KeyCrime’s Brad Hathaway took a practical approach to #ResponsibleAI in his talk on December 10th, 2021 at the Internet Governance Forum Poland’s International Security & Digital Council remote hub in Rome. The exchange with Avv. Benvegnù, especially during the question and answer at the end was thought provoking.

Abstract:
Facts are the basis of criminal investigations. Facts lead to clues and clues lead to connections and, with time, enough connections and the finely tuned mind of an experienced investigator turns those connections into cleared cases. In computing terms, facts are data. But not all data is created equal in terms of “analyzability”, especially when it comes to being usable for crime analysis. Most crime data is created and archived in police IT systems in the form of video, free form text, images and audio. This is all, of course, usable data for analysis and comparison – if you’re a human investigator. If you’re a computer algorithm, though, either a human needs to “translate” this data into the granular, detailed data that analysis algorithms require or … we can get the computers to do it. This talk explores where we are, where we need to be, how to get from point A to point B and the ramifications of applying Machine Learning techniques to data in the context of public safety. The talk also acts as a call to action to police investigative units everywhere to start considering the problem and work together to imagine the solution!