Google’s TensorFlow software kits are being used by the U.S. military to both study and intepret drone footage in hopes of both training and creating its own drone artificial intelligence (AI).
Known as Project Maven, the objective of the program (at least, from Google’s side) is to train the U.S. drones to identify specific objects in drone footage; this is where Google’s TensorFlow comes in. The programming kit will help the drone AI in question to sift through what Engadget describes as a “glut of drone footage,” as quickly and as efficiently as possible.
Now, Google has stressed that its open-source AI program that is helping the drone AI is strictly for ‘non-offensive use’ and that it will not help future U.S. drones with their long-range striking capabilities.
Despite its open-source nature, however, the action raised moral concerns among Google staff, many of which felt angry at the fact that the software was being used by a branch of the U.S. military to better train an automated surveillance aircraft that has been effectively used to conduct precision strikes at targets deemed a threat.