Benjamin Miller

Grant to support an inter-institutional network of Atlanta-based researchers to develop and promote humanities-forward, justice-oriented approaches to artificial intelligence

Planning at Georgia State University focused on several themes, including: 1) creating and funding non-teaching-based research assistantships at the University and with local employers; 2) establishing an advising and mentoring program with industry, arts, and NGO-based alumni partners; 3) collecting alumni data; and 4) developing a digital humanities certificate and/or degree program. Georgia State University’s intention is to build an infrastructure capable of encouraging and sustaining deep transformations in the training of humanists and in the attitudes of students, faculty, and administrators.

The Mobilizing Media project, funded by the U.S. Department of Defense Minerva Initiative (2015–2020), investigated how extremist organizations—including ISIS, ISIL, al-Qaeda in the Islamic Maghreb (AQIM), and related actors—used media, language, and digital communication to recruit supporters, shape narratives, and mobilize political action. The interdisciplinary project combined political science, communication studies, ethnomusicology, area studies, language expertise, and computational methods to analyze extremist propaganda and online media ecosystems. It developed large-scale textual datasets and applied NLP and text-mining approaches to examine patterns of messaging, identity formation, persuasion, and radicalization across multilingual extremist communications.

A cross-disciplinary workshop and follow-up activities to develop a set of essays and a metadata schema to understand and describe toxic rhetoric in online spaces, with an emphasis on large-scale multiplayer computer games. A one-year collaboration and two-day working group meeting of scholars from English, Linguistics, Law, Psychology, Education, Game Studies, Communication, and Justice Studies in consultation with industry experts from game development, documenting: 1) best practices for studying and moderating online toxicity, 2) conceptual and legal frameworks for addressing online hate speech, dangerous speech, and toxic speech, 3) patterns of toxic language in digital media, 4) next steps for building a reference corpus of toxicity types and a descriptive taxonomy, and 5) a humanistic perspective on consequences of online toxicity and its moderation procedures.

This project is part of Round Two of the international Digging into Data Program and was fully funded by the following funding partners: NSF [US] (US 175,000); SSHRC [Canada] (CAN 60,700). Digging into Human Rights Violations will develop an automated reader for large text archives of human rights abuses that will reconstruct stories from fragments scattered across a collection, and an interface for navigating those stories. Unidentified victims, perpetrators, and details of human rights violations are camouflaged by the scale of archival records of witness reports; keyword based search is insufficient to extract them. Digging into Human Rights Violations will develop an automated reader for large text archives of human rights abuses that will reconstruct stories from fragments scattered across a collection, and an interface for navigating those stories. By improving on anaphora resolution techniques in Natural Language Processing for the connection of pronouns to specific nouns, this system will help researchers and courts reveal witnesses and patterns contained in their own collections. This project will read reports of historic disappearances and systemic violence in Guatemala, of abuse against women and children in Burma, and of violent incidents and rights violations in Chechnya, and provide new methods and free tools for human rights advocates.