In order to create a more inclusive training datasets, we want to incentivize the creation of a global data repository to consolidate public/private data. To achieve this, policy to mandate the sharing of data and security of said data (access tiers, etc.) would need to be enacted. Also, new methods of data collection would need to be created that incentivize the inclusion of rare data, increase accessibility of sampling sites.
Beyond policy, there could be methods to incentivize companies to pool their data by giving them access to a data repository provided they contribute, giving them access to a larger, hopefully more inclusive, dataset.