1. Home
  2. News
  3. Latest from the institute
  4. Twitter as a new data basis for researching public opinion

Who talks about Artificial Intelligence?

Twitter as a new data basis for researching public opinion

The project Opinion monitoring Artificial Intelligence [MeMo:KI] publishes network analyses on communication about artificial intelligence (AI) on Twitter for the year 2021.

10. March 2022

Visually, it becomes comprehensible which accounts of organisations and individuals express themselves most frequently on the topic of AI, who mentions each other and who is referred to most in the discourse, i.e. which voices shape the debate. In addition to a monthly population survey and a survey of media coverage, Twitter communication is the third source of data for analysing opinion formation on AI.

The word “AI” or “artificial intelligence” appeared in 151,325 tweets in 2021. The research team from Heinrich Heine University Düsseldorf can trace these tweets back to 49,749 Twitter accounts and asks the question: How do these voices relate to each other? Through mentions (so-called “@-mentions”) and the sharing of other content (“retweets”), “discourse networks” of the different accounts can be traced. Various visualisations offer users of the website the opportunity to explore the voices involved themselves: The network graphics show which accounts were most frequently referred to in the Twitter discourse on artificial intelligence in 2021. This can be supplemented by tables listing the Twitter accounts that produce the most content, as well as those accounts that refer to others and thus help them gain more visibility. The comparison over time shows how the networks change over time: The data is available for the first and second half of last year and will be added every six months in the future.

A look at the last half-year in 2021 shows that especially journalistic media, private individuals, but also accounts that are committed to teaching AI skills appear in the debate. The strength of the connections in the network graph indicates how often the accounts refer to each other.

Do the same actors have their say on Twitter as in the media coverage?

The analyses of Twitter communication show that the networks and voices involved are very diverse. For print and online reporting, it could be shown in the past that economic topics are often highlighted in journalistic terms. However, the new visualisations of reciprocal mentions in the network graphs point to a stronger mixing of the discourse between science, business, civil society and journalism.

At the current time, the representations focus on the interconnectedness of the different accounts. The analysis of the content and topics of the tweets will be carried out in the course of the year.