RightsCon Toronto 2018

AI, Machine Learning and Image Wars in Journalism and Human Rights (WITNESS)

Speakers: Sam Gregory

The nightmare scenario: What happens when are public sphere, our social media feeds and our messaging apps are flooded with fake images and audio that make it near impossible for us to discern truth from falsehood, partial truth from partial lie? The rapid spread and consumerization of AI, computer vision and machine learning in an array of fields is facilitating increasing degrees of search, discovery, censorship and manipulation of the vast quantities of visual data in the world. While this has tremendous potential for good, it also has the potential to exacerbate existing declines in public trust and to polarize further. One particular dimension of this is the increasing capacity to create fake audio and video and other forms of 'synthetic media' or 'deep fakes.'

As commercial and consumer-level technologies are increasingly able to use machine learning and computer vision to create convincing simulations of authentic video and audio content including fake representations of public figures and of everyday events, researchers, technologists, journalists and human rights fact-finders will have to do better at exposing fakes, utilizing new tools and challenging authoritarian and other usages that potentially present an existential threat to already declining public trust as well as reliable journalism and trustworthy human rights documentation.

In this talk I'll explain this potential threat, and what we can do about it if technologists, journalists and human rights fact-finders come together in search of pragmatic old and new solutions to fight the potential dark side of a fake audio and video future.

The future of press freedom online (Committee to Protect Journalists)

Speakers: Courtney Radsch

From counterfeit news to misinformation to undesirable communication, the "fake news" trope has rocketed to the forefront of policy agendas around the world. This session will look at how "fake news" is being used by governments around the world, from the US and the EU to Cambodia, Russia and Egypt, to name but a few, as an excuse to clamp down on legitimate speech online. This talk will explore how policy responses to fake news in a variety of countries threaten to privatize censorship and undermine the fabric of a global internet. It examines the similarities shared by several other policy issues du jour - from countering violent extremism online to dealing with "fake news" - and the pitfalls that accompany responses by policymakers fearful of cyber manipulation or eager to find cover for their own repressive policies. But it also highlights the dangers of outsourcing censorial power to unaccountable private behemoths, offering some ideas for how to ensure some level of transparency and accountability. Drawing on a range of examples from around this world, this talk will provide an evidence-based assessment of the fake news phenomenon and how the internet freedom and press freedom communities can respond.

Citizen journalism in Elections: Lessons from Uchaguzi 2017 (Ushahidi Inc)

Speakers: Angela Oduor Lungati & Monica Nthiga

In 2017, Kenya lived through one of the longest election cycles following the annulment of the August 8th Presidential results and the repeat presidential polls that were held in on 26th October, 2017. During both instances Ushahidi deployed the Uchaguzi platform for nation -wide citizen led monitoring with capacity for both citizen and trained monitor reporting via mobile app, embedded web app, SMS and for social media monitoring on Twitter and Facebook messenger bot.

Real-time reporting coupled with uncertainty caused by the volatile political environment led to an increased circulation of unverified information and underreporting of incidence.

This talk will highlight the complexities of the data aggregated from general public/ trained monitors and the wins/failures of different social media tools integrated into the Ushahidi platform, based on our experience in 2017

Blockchain as factchecker, designing a system of trust (Associate Reader Readership (research lab) Strategic Creativity Design Academy Eindhoven)

Speakers: Danielle Arets & Martijn Rademakers

How can we use blockchain -as a system of distributed trust- to collectively take more responsibility for fact checking. With that question in mind we (a multidiscipinary team of design researchers, technology experts and journalist) started, 6 months ago our design research, funded by SIDN and GOOGLE news initiative. By diving into blockchain technology and simultaneously doing qualitative user- research with the ream of a rapidly changing journalistic landscape, we came to understand that blockchain as a technology might not be the best fact checker, however as an analogue system of distributed trust- it can become very meaningful as we found out during our project "kiesfeiten.nl" (choose facts). In the running up to the Dutch City council elections we monitored the newsfeed of the 5 largest Dutch Cities together with a team of 15 peer reviewers. By using peer review in journalism, we came to understand that we can regain journalistic trust by empowering journalist in the editorial phase with input from experts. During our session we will explain our research with a public experiment. Following our 'thinking trough making approach we will do a small workshop explaining this analogue system of trust.