University of Amsterdam · Amsterdam Business School · 2026

PhD in Digital Marketing & Applied AI

Voices in Action: AI-Powered Insights from Corporate Messaging During Societal Crises

The dissertation examined how organizations communicate during high-stakes societal moments—and how those messages shape public engagement and action. It combined large-scale computational analysis with theory-driven human validation and controlled experiments.

Kedma Hamelberg presenting her PhD research at the University of Amsterdam Agnietenkapel
UvA University of Amsterdam ABS Amsterdam Business School 30 Jun 2026 public doctoral defense Method computational + experimental triangulation

The research in one minute

Three contexts. One question: whose voice and message design shape what happens next?

Study 01 · Geopolitical crisis

CEO voice or brand voice?

During the Russia–Ukraine war, CEO and brand accounts behaved as different communication tools. CEO voices generated stronger engagement, while effects on action depended on framing and psychological distance.

Study 02 · DEI

When should brands stand firm?

Public response depended not only on whether brand and citizen positions aligned, but on the direction of that alignment. Pro-DEI alignment was rewarded; anti-DEI alignment could be punished.

Study 03 · SDGs & multimodal meaning

Can a tiny visual cue change engagement?

When brands communicated about socially distant causes, engagement fell. Symbolic emojis that expanded meaning could narrow that gap; repetitive icons did not produce the same effect.

Methodological contribution

AI scaled the measurement. Humans anchored the meaning.

Across the empirical chapters, computational methods helped identify and classify more than 250,000 social-media posts against theory-driven rubrics. The categorical decisions were not treated as model truth: the coding logic was developed from theory and validated with human raters.

Field evidence was then triangulated with controlled experiments. This combination of computational scale, human-validated measurement, and experimental replication is the methodological signature that connects the dissertation to my current applied-AI work.

Academic validation

How the committee described the contribution.

“I really like the systematic analysis of social media data because it's novel but also that it's combined with your experimental studies to address mechanisms.”

Prof. dr. Corine Boon
University of Amsterdam

“It stands for everything that the business school is standing for. It's very innovative. You took some risk using, at that time, very modern tools like large language models.”

Prof. dr. Marc Salomon
Dean, Amsterdam Business School

Doctoral examination

Defense committee

Prof. dr. Ronald Does · Chair Prof. dr. Marc Salomon · Amsterdam Business School Prof. dr. Corine Boon · University of Amsterdam Prof. dr. Raoul Kübler · ESSEC Business School Prof. dr. Francesca Sotgiu · Vrije Universiteit Amsterdam Dr. Bernadette van Ewijk · University of Amsterdam

From dissertation to current work

The PhD is a foundation for what comes next.

The same principles now extend into multimodal digital behavior, human–AI systems, auditable LLM workflows, decision intelligence, and the AI Empowerment Observatory.

About the Dutch public PhD defense format

Dutch doctoral defenses are formal public academic examinations. At the University of Amsterdam, the candidate presents the work publicly and is questioned by the doctoral committee during a timed formal opposition. My defense took place at the historic Agnietenkapel in Amsterdam.