← Projects

AI to detect online propaganda

2018

Online propaganda has shifted from simple "fake news" to highly sophisticated, AI-generated text, deepfakes and coordinated bot networks, and researchers and tech companies are deploying AI to fight back. Could the skill of spotting online propaganda qualitatively be transferred into a process a machine could do at scale?

Partner
Baidu × Tang Consulting × iCenter, Tsinghua University
Role
Design & innovation consultant
Type
Design research
Markets
🇨🇳 China
Confidential project · research detail available on request
Context

Baidu is a Chinese multinational technology company specializing in Internet services and artificial intelligence. As the Chinese equivalent of Google/Alphabet, it owns the biggest search engine in the People's Republic of China. As part of their digital innovation efforts, they approached Tang Consulting to explore new pathways to surface propaganda, with human-centered design principles in mind.

Approach

Exploratory research to train an AI system on how to learn from qualitative, context-based online propaganda. Using a global mega-event as a testbed (the 2018 World Cup in Russia), the project compared how China-accessible international media outlets across three regions covered the same event, then codified the qualitative tells of propaganda into structured, measurable signals: bias, selective omission, framing emphasis, and the sentiment polarity around national references. This qualitative review bridged to an AI system, turning a manual task into automated features an NLP model would be trained to detect.

Shared in confidence. Further detail available on request.