How do we make sense of changing human social norms? Ask a bot, of course | Torsten Bell

Sceptics should note that machine learning and big data analysis is supporting social science progress. Take the debate about cultural norms, where some emphasise the persistence of views passed between generations, while others argue ideas converge between places over time. We struggle to know which view is accurate (surveys of public attitudes are relatively recent or only national).

But fear not. Recent research used machine learning to examine 193m pages of local US newspapers from the past 160 years to measure attitudes towards women and how they vary across the US. For example, it measured whether the language in those pages associates men with careers and women with caring, or whether feminism is seen as dangerous extremism or desirable equality campaigning in different places.

Torsten Bell is chief executive of the Resolution Foundation. Read more at resolutionfoundation.org