The usage of conversational agents (CAs) such as Google Assistant and Alexa doubles every year. CAs can be used to access information and to consume news via voice by allowing users to (1) decide the news sources of their preference or (2) let the algorithm decide the news outlets and, in many cases, the snippets of information they will hear. By now, one in every five individuals asks CAs for news regularly and those numbers are expected to rise further in the upcoming years.
Traditionally, news organizations were the main news gatekeepers, in other words, they had control over the information that reached the public. Nowadays, algorithms and the institutions behind them take over this gatekeeping role and control the information the public is exposed to. If now most people, who use a CA to consume news, let the algorithms decide what information to present to them (previous studies have shown that this is what happens most often) then this can have major consequences on what citizens believe are the most important issues that public policies should address.
Under the premise that consuming information prioritized by algorithms could lead to more personalized information streams, our study tried to identify the distinct perceptions of the most important issues, such as corona cases, refugees, or cabinet formation. We compared citizens’ perceptions in the Netherlands when consuming news through conversational agents, social media, and news websites. This way we could generate an overview of users’ priorities based on the channel they used most frequently to get news (see figure 1 for the general overview of priorities).
What did we find?
Although our effects are small and mostly nonsignificant, our results show some patterns that might influence which public issues citizens believe are most important. For instance, participants consuming information through CAs are less likely to agree on the most important issues in the country in comparison to citizens consuming news through news websites or social media. This could mean that individuals, who mainly get informed through CAs, might have a slightly different view on the most relevant issues in the country than citizens using news websites. A possible explanation could be that the news that is presented through CAs is already personalized more to the user’s characteristics (e.g., age, gender, location) and their previous input. Surprisingly, citizens who mainly get the news through social media and news websites had similar issue priorities. For example, users consuming news mainly through social media and news websites described income allowance problems as one of the most important issues in the country, while users consuming news through CAs prioritized immigration as well as corona or the housing crisis as the most important issues. A possible explanation could be that since news is a social event, social media could be used as a space for discussion, creating a shared understanding of the relevant topics in the country.
Figure 1. General issue prioritization in the Netherlands, October 2021
These findings show how important it is to study the effects of CAs on news consumption. With more and more people speaking to Google Assistant or Alexa at home, it becomes increasingly crucial to understand what effects this can have on us as citizens and our shared knowledge.