On 28-30 October, Tetra Tech’s Evaluation and Research team will join the global evaluation community at the 16th European Evaluation Biennial Conference in Lille, France, to share our perspectives and contribute to discussions on the future of evaluation.
Under the conference theme Evaluation for Vibrant Democracies, our monitoring, evaluation, research and learning specialists will examine how artificial intelligence is changing the evaluation practice. Their contributions will explore the opportunities AI offers to improve the speed, scale and consistency of analysis, while highlighting the importance of transparency, professional judgement and democratic accountability.
We will present two insightful sessions, drawing on our range of expertise, thematic knowledge and lessons from our work on the FCDO-funded Learning and Inclusion for Transformation (LIFT) Monitoring, Evaluation, Research and Learning project in Rwanda, and the study for the European Commission examining how AI is used to measure external communication.
You can find out more about the conference programme here.
Our presentations
AI vs Human Coding in Evaluation: Trade-offs, Complementarities, and the Limits of Delegation in Qualitative Analysis
The growing availability of AI-assisted qualitative analysis tools raises a practical and methodological question for evaluation: what can be reliably delegated to AI, and what must remain under professional judgment? This paper presents learning from a structured pilot conducted within the UK-funded Learning and Inclusion for Transformation (LIFT) Monitoring, Evaluation, Research and Learning (MERL) project in Rwanda. The examined the trade-offs and complementarities between AI-assisted qualitative coding and manual coding conducted by trained researchers in an education study involving interviews and classroom observations.
The study was designed as a comparative exercise to assess three questions: (1) how AI-generated coding structures differ from a human-developed coding framework; (2) the extent to which AI can reproduce human coding decisions when applying the same framework; and (3) what types of analytical tasks AI may be better or less suited to within qualitative workflows. A subset of pilot transcripts was coded manually using a codebook anchored in research questions on early-grade teaching, gender-responsive pedagogy, and inclusive education. The same transcripts were analysed using the AI platform in two ways: by applying the human-developed framework, and by examining the inductive coding structure generated by the tool itself.
28 October, 13:00 – 14:30 CET
Presenter: Youngjin Kim
Using AI to Evaluate Public Communication: Practical Implications for Democratic Accountability
Public institutions are increasingly adopting artificial intelligence (AI) tools to monitor and evaluate their external communication. This paper presents findings from an ongoing study for the European Commission examining how AI is used to measure communication performance, focusing on tools such as automated sentiment analysis, topic detection, and engagement metrics.
The paper documents how these tools are currently applied in practice, including: (1) real-time monitoring of public reactions across platforms, (2) large-scale analysis of text data to identify dominant narratives, and (3) performance benchmarking based on engagement indicators. It assesses their added value compared to traditional evaluation approaches, particularly in terms of speed, scale, and consistency.
At the same time, the paper identifies three concrete limitations observed in current practice:
- Systematic bias in sentiment and topic classification, especially across languages and political contexts
- Over-reliance on engagement metrics as proxies for communication effectiveness
- Limited transparency of AI models, making it difficult to validate or challenge findings
These limitations have direct implications for democratic accountability, as they affect how public feedback is interpreted and used in decision-making.
Based on these findings, the paper proposes practical adjustments for evaluators using AI in communication assessment, including: combining AI outputs with qualitative validation, diversifying indicators beyond engagement metrics, and introducing minimum transparency requirements for AI-supported evaluation.
The paper contributes by translating emerging AI practices into concrete methodological guidance for evaluators working in public communication, with the aim of supporting more reliable and accountable evaluation in democratic contexts.
30 October, 14:00-14:20 CET
Presenters: Aimilina Sarafi and Aurélie Heetman
Meet our team
Youngjin Kim
Youngjin Kim is a Senior Consultant at Tetra Tech International Development in London, specialising in Monitoring, Evaluation and Learning (MEL) for international development programmes.
Her work focuses on mixed-methods research, with a growing interest in integrating AI-assisted approaches into evaluation practice. In her recent work on FCDO’s Learning and Inclusion for Transformation (LIFT) and Monitoring, Evaluation, Research and Learning for Yemen (MERLY) programmes, she has explored how AI can be used to make MEL processes more effective.
Aimilina Sarafi
Aimilina Sarafi is a Consultant at Tetra Tech International Development’s Research and Evaluation EU Public Policy team with thematic expertise in fundamental rights and criminal justice, EU law and policy.
Aimilina has supported multiple studies, evaluations and impact assessments for clients from the EU Institutions, including the European Union Agency for Fundamental Rights (FRA), the European Public Prosecutor’s Office (EPPO) and the European Anti-Fraud Office (OLAF), the European Commission’s Directorate-General for Justice and Consumers (DG JUST), DG Communications (COMM) and others.
Her work includes conducting legal research and analysis, drafting project findings, engaging EU staff, civil society actors and national authorities in stakeholder consultations, such as interviews with high-level EU officials and thematic human rights focus groups, , as well as drafting in-depth case studies on topics of her expertise.
Aurélie Heetman
Aurélie Heetman is a Principal Consultant within the EU research and evaluation practice at Tetra Tech International Development, leading the EU justice and home affairs policy portfolio.
For the past 14 years, she has led and supported evaluations, impact assessments and other Better Regulation research projects. These projects have involved the collection and analysis of qualitative and quantitative data across the EU-27 and have been carried out for EU institutions, bodies and agencies, including DG JUST, DG HOME, DG CNECT, DG RTD, DG TAXUD, the EEAS, Eurojust, the EPPO, EIGE, the FPI, the FRA and Frontex, as well as the European Parliament.
She holds an LLM in International and European Law from the University of Amsterdam.