Managing Risk
Handling Information Biases and "Fake News" across the Digital World
Paper Presentation in a Themed Session Vered Elishar-Malka, Yaron Ariel, Dana Weimann Saks
In the current information ecosystem, it has become utterly challenging - while at the same time - extremely essential, to identify misinformation, information biases and manipulations. The 2016 U.S presidential elections, for example, were characterized with harsh accusations from both sides, blaming the media, political players, and even foreign governments, for deliberately spreading "fake news" to influence election outcomes. Content copying and editing procedures seem to become more accessible than ever, and despite the attempt to prevent it numerous "fake" copies exist. Our paper argues that it is imperative to consider more efficient ways to track content unit across different digital platforms, whether for keeping track of the agenda-setting building in Hybrid media systems or for the broader goal of keeping democratic processes uncontaminated. Weinberger (2007) suggested that an efficient way to deal with information overload is dynamic tagging, predetermined by the content creator or by post-evaluation of editors, users, and automated software. Thus, we suggest that hashtags or similar features (e.g., Barcodes) should be used to enable a reliable tracking system. As Blockchain mechanism thought us, tracking is not equal to governmental or industrial surveillance, thus, such system will make it possible for anyone of interest to identify the source, as well as the whole "life circle" of any piece of information and idea, which might have been traveling for a while across various social networks and the internet. This tracking mechanism might also drive some players away from any attempt to spread "alternative facts," lies, and biased information.
Understanding Patterns of Terrorism in India Using AI Machine Learning
Paper Presentation in a Themed Session Scott Gartner, Diane Felmlee, Rithvik Yarlagadda, Dinesh Verma
Terrorism represents an undesirable but seemingly inevitable part of the modern social landscape, and understanding terrorism dynamics can provide useful insights for developing governance structures and policies that are both more effective at reducing violence and less invasive on general society. With the tremendous increases that are happening in Artificial Intelligence capabilities in computing technology, application of AI technologies to terrorist data can yield useful insights regarding the interaction of terrorists, governance and society. Generally, there have been few applications of machine learning techniques to understanding patterns of terrorist behavior. Specifically, little work has been done to use AI to analyze terrorism patterns in India, which experiences among the world’s highest levels of terrorism. Using the Global Terrorism Database and the South Asian Terrorism Portal we apply "shallow machine learning models" that require only a modest amount of data to train themselves and can facilitate our exploration of three questions crucial to understanding the complex dynamics of terrorism, state and society: From a description of the attack can we figure out who the likely terrorist group is? Can we predict the likely location for next attack from a history of past attacks? Can we identify the principal factors that cause a city to be targeted? We believe that this project will: provide an example of socially-relevant AI research; expand our understanding of the factors that shape counterterrorism policy, and contribute to our greater recognition of the interwoven relationship of technology, knowledge and society.
