Facebook Is Making a Map of Everyone in the World

And they’re using a brand new A.I. technique and a whole lot of computing power to do it.

Facebook developers taught their neural-net algorithms to recognize what a building looked like from above in this data. Then, they turned the programs loose. The software estimated urban density by the number of buildings it could see, and extrapolated the best-available population data onto the settlements.

.. These photos will be cheaper than ever before, and they’ll be more recent: Thanks to cheaper rockets and advances in small satellite-building, some companies are promising “revisit rates” of six or seven times per week.

.. This new data won’t be valuable unless companies—especially financial companies—get good at deciphering it without human labor. Already, startups like Skybox (owned by Alphabet) and Descartes Labs say they’ve made advances in pulling information out of pictures without a person sitting at the table.