The thought police are not inevitable

This week’s Economist carries a leader about the power of data analytics to improve government services, particularly in the area of using data to predict behaviour. Research and real-world experience suggest applications as diverse as anticipating who will break bail or which restaurants will fail hygiene inspections.

The article is very much about the “big brother” concerns that accompany work in this area, as well as the potential. It concludes that a fairly non-controversial first step would be to bring data analytics in through the education system, identifying students that need extra help at school.

It is hard to imagine things that don’t exist yet. The science fiction touchpoint the article chose was Minority Report, which apart from being a great film, is 100% a vision of a central, “superhuman” intelligence assisting the state to make decisions on its citizens’ behalf. On the contrary, I struggle to think of a film where an intelligence is portrayed as assisting individual people, without it coming in the form of cute-but-flawed robots (Robot & Frank, Wall-E, et al).

What scenarios such as Minority Report fail to anticipate is the extremely low cost of putting artificial intelligence in everybody’s pocket. It’s become a cliché to talk about modern smartphones being as powerful as supercomputers from past decades, but crucially they are also almost constantly connected to networks that allow for constant updates and improvements to the data sets and algorithms behind any such “intelligence”.

The consequence of this is that the difference between giving a government an artificial intelligence and giving its citizens their own artificial intelligence, is negligible. The big win does not come from using public funding to create the thought police, but to ensure that the digital tools made available to citizens are of the highest quality they can be. The development of tools to support individuals and their infinitely complex individual circumstances is the challenge to address.

A promising model is one of open access to data and tools, with open source projects in charge of the development behind the tooling. Funding and governance models that would allow this to work are already functioning to support other open source projects, as explored in great detail by Nadia Eghbal in her recent report, “Roads and Bridges”. Government funding can obviously make a huge difference to the development of an open source project, but by funding the ecosystem these projects live in and enforcing open access to data and tools, governments stand the best chance of putting the power to solve their own problems into the hands of its citizens.