Research
Research
Computational research for decisions that matter.
I combine large-scale digital data, machine learning, language models, experiments, and causal inference to study communication, behavior, AI adoption, and human agency.
AI, Communication & Digital Behavior
How do organizations and people communicate when the stakes are high?
This programme treats unstructured digital behavior—text, image, audio, and video—as an early signal of how narratives, preferences, and public responses are changing. It studies brands, CEOs, institutions, and consumers across crises, controversial issues, sustainability, inclusion, and AI-mediated environments.
Questions
- Which forms of communication drive engagement, support, and action?
- How do language, imagery, timing, and messenger identity interact?
- How do digital platforms reshape public narratives around contested issues?
Methods
- NLP, transformers, multimodal classification, LLM-assisted coding
- 30M+ social-media posts across projects plus multimodal text, image, audio, and video data
- Early-signal analysis connecting unstructured behavior to later structured outcomes
- Experiments, panel models, prediction, and causal inference
CEO communication can generate stronger public engagement than brand communication during geopolitical crisis, with effects depending on framing and appeal.
Read the paper ↗Human–AI Empowerment & Decision Systems
When does AI expand capability—and when does it transfer power away from people?
This programme examines human–AI relationships through capability, creativity, autonomy, control, rights, protection, and public influence. It also develops auditable AI workflows for turning noisy digital evidence into structured decision signals.
Questions
- When does AI augment people rather than merely automate tasks?
- How do AI systems shift autonomy, ownership, and decision authority?
- How can model output be validated without treating it as ground truth?
Methods
- LLMs, embeddings, model-auditor architectures, human validation
- Event resolution and multilingual evidence pipelines
- Reliability testing, calibration, governance, and human-in-the-loop review
AI Empowerment Observatory
A live public system distinguishing AI news volume from unique developments and tracking whether those developments expand or constrain human capability and control.
Explore the Observatory ↗Methods & capabilities
A hybrid toolkit: machine learning + causal reasoning + research design.
Research collaboration