sections of the manuscript where modifications were made based on reviewer feedback are indicated in red]

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

Thank you for submitting your manuscript and for the thoughtful work invested in developing an integrative framework for systems thinking and the responsible use of generative AI in professional education. The topic is timely and of clear importance. However, I still have some concerns about the contribution to the field.

- The document states that it integrates several bodies of literature but does not specify the selection criteria, scope, or methodological boundaries of this synthesis. Without such justification, the conceptual framework lacks transparency and reproducibility.

- The manuscript attempts to cover foundational theory, curriculum design, assessment rubrics, governance mechanisms, testable propositions, and a research agenda. This breadth results in limited depth in key areas, making it difficult to evaluate the robustness and distinctiveness of the proposed architecture.

- Although the paper synthesizes diverse traditions in systems thinking and education, it does not convincingly demonstrate what specific gap in current scholarship it fills or how the proposed framework substantively advances beyond existing integrative models. I keep in mind that a clearly framed gap is essential for a conceptual contribution.

- The section on GenAI is extensive and well-documented, but it remains loosely connected to the core pedagogical architecture. As currently written, AI appears as an add-on rather than a conceptually embedded element of the proposed model. 

- Even as a conceptual paper, the manuscript would greatly benefit from concrete examples, cases, or hypothetical scenarios demonstrating how the proposed framework would operate in practice.

Author Response

Comment 1: The document states that it integrates several bodies of literature but does not specify the selection criteria, scope, or methodological boundaries of this synthesis. Without such justification, the conceptual framework lacks transparency and reproducibility.

Response 1: We thank the reviewer for this important observation. We agree that the earlier wording did not make the scope and methodological boundaries of the synthesis sufficiently explicit. At the same time, we would like to clarify that this manuscript is a conceptual theory-building paper rather than a systematic review, and therefore its aim is transparency of selection logic and analytical scope rather than reproducibility in the review-based sense.

In response, we revised the Methodological Approach section to clarify more explicitly how the literature was sourced, what bodies of scholarship were included, and what boundaries define the synthesis. The revised text now states that the paper draws purposively on four bodies of scholarship, that relevant literature was identified primarily through Google Scholar, and that source selection was guided by conceptual relevance, theoretical centrality, and usefulness for framework construction. We also clarify that the synthesis is selective rather than exhaustive, oriented toward framework building rather than corpus mapping, and intended to integrate seminal and analytically productive contributions rather than reproduce the procedures of a systematic review.

We believe this revision improves transparency and makes the scope, logic, and limits of the conceptual synthesis clearer.

 

Comment 2: The manuscript attempts to cover foundational theory, curriculum design, assessment rubrics, governance mechanisms, testable propositions, and a research agenda. This breadth results in limited depth in key areas, making it difficult to evaluate the robustness and distinctiveness of the proposed architecture.

Response 2: We thank the reviewer for this thoughtful observation. We agree that the manuscript spans several domains—foundational theory, curriculum, assessment, governance, and empirical development—and that, in the previous version, this breadth could make it more difficult to identify the paper’s main contribution and to judge the distinctiveness of the proposed architecture.

In response, we revised the manuscript to clarify the hierarchy of the argument. The revised version now states more explicitly that the paper’s primary contribution is the proposed four-practice conceptual architecture for systems thinking in professional education. The discussions of curriculum design, assessment, GenAI governance, and future empirical development are now framed more clearly as translational extensions of that architecture rather than as equally elaborated stand-alone frameworks.

To address this point, we also revised the Introduction, the opening of Section 5, and Section 8. These revisions moderate the scope of the claims, clarify the derivative status of the later sections, and present the propositions and research agenda as an initial empirical extension rather than as a comprehensive programme. We believe these changes improve the robustness, distinctiveness, and evaluability of the proposed architecture while preserving the integrative purpose of the manuscript.

 

Comment 3: Although the paper synthesizes diverse traditions in systems thinking and education, it does not convincingly demonstrate what specific gap in current scholarship it fills or how the proposed framework substantively advances beyond existing integrative models. I keep in mind that a clearly framed gap is essential for a conceptual contribution.

Response 3: We thank the reviewer for this important observation. We agree that, for a conceptual paper, the contribution must rest on a clearly framed gap and on a convincing explanation of how the proposed framework advances beyond existing scholarship. In the previous version, although the manuscript identified relevant challenges in systems thinking education, assessment, and GenAI integration, it did not sufficiently clarify how these strands had been addressed separately in prior work or why their integration into a single architecture constitutes a substantive advance.

In response, we revised the Introduction to sharpen both the gap statement and the distinctiveness of the contribution. The revised text now explains more explicitly that existing integrative contributions have tended to advance one dimension at a time—for example, curricular guidance, assessment design, or GenAI-related educational concerns—without connecting these within a single end-to-end design logic for professional systems thinking education. We therefore clarify that the specific gap addressed by this paper is not the absence of relevant scholarship per se, but the absence of a design-oriented architecture that integrates these strands into a coherent account of how systems thinking can be learned, assessed, and governed in professional education.

We also revised the contribution statement so that the framework’s distinctiveness is more visible. In particular, the revised version emphasises that the paper’s core contribution is the proposed four-practice architecture, while the discussions of curriculum, assessment, governance, and empirical development are presented as translational extensions derived from that central framework.

 

Comment 4: The section on GenAI is extensive and well-documented, but it remains loosely connected to the core pedagogical architecture. As currently written, AI appears as an add-on rather than a conceptually embedded element of the proposed model. 

Response 4: We appreciate the reviewer’s comment and agree that the manuscript needed to show more explicitly how the GenAI discussion is internal to the proposed pedagogical architecture rather than adjacent to it.

To address this, we revised both the text and Figure 1. The revised manuscript now clarifies that GenAI is not conceptualised as an external add-on or a separate fifth practice, but as a bounded cross-cutting layer of support, risk, and governance operating across the four core practices of the framework: sensemaking and boundary setting, co-modelling and causal representation, intervention reasoning, and meta-learning. We also revised the surrounding discussion so that the productive roles of GenAI, its predictable failure modes, and the governance mechanisms required for responsible use are each linked directly to these four practices.

In addition, Figure 1 has been revised to make this integration visually explicit, and Table 6 was restructured so that GenAI roles, risks, and governance responses are mapped directly onto the four-practice architecture. We believe these revisions address the reviewer’s concern by making the GenAI component more conceptually embedded and by strengthening the coherence between Figure 1, Section 4.2, and Section 7.

 

Comment 5: Even as a conceptual paper, the manuscript would greatly benefit from concrete examples, cases, or hypothetical scenarios demonstrating how the proposed framework would operate in practice.

Response 5: We thank the reviewer for this helpful suggestion. We agree that, even in a conceptual paper, the framework becomes easier to evaluate when readers can see how it would operate in a concrete professional-learning context.

In response, we revised the manuscript to include a more explicit illustrative example of framework enactment. The revised text now presents a health-professions module focused on emergency department overcrowding and delayed discharge, showing how learners move through sensemaking and boundary setting, co-modelling and causal representation, intervention reasoning, and meta-learning within one authentic professional task. We also clarify how GenAI may be used within this example under bounded conditions of documentation, critique, and stakeholder validation.

In addition, we strengthened the practice-oriented dimension of the manuscript by elaborating the section on authentic tasks and conditions for learning and by adding a brief illustrative example after Table 6 showing how GenAI may operate across the four practices in an applied task. We believe these revisions make the framework more concrete, more interpretable for educators, and easier to envision in practice.

[All sections of the manuscript where modifications were made based on reviewer feedback are indicated in red]

Reviewer 3 Report

Comments and Suggestions for Authors

• The paper does a good job of synthesizing the literature on systems thinking. But it's not entirely clear what the actual novel contribution is here. The authors should explicitly state what is new in their framework beyond just combining existing concepts.
• Generative AI is highlighted in the title as a core focus, but it’s missing from the proposed architecture in Figure 1. Figure 1 and the related text need to be updated to show exactly where GenAI fits into the model.
• The Methodological Approach section is missing details on data collection. The authors need to clearly explain how the literature was sourced, including the databases used and their specific selection criteria.
• Section 3 gives a useful overview of systems thinking traditions, but the link to the proposed framework is a bit weak. Please clarify exactly how these traditions inform the rest of the paper.
• A lot of the manuscript (especially the table explanations) is still very conceptual and abstract. Adding concrete examples, like a specific learning activity or curriculum implementation, would make the framework much easier to understand and apply.
• Several key terms (like sensemaking, co-modelling, intervention reasoning, and meta-learning) appear in the tables before they are defined. Adding their brief definitions in the text beforehand would really help with readability.

Comments on the Quality of English Language The manuscript is well-written, clear, and uses appropriate academic vocabulary.

Author Response

Comment 1: The paper does a good job of synthesizing the literature on systems thinking. But it's not entirely clear what the actual novel contribution is here. The authors should explicitly state what is new in their framework beyond just combining existing concepts.

Response 1: We thank the reviewer for this important observation. In response, we now state the paper’s novel contribution more explicitly. The revised manuscript clarifies that its contribution lies not only in synthesis, but in architectural integration. More specifically, the Introduction now presents four linked advances: (1) an iterative four-practice architecture for systems thinking in professional education; (2) a cross-level design model linking systems concepts to professional capabilities, learning artefacts, curriculum sequencing, and assessment criteria; (3) the integration of GenAI into the framework as a governance condition internal to the teaching and learning of systems thinking; and (4) a pathway for empirical development through propositions, measurable constructs, and candidate instruments.

These clarifications are incorporated most explicitly in the Introduction and are reinforced in the Conclusion, with the Methodological Approach section supporting the logic of framework construction. We believe these revisions make the manuscript’s novel contribution much clearer.

 

Comment 2: Generative AI is highlighted in the title as a core focus, but it’s missing from the proposed architecture in Figure 1. Figure 1 and the related text need to be updated to show exactly where GenAI fits into the model.

Response 2: We appreciate the reviewer’s comment and agree that the earlier version of Figure 1 underrepresented the role of GenAI relative to the paper’s title and conceptual framing. To address this, we revised Figure 1 and the related discussion in Section 4.2 so that GenAI is now explicitly incorporated into the model.

Specifically, GenAI is represented as a cross-cutting sociotechnical layer that operates across the four learning practices, rather than as an additional fifth practice. This revision is theoretically important because the paper does not conceptualise GenAI as a discrete stage in systems thinking, but as a mediating condition that can both support and distort learning processes across the cycle.

These revisions make the visual model more consistent with the title, the introduction, and the paper’s later discussion of GenAI-enabled learning.

 

Comment 3: The Methodological Approach section is missing details on data collection. The authors need to clearly explain how the literature was sourced, including the databases used and their specific selection criteria.

Response 3: We thank the reviewer for this important observation and apologise for the previous wording, which may have created confusion regarding the methodological status of the paper. Our intention was not to suggest that the manuscript followed a systematic review or formal data-collection protocol. Rather, this is a conceptual theory-building article supported by purposively selected literature.

In response, we revised the Methodological Approach section to make this explicit. The revised version now clarifies that relevant literature was identified primarily through Google Scholar and selected on the basis of conceptual relevance to the focal educational design problem, theoretical centrality within the field, and usefulness for framework construction. We also state directly that no formal inclusion/exclusion criteria were used, because the aim of the paper is not exhaustive evidence synthesis but conceptual integration and framework development.

We believe this revision improves transparency while also aligning the methodology section more closely with the actual genre and purpose of the manuscript.

 

Comment 4: Section 3 gives a useful overview of systems thinking traditions, but the link to the proposed framework is a bit weak. Please clarify exactly how these traditions inform the rest of the paper.

Response 4: We thank the reviewer for this helpful observation. We agree that the previous version of Section 3 described the main systems thinking traditions clearly, but did not make sufficiently explicit how those traditions were translated into the proposed framework. In response, we revised the end of Section 3 and the opening of Section 4 to clarify this link more directly.

More specifically, we now explain that system dynamics informs the framework’s emphasis on causal representation, feedback, delays, and leverage-oriented intervention reasoning; soft systems and related problem-structuring traditions inform sensemaking and boundary setting, stakeholder engagement, and the treatment of professional problems as contested rather than given; and critical systems thinking informs the framework’s focus on boundary critique, power, inclusion, and the ethical defensibility of interventions.

These revisions strengthen the theoretical throughline between Section 3 and the rest of the paper and make the derivation of the proposed framework more explicit.

 

Comment 5: A lot of the manuscript (especially the table explanations) is still very conceptual and abstract. Adding concrete examples, like a specific learning activity or curriculum implementation, would make the framework much easier to understand and apply.

Response 5: We thank the reviewer for this helpful suggestion. We agree that, although the manuscript already includes activity sequences, authentic tasks, assessment artefacts, and governance routines, much of this material was presented at a relatively abstract level, especially in relation to the tables. In response, we added a brief illustrative implementation example showing how the framework can be enacted in a concrete professional-learning context.

More specifically, the revised text now walks the reader through a sample module in which learners move through sensemaking and boundary setting, co-modelling and causal representation, intervention reasoning, and meta-learning around a complex professional case, while also showing the associated artefacts, evaluative logic, and, where relevant, GenAI-governance conditions. This addition is intended to make the framework easier to understand, apply, and translate into course design without changing the conceptual scope of the paper.

 

Comment 6: Several key terms (like sensemaking, co-modelling, intervention reasoning, and meta-learning) appear in the tables before they are defined. Adding their brief definitions in the text beforehand would really help with readability.

Response 6: We thank the reviewer for this helpful suggestion. We agree that, although the four core practices were defined in Section 4.1 before the tables, their meanings could be made more explicit for ease of reading. In response, we revised Section 4.1 so that the four practices are defined more clearly in the running text before the tabular translation of the framework.

More specifically, the revised text now clarifies that sensemaking and boundary setting refers to defining the system of interest and making assumptions, stakeholders, and boundary judgements explicit; co-modelling and causal representation refers to developing shared accounts of causal structure, feedback, and delays; intervention reasoning refers to connecting models to leverage points, scenarios, feasibility constraints, and ethical trade-offs; and meta-learning refers to reflection, revision, and uncertainty-aware judgement across the cycle.

[All sections of the manuscript where modifications were made based on reviewer feedback are indicated in red]

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

Dear Authors,

Thank you for submitting the revised version of your manuscript. I appreciate the careful attention given to the reviewers’ comments and the substantial revisions undertaken.

The manuscript has improved significantly since the initial submission. The research gap is now clearly articulated, the scope of the contribution is well delimited, and the four‑practice learning architecture is presented as the central conceptual contribution.

The paper's methodological positioning as a theory‑building contribution is now explicit and transparent, and the integration of generative AI as a cross‑cutting governance condition is conceptually coherent and well-aligned with the framework.

Importantly, the revisions go beyond textual adjustments. The manuscript now demonstrates stronger internal consistency across the Introduction, conceptual model, illustrative example, and implications for curriculum, assessment, and governance. The added example enhances interpretability without compromising the study's conceptual nature, and the tables and figures appropriately reflect the revised argument.

Only minor editorial issues remain, primarily related to typographical consistency and final copy‑editing. These do not affect the substance or validity of the contribution.

Thank you for your thoughtful engagement with the review process and for strengthening the manuscript accordingly.

Author Response

Comment 1: Thank you for submitting the revised version of your manuscript. I appreciate the careful attention given to the reviewers’ comments and the substantial revisions undertaken.

The manuscript has improved significantly since the initial submission. The research gap is now clearly articulated, the scope of the contribution is well delimited, and the four‑practice learning architecture is presented as the central conceptual contribution.

The paper's methodological positioning as a theory‑building contribution is now explicit and transparent, and the integration of generative AI as a cross‑cutting governance condition is conceptually coherent and well-aligned with the framework.

Importantly, the revisions go beyond textual adjustments. The manuscript now demonstrates stronger internal consistency across the Introduction, conceptual model, illustrative example, and implications for curriculum, assessment, and governance. The added example enhances interpretability without compromising the study's conceptual nature, and the tables and figures appropriately reflect the revised argument.

Only minor editorial issues remain, primarily related to typographical consistency and final copy‑editing. These do not affect the substance or validity of the contribution.

Thank you for your thoughtful engagement with the review process and for strengthening the manuscript accordingly.

Response 1: We sincerely thank Reviewer for this positive assessment of our manuscript and for noting that only minor editorial issues remained. In response, we carried out a careful final editorial revision of the manuscript, focusing on typographical consistency and copyediting. This final check included the review of minor wording and formatting details, the standardization of internal stylistic choices where needed, and the correction of small inconsistencies in the reference list and related presentation details. These revisions are editorial in nature and do not affect the substance, argument, or validity of the contribution.

Reviewer 3 Report

Comments and Suggestions for Authors

• The revised manuscript shows improvements reflecting comments in the first round. The framework's contribution stands out clearly now in the abstract and conclusion.
• The approach used to do the research is clear and makes sense for a study that builds concepts and theory. The way the method is explained is good.
• It would be helpful to give a more idea about how broad the literature search was (e.g., indicative search terms or scope).

Comments on the Quality of English Language

The manuscript is well-written, clear, and uses appropriate academic vocabulary.

Author Response

Comment 1: The revised manuscript shows improvements reflecting comments in the first round. The framework's contribution stands out clearly now in the abstract and conclusion.

The approach used to do the research is clear and makes sense for a study that builds concepts and theory. The way the method is explained is good.

It would be helpful to give a more idea about how broad the literature search was (e.g., indicative search terms or scope).

Response 1: We sincerely thank Reviewer for this helpful suggestion and for the positive assessment of the revised manuscript. In response, we have expanded the methodological section to provide a clearer indication of the breadth and thematic scope of the literature search. Specifically, we now include indicative examples of the search terms used across the four bodies of scholarship informing the framework, while clarifying that the search was broad and conceptually guided rather than exhaustive or systematic. This addition is intended to make the scope of the literature identification process more transparent while remaining fully consistent with the conceptual and theory-building nature of the study.