Data Visualization Report

Executive Summary

Key Insights

  1. AI Role Distribution: The portfolio spans 13 "implicit", 11 "opportunity", and 9 "explicit" AI initiatives. Explicit AI clusters around governance and health (SDG 16, SDG 3), while implicit and opportunity projects emphasize environment, water, urban systems, and operational digital government (SDG 13, SDG 6, SDG 11).
  2. SDG Concentration and Gaps: SDG 13 (climate) and SDG 16 (governance) are the most represented, followed by SDGs 3, 11, 12, 15, and 10. Underrepresented areas include SDG 14 (Life Below Water) and SDG 1 in explicit initiatives, indicating opportunities to expand ocean and poverty-focused AI applications.
  3. Methodological Diversity: NLP, predictive modeling, geospatial AI, network analysis, computer vision, and generative design are well-distributed across themes. Governance-heavy institutes (UNU-CPR, UNU-CRIS, UNU-EGOV) leverage NLP, simulation, and network analysis; environment-focused institutes (UNU-EHS, UNU-FLORES, UNU-INRA/INWEH) use geospatial and computer vision; economics-focused units (UNU-MERIT, UNU-WIDER) utilize causal and econometric ML.
  4. Collaboration Potential: Strong method overlap suggests high collaboration potential: geospatial AI links EHS–INRA–INWEH–FLORES; governance/NLP links CPR–CRIS–EGOV–IIGH; equality/economics ML links MERIT–WIDER. UNU Headquarters operational digital initiatives can benefit from shared tooling developed in EGOV and CPR.
  5. Impact Narratives vs. Metrics: Impacts are predominantly qualitative ("enhances", "accelerates"). Few standardized KPIs are present, indicating a need for a cross-portfolio measurement framework capturing policy influence, predictive accuracy, operational gains, and beneficiary reach.

Strategic Recommendations

  1. Establish a shared AI capability hub focused on geospatial ML, NLP for policy analysis, and network modeling; prioritize cross-institute tool repositories and training.
  2. Introduce a standardized impact measurement framework with quantitative KPIs (e.g., policies influenced, model performance, efficiency gains, beneficiaries) and track project maturity stages.
  3. Expand portfolio into underrepresented SDGs (14, 1 in explicit) and regions (Latin America beyond BIOLAC, MENA beyond biodiversity) through targeted calls and partnerships.
  4. Launch cross-institute consortia: Governance & NLP (CPR–CRIS–EGOV–IIGH), Climate & Geospatial (EHS–INRA–INWEH–FLORES), Inclusive Economies & Causal ML (MERIT–WIDER).
  5. Integrate operational AI in administrative functions via UNU-AI Institute; leverage lessons from EGOV for chatbots, data governance, and resource optimization.

Key Metrics

  1. Portfolio Size: [33 initiatives] - Indicates a broad and diverse AI-related activity set. Compared to benchmark: Robust coverage across governance, climate, and digital domains.
  2. AI Role Mix: [Explicit: 9, Implicit: 13, Opportunity: 11] - Indicates balanced exploration and operationalization. Compared to benchmark: Healthy pipeline from opportunity to implicit integration.
  3. Top SDGs: [SDG 13: 13 mentions, SDG 16: 12 mentions] - Indicates strong climate and governance focus. Compared to benchmark: Above expected representation in climate/governance.
  4. Method Overlap Clusters: [3 major clusters] - Indicates high collaboration potential in geospatial, NLP/policy, and econometrics/causal ML. Compared to benchmark: Strong foundations for shared tooling.
  5. Underrepresented Domains: [SDG 14 minimal] - Indicates ocean-related gaps. Compared to benchmark: Below expected; opportunity for targeted initiatives.


Charts

SDG Frequency by AI Type

Analysis Note:
Description: This chart aggregates the count of initiatives mentioning each SDG by AI role type (explicit, implicit, opportunity). Counts were derived by tallying SDGs listed per initiative and grouping by aiType. Key insights: SDG 13 (climate) and SDG 16 (governance) lead across the portfolio, with explicit AI concentrated in SDG 16 and SDG 3 (health), while implicit and opportunity AI emphasize SDG 13, SDG 6 (water), SDG 11 (cities), and SDG 12 (consumption). Outliers include minimal representation of SDG 14. Recommendations: Launch targeted calls to address SDG 14 and reinforce explicit AI in poverty alleviation (SDG 1) and oceans; align capacity building to dominant SDGs to scale impact.

Institute vs. AI Method Usage Heatmap

Analysis Note:
Description: This heatmap encodes the frequency of AI method mentions per institute. Methods were extracted from aiEnhancements and mapped to a standardized taxonomy (NLP, predictive modeling, geospatial AI, computer vision, network analysis, generative design, causal/econometric ML, simulation, anomaly detection). Missing combinations are represented as 0. Key insights: Governance-oriented units (CPR, CRIS, EGOV) emphasize NLP, simulation, and network analysis; environment-focused units (EHS, INRA, INWEH, FLORES) show strong geospatial and predictive modeling; economics units (MERIT, WIDER) use causal/econometric ML. Recommendations: Build shared tooling in geospatial ML and NLP; cross-train institutes with low method diversity; establish repositories of reusable models.

System-Wide Knowledge Flow: Institute → AI Type → SDG

Analysis Note:
Description: What it shows: A portfolio flow from each institute to AI maturity stages and onward to SDGs, with link thickness proportional to project counts. Institute→AI links use counted projects; AI→SDG links aggregate SDG tags by AI type. Key insights: Strong opportunity flow to SDG 13/15 indicates high-impact climate-environment potential awaiting activation. Explicit flows cluster around SDG 16 and SDG 3, reflecting mature governance and health analytics. FLORES and IAS dominate opportunity pipelines to urban/circular SDGs (11/12). Recommendations: Allocate catalytic resources to convert opportunity-heavy climate and urban streams to explicit pilots; replicate explicit governance-analytics assets across regions; use this flow map quarterly to monitor rebalancing and impact.

AI Type → Theme → SDG Sankey

Analysis Note:
Description: Flows were constructed by categorizing each initiative into themes (Governance/Policy, Health, Climate/Environment, Agriculture/Food, Urban/Infrastructure, Digital Government, Migration/Displacement) and mapping to listed SDGs. Link values approximate counts based on initiative distribution. Key insights: Explicit AI strongly feeds Governance/Policy and Health; implicit and opportunity streams feed Climate/Environment, Urban/Infrastructure, and Digital Government. SDG 13 and SDG 16 capture the largest downstream flows. Recommendations: Balance flows by initiating ocean-focused projects (SDG 14) and reinforcing education equity in displacement (SDG 4) with explicit AI pilots.

Institute Collaboration Network by Shared Methods and SDGs

Analysis Note:
Description: Nodes represent institutes sized by approximate portfolio relevance; edges indicate shared AI methods and overlapping SDG coverage, with weights reflecting the number of shared attributes. Key insights: Three clusters emerge—Governance/NLP (CPR–CRIS–EGOV–IIGH–HQ), Climate/Geospatial (EHS–INRA–INWEH–FLORES), and Inclusive Economies (MERIT–WIDER). Recommendations: Formalize cross-institute working groups per cluster; establish shared repositories and training tracks; use HQ as a conduit for operational adoption.

Institute Theme Treemap

Analysis Note:
Description: This treemap shows the distribution of initiatives across themes within each institute. Theme counts were derived from project titles and descriptions and grouped into coherent categories. Key insights: FLORES is diversified across urban, climate, agriculture, and biodiversity; CPR and CRIS are governance-heavy; EGOV focuses on digital government and data governance; EHS and INRA emphasize climate and migration/agriculture. Recommendations: Encourage institutes with narrow thematic portfolios to collaborate across themes; leverage diversified institutes (FLORES, IAS) to bridge environment, urban, and education initiatives.

AI Enhancements and Impact Keywords Wordclouds

Analysis Note:
Description: Keywords were extracted from aiEnhancements and impact text, normalized, and counted; sizes reflect relative frequency (approximate scaling). Key insights: Dominant method terms include predictive, NLP, geospatial, network, simulation, and computer vision; impact narratives emphasize governance, efficiency, resilience, capacity building, and health. Recommendations: Introduce keywords tied to measurement (accuracy, F1, beneficiaries) to shift narratives toward quantifiable outcomes; expand domain-specific terms (oceans, fisheries) to cover SDG 14.

SDG Breadth per Project by Institute

Analysis Note:
Description: This boxplot visualizes the distribution of the number of SDGs referenced per project, grouped by institute. Data points represent per-project SDG counts derived from the dataset. Key insights: IAS and FLORES exhibit broader SDG coverage, reflecting integrative nexus approaches; CPR and CRIS show moderate breadth consistent with governance-focused initiatives; EGOV maintains focused coverage aligned with digital government tasks; Headquarters has internal operational focus with N/A SDGs. Recommendations: Encourage institutes with narrow breadth to integrate cross-SDG linkages (e.g., governance + climate + equity) and standardize SDG tagging to reflect multi-dimensional impacts.