Factors Complicating the Use of AI in Detecting Disinformation

And the final speaker for this session and this day at IAMCR 2023 is Aline Duelen, whose focus is on the use of AI in combatting disinformation. Disinformation is a major problem in online communication spaces today, of course, and there is some existing research that identifies factors that play a role in detecting disinformation – but these cannot easily be automated, as their application usually requires the application of critical thinking skills. The development of more automated systems therefore requires citizen co-creation approaches.

The present project undertook such a workshop with 26 Belgian citizens. It first explored their news and engagement consumption habits; then worked with them to explore the signals they used for identifying disinformation; then worked through a scenario of using a disinformation detection system; and finally explored potential hurdles to the successful implementation of such a system.

Key platforms that emerged from the first session were Instagram, Facebook, online newspapers and apps, radio, and TV, but they were unevenly distributed through the day. Key signals for disinformation that emerged from the second session were title of the story, source, content, writing style, layout, social media interactions, as well as various general factors. Such signals have a very subjective character, of course, and cannot be automated straightforwardly.

And that’s it for today. And for those of you playing at home, the temperature in Lyon now is 34°, so time for a cold shower and/or drink, I think…