No. 115/2023

31 HUMBOLDT KOSMOS 115/2023 is scheduled to take five years, and Roberts intends to commute between California and Germany. Most of the work is done at the computer anyway, she notes. And the tools that she developed in the past will stand her in good stead here as well. “We’re talking huge volumes of data,” says Roberts. “They can only be tackled with the help of artificial intelligence and machine learning.” PROFESSOR DR MARGARET E. ROBERTS teaches and conducts research at the Department of Political Science and the Halıcıoğlu Data Science Institute at the University of California, San Diego, United States. There, she is also co-director of the China Data Lab at the 21st Century China Center. Her book “Censored: Distraction and Diversion Inside China’s Great Firewall” (Princeton University Press, 2018) received multiple awards. In 2022, she was granted the Max Planck-Humboldt Research Award, a joint award by the Humboldt Foundation and the Max Planck Society valued at 1.5 million euros. Roberts got involved in artificial intelligence through statistics. For her final project on a machine learning course at Harvard, she developed a digital tool for topic analysis that she still uses to this day. “The analysis examines the words in the various documents and then calculates which topics are the most probable,” Roberts explains. In this way huge datasets can be evaluated with the help of a computer. Together with a fellow student, Brandon Stewart, Roberts wrote a programme which also allows you to trace how topics change in the course of time. For evaluating social media platforms, this is now worth its weight in gold because you can follow online debates using the programme without having to read every single post. According to Roberts, the fascinating thing about it is that the machine more or less does it unsupervised. This means that now, unlike in her earlier research on Chinese blog posts, neither the topics nor the valuations are determined in advance. “The tool does the filtering on its own using word probability distribution,” says Roberts. “I only interpret which topic the terms belong to afterwards.” The programme is now also used by other researchers as well as journalists for evaluating social media posts. THE CENSOR AS MODERATOR Thanks to receiving the Max Planck-Humboldt Research Award, Roberts now wants to channel these diverse findings into a completely new project: she intends to investigate how users are influenced on social media platforms and analyse their content moderation methods, which are anything but transparent. Molly Roberts will cooperate, amongst others, with the Konstanz political scientist Nils B. Weidmann, who conducts research on protest movements and civil wars as well as digital communication and political mobilisation. Weidmann originally came to Konstanz sponsored by the Humboldt Foundation: in 2012, he relocated there from Norway on the strength of a Sofja Kovalevskaja Award and used the funding for outstanding junior researchers to build up his own research project and working groups. “The emergence of social media platforms has produced a raft of new, interesting phenomena,” says Roberts and lists just some of them: disinformation, online harassment, hate speech, the influence of foreign governments on elections. She now wants to explore the tools that are necessary to guarantee freedom of information as well as the influence that internal moderation methods on social media platforms have on democracies. Her project Photos: picture alliance/AP Images (l.), Humboldt Foundation/David Ausserhofer (r.)