See below for additional RedTail data analysis added 2.24.19.
Recently, the U.S. government published two AI strategy documents, one a White House Executive Order and the other a lengthier strategy from the Pentagon. RedTail has a few things to say about both, and those comments will appear in a different publication soon. Meantime, the late-to-the-game appearance of U.S. strategies provides an opportunity to share some preliminary data evaluating what other regional and national governments have done when it comes to incorporating ethical considerations in their own AI plans.
So, what do governments talk about when they talk about AI Ethics? It varies, of course, but there definitely are issues that show up more frequently than others.
RedTail analyzed 22 non-binding government directives and reports related to AI strategies and investment published since 2016, all of which also address ethics-related concepts and principles. These included eight documents from the EU and 14 from nine countries in Europe, the UK, Asia and the Middle East: Dubai, Finland, France, Germany, India, Italy, Japan, Singapore and the UK.. Analysis of European government strategies by AccessNow was immensely helpful in this pursuit.
Here’s a quick snapshot, part of a larger overview of the AI Ethics Landscape currently in development here at RedTail Labs.
With eight documents coming out of one place — Brussels — RedTail Labs researchers worried that the EU might have had too much influence over the results. After all, the goal here was to try to understand which issues resonate most among government AI strategies that include ethical considerations. What would it look like if all eight EU government directives were consolidated into one? This meant a total of 15 documents were evaluated, with all eight from the EU combined so all issues reflected in those communications were reflected in that singular EU consolidation.
Well, it turned out there were few differences. The top six issues remain, in the same order. Workers’ Rights — RedTail’s catchall for Future of Work and labor-related topics — rose higher in prominence, surfacing it on a list of eight. Safety and Autonomous Weapons Issues tied in the last slot, each showing up in four of 15 documents, or 27%.
0 comments on “What Governments Talk about When They Talk about AI Ethics”