Hot Knowledge, Cool Action? Designing Digital Knowledge Platforms in the Age of Artificial Intelligence

1. Introduction: Knowledge Platforms and the Limits of Information Infrastructure

Over the past two decades, digital knowledge platforms have become ubiquitous within the sustainable development ecosystem, absorbing substantial public and donor investment as core instruments for evidence-based policymaking, capacity building, and knowledge democratization. Across climate governance, circular economy, sustainable consumption and production, and green growth, platforms are routinely expected to bridge gaps between research, policy, and practice by aggregating, curating, and disseminating information at scale. Conservative estimates suggest the existence of several hundred sustainability-oriented knowledge platforms globally, although precise quantification remains elusive due to definitional ambiguities and the fluid nature of digital infrastructures. ¹

Beyond their instrumental role, however, these platforms increasingly function as media environments in the sense articulated by McLuhan and later elaborated by Postman. In other words, not neutral tools, but environments that reorganize attention, authority, and participation by shaping how knowledge is encountered, interpreted, and acted upon. As media environments, platforms do not simply host information; they structure what is visible, what is credible, and who is positioned as a legitimate knower within sustainability discourse.

Despite this proliferation, concerns regarding the actual effectiveness of knowledge platforms have grown steadily within both academic and practitioner communities. A recurring critique is that many platforms function primarily as static repositories; well-organized archives of reports, tools, and case studies, rather than as living systems that actively support learning, interpretation, and action. This gap between institutional ambition and operational impact raises a foundational question: why do platforms designed to democratize knowledge so often struggle to catalyse engagement, collaboration, or behavioural change?

Empirical evidence from a series of structured knowledge exchanges co-organised by the EU-funded SWITCH-Asia programme and the Green Growth Knowledge Partnership (GGKP) between October 2024 and May 2025 provides insight into this question (EU SWITCH-Asia & GGKP, 2025). The initiative engaged over 450 stakeholders, including policymakers, business leaders, researchers, platform managers, and civil society actors from Asia-Pacific, Africa, Europe, and the Americas. Survey results revealed persistent structural weaknesses: 62 percent of respondents identified lack of inter-platform connectivity as the most significant operational challenge, while 48 percent cited insufficient understanding of target audience needs as the primary barrier to effective knowledge sharing. These findings mirror patterns identified in the broader knowledge management literature, where fragmentation, duplication, and low user engagement continue to undermine platform effectiveness.

However, these challenges point beyond technical shortcomings or coordination failures. They suggest a deeper issue concerning how digital platforms conceptualize knowledge itself and how users are positioned in relation to it. Most existing analyses focus on improving content quality, metadata standards, interoperability, or governance arrangements. While necessary, such interventions often overlook the media effects of platform architecture itself; namely, its effects on attention, trust, interpretive agency, and perceived legitimacy. In this sense, platform design operates as a form of communication that shapes user behaviour and meaning-making independently of content quality.

This article advances a different argument: that many sustainability knowledge platforms fail because their underlying media architectures implicitly frame knowledge as something to be transmitted rather than co-produced. To interrogate this problem, the study applies Marshall McLuhan’s media ecology framework, particularly his assertion that “the medium is the message” and his distinction between “hot” and “cool” media. From this perspective, platform architecture is not neutral; it conditions the possibilities of engagement by scripting roles for users before any content is accessed. Platforms designed as “hot” media (characterised by high-definition, polished information requiring minimal user participation) tend to position users as passive consumers. In doing so, they constrain the dialogic and participatory processes necessary for translating knowledge into situated action.

This analysis adopts a non-determinist perspective, treating platform architectures as conditioning rather than determining engagement, an approach elaborated in Section 2. Even recent advances in AI-enabled knowledge infrastructures have struggled to close the persistent knowledge-action gap, raising questions about whether the limitation lies in technological capability or in the underlying design logic of platforms themselves. From a media ecology perspective, many contemporary knowledge platforms continue to operate as hot media, privileging authoritative transmission and completeness over interpretive agency and participation.

While platforms frequently promise global connectivity and shared understanding, their architectures often reproduce asymmetries of visibility and authority. Questions of representation, epistemic inclusion, and participation are therefore central to platform design and are examined in detail in the theoretical discussion that follows.

By bringing media ecology into dialogue with empirical evidence from contemporary sustainability platforms, this research reframes knowledge platforms as inherently pedagogical and epistemological systems. It sets the stage for a critical examination of how platform design choices, particularly in the context of emerging AI-enabled knowledge infrastructures, can either reinforce passive consumption or support participatory, “cool” environments that invite interpretation, co-creation, and collective learning.

1.1 Hot Repositories and the Persistence of the Knowledge-Action Gap

The challenge transcends technical limitations, encompassing fundamental epistemological questions regarding knowledge creation, validation, dissemination, and application. The knowledge-action gap (the persistent disconnect between available information and implemented practice) has been extensively documented in sustainable development literature (Cash et al., 2003; van Kerkhoff & Lebel, 2006; Reed et al., 2014). From a media ecology perspective, this gap can be understood not only as a failure of information transfer, but as an effect of how knowledge is mediated, structured, and rendered actionable within specific media environments.

Contemporary knowledge platforms, despite increasing reliance on artificial intelligence tools such as semantic search, automated classification, and recommendation systems, continue to struggle to support meaningful engagement and action. This recurring failure challenges assumptions of technological solutionism and suggests that optimizing information delivery alone is insufficient. If increasingly sophisticated technologies do not close the knowledge-action gap, the problem may lie less in computational capacity than in how platforms are designed to position users and structure participation.

McLuhan’s distinction between hot and cool media offers a complementary explanation. Platforms dominated by comprehensive reports, polished dashboards, and AI-curated content overwhelmingly function as hot media, positioning users as passive recipients rather than active participants. Such architectures privilege authoritative transmission over interpretive agency, shaping engagement patterns that undermine the dialogic processes necessary for knowledge to become meaningful in practice.

1.2 Toward Cool Knowledge Commons: Scope and Contribution

This investigation proposes that effective knowledge platforms must be reconceived as cool media environments, namely, systems designed with strategic incompleteness that invites user participation, co-creation, and contextual interpretation. Within McLuhan’s media ecology, such “coolness” is not a stylistic choice but a structural condition that redistributes interpretive labor from the medium to the user, thereby reshaping power relations in knowledge production. Drawing on McLuhan’s media ecology framework, supplemented by Neil Postman’s critical extensions, Raymond Williams’ attention to political economy, and Douglas Engelbart’s vision of cognitive augmentation, the research develops both theoretical understanding and practical design principles for participatory knowledge infrastructures. In doing so, it explicitly rejects technologically determinist accounts by situating platform design within institutional, economic, and communicative contexts that mediate how technologies are adopted and used.

The research recognizes that knowledge platforms are inherently pedagogical environments. They structure relationships between knowledge and knowers, model particular epistemologies, and shape how participants learn to engage with information. From a cultural studies perspective, platforms do not merely transmit knowledge but encode preferred ways of seeing, valuing, and interpreting sustainability challenges, while simultaneously constraining alternative readings. McLuhan’s critique of traditional education as “intellectual penal institutions” delivering packaged content to passive recipients applies with equal force to knowledge platforms designed as hot media repositories. His alternative vision, articulated in City as Classroom (1977), emphasized discovery learning through probing questions and active engagement with environments. For NGO communications and sustainable development practice, fields that are inherently activist and community-based, platforms must model curiosity, invite participation, and build shared understanding through dialogue rather than transmission.

Read through this lens, knowledge platforms become sites where epistemic authority is either reproduced or contested; where some forms of knowledge are rendered visible and legitimate, while others risk marginalization through architectural design choices. This insight aligns with critical media scholarship emphasizing that representation and participation are structured not only by content but by the formal conditions of visibility and voice embedded in media systems (Said, 1978; Mulvey, 1975). 

The article is organized to guide the reader from theory to empirical analysis and applied implications. To begin, Section 2 establishes the theoretical foundations by positioning McLuhan’s media ecology framework as the primary analytical lens, complemented by cognitive augmentation theory and critical perspectives on artificial intelligence. Section 3 then outlines the mixed-methods research design underpinning the stakeholder consultation initiative. Building on this methodological basis, Section 4 presents the empirical findings, which are interpreted through the hot–cool media framework. Subsequently, Section 5 examines the development of the GGKP taxonomy and API as illustrative case studies of platform architecture design. Drawing these strands together, Section 6 articulates a set of design principles for “cool” knowledge platforms derived from both theoretical analysis and empirical validation. Section 7 translates these principles into practical recommendations for diverse stakeholder communities. Finally, Section 8 concludes by discussing the implications for sustainable development practice and identifying directions for future research.

2. Theoretical Foundations

This investigation draws on three complementary theoretical traditions: artificial intelligence research, which clarifies current technological capabilities and limitations; cognitive augmentation theory, which provides design principles for human-machine collaboration; and media ecology, which sheds light on how platform architecture shapes user engagement independent of content or technological sophistication. Taken together, these traditions enable an analysis of digital knowledge platforms not merely as technical systems but as sociotechnical media environments whose effects emerge from the interaction of technology, institutional context, and communicative form. Furthermore, these frameworks explain why knowledge platforms fail despite substantial investment and suggest pathways toward more effective design.

2.1 Artificial Intelligence: Current Capabilities and Future Speculation

The demarcation between artificial narrow intelligence and artificial general intelligence represents a critical distinction frequently obscured in development discourse. Current AI systems demonstrate superior performance in circumscribed domains: image recognition algorithms achieving superhuman accuracy, natural language processing systems generating coherent text, and game-playing agents defeating world champions (Silver et al., 2018; Brown et al., 2020). However, these achievements occur within bounded problem spaces with clearly defined parameters and success metrics. From a media ecology perspective, such systems excel at operating within predefined representational frames but do not themselves interrogate or transform the epistemic assumptions embedded in those frames.

Within knowledge platform contexts, AI technologies manifest through specific functional implementations:

• Natural Language Processing (NLP): Extraction of entities, topics, and semantic relationships from unstructured text using transformer-based architectures and word embedding techniques (Devlin et al., 2019);
• Semantic Search Capabilities: Implementation of concept-based retrieval systems transcending keyword matching through vector space representations and similarity metrics (Reimers & Gurevych, 2019);
• Recommendation Algorithms: Deployment of collaborative and content-based filtering techniques to suggest relevant resources based on user behavior patterns and content characteristics;
• Knowledge Graph Construction: Development of structured representations capturing relationships between concepts, documents, entities, and metadata using graph database technologies (Hogan et al., 2021);
• Automated Classification Systems: As exemplified by the GGKP API’s classifier, which processes knowledge products to generate titles, descriptions, summaries, and taxonomy-based categorizations through supervised learning approaches.

These technological implementations, while sophisticated, operate within constrained parameters. They function primarily as instruments of optimization and efficiency, reinforcing what Neil Postman described as a tendency toward technological solutionism, wherein complex social and epistemic challenges are reframed as problems of technical refinement. The knowledge exchanges conducted with multi-sector stakeholders revealed practical limitations that challenge prevailing assumptions about AI requirements and capabilities. Participants observed that AI tools demonstrate effectiveness even with limited datasets when supported by human expert curation, contradicting widespread beliefs about big data prerequisites for AI functionality. Simultaneously, in the context of the circular economy knowledge taxonomy development, concerns regarding “circular washing” – the misrepresentation or exaggeration of circular economy credentials – highlighted the necessity for human expertise in claim verification, as AI systems lack the contextual understanding to assess credibility comprehensively.

2.2 Artificial General Intelligence (AGI): Theoretical Constructs and Temporal Uncertainties

Artificial general intelligence, defined as machine intelligence equalling or exceeding human cognitive capabilities across all domains, remains a theoretical construct with substantial uncertainty regarding feasibility and timeline. Expert surveys reveal divergent predictions, with median estimates ranging from 2040 to 2100, though significant minorities consider AGI either imminent or impossible (Müller & Bostrom, 2016; Grace et al., 2018). The speculative status of AGI reinforces the importance of resisting determinist narratives that attribute future epistemic transformation to technological inevitability rather than to design choices, governance structures, and social values. This temporal uncertainty necessitates pragmatic focus on optimizing current technologies while establishing flexible frameworks capable of accommodating potential future developments.

2.3 Cognitive Augmentation Theory and Implementation – Historical Foundations and Contemporary Relevance

Douglas Engelbart’s conceptualization of intelligence augmentation, articulated in his seminal 1962 report “Augmenting Human Intellect: A Conceptual Framework,” provides enduring theoretical guidance for human-computer interaction design (Engelbart, 1962). Engelbart envisioned computational systems that would amplify rather than replace human cognitive capabilities, enabling collaborative problem-solving for increasingly complex challenges. His vision stands in sharp contrast to automation-centric imaginaries, aligning instead with McLuhan’s emphasis on media that extend human perception and participation rather than displacing them. This theoretical framework maintains remarkable relevance for contemporary knowledge platform design.

The GGKP-SWITCH-Asia collaboration exemplifies practical implementation of augmentation principles through their development of tools designed to enhance rather than substitute human judgment. The ‘Common Circular Economy Knowledge Taxonomy’, developed through iterative consultation with leading knowledge platforms globally, demonstrates how standardized frameworks can facilitate knowledge organization while preserving flexibility for contextual adaptation. In this sense, the taxonomy functions as an enabling medium rather than a closed classificatory system, inviting human interpretation and revision rather than enforcing epistemic closure. Moreover, the taxonomy’s architecture, incorporating dynamic expansion capabilities and custom domain provisions, acknowledges the necessity of accommodating diverse epistemological frameworks while maintaining operational coherence.

2.4 Empirical Validation Through Stakeholder Engagement

Contemporary research in human-AI collaboration provides empirical support for augmentation approaches. Studies demonstrate that human-AI teams consistently outperform either humans or AI systems operating independently in complex knowledge work (Brynjolfsson & McAfee, 2017; Daugherty & Wilson, 2018). These findings reinforce Engelbart’s original insight that intelligence emerges relationally, through interaction between humans, tools, and environments. The findings from the series of knowledge exchanges reinforce this perspective, with participants emphasizing that knowledge hubs must evolve beyond information repositories to become active intermediaries facilitating evidence synthesis, partnership brokerage, and metric standardization, functions requiring sophisticated human-machine collaboration rather than full automation.

2.5 The Medium is the Message: Platform Architecture as Communication

While AI capabilities and cognitive augmentation principles address what platforms can do, media ecology theory addresses how platform design shapes what users experience and how they engage. Marshall McLuhan’s (1964) foundational insight that “the medium is the message” directs attention to the structural characteristics of communication technologies themselves. McLuhan argued that media forms shape social outcomes independent of content: “The medium is the message because it is the medium that shapes and controls the scale and form of human association and action.” For knowledge platforms, this means that interface design, information hierarchies, and interaction features communicate values and shape engagement possibilities before any content is accessed. In this sense, platform architecture functions as a form of implicit pedagogy, encoding assumptions about authority, participation, and the role of users in knowledge production.

McLuhan’s distinction between hot and cool media provides a valuable lens for analyzing engagement dynamics. Hot media, characterized by high-definition content and low audience participation, correspond closely to how many platforms currently operate. Comprehensive reports, polished dashboards, and structured datasets position users primarily as passive consumers of information. Such designs privilege visual space that is linear, hierarchical, and completion-oriented, over relational or dialogic modes of engagement. The abundance of such “hot” outputs fosters one-way communication flows; top-down dissemination that leaves little room for feedback, reinterpretation, or collaborative meaning-making. In practice, this reduces platforms to static repositories that showcase information rather than enabling its transformation into action.

By contrast, cool media invite users into an active role, supplying incomplete or open-ended content that requires participation to generate meaning. Discussion forums, collaborative wikis, and interactive simulations embody this coolness by demanding user input, exchange, and interpretation. In the context of digital knowledge platforms, cool media affordances open space for co-creation, deliberation, and adaptive learning. The difference is not merely technical but epistemic: where hot media reinforce linear transmission and accountability-driven communication, cool media cultivate relational engagement and knowledge-in-use. From a cultural studies perspective, this shift also redistributes interpretive agency, enabling users to negotiate or rearticulate meaning rather than merely decoding preferred readings embedded in platform design.

Insights from the knowledge exchange sessions reinforce this interpretation. Participants consistently emphasized the need to transform platforms from static libraries into dynamic service environments. Their calls for practical case study curation, deployment of AI-driven content tools, and facilitation of cross-sector partnerships highlight a clear preference for cool media characteristics. Notably, these demands reflect a desire for platforms to act as mediating environments rather than authoritative broadcasters. What stakeholders seek is not only access to polished outputs but also opportunities to shape, contextualize, and apply knowledge collaboratively. This demand for interactivity and participation illustrates the limits of hot-media-dominated designs.

2.6  Epistemic Pluralism in Digital Systems

The collaborative development of a ‘Common Circular Economy Knowledge Taxonomy’ illustrates practical approaches to epistemological challenges in global knowledge systems. By enabling custom domains for partners with unique terminological requirements while maintaining global structural coherence, the taxonomy acknowledges that knowledge systems must accommodate multiple ways of knowing, a concept well-established in science and technology studies literature (Haraway, 1988; Harding, 1998; Ludwig & El-Hani, 2020). This epistemic pluralism aligns with cool media principles: rather than delivering authoritative, complete classification systems, the taxonomy creates structured incompleteness that invites contextual completion by diverse knowledge communities.

2.7 Integrating Perspectives: Technology, Augmentation, and Medium

These three theoretical frameworks are complementary rather than competing. AI research clarifies what current technologies can and cannot accomplish; cognitive augmentation theory provides principles for designing human-machine collaboration; media ecology explains how platform architecture shapes engagement regardless of technological sophistication. A platform may deploy state-of-the-art AI and embody augmentation principles yet still fail if its hot media design positions users as passive recipients. Conversely, cool media architecture alone cannot overcome fundamental technological limitations or poorly conceived human-machine workflows. Effective knowledge platform design requires attending to all three dimensions simultaneously; in other words, leveraging AI capabilities, supporting human cognitive augmentation, and structuring participatory engagement through appropriate media design. Only through this integrative approach can platforms evolve from information delivery systems into environments that actively support collective intelligence and action.

3. Methodology

3.1 Research Design and Data Collection

The Circular Economy Knowledge Hubs Webinar Series employed a structured mixed-methods approach to gather empirical evidence regarding platform effectiveness and user requirements. The series comprised four distinct events, each with specific objectives and methodological approaches:

Workshop 1 (16 October 2024): An invitation-only workshop engaging 80 participants from selected circular economy platforms. Representatives included stakeholders from the Central Asia Climate and Information Portal (CACIP), European Circular Economy Stakeholder Platform (ECESP), ASEAN Circular Economy Stakeholder Platform (ACESP), Circular Economy Coalition for Latin America and the Caribbean (CEC LAC), and African Circular Economy Network (ACEN). This workshop employed interactive polling and facilitated discussion to identify operational gaps and challenges.

Webinar 2 (28 November 2024): A public webinar attracting 114 participants, structured around two panel discussions examining regional perspectives and user needs. Panel composition included platform administrators, policymakers, and business representatives providing diverse sectoral insights.

Webinar 3 (27 February 2025): Engaging 128 participants in examining knowledge hub potential for facilitating circular economy transitions. Discussion focused on strengthening policy implementation through enhanced knowledge management and cross-sector collaboration.

Webinar 4 (6 May 2025): The series culmination, involving 180 participants exploring artificial intelligence applications in knowledge exchange and accessibility. This session provided critical insights regarding AI capabilities and limitations in knowledge management contexts.

The progressive increase in participation across the series (from 80 to 180 participants) reflects growing interest in knowledge platform challenges and suggests that the topics addressed resonated with practitioner communities. The total of 502 participant engagements across all four events, representing 450 unique individuals, provided a substantial empirical base for analysis.

3.2 Data Collection Instruments and Procedures

A combination of methods was used to capture stakeholder perspectives during the knowledge exchange sessions. Real-time digital polling provided quantitative insights into participants’ perceptions of platform gaps and challenges. The use of forced-choice questions with predetermined response categories allowed results to be statistically aggregated, though this approach inevitably limited the ability to capture more nuanced perspectives. Complementing these polls, structured panel discussions brought together regional platform managers, policymakers, and practitioners to share qualitative insights on implementation challenges and opportunities. The panels were deliberately designed to ensure geographic diversity, sectoral representation, and the inclusion of participants with practical operational experience. Finally, open dialogue sessions offered a space for participant-driven inquiry and collaborative problem-solving. Guided by professional moderators, these sessions encouraged balanced participation while maintaining thematic coherence, enabling the co-creation of solutions and the articulation of shared priorities.

3.3 Analytical Framework

3.3.1 Quantitative Analysis

The polling data were analyzed using descriptive statistics to identify central tendencies and distribution patterns. The results revealed several recurring concerns among participants. A majority, 62 percent of respondents in Workshop 1 (n=80), identified connectivity gaps as a major barrier to effective platform use. Nearly half of respondents pointed to challenges in audience understanding (48 percent) and expressed a demand for sectoral customization (46 percent). A similar proportion (45 percent) indicated a preference for lifecycle-based organization of content in the context of circular economy knowledge. While these figures provide a useful snapshot of stakeholder priorities, they should be interpreted with caution. The sample consisted of engaged stakeholders rather than a randomly selected population, introducing the possibility of self-selection bias. Moreover, the reliance on forced-choice questions may have constrained the expression of more nuanced views. The quantitative findings are therefore treated as indicative of patterns among active platform users and operators rather than representative of all potential stakeholders.

3.3.2 Qualitative Analysis

Thematic analysis of the panel discussions and open dialogue transcripts was conducted using an inductive coding approach to capture recurring patterns and emergent themes (Braun & Clarke, 2006). The process unfolded in several stages, beginning with familiarization with the transcript data, followed by the generation of preliminary codes. These codes were then refined into broader themes, which were systematically reviewed and validated against the raw data to ensure consistency. The final stage involved defining and naming the themes to reflect their core meaning. Through this iterative process, three overarching themes emerged across all sessions: persistent challenges of knowledge fragmentation, a strong demand for more active intermediation by platforms, and the recognized necessity of inclusive approaches to ensure broad stakeholder engagement. These themes proved consistent across geographic regions and stakeholder types, suggesting they reflect systemic rather than context-specific challenges.

3.4 Methodological Limitations

A series of limitations in this study should be acknowledged. Participant recruitment relied primarily on existing knowledge platform networks, which may have introduced selection bias by favoring stakeholders already familiar with or engaged in digital technologies, while potentially underrepresenting marginalized perspectives. Moreover, while knowledge platform implementers and users were selected as panelists for the knowledge exchange sessions, this did not necessarily mean they were experts in platform management, communications, information technology, or software development. Their perspectives were, therefore, valuable for capturing user experiences and challenges, but they did not always reflect the full range of technical or managerial expertise required for platform design and operationalization. Furthermore, the predominance of English-language facilitation may have constrained the depth of participation from non-Anglophone regions. 

4. Empirical Findings and Analysis

4.1 Knowledge Fragmentation: Structural and Operational Dimensions

Findings from the knowledge exchanges made it clear that knowledge fragmentation continues to pose a critical operational barrier, even after substantial investments in digital infrastructure. Notably, 62 percent of stakeholders identified inter-platform connectivity deficits as the most significant gap, a result that reinforces patterns already documented in the knowledge management literature (Fazey et al., 2013; Cornell et al., 2013). Participants explained that valuable information remains dispersed across disconnected repositories, constrained by institutional silos, technological incompatibilities, and inconsistent data standards. From a media ecology perspective, this fragmentation can be understood as an effect of multiple, non-communicating media environments rather than a simple absence of information or tools. 

This fragmentation is not limited to a single aspect of platform design but manifests across multiple dimensions. At the technical level, challenges arise from incompatible data formats, proprietary systems, and the absence of standardized application programming interfaces. Semantic fragmentation adds another layer, as divergent terminologies, classification schemes, and metadata standards undermine the possibility of effective cross-platform search and retrieval. Institutional fragmentation further complicates matters, with organizational boundaries, competitive dynamics, and donor-driven funding structures discouraging collaboration and information sharing. These institutional dynamics resonate with Raymond Williams’ insight that media systems are shaped not only by technical affordances but by political-economic arrangements that condition how technologies are deployed and governed. Finally, epistemological fragmentation reflects differences in knowledge validation criteria, quality standards, and evidentiary requirements, which make integration across platforms difficult. Fragmentation is simultaneously technical, institutional, and epistemic, reinforcing itself across levels.

These patterns of fragmentation help explain the persistent knowledge-action gap: when practitioners must navigate multiple disconnected platforms with incompatible classification systems, the cognitive burden of locating and synthesizing relevant information often exceeds available time and resources. The resulting overload mirrors what Postman described as a shift from information scarcity to meaning scarcity, where the abundance of unintegrated information undermines its practical usability. The result is that even high-quality knowledge remains underutilized.

4.2 The GGKP API Response

The Green Growth Knowledge Partnership’s development of the AI Knowledge Brokerage API represents a significant technical achievement addressing connectivity challenges. The system, now operational with full documentation available at docs.api.ggkp.org, provides comprehensive endpoints for knowledge management, automated content classification, and cross-platform integration. Viewed through McLuhan’s lenses, the API can be interpreted as an infrastructural intervention aimed at reconfiguring the medium through which knowledge circulates, rather than merely increasing the volume of content available. 

The API architecture incorporates:

• Centralized multi-domain taxonomy management: A harmonized green and circular economy knowledge taxonomy, accessible at taxonomy.ggkp.org, enables consistent categorization and retrieval of resources aligned with globally recognized frameworks;
• Automated content classification using natural language processing: The system processes knowledge products to generate titles, descriptions, summaries, and taxonomy-based categorizations through machine learning approaches;
• Intelligent search capabilities: An AI-powered search interface at search.ggkp.org, utilizing a custom Mistral-3 8B model optimized for low energy consumption, transcends keyword matching to help practitioners quickly locate actionable insights;
• Federated query distribution across partner platforms: The API enables automated knowledge sharing and real-time synchronization, connecting GGKP’s knowledge libraries with other knowledge hubs;
• AI-powered knowledge brokerage: Intelligent conversational agents serve as virtual librarians, guiding users through the knowledge ecosystem, answering questions about green growth policies and practices, and providing personalized resource recommendations.

This technical infrastructure directly addresses the connectivity gap identified as most significant by workshop participants. Partner platforms can now integrate GGKP’s knowledge resources directly into their own systems, enabling the kind of seamless information flow that stakeholders identified as essential yet largely absent from current practice. At the same time, the API illustrates the limits of purely technical solutions. While it mitigates fragmentation at the infrastructural level, its effectiveness ultimately depends on whether connected platforms adopt participatory, “cool” media logics that support interpretation, contextualization, and use, rather than merely accelerating the circulation of already “hot” content.

4.3 Evolution from Repositories to Intermediaries – Three Emergent Themes

Analysis of the webinar discussions and exchanges revealed three consistent themes regarding the future evolution of knowledge platforms. The first concerns the need for active intermediation. Stakeholders repeatedly stressed that platforms should serve not simply as repositories of information, but as intermediaries capable of translating research into actionable insights. This requires curated content, relevant performance indicators, and decision-support tools that go beyond static provision of data. From a media ecology perspective, this shift reflects a movement away from “hot” broadcast-oriented architectures toward “cooler” environments that support dialogue, interpretation, and situated use. Such expectations resonate with the boundary organization literature, which highlights the critical role of active knowledge brokerage in linking science, policy, and practice (Guston, 2001; Parker & Crona, 2012). 

A second theme centered on the imperatives of interoperability and standardization. Participants identified persistent bottlenecks in data quality and technical integration that continue to constrain platform effectiveness. Key challenges included the absence of standardized protocols for material flow data, incompatibilities among taxonomic systems across regions, limited cross-border comparability of datasets, inconsistent quality assurance mechanisms, and insufficient communications and knowledge management expertise. These challenges point not only to technical misalignment but to the absence of shared media conventions that would allow platforms to “speak” to one another in meaningful ways. These findings echo earlier research documenting the difficulties of achieving standardization in environmental data systems (Michener & Jones, 2012; Wilkinson et al., 2016).

Finally, participants emphasized inclusivity as a precondition for effectiveness. In the context of the circular economy, they argued that knowledge platforms can only drive meaningful transitions if they engage small and medium-sized enterprises, informal sector workers, and marginalized communities. This perspective aligns with justice-oriented sustainability frameworks that emphasize both procedural and distributive equity (Agyeman et al., 2016; Bennett et al., 2019). Read through critical media theory, these concerns foreground questions of voice, visibility, and representational power: whose knowledge is invited, whose is validated, and whose remains peripheral within platform architectures. Importantly, participants argued that inclusivity must be embedded at multiple levels; not only in the design of platforms, shaping their architecture, accessibility, and governance, but also in the processes of content submission, so that diverse voices, experiences, and knowledge contributions are fully represented and valued.

4.4 Implications for Platform Design

These themes suggest the need for a reconceptualization of platform architecture and governance. Rather than functioning as passive repositories, effective platforms must be designed to actively facilitate knowledge synthesis, quality assurance, and stakeholder engagement. Achieving this transformation requires not only robust technical infrastructure and institutional capacity for curation, validation, and facilitation, but also dedicated communications expertise and strategy. Clear communication is essential to defining the platform’s purpose, aligning it with user needs, and ensuring that its design supports meaningful interaction rather than one-way dissemination. Such communicative work can be understood as a form of mediation that actively structures how knowledge circulates and is made actionable. In addition, communications expertise brings in mechanisms for monitoring and evaluating whether the platform is meeting its objectives, making results measurable and progress trackable over time. By embedding the latter into both the design and implementation phases, platforms can evolve into results-oriented systems that are not only technically sound but also purposeful, user-centered, and adaptive to changing contexts.

The stakeholder preference for active intermediation over passive storage has implications beyond technical design. It suggests that the dominant model of knowledge platforms may itself be part of the problem. What is being challenged here is not the absence of information, but the media logic through which information is organized and encountered. What practitioners seek is not more polished content but more opportunities to engage with knowledge: to question, contextualize, adapt, and apply it within their specific circumstances.

4.5 Artificial Intelligence Implementation: Empirical Insights

The webinar on Harnessing AI to Scale the Impact of Circular Economy Knowledge generated a set of empirical insights that challenged several prevailing assumptions about artificial intelligence. One of the most striking findings concerned the relationship between data volume and system effectiveness. Contrary to the dominant big-data paradigm, participants reported that successful AI applications were achieved with relatively modest datasets, provided these were paired with expert curation. This suggests that in specialized contexts, domain expertise and data quality may offset the need for vast amounts of data, a view consistent with recent advances in few-shot and transfer learning (Wang et al., 2020). 

A second finding related to the advantages of domain-specific approaches. Participants highlighted how specialized taxonomies and controlled vocabularies can enhance AI performance in knowledge classification tasks. The experience of the ‘Common Circular Economy Knowledge Taxonomy’ illustrated how tailoring classification structures to sectoral needs significantly improves accuracy compared to reliance on general-purpose systems. From a media ecology perspective, such taxonomies function as structuring environments that shape what can be seen, retrieved, and connected within the platform.

The discussions also reaffirmed the continued importance of human oversight. Participants stressed that, while AI can accelerate classification and retrieval, human judgment remains indispensable for ensuring contextual appropriateness, ethical compliance, and quality assurance. In the context of the circular economy, concerns about “circular washing” exemplify situations where algorithmic outputs alone are insufficient, and credibility depends on human evaluation.

Finally, the webinar highlighted transparency and explainability as critical requirements for stakeholder trust. Participants expressed strong preferences for AI systems that allow users to understand both the logic of classification and the rationale for recommendations. This emphasis aligns with the growing recognition in the literature that explainability is essential in high-stakes decision contexts (Arrieta et al., 2020).

4.6 Balancing Automation and Participation

The stakeholder insights on AI implementation reveal a nuanced understanding of technology’s role in knowledge systems. Participants did not reject AI indeed, they expressed enthusiasm for its potential to reduce search costs, improve classification consistency, and enable cross-platform integration. However, they consistently framed AI as a tool for augmenting human capabilities rather than replacing human judgment.

This perspective has design implications. AI systems that deliver complete, authoritative outputs without revealing their reasoning may undermine user engagement by positioning practitioners as passive recipients of algorithmic determinations. By contrast, AI systems that generate provisional suggestions, expose their confidence levels, and invite human validation can support more active user engagement with knowledge. The GGKP API’s architecture, combining automated classification with human review workflows and transparent quality assurance mechanisms, reflects this augmentation-oriented approach.

The finding that AI effectiveness depends more on expert curation than data volume also has strategic implications for resource-constrained organizations. Rather than pursuing comprehensive data collection as a prerequisite for AI deployment, platforms may achieve greater impact by investing in domain expertise and carefully curated training datasets. This approach democratizes AI capabilities, making them accessible to organizations that lack the resources for large-scale data infrastructure.

5. Case Analysis: Developing a Common Knowledge Taxonomy on the Circular Economy  

5.1 Process, Features and Innovation 

The development of the ‘Common Circular Economy Knowledge Taxonomy’ was the result of a structured collaboration between the GGKP and the EU SWITCH-Asia programme, enriched by contributions from a range of global knowledge platforms. The process unfolded through a series of iterative consultation cycles designed to balance methodological rigor with responsiveness to stakeholder needs. It began with the synthesis of existing classification systems, through which the core team identified both commonalities and divergences across current approaches. This preliminary framework was then subjected to regional consultations, where knowledge platforms from different contexts provided feedback on its applicability and highlighted specific requirements relevant to their local realities.

Building on this input, the framework underwent several rounds of refinement, ensuring that stakeholder perspectives were integrated while preserving structural coherence. To further test its robustness, the taxonomy was piloted across a variety of content types, allowing the team to assess its effectiveness in real-world application. Importantly, the framework was not conceived as static. Such openness aligns with cool media principles, where meaning is co-produced through participation rather than delivered as a finished product. Mechanisms were incorporated to enable continuous adaptation through ongoing community input and engagement. This participatory methodology reflects best practices in collaborative ontology development, as emphasized in the knowledge engineering literature (Suárez-Figueroa et al., 2012), ensuring that the taxonomy remains both scientifically sound and practically relevant.

The architecture of the taxonomy integrates several innovative features designed to address the challenges raised during stakeholder consultations. At its core, it employs a hierarchical structure with flexible depth, allowing for both broad, high-level categorization and more granular classification depending on the needs of different users. This flexibility mitigates the risk of technological determinism by acknowledging that classification systems do not simply reflect reality but actively shape how reality is interpreted and acted upon. Recognizing the tension between global standardization and local relevance, the framework also provides for custom domain provisions, in other words, partner organizations are able to develop specialized sub-taxonomies while maintaining overall compatibility with the global system. To enhance accessibility, the design incorporates a multilingual support architecture that enables terminology translation and cultural adaptation, although full implementation across languages is still in progress. Finally, a system of version control and evolution tracking ensures that taxonomic changes are systematically documented, allowing for longitudinal analysis and preserving backward compatibility as the taxonomy continues to evolve.

5.2 GGKP API Implementation: Technical Architecture and Functionality

The GGKP API represents a comprehensive attempt to address platform interoperability through technical standardization. Key architectural components include:

Automated Classification Engine: Utilizing natural language processing techniques, the system analyzes knowledge products to generate:

• Descriptive titles optimized for search discovery
• Executive summaries capturing key insights
• Taxonomic classifications based on content analysis
• Metadata extraction for bibliographic purposes

Federated Search Infrastructure: The API enables distributed queries across partner platforms through:

• Query translation and optimization for different platform architectures
• Result aggregation and relevance ranking
• Duplicate detection and consolidation
• Source attribution and provenance tracking

Quality Assurance Mechanisms: Incorporating both automated and manual validation:

• Automated consistency checking for classification accuracy
• Human review workflows for high-stakes content
• Citation verification to address “circular washing” concerns
• Periodic audit processes ensuring system reliability

Together, these components position the API not merely as a backend utility, but as a mediating infrastructure that shapes how knowledge circulates, is evaluated, and gains legitimacy across platforms.

5.3 Pilot Implementation Outcomes

The pilot phase, carried out in the final quarter of 2025, provided preliminary insights into the system’s effectiveness. Partner platforms reported improvements in their ability to access distributed knowledge resources, addressing, at least in part, the connectivity gaps previously highlighted by 62 percent of stakeholders. Automated classification also showed promising results, demonstrating a 70-80 percent level of agreement with human expert judgment, although performance varied depending on the content type and degree of domain specificity. Early user engagement data indicated modest increases in participation, though a more comprehensive assessment of impact will require longer observation periods. This suggests that infrastructural change is a necessary but insufficient condition for transforming user engagement patterns.

5.4 Lessons from Implementation Experience

5.4.1 Technical Insights

The implementation experience of GGKP and SWITCH-Asia offers several valuable technical lessons. First, achieving genuine interoperability is far more complex than simply aligning technical protocols. It also requires semantic consistency, agreed quality standards, and governance arrangements that allow platforms to communicate meaningfully across institutional boundaries. Technological systems are inseparable from the social and institutional contexts in which they are embedded. Technical solutions on their own have proven insufficient without the parallel development of these institutional frameworks and effective communications support.

Second, the pilot highlighted significant scalability challenges. While the API demonstrated its technical feasibility, extending the system to support platforms built on diverse architectures poses considerable engineering hurdles and demands substantial resource investment. Finally, the experience highlighted the importance of ongoing maintenance. Sustaining effective operation requires continuous updates to the taxonomy, rigorous quality assurance, and consistent system upkeep, costs that are frequently underestimated during initial planning stages.

5.4.2 Institutional and Social Dimensions

Beyond the technical dimensions, the implementation process also revealed a set of critical institutional factors. Effective taxonomy management depended on the establishment of clear governance structures capable of balancing the need for standardization with respect for stakeholder autonomy. Multi-stakeholder governance bodies played a particularly important role, providing both legitimacy and a forum for sustained engagement. The experience further emphasized the necessity of capacity building. Many platform partners required substantial technical support to implement API integration, illustrating that technology transfer is only successful when accompanied by investment in training and assistance. Finally, the question of trust emerged as a decisive factor in adoption. Stakeholders emphasized that transparent documentation of classification algorithms and quality assurance processes was essential to building confidence in the system and ensuring long-term participation. Trust, in this sense, emerges as a media effect of transparency and participation rather than a by-product of technical sophistication alone.

6. Theoretical Implications and Design Principles

6.1 Validated Design Principles for Cognitive Augmentation

Drawing on empirical findings and implementation experience, five design principles can be identified as central to the development of effective knowledge platforms. These principles articulate what it means to “cool” digital knowledge infrastructures; to shift them from authoritative, closed systems toward participatory media environments that support collective sense-making and action.

The first is epistemic pluralism. Platforms must be able to accommodate multiple epistemological frameworks while still maintaining operational coherence. The success of the Common Taxonomy’s custom domain provisions illustrates how this can be achieved in practice. Flexible classification systems allow for the coexistence of different organizational logics, while metadata structures can capture varying evidence types and quality criteria. Interfaces that enable users to filter content according to their epistemological preferences, alongside validation mechanisms tailored to different knowledge traditions, ensure that diverse ways of knowing are recognized and integrated. This principle directly counters the tendency of global platforms to universalize particular epistemologies under the guise of neutrality, a dynamic long critiqued in postcolonial and feminist scholarship.

The second principle is participatory architecture. Effective platforms do not simply disseminate information but actively involve users in knowledge creation, validation, and application. Mechanisms for user-generated content, supported by quality controls, provide opportunities for contribution. Annotation and commentary systems encourage collaborative construction of knowledge, while peer review and distributed validation workflows help to share responsibility for quality assurance. Recognition and incentive structures further motivate sustained participation, transforming users from passive recipients into active co-creators. Such architectures redistribute interpretive agency, enabling users to negotiate meaning rather than merely decode preferred readings embedded in platform design.

The third principle is transparent augmentation. Stakeholders need to understand not only the outputs of algorithms but also their operational logic and limitations. This requires the use of explainable AI techniques that reveal classification and recommendation rationales, as well as confidence indicators that communicate levels of uncertainty. Provenance tracking supports verification of sources, while user controls allow individuals to adjust algorithmic parameters, reinforcing trust and accountability. Transparency here functions as a media effect rather than a purely technical feature in that it shapes users’ willingness to engage with, question, and contest algorithmic authority.

The fourth principle is contextual adaptation. Platforms must serve diverse use contexts while remaining interoperable. Beyond simple language translation, localization efforts need to extend to cultural adaptation. Configurable interfaces should account for different levels of user expertise, while flexible data models can accommodate regional variations in metrics and standards. Progressive disclosure techniques, which tailor the level of detail according to user needs, can help manage complexity without overwhelming different audiences. This principle reflects McLuhan’s distinction between visual and acoustic space. Rather than imposing a single linear pathway, platforms must allow users to navigate knowledge relationally, according to situational relevance.

Finally, human-AI collaboration is central to realizing the promise of cognitive augmentation. Rather than replacing human judgment, AI must be positioned as a complement. Human-in-the-loop workflows remain critical for key decisions, while AI-assisted search and automated suggestions can accelerate processes, provided they remain subject to human validation. Effective task allocation, assigning repetitive, large-scale classification to machines and interpretive, strategic functions to humans, ensures that the strengths of each are leveraged. This approach operationalizes Engelbart’s vision of augmentation while avoiding the automation-centric logic that risks intensifying hot media dynamics.

Across all five principles, the role of communication and knowledge management expertise emerges as indispensable. Skilled professionals are needed to oversee content development, curation, management, and dissemination so that platforms remain relevant, inclusive, and user-centered. These experts bridge the gap between algorithmic outputs and meaningful knowledge products, ensuring data is translated into accessible formats, contextualized appropriately, and shared effectively. Moreover, they function as mediators within the system, sustaining dialogue, feedback, and learning. Without this human layer of expertise, even the most advanced AI-enabled platforms risk producing fragmented or underutilized outputs that fail to meet the evolving demands of diverse stakeholders.

6.2 Implications for Future Platform Development

The research demonstrates that viewing platforms merely as repositories of information fundamentally constrains their potential. To remain relevant and effective, future platforms must evolve into active intermediaries that facilitate knowledge synthesis, provide quality assurance, and support the application of insights in real-world contexts. This shift entails a transition from hot media logics of transmission and authority toward cool media environments that invite participation by users. This shift requires moving beyond the conception of platforms as purely technical infrastructures and instead recognizing them as sociotechnical systems, embedded in broader institutional, cultural, and communicative environments.

A central design challenge lies in reconciling the need for standardization with the demand for contextual flexibility. On one hand, standardized frameworks are essential for ensuring interoperability across diverse systems; on the other, rigid uniformity risks undermining relevance in specific local or sectoral contexts. Our experience demonstrates a pragmatic approach to this dilemma, namely, adopting standardized taxonomic frameworks while allowing for customization through specialized provisions. Although this model offers a workable compromise, it also introduces considerable implementation complexity that must be carefully managed.

Finally, the findings highlight that technical innovations alone are insufficient to resolve the problem of knowledge fragmentation. Effective and sustainable platforms depend equally on governance and funding arrangements. Multi-stakeholder governance bodies are necessary to guarantee legitimacy and inclusivity, while transparent decision-making processes help build and maintain trust among diverse partners. At the same time, diversified and stable funding models are critical to sustaining platform operations over the long term. Together, these elements highlight that the durability and effectiveness of digital knowledge infrastructures hinge as much on governance and sustainability as on technical design.

7. Recommendations for Stakeholder Communities

7.1 Knowledge Platform Operators

Based on the empirical findings, several strategic initiatives are recommended for platform operators seeking to enhance the effectiveness and sustainability of their systems. These recommendations translate media ecology insights into actionable governance and design choices, recognizing operators as media designers rather than mere custodians of information.

First, strategic planning must place communication and knowledge management at the center of platform design and operation. Comprehensive strategies should align platform capabilities with stakeholder needs by articulating clear value propositions, identifying priority audiences, and establishing measurable impact metrics. This ensures that platforms remain user-oriented and results-driven rather than supply-driven. This step is key to counter “hot” media tendencies that prioritize output volume over meaning, use, and engagement.

Second, operators should pursue service transformation initiatives that move platforms beyond static repositories toward dynamic service environments. Such platforms would not only curate knowledge enriched with contextual metadata but also facilitate partnerships and matchmaking among stakeholders. Additional services such as performance benchmarking and comparison tools, along with rigorous quality assurance and validation mechanisms, would further enhance credibility and usefulness. These intermediary functions correspond to “cool” media environments that invite participation, interpretation, and situated application rather than passive consumption.

Third, sustained investment in technical infrastructure is essential. Implementing standardized taxonomies, application programming interfaces (APIs), and metadata schemas will improve interoperability across platforms. At the same time, integrating citation tracking and source verification safeguards the credibility of shared knowledge and supports transparent use of evidence. Such infrastructural investments should be understood not as efficiency upgrades alone, but as interventions that reshape the media conditions under which knowledge circulates and gains legitimacy.

Finally, platforms must be anchored in inclusive governance arrangements. This involves establishing governance bodies that reflect diverse stakeholder perspectives, with deliberate efforts to include marginalized communities. Such inclusivity not only enhances legitimacy but also ensures that platforms respond to the realities and needs of those most affected by sustainability transitions. As critical media scholarship reminds us, participation without voice risks reproducing existing hierarchies rather than democratizing knowledge.

7.2 Policy and Regulatory Stakeholders

Policymakers have an important role to play in addressing systemic barriers that currently limit the effectiveness of digital knowledge infrastructures. Harmonization initiatives should be prioritized through regional agreements that standardize data formats, terminologies, and quality criteria, while still respecting national sovereignty and accommodating contextual variations. At the same time, investment in evidence infrastructure is needed to fill critical knowledge gaps, in the context of the circular economy, this includes: support for material flow analyses, mapping of informal sector activities, and the development of AI-ready datasets.

In parallel, policymakers should pursue adaptive policy frameworks capable of accommodating rapid technological evolution while upholding accountability and transparency requirements. Such flexibility ensures that regulatory environments remain both enabling and responsible. Finally, pilot program development offers a practical way forward by testing AI applications in policy analysis under controlled conditions, with safeguards for human oversight and ethical integrity.

7.3 Development Partners and Funding Organizations

Development partners and funding bodies should align their support strategies to reflect the needs identified through platform implementation. First, core funding provision must be recognized as essential for sustainability. Knowledge platforms require ongoing operational support that extends well beyond initial development investments, including dedicated resources for communications management, regular maintenance, system updates, and capacity building. Without such support, platforms risk becoming “hot” showcases that prioritize visibility over long-term usability and learning.

Second, targeted investments in interoperability are necessary to ensure that platforms can connect and exchange information seamlessly. This includes support for standardization initiatives, API development, and the creation of common vocabularies. Such investments should be framed as contributions to shared media environments rather than isolated project outputs.

Third, capacity-building programs should be expanded to equip stakeholders with the technical skills, knowledge management competencies, and collaborative governance capabilities required for effective platform use and maintenance.

Finally, partners should invest in the creation of digital commons, such as AI training datasets, evaluation benchmarks, and open-source tools. These shared resources reduce barriers to entry for resource-constrained organizations, fostering more inclusive participation in global knowledge ecosystems.

7.4 Knowledge Practitioners and End Users

Practitioners are encouraged to adopt approaches that maximize both the reliability and the impact of knowledge platforms. One key strategy involves the use of hybrid intelligence, where AI-assisted discovery is systematically combined with human expert validation. This balance ensures that results are not only efficient but also contextually appropriate and subject to rigorous quality assurance.

Equally important is community engagement. Active involvement in peer learning networks, the exchange of case studies, and participation in collaborative problem-solving initiatives help strengthen the relevance of platforms while building shared ownership among users.

Finally, practitioners should place strong emphasis on value communication. The effectiveness of knowledge platforms depends not only on making information available, but on demonstrating how that information can be translated into action. By clearly highlighting the benefits for decision-making, collaboration, and problem-solving, practitioners can enhance user engagement and ensure that knowledge is actively applied rather than left unused.

8. Conclusions and Future Directions

8.1 Synthesis of Key Findings

This investigation reveals several challenges in current knowledge platform architectures and operations. The empirical evidence, derived from engagement with over 450 global stakeholders, demonstrates that despite substantial technological investments, platforms fail to effectively bridge the persistent knowledge-action gap characterizing sustainable development practice.

The identification of platform connectivity deficits by 62 percent of stakeholders and recognition of audience understanding gaps by 48 percent of participants indicate systemic rather than isolated challenges. These findings, while derived from self-reported assessments requiring cautious interpretation, align consistently with patterns documented in knowledge management and sustainable development literature, lending credibility to their validity.

The collaborative development of the ‘Common Circular Economy Knowledge Taxonomy’ and GGKP API implementation represents meaningful progress toward addressing identified challenges. These initiatives demonstrate that effective knowledge systems require integration of technical infrastructure (standardized APIs, automated classifiers, federated search capabilities) with social infrastructure (inclusive governance, good communication management, community participation, quality assurance mechanisms, among others). Their partial success reinforces the argument that infrastructural change must be accompanied by shifts in media logic toward participation and mediation. The importance of maintaining human oversight alongside AI capabilities confirms that cognitive augmentation rather than automation should guide platform development strategies.

8.2 Theoretical Contributions

This research aims at contributing to the theoretical debates in knowledge management and human-computer interaction. By applying McLuhan’s distinction between hot and cool media to the design and functioning of digital knowledge platforms, it offers possibly new insights into user engagement dynamics. The empirical finding that users show a preference for participatory, “cool” media characteristics over passive “hot” media consumption provides an important extension of media ecology theory in digital contexts. In doing so, the study responds to calls to operationalize media theory within contemporary digital infrastructures rather than treating it as purely descriptive.

At the same time, the study advances cognitive augmentation scholarship by extending Engelbart’s original concepts through practical implementation examples. These cases demonstrate how human-AI collaboration can be operationalized in practice within knowledge management environments. A further contribution lies in the operationalization of epistemic pluralism. The flexible taxonomic structures developed here accommodate diverse knowledge traditions while preserving structural coherence, offering a practical model for integrating plural epistemologies within global digital systems.

A number of limitations in this study should also be acknowledged. First, the reliance on self-reported stakeholder assessments without independent verification constrains the ability to make strong causal inferences about platform effectiveness. Future studies should complement stakeholder perspectives with objective usage analytics and independent impact assessments. Second, the research process was mediated largely in English and conducted through existing professional networks, which may have introduced geographic and linguistic biases. Future research should address these constraints through multilingual methodologies and purposive sampling strategies to ensure more inclusive representation. Finally, while the study carefully distinguishes between current AI applications and the theoretical possibilities of artificial general intelligence, the longer-term implications of AGI for knowledge systems remain speculative. Ongoing theoretical development will be required to more fully assess the risks and opportunities associated with AGI in this domain.

8.3 Implications for Sustainable Development Practice

The path forward requires fundamental reconceptualization of knowledge platforms as cognitive augmentation systems that amplify human intelligence rather than attempting to replace it. This transformation demands moving beyond hot media paradigms toward participatory architectures enabling active knowledge creation, contextual adaptation preserving epistemic diversity, and transparent AI systems supporting rather than substituting human judgment. It requires redesigning platforms as cool media environments oriented toward collective sense-making. The sustainable development community faces a critical decision point. Current trajectories suggest continued proliferation of sophisticated but ultimately ineffective platforms, perpetuating fragmentation and failing to catalyze necessary transformations. Alternatively, the community can embrace fundamental reconceptualization of how digital systems support collective intelligence for sustainability. The empirical evidence, validated through extensive stakeholder consultation, unequivocally supports the latter approach.

Success in bridging the knowledge-action gap requires more than the deployment of increasingly sophisticated technologies. It demands a fundamental reconceptualization of the relationship between human cognition, artificial intelligence, and collective problem-solving, anchored in robust communication and knowledge management expertise. This expertise is needed not only for designing platforms that are purposeful and user-centered, but also for ensuring that their objectives are clearly articulated, their value effectively communicated, and their results systematically monitored and evaluated. Only by treating platforms as media environments and partners in cognition, rather than neutral tools, can the sustainable development community harness their full potential. Such an approach ensures that the right knowledge reaches the right people in accessible and actionable forms, empowering stakeholders to access, digest, apply, and transform knowledge into practice. In doing so, platforms can finally move beyond static information repositories to become engines of implementation, helping to close the persistent gap between what we know and what we are able to do.

References

¹ Various reports reviewed in this study frequently cite an estimate of ‘over 500 sustainability-focused knowledge platforms globally,’ despite the absence of systematic enumeration. This figure should be interpreted as indicative rather than definitive, highlighting the scale of platform proliferation rather than providing precise quantification.

Agyeman, J., et al. “Exploring the Nexus: Bringing Together Sustainability, Environmental Justice and Equity.” Space and Polity, vol. 6, no. 1, 2016, pp. 77–90.

Arrieta, A. B., et al. “Explainable Artificial Intelligence (XAI): Concepts, Taxonomies, Opportunities and Challenges toward Responsible AI.” Information Fusion, vol. 58, 2020, pp. 82–115.

Bennett, N. J., et al. “Communities and Change in the Anthropocene: Understanding Social-Ecological Vulnerability and Planning Adaptations to Multiple Interacting Exposures.” Regional Environmental Change, vol. 16, no. 4, 2019, pp. 907–926.

Braun, V., and V. Clarke. “Using Thematic Analysis in Psychology.” Qualitative Research in Psychology, vol. 3, no. 2, 2006, pp. 77–101.

Brown, T., et al. “Language Models Are Few-Shot Learners.” Advances in Neural Information Processing Systems, vol. 33, 2020, pp. 1877–1901.

Brynjolfsson, E., and A. McAfee. Machine, Platform, Crowd: Harnessing Our Digital Future. W. W. Norton & Company, 2017.

Cash, D. W., et al. “Knowledge Systems for Sustainable Development.” Proceedings of the National Academy of Sciences, vol. 100, no. 14, 2003, pp. 8086–8091.

Cornell, S., et al. “Opening up Knowledge Systems for Better Responses to Global Environmental Change.” Environmental Science & Policy, vol. 28, 2013, pp. 60–70.

Daugherty, P. R., and H. J. Wilson. Human + Machine: Reimagining Work in the Age of AI. Harvard Business Review Press, 2018.

Devlin, J., et al. “BERT: Pre-Training of Deep Bidirectional Transformers for Language Understanding.” Proceedings of NAACL-HLT, 2019, pp. 4171–4186.

Engelbart, D. C. Augmenting Human Intellect: A Conceptual Framework. Stanford Research Institute, 1962.

EU SWITCH-Asia Policy Support Component and Green Growth Knowledge Partnership. Bridging Circular Economy Knowledge and Action through Enhanced Knowledge Generation, Management and Sharing: Proceedings Report of the Circular Economy Knowledge Hubs Webinar Series 2024–2025. EU SWITCH-Asia Programme, 2025.

Fazey, I., et al. “Knowledge Exchange: A Review and Research Agenda for Environmental Management.” Environmental Conservation, vol. 40, no. 1, 2013, pp. 19–36.

Grace, K., et al. “When Will AI Exceed Human Performance? Evidence from AI Experts.” Journal of Artificial Intelligence Research, vol. 62, 2018, pp. 729–754.

Guston, D. H. “Boundary Organizations in Environmental Policy and Science: An Introduction.” Science, Technology, & Human Values, vol. 26, no. 4, 2001, pp. 399–408.

“GGKP API.” 2025, https://docs.api.ggkp.org

“GGKP Taxonomy.” 2025, https://taxonomy.ggkp.org

Hall, S. “Encoding/Decoding.” Culture, Media, Language: Working Papers in Cultural Studies, 1972–79, edited by S. Hall et al., Hutchinson, 1980, pp. 128–138.

Haraway, D. “Situated Knowledges: The Science Question in Feminism and the Privilege of Partial Perspective.” Feminist Studies, vol. 14, no. 3, 1988, pp. 575–599.

Harding, S. Is Science Multicultural? Postcolonialisms, Feminisms, and Epistemologies. Indiana University Press, 1998.

Hogan, A., et al. “Knowledge Graphs.” ACM Computing Surveys, vol. 54, no. 4, 2021, pp. 1–37.

Ludwig, D., and C. N. El-Hani. “Philosophy of Ethnobiology: Understanding Knowledge Integration and Its Limitations.” Journal of Ethnobiology, vol. 40, no. 1, 2020, pp. 3–20.

Marcus, G., and E. Davis. Rebooting AI: Building Artificial Intelligence We Can Trust. Pantheon Books, 2019.

McLuhan, M. Understanding Media: The Extensions of Man. McGraw-Hill, 1964.

McLuhan, M., and G. B. Leonard. City as Classroom: Understanding Language and Media. Agincourt, 1977.

Michener, W. K., and M. B. Jones. “Ecoinformatics: Supporting Ecology as a Data-Intensive Science.” Trends in Ecology & Evolution, vol. 27, no. 2, 2012, pp. 85–93.

Mitchell, M. Artificial Intelligence: A Guide for Thinking Humans. Farrar, Straus and Giroux, 2019.

Müller, V. C., and N. Bostrom. “Future Progress in Artificial Intelligence: A Survey of Expert Opinion.” Fundamental Issues of Artificial Intelligence, edited by V. C. Müller, Springer, 2016, pp. 555–572.

Mulvey, L. “Visual Pleasure and Narrative Cinema.” Screen, vol. 16, no. 3, 1975, pp. 6–18.

Parker, J., and B. Crona. “On Being All Things to All People: Boundary Organizations and the Contemporary Research University.” Social Studies of Science, vol. 42, no. 2, 2012, pp. 262–289.

Postman, N. Amusing Ourselves to Death: Public Discourse in the Age of Show Business. Viking, 1985.

Postman, N. Technopoly: The Surrender of Culture to Technology. Knopf, 1992.

Reed, M. S., et al. “Five Principles for the Practice of Knowledge Exchange in Environmental Management.” Journal of Environmental Management, vol. 146, 2014, pp. 337–345.

Reimers, N., and I. Gurevych. “Sentence-BERT: Sentence Embeddings Using Siamese BERT-Networks.” Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing, 2019.

Said, E. W. Orientalism. Pantheon Books, 1978.

Silver, D., et al. “A General Reinforcement Learning Algorithm That Masters Chess, Shogi, and Go through Self-Play.” Science, vol. 362, no. 6419, 2018, pp. 1140–1144.

Suárez-Figueroa, M. C., et al. “The NeOn Methodology for Ontology Engineering.” Ontology Engineering in a Networked World, Springer, 2012, pp. 9–34.

van Kerkhoff, L., and L. Lebel. “Linking Knowledge and Action for Sustainable Development.” Annual Review of Environment and Resources, vol. 31, 2006, pp. 445–477.

Wang, Y., et al. “Generalizing from a Few Examples: A Survey on Few-Shot Learning.” ACM Computing Surveys, vol. 53, no. 3, 2020, pp. 1–34.

Wilkinson, M. D., et al. “The FAIR Guiding Principles for Scientific Data Management and Stewardship.” Scientific Data, vol. 3, no. 1, 2016, pp. 1–9.