Ai and Monitoring & Evaluation in Kenya’s Public Sector
How AI Could Strengthen Monitoring and Evaluation, Transparency and Accountability in Kenya’s Public Sector
An EvalCommunity interpretation of research on AI-driven monitoring and evaluation in public project implementation.
This case study is based on a published research article by Bildad Awere and Daniel Mishael Masetu. It presents the study’s reported findings alongside EvalCommunity’s interpretation of their implications for monitoring and evaluation (M&E) practice. The practical workflow and lessons below are EvalCommunity’s analysis and recommendations; they are not presented as interventions tested by the original researchers.
1. The case at a glance
Public projects require reliable information about implementation progress, resource use, results and risks. Yet monitoring systems can struggle when information is collected manually, stored in disconnected systems or reported too slowly to support timely decisions.
The research examined how artificial intelligence (AI)-supported M&E could contribute to more transparent and accountable public project implementation, with a focus on the Kenyan public-sector context.
Its findings point to potential benefits from predictive analytics, real-time data processing and automated project tracking. They also highlight institutional and infrastructure constraints that can limit adoption.
These figures are reported in the research article. They should be interpreted in the context of its sample, methods and limitations.
2. The M&E challenge: when information arrives too late
The article describes challenges associated with traditional public-sector monitoring and evaluation, including manual processes, fragmented information systems, delays in reporting and limited analytical capacity.
These challenges matter because public project managers and oversight institutions need timely, credible information to understand whether activities are progressing as planned, whether resources are being used appropriately and whether corrective action is required.
The problems identified in the research context
- Manual monitoring and reporting processes.
- Fragmented or inconsistent departmental data.
- Delays in identifying implementation problems.
- Limited ICT infrastructure and connectivity.
- Insufficient technical skills and institutional capacity.
- Challenges in ensuring transparency, oversight and accountability.
EvalCommunity interpretation: The central issue is not simply a lack of technology. It is the gap between the information a public project generates and the information decision-makers can access, verify, interpret and act upon. AI may help narrow that gap, but only if the underlying M&E system is sufficiently reliable.
3. What did the research examine?
The article, AI-Driven Monitoring and Evaluation: The Future of Transparent and Accountable Governance in Public Project Implementation, set out to assess how AI-driven M&E could promote transparent and accountable governance in public project implementation.
Research objectives
- Analyze the role of AI in strengthening innovation and transparency in M&E systems.
- Examine institutional and infrastructural factors affecting AI adoption in project monitoring.
- Evaluate the relationship between AI-enabled M&E, corruption risks and governance outcomes.
Research methodology
The authors report using a mixed-methods approach, combining questionnaires, interviews, document analysis and AI data mining.
| Participant group | Target population | Sample |
|---|---|---|
| Public officers | 500 | 40 |
| ICT experts | 150 | 30 |
| Project managers | 100 | 30 |
| Total | 750 | 100 |
The article reports a 100% questionnaire response rate among the 100 selected participants. Quantitative analysis included descriptive statistics, correlation, regression and ANOVA. Qualitative information was analyzed using thematic coding.
The authors also describe informed consent, anonymization, secure data storage and attention to privacy, fairness and potential AI bias as ethical considerations.
4. What did the study report?
The article presents several findings concerning AI applications in M&E and public project management. The percentages below are the frequencies reported by respondents in the study; they should not be interpreted as independently measured improvements in project outcomes.
A. AI innovations in M&E and project tracking
| AI application | Frequency reported |
|---|---|
| Predictive analytics | 65% |
| Real-time data processing | 55% |
| Automated monitoring | 45% |
| Risk identification | 35% |
EvalCommunity interpretation: Predictive analytics and real-time processing are particularly relevant to the M&E cycle because they can help teams move beyond periodic reporting. Rather than waiting for a scheduled report to reveal a problem, teams may be able to identify unusual patterns or emerging risks earlier. However, an AI-generated alert is a signal for investigation—not proof that a problem exists.
B. Applications in infrastructure and service delivery
| AI application | Frequency reported |
|---|---|
| Resource allocation optimization | 70% |
| Automated project management | 60% |
| Infrastructure monitoring | 50% |
EvalCommunity interpretation: These findings point to an important opportunity: AI-supported M&E should not be limited to producing reports. It can potentially connect monitoring information to operational decisions, such as reviewing resource allocation, investigating implementation delays and prioritizing management attention.
C. Human resources and institutional capacity
| Reported area | Frequency reported |
|---|---|
| Workforce planning | 55% |
| Enhanced decision-making | 60% |
| Operational streamlining | 65% |
EvalCommunity interpretation: The human dimension is central. AI adoption is not simply a technical upgrade; it changes the information available to staff and the way teams organize their work. M&E professionals need the competence to assess AI-generated information, explain its limitations and connect analytical findings to programme decisions.
D. Knowledge management and organizational learning
| Knowledge management area | Frequency reported |
|---|---|
| Knowledge capture and storage | 68% |
| Sharing best practices | 72% |
| Continuous learning | 65% |
EvalCommunity interpretation: Knowledge management deserves a more prominent place in AI-enabled M&E. Evaluation reports, implementation lessons, monitoring records and management responses can become more useful when they are organized and accessible. AI may help teams retrieve and synthesize this knowledge, but the accuracy and context of the underlying records remain essential.
5. What do the statistical results suggest?
The article reports positive correlations between AI-related practices and governance outcomes. The reported correlations include:
| AI-related practice | Efficiency | Transparency | Accountability |
|---|---|---|---|
| Predictive analytics / AI adoption | 0.82 | 0.75 | 0.79 |
| Project tracking automation | 0.79 | 0.77 | 0.80 |
| Real-time data processing | 0.85 | 0.80 | 0.83 |
The article reports these correlations as statistically significant at p < 0.01.
The study also reports a regression model with R = 0.91, R² = 0.83 and adjusted R² = 0.82. The authors interpret the R² value as indicating that the model explains 83% of the variation in governance performance.
Important statistical interpretation
An R² of 0.83 describes the proportion of variation explained by the fitted regression model in the data analyzed. It does not mean that AI caused an 83% improvement in governance, nor does it establish that corruption was reduced by 83%. Correlation and regression results should be interpreted in light of the research design, measurement approach and possible alternative explanations.
The article also reports an ANOVA result of F = 25.78 and p = 0.000, conventionally expressed as p < 0.001. The authors interpret this as evidence of statistically significant differences in the comparison of AI-supported and traditional M&E approaches.
EvalCommunity interpretation: The statistical findings are encouraging, but the practical question for M&E professionals is how these relationships translate into better-quality evidence, stronger management responses and measurable results. Statistical association is a starting point for further investigation—not a substitute for evaluating actual programme outcomes.
6. EvalCommunity interpretation: five implications for M&E practice
1. AI can help shift M&E from periodic reporting to more continuous monitoring
Traditional reporting cycles can leave decision-makers working with information that is already outdated. Real-time data processing and automated monitoring may help teams identify emerging implementation issues sooner.
For M&E professionals, this means thinking beyond the production of periodic reports and considering how reliable data can support ongoing performance monitoring and management decisions.
2. Predictive analytics can strengthen risk-based monitoring
Predictive analytics may help identify patterns associated with delays, cost pressures or other implementation risks. This can support more focused monitoring and help teams decide where additional verification or management attention is needed.
However, predictions are not findings. They should be checked against project records, field observations, stakeholder feedback and other independent evidence before decisions are made.
3. Better data integration is a prerequisite for meaningful AI
AI cannot reliably compensate for incomplete, inconsistent or poorly defined data. Fragmented information systems and weak data quality can produce misleading analysis, even when sophisticated tools are used.
Our interpretation is that organizations should strengthen data governance, indicator definitions, data validation and interoperability before scaling AI-supported M&E.
4. Human oversight remains essential for accountability
AI can help detect anomalies, summarize information and identify patterns. It cannot independently determine whether a public project is fair, whether an explanation is credible or who should be held accountable for a management decision.
M&E professionals and responsible institutions must retain ownership of interpretation, verification, recommendations and decisions. AI outputs should be traceable to their underlying evidence and reviewed by qualified people.
5. Knowledge management should connect evidence to organizational learning
The knowledge-management findings reinforce the importance of preserving and sharing lessons. AI-supported search and synthesis may help teams identify recurring issues across reports and projects, but lessons must be validated and interpreted in their context.
For evaluation practice, the opportunity is to make evidence more accessible and useful—not simply to produce more automated summaries.
7. An AI-supported M&E workflow: EvalCommunity’s practical interpretation
Building on the issues and findings discussed in the article, EvalCommunity proposes the following practical workflow. This is an illustrative framework for M&E teams; it is not a workflow tested or validated by the original study.
The key principle: AI can assist the M&E process, but accountability remains with the people and institutions responsible for the work.
8. Risks and conditions that organizations must address
The research identifies infrastructure, institutional capacity and ethical concerns as important constraints to AI adoption. EvalCommunity considers these issues essential to any responsible implementation strategy.
Infrastructure and connectivity
Reliable systems, connectivity, appropriate hardware and technical support are needed for effective adoption.
Skills and institutional capacity
Staff need practical AI literacy, data skills and the ability to interpret outputs critically.
Data privacy and security
Sensitive project and personal data must be protected through appropriate governance and security measures.
Bias and fairness
AI systems and their outputs should be reviewed for bias, exclusion and misleading patterns.
Institutional integration
AI tools must fit real workflows, roles, reporting structures and decision-making processes.
Human accountability
Organizations must define who verifies outputs, approves decisions and responds when systems fail.
9. Recommendations for M&E stakeholders
The following recommendations combine the practical recommendations made by the article’s authors with EvalCommunity’s interpretation for M&E implementation.
For governments and public institutions
- Invest in reliable digital infrastructure and modern data collection systems.
- Develop clear institutional policies for responsible AI use in M&E.
- Strengthen data governance, privacy, cybersecurity and accountability arrangements.
- Involve relevant stakeholders when selecting and integrating AI tools.
- Build staff capacity through continuous training and practical support.
For project managers
- Identify project processes where AI-supported monitoring could add practical value.
- Improve the timeliness and quality of project implementation data.
- Use predictive analysis as an input to risk management, not as an automatic decision-maker.
- Ensure that identified risks lead to documented management responses.
For M&E officers and evaluators
- Develop the skills to assess AI-generated analysis and recognize its limitations.
- Verify AI-generated findings against original data and independent evidence.
- Maintain transparent documentation of where and how AI was used.
- Protect confidentiality and ensure that data use is appropriate and authorized.
- Retain professional responsibility for conclusions, recommendations and reporting.
For ICT teams
- Assess system compatibility, data integration and cybersecurity requirements.
- Work with M&E teams to select tools that respond to actual operational needs.
- Monitor system performance and review outputs for errors and bias.
- Establish processes for technical support, maintenance and continuous improvement.
10. A critical reading: what this research does—and does not—establish
Research-based case studies are most useful when their findings are interpreted carefully. The article offers relevant insights into the potential of AI-supported M&E, but readers should distinguish the study’s reported associations from demonstrated changes in real-world project outcomes.
What the article reports
- Responses from a sample of 100 public officers, ICT experts and project managers.
- Reported use and perceived benefits of several AI applications in M&E and project management.
- Positive statistical relationships between AI-related practices and governance indicators.
- Infrastructure, capacity and ethical challenges affecting AI adoption.
What should not be concluded from these findings alone
- That AI has caused a specific percentage improvement in public-sector governance.
- That corruption was demonstrably reduced by a measured amount in a named project.
- That every public institution in Kenya has the infrastructure or capacity to adopt AI successfully.
- That an AI-generated alert or prediction is sufficient evidence for an audit finding or management sanction.
The article is not presented here as a documented evaluation of one named public project or a controlled assessment of a particular AI intervention. Its findings should therefore be treated as research evidence about reported applications and relationships, rather than proof of a specific deployment’s impact.
For future research and implementation assessments, M&E professionals would benefit from more detailed evidence on the AI systems used, data quality, implementation costs, baseline conditions, comparison groups, independent verification and longer-term project outcomes.
11. Questions for M&E professionals
Use these questions to connect the research findings to your own organization or programme. They are reflection questions developed by EvalCommunity, not questions from the original research instrument.
1. Where could AI help us identify project implementation risks earlier?
Consider delays, budget variances, incomplete activities, unusual indicator patterns and gaps in reporting. Identify which data sources could support these analyses and who would verify the results.
2. Are our data systems ready for AI-supported analysis?
Review data completeness, consistency, indicator definitions, interoperability, access permissions and data protection arrangements before selecting tools.
3. Who is responsible for validating AI-generated findings?
Define the roles of M&E officers, project managers, technical teams and decision-makers. Establish a clear process for checking evidence and documenting approval.
4. How will AI-supported analysis lead to management action?
Connect verified findings to management responses, assigned responsibilities, deadlines and follow-up monitoring.
5. How will we know whether AI is actually improving M&E performance?
Define measurable criteria such as reporting timeliness, data quality, verification rates, time required to identify risks, management-response completion and the usefulness of evidence for decisions. Establish a baseline and evaluate changes over time.
12. Conclusion: AI should strengthen the M&E system—not replace it
The research by Awere and Masetu contributes to the discussion about how AI-supported monitoring and evaluation could strengthen public-sector project oversight. Its findings highlight the potential relevance of predictive analytics, real-time data processing, automated tracking and knowledge management, while also drawing attention to infrastructure, institutional capacity and ethical challenges.
EvalCommunity’s interpretation is that the most important opportunity is not automation for its own sake. It is the possibility of building M&E systems that produce more timely information, support better-informed decisions, identify issues for verification and strengthen the connection between evidence and management action.
But technology alone does not create accountability. Reliable data, transparent methods, professional judgment, independent verification and clear institutional responsibility remain essential.
The EvalCommunity takeaway
Use AI to support the work of M&E professionals—not to transfer professional responsibility to a machine. The quality of an AI-enabled M&E system ultimately depends on the quality of its evidence, the competence of its people and the accountability of the institutions using it.
Original research source
Awere, B., & Masetu, D. M. (2025). “AI-Driven Monitoring and Evaluation: The Future of Transparent and Accountable Governance in Public Project Implementation.” International Journal of Research and Innovation in Applied Science (IJRIAS), Volume X, Issue X, October 2025, pp. 722–737.
Editorial note: This case study was prepared by EvalCommunity as an independent interpretation of the cited research. The original research findings are attributed to the authors. EvalCommunity’s analysis, proposed workflow, reflection questions and practical interpretations are editorial contributions and should not be attributed to the original authors.
Build your AI skills for the future of M&E
Explore practical learning opportunities to understand AI applications, strengthen analytical skills and use AI responsibly in monitoring, evaluation and international development.
