Masterclass: AI for Evidence: The Good, The Bad, and The Ugly - Practical Uses in the Development Sector
Master Class | Online
-
Organized by:
Sambodhi Research and Communications
About the Event
In the context of Global Evaluation Week 2026, we at Sambodhi propose a 2–3-hour masterclass that examines how artificial intelligence is currently being deployed across monitoring, evaluation, research, and learning functions in the development sector.
There is significant noise around AI right now. On one end, it is seen as a tool that can dramatically improve efficiency, speed, and scale in evidence generation and program monitoring. On the other, there are valid concerns around accuracy, bias, privacy, and the dilution of critical and methodological rigor. For professionals working in monitoring and evaluation, research, communications, and program implementation, the question is no longer whether to use AI, but how to use it responsibly and effectively.
This session will move beyond theory and showcase practical case studies of AI applications in M&E and research contexts. We will map major use cases of AI across the evaluation cycle—from data collection and cleaning to analysis, reporting, and dissemination—and critically assess where these tools perform well and where they fall short. Particular attention will be given to the risks of using AI in evidence-based sectors where credibility and trust are paramount.
The masterclass will also introduce participants to practical AI tools and demonstrate how they can support different functions across the M&E and development ecosystem.
Objectives
To identify areas within monitoring, evaluation, research, and learning where AI can improve efficiency and quality.
To examine challenges in the use of AI, including inclusivity, ethics, bias, privacy, and skill gaps.
To equip participants with practical knowledge of AI tools and frameworks for deploying them responsibly in their work.
Perspectives on AI and Trust in Evidence
1. The Good
This segment will cover areas where AI is already improving workflows in M&E and research, such as:
Survey design and instrument development
Transcription and qualitative coding
Data cleaning and analysis
Literature reviews and evidence synthesis
Report drafting and visualization
Knowledge management and dissemination
The focus will be on efficiency gains, accessibility, and enhanced productivity.
2. The Bad
We will examine common pitfalls, such as:
Hallucinations and factual inaccuracies
Shallow or generic analysis
Over-reliance on AI-generated insights
Weakening of methodological rigor
Poor interpretation of context-specific data
This section will emphasize the importance of human oversight in evaluation and research workflows.
3. The Ugly
This segment will address deeper concerns, including:
Algorithmic bias and exclusion
Data privacy and security risks
Misinformation and fabricated evidence
Ethical implications of AI-generated findings
Risks to credibility in research and evaluation contexts
These three lenses will anchor a broader discussion on how trust in evidence can be built and maintained in an AI-enabled world.
Practice Lab
The session will conclude with a hands-on exercise where participants will:
Identify 2–3 M&E, research, or programmatic tasks from their own work where AI could be deployed.
Match these tasks with relevant tools introduced during the session.
Reflect on risks and limitations associated with those tools.
Define basic checks, safeguards, and validation mechanisms to ensure responsible use.
There is significant noise around AI right now. On one end, it is seen as a tool that can dramatically improve efficiency, speed, and scale in evidence generation and program monitoring. On the other, there are valid concerns around accuracy, bias, privacy, and the dilution of critical and methodological rigor. For professionals working in monitoring and evaluation, research, communications, and program implementation, the question is no longer whether to use AI, but how to use it responsibly and effectively.
This session will move beyond theory and showcase practical case studies of AI applications in M&E and research contexts. We will map major use cases of AI across the evaluation cycle—from data collection and cleaning to analysis, reporting, and dissemination—and critically assess where these tools perform well and where they fall short. Particular attention will be given to the risks of using AI in evidence-based sectors where credibility and trust are paramount.
The masterclass will also introduce participants to practical AI tools and demonstrate how they can support different functions across the M&E and development ecosystem.
Objectives
To identify areas within monitoring, evaluation, research, and learning where AI can improve efficiency and quality.
To examine challenges in the use of AI, including inclusivity, ethics, bias, privacy, and skill gaps.
To equip participants with practical knowledge of AI tools and frameworks for deploying them responsibly in their work.
Perspectives on AI and Trust in Evidence
1. The Good
This segment will cover areas where AI is already improving workflows in M&E and research, such as:
Survey design and instrument development
Transcription and qualitative coding
Data cleaning and analysis
Literature reviews and evidence synthesis
Report drafting and visualization
Knowledge management and dissemination
The focus will be on efficiency gains, accessibility, and enhanced productivity.
2. The Bad
We will examine common pitfalls, such as:
Hallucinations and factual inaccuracies
Shallow or generic analysis
Over-reliance on AI-generated insights
Weakening of methodological rigor
Poor interpretation of context-specific data
This section will emphasize the importance of human oversight in evaluation and research workflows.
3. The Ugly
This segment will address deeper concerns, including:
Algorithmic bias and exclusion
Data privacy and security risks
Misinformation and fabricated evidence
Ethical implications of AI-generated findings
Risks to credibility in research and evaluation contexts
These three lenses will anchor a broader discussion on how trust in evidence can be built and maintained in an AI-enabled world.
Practice Lab
The session will conclude with a hands-on exercise where participants will:
Identify 2–3 M&E, research, or programmatic tasks from their own work where AI could be deployed.
Match these tasks with relevant tools introduced during the session.
Reflect on risks and limitations associated with those tools.
Define basic checks, safeguards, and validation mechanisms to ensure responsible use.
Speakers
| 名称 | 标题 | Biography |
|---|---|---|
| Dr. Anuradha Katyal | Deputy Vice President - Public Health Practice at Sambodhi | Dr. Anuradha Katyal has over 15 years of experience at the intersection of health systems research and data science, her expertise spans health financing, primary care, urban health, and service delivery. She holds dual master’s degrees in data science from IIIT - Bangalore and Liverpool John Moores University, which have prompted her to explore the application of machine learning in health systems and to work towards advancing ethical, equity-focused uses of AI in public health. |