iWitness: Design For Human Rights

Maxim Dedushkov

February 10, 2021

Crimes against humanity and mass atrocity are as old as humanity itself, which is an unfortunate fact. However, in the last decade or so we have been witnessing a shift in the way that we evidence international crimes and the availability of relevant proof of human rights abuse. From the Arab Spring to Black Lives Matter, people are picking up their mobile phones documenting in real-time atrocities ranging from war crimes to civic oppression and police brutality.

How can we make sure that this type of evidence is used to its fullest potential?

How might we apply big data approaches such as collective intelligence coupled with artificial intelligence to increase criminal convictions and reduce the incidence of crime?

In this project, we explored how communities and legal experts might use big data and AI to provide a deterrent against, and increase convictions for, human rights abuses.

We worked with Dr John Stevens and Iulia Ionescu from the Royal College of Art (RCA) and Roland Harwood from Liminal to design and deliver this workshop for RCA Global Innovation Design students. To make the workshop real and uptodate special advisors helped our work: Prof Yvonne MacDermott-Rees (Professor of Law, War Crimes at Swansea University), Jeff Deutch (Syrian Archive) and Dr David Boyle (AI / ML technology at Imperial College.

Read more about the developed projects and get in touch if you are interested to discuss those!

CEPI is platform where victims can trigger audio recording by saying a predefined hot word and get insights about validity of the evidence calculated by machine learning.

J{AI}NE DOE is a tool to extend law enforcement’s ability to reach, identify and convict online sex traffickers by streamlining, automating, and scaling the deployment of AI victim personas to engage online sex traffickers.

Project X is an anonymising service that aims to enable protestors to record/publish videos of the police/authorities without fear of being targeted.