Skip to content

What ValueMonitor can do

Monitoring values for technology governance

To responsibly develop and introduce new technologies into society, organizations need to address moral and societal values pro-actively. Governmental organizations might want to evaluate if policy programs had a significant impact on the consideration of privacy within target groups. Research councils might want to assess whether values voiced in society for AI and Blockchain Technology (e.g. democracy and environmental sustainability) are adequately reflected in research programs.

A dilemma between resources intensive assessments and imprecise tools

To perform such value-based assessments, one could choose to perform (frequent) interviews and surveys in order to monitor values in target groups. Yet, this approach is time intensive and may not be within the budget of most organisations. Moreover, standard survey methods for monitoring values are often limited to general psychological and sociological values, while technology governance requires attention to technology-specific values (like explainability for AI).

Another possibility is to do quantitative analyses that use other sources of empirical data, for example written sources (e.g. newspaper articles, policy documents, scientific articles). A major difficulty for such an approach is that values tend to be ‘latent’ concepts. Often texts do not mention values explicitly, but they are reflected by a broader range of words that, depending on the context, refer to the value. This largely limits the effectiveness of most text analysis tools to trace values adequately.

The solution provided by ValueMonitor

ValueMonitor solves this dilemma. ValueMonitor is a text-mining tool based on probabilistic topic models. Using this tool, we define values as distribution of words instead of keywords. This means that we are able to capture values voiced in documents, even if these values are not explicitly mentioned. A quick scan of values in a technological domain can be done with ValueMonitor in about one day. Such a quick scan be the basis for more detailed investigation focusing on specific values or questions.