Honored to be part of the panel “Political Communication in 2020: Assessing the state of the field” organized by Kim Gross @poli_com with my brilliant colleagues @prof_vaccari @ashleymuddiman @s_soroka @tkobyashi and @ekvraga to talk about AI and algorithms. #ICA20 Thread (1/10)
Her is a short summary of my main points: AI in pol com is a lot less exciting (in both a good and a bad way) than the robots roaming the parks of Singapore telling everyone to keep their distance. (2/10) https://twitter.com/Techaushad/status/1262946947696115713?s=20
We use it AI as both a tool (for example word embeddings or @annekroon et al. exciting work on bias) and object of study. As an object of our research, there has been a lot of focus on ideological bias and filter bubbles. (3/10) https://journals.sagepub.com/doi/abs/10.1177/0093650220905948
In a nutshell: evidence on functional filter bubbles is limited to experiments and fringe bubbles, to most of us algorithms actually increase diversity. (4/10) https://pure.uva.nl/ws/files/42625282/Filter_bubbles_in_the_Netherlands.pdf or @dragz: https://reutersinstitute.politics.ox.ac.uk/risj-review/truth-behind-filter-bubbles-bursting-some-myths
We are only at the beginning of understanding how algorithms structure public discourse. An important aspect is how they amplify knowledge gaps. @kthorson and @kuempelanna recent work shines a light on how the specific affordances of algorithms make the rich-get-richer in news use
Let me add to that that without access to diverse and well-researched news citizens are more vulnerable to disinformation and manipulation, and AI can be used to target exactly these profiles with such content. (6/10)
The overwhelming majority of users of AI systems feels no agency when interacting with algorithms. Without agency raising levels of literacy is almost pointless. So we need to contribute to work on designing reflective AI to create systems that create efficacy among users. (7/10)
The popular debate on AI and journalism often pitches AI as a threat because it replaces human journalists. While this might be true, there is a second threat emerging from the competition between AI and humans when it comes to creating and selecting news. (8/10)
Journalists are slowly adopting the algorithmic logic of optimization, changing journalistic norms and values to optimize audience engagement measured in click-through-rates and time spend with the articles. (9/10)
All in all, AI in pol com has many different faces and it is high time the political communication community engaged more vividly in the constructive debate to build AI that makes our lives and democracies better rather than worse. (10/10)