Thanks for this post,
While ML has many potentials, I believe we should also adopt strong ethical principals within the WBG to ensure those technologies are not used to reinforce power inequalities, in particular when those technologies are used to capture data from the most vulnerable people.
When you use ML services from the cloud, you do not have full control and certainty over the data you share with the company, and - as we have seen - you do no have certainty on how the data could be re-used.
That being said, I believe it is the role of an institution such as the WBG to explore new technologies for development and I hire experts for that, but we should also be very careful with AI applied to surveys or identification as the risks of backlash are very high. This means having not only data experts but also legal and ethical experts, not the typical lawyers, but the one who are willing to innovate in an responsible manner.
Because the main issue is not about scaling up those technologies - in the end private sector can do it - but about ensuring we are not creating a new bigger threats that would undermine other efforts and that could also have an impact money-wise.
In an way we need to think about Environment and Safeguard Policies applied to data.
All the best