MeMo:KI

Opinion Monitor Artificial Intelligence. What do people think about artificial intelligence? How do the media report on AI?

There are many different ideas and opinions about artificial intelligence in public discourse. Decision-makers from politics, the economy and society should know what ideas people have for designing technology that is compatible with society and democracy. Opinion Monitor Artificial Intelligence will provide a long-term and reliable picture of the public debate.

In discussions on digitalization and artificial intelligence (AI), people often use individual observations from public opinion and opinion propagated in the mass media to strengthen their own position. However, this often ignores the fact that these opinions are usually not based on a systematic and continuous observation of the discourse. Moreover, a glance at various demoscopic studies shows that the findings on attitudes of people towards AI are by no means uniform. For example, the Eurobarometer (2017) indicates that the population tends to have a positive basic attitude towards AI technologies, while a survey conducted by bitkom (2018) shows that the majority of respondents believe that AI offers opportunities for innovation. However, other studies have shown that perceptions of risk predominate, both in general and in relation to specific fields of application (YouGov 2018).

There has not yet been a long-term study of public opinion in Germany that allows for developments to be compared and for an analysis of opinion dynamics to be carried out, a gap that will be closed by the project Opinion Monitor Artificial Intelligence. This project focuses on artificial intelligence as an identified key technology in the digitalization debate.

Towards a politicization of AI?

Given the significance and sheer scope of the digital transformation, the focus in politics has been shifted to the social and democratic compatibility of AI. Political institutions have positioned themselves in various papers from the national to the EU level, demanding an orientation towards people and a European path to AI that is based on ethical principles. If the political sphere takes its own demands seriously, then it requires the support of an active and enlightened civil society, one that can contribute its needs and ideas to the shaping of digitalization policy in a responsible and independent manner. Otherwise, the development and implementation of AI technologies will remain the preserve of industry, whose primary goal is to maximize profits and increase efficiency. However, to realize the demand that digitalization policy be shaped by society as a whole requires politicizing the issue, which means first of all recognizing AI and digitalization as an object of political debate and decision-making. Research from other areas such as EU policy (e.g. De Wilde 2011; Schattschneider 1957) shows that three conditions must be met:

• Polarization: Alternative positions must be represented in order to make a decision.
• Intensity: Broad media attention ensures that the issue is popularized, and the positions are propagated.
• Resonance: Once an issue finally reaches the agenda of the general public and organizations and is considered relevant for decisions, then it can be considered politicized.

An excellent example of such a development is the recent thematic and political trend of climate change. Opinion Monitor Artificial Intelligence will systematically observe public and published opinion to gauge the extent to which AI will develop in a similar way.

Using a combination of methods to achieve the goal

The study uses a combination of different empirical methods: both the tried and tested methods of research in the social sciences (such as media content analysis and surveys), and new computer-based methods.

• Relevant issues, opinions and perceptions are continuously monitored by means of a 14-day survey. At the same time, a semi-automated theme and sentiment analysis (investigation of tonality) of media coverage is carried out. This allows the mapping of public and published opinions on AI.
• The second step deepens the initial findings in a panel study (survey with later re-questioning of the same respondents) and with a manual analysis of the media content. These methods allow us to capture both more complex perceptions and interpretations or arguments better.
• This is complemented by an analysis of social media communication on the issue of AI, which aims to identify actors and issues beyond communication that is influenced by journalists, and to complete the picture of public and published opinion.

Goals of Opinion Monitor Artificial Intelligence

Besides investigating the dynamics of opinion and politicizing AI, Opinion Monitor Artificial Intelligence pursues three further goals:

• Developing and testing an instrument that can continuously monitor the discourse on AI
• Providing a broad and interested public with information on data and results so that they can achieve self-observation in their dealings with AI
• Quickly recognizing developments that require action and research in how society deals with digital transformation