On 16 and 17 June 2021, the WeVerify project hosted a two-day workshop. Focus: the role of technology and artificial intelligence in the disinformation sphere. In particular, presentations, talks and discussions concentrated on how technology can support in data analysis and its verification, and therewith support in countering disinformation.
The workshop was organised over two days. Day 1 lasted four hours and was attended by over 140 people – virtually, of course, as the Corona pandemic still restricted travel and physical meetings.
Here, we provide a brief recap and link to video recordings of all presentations but the first session (which failed to record).
The policy perspective: approaches and initiatives
We kicked off the day with an introduction by the day’s moderator (myself – Jochen Spangenberg) and a subsequent project overview by Zlatina Marinova to the set the scene of what was to come and also provide context as well as a WeVerify perspective on things.
This was followed by a policy session (Session 1) featuring Dr. Alberto Rabacchin, Scientific Project Officer at the European Commission, and Paula Gori, Secretary-General and Project Manage of EDMO, the European Digital Media Observatory. The two speakers outlined their organisation’s respective approaches and activities, and also portrayed how they see the role of Artificial Intelligence and technology in the fight against disinformation. They furthermore provided insights into policy aspects such as the Code of Practice on Disinformation.
(Sadly, no recording is available of this session due to technical issues – hence only a screenshot of the discussion that followed).
The practical side of verification: challenges and opportunities
Session 2 then concentrated on more practical matters and dealt with actual verification work and practices in both the journalistic field and the human rights sector.
This was started off by Hazel Baker, Global Head of UGC Newsgathering at Reuters / Reuters Fact Check. Hazel gave deep insights into the news agency’s working practices, also pointing to existing challenges and pain points.
(You can view Hazel’s entire talk in the video below.)
After Hazel we were delighted to welcome Sophie Nicholson, Deputy Chief Editor – Digital Verification at AFP Agence France Press / AFP FactCheck. Sophie outlined from her perspective what challenges exist in the day-to-day business of a fact-checker, and how the French news agency operates in this space, across borders and languages.
(Again, watch the video blow – and btw: the lady pictured in the screenshot is not Sophie Nicholson :).
Session 2 was concluded by a talk of Sam Dubberley, Head of the Evidence Lab at Amnesty International’s Crisis Response Team. Sam presented a number of case studies, outlining what great assets and opportunities there are using OSINT (Open Source Intelligence) and various technical solutions, while also pointing to the shortcomings and needs of the human rights sector when it comes to checking and verifying digital content. It became obvious that there are many similarities in the verification and content analysis work of journalists and human rights workers. However, not only motives differ slightly. So do certain foci and approaches, as well as priorities.
(In the recording below, you can find out lots more.)
WeVerify outcomes and developments
In the final session of the day, session 3, speakers from the project showcased some of the work and results coming out of WeVerify.
First, Prof. Kalina Bontcheva, the scientific director of WeVerify and Head of the Natural Language Processing Unit at The University of Sheffield, presented the WeVerify verification assistant, which embeds AI technology for analysing memes, matching claims against an existing database of known fakes, and flagging content from unreliable sources.
The second talk was given by Dr. Symeon Papadopoulos, Head of the Media Verification (MeVer) team at CERTH. He presented an overview of the emerging Internet threat of so-called “Deep Fakes” and summarised the efforts towards building the WeVerify Deep Fake detection service.
Session 3 was concluded by Antoine Grégoire, OSINT investigator at the EU Disinfolab, who presented WeVerify research in analysing Coordinated Inauthentic Behaviour (CIB) and the elaboration of the so-caled CIB Detection Tree. His talk summarised the criteria used by platforms and non-platforms to define CIBs. It furthermore elaborated on a hybrid tool to help humans assisted by AI to detect and prove Coordinated Inauthentic Behaviours, all of relevance when it comes to assessing the validity of claims, for example.
We are very grateful with the way Day 1 of the workshop went. Looking back, the project team considers the effort invested very worthwhile. This is supported by audience feedback: judging from the polls conducted live during the event, and following feedback received afterwards, it seems that we “hit a spot” and provided a half-day worthwhile to attend. It featured a great variety of fantastic speakers coming from a variety of fields. So were the topics presented and approaches that were showcased. Something all participating speakers shared were their efforts to work against disinformation – doing so with different approaches and foci, and from different perspectives.
Wrapping up: all that remains to be said is a big THANK YOU by the WeVerify project team and everyone else who was involved on a variety of levels. Also thanks a lot to all participants who attended on the day and made it a success (knowing that there were more than 100 people out there at any point in time certainly added to the enthusiasm of all involved).
If you did not manage to attend on the day itself and capture the live athmosphere, you now have the opportunity to watch the video recordings here, on this page. Alternatively, you can do so on the WeVerify YouTube channel, where you also find additional recordings and videos related to the project and its activities.
And finally: as the WeVerify workshop lasted for two days, you can read and watch more on this separate page (due to be published shortly) where we provide the contents of Day 2, which focussed more on technical aspects and work carried out in and coming out of the project.
Author: Jochen Spangenberg (DW), co-author and video editor: Zlatina Marinova (Ontotext / Sirma AI)