I often hear people ask on the subject of AI: “Why are you doing this when Facebook, Google etc have an unfair advantage because of the amount of data they have?” As Dennis Mortensen from x.ai put it, its one thing have data sets, another having ready to use data sets on which you can actually do some work with. Incumbents can make their data sets useful, but this requires the same, if not more, amount of work it takes a startup to build a useful data set.