Google's TensorFlow Helps The Pentagon Analyze Drone Footage


Tech giant Google provides its TensorFlow AI framework to the Pentagon and the United States Department of Defense for use in analysis of drone imagery, according to a report from Bloomberg. The versatile machine learning framework is fully open-source and can be used by anybody, but Google is working closely with the Pentagon on this particular project as part of an ongoing effort to score more government contracts for its cloud and AI service businesses. Google’s expertise in this field could be put to work for any number of uses, but according to a Google spokesperson, all that this particular AI is being trained to do is analyze drone footage and flag certain things for human review. Furthermore, the Google spokesperson specified that TensorFlow is by no means meant for use in any kind of offensive combative capacity.

In relation to this development, Secretary of Defense James Mattis visited Alphabet to discuss possible business interests involving AI and the cloud back in August of 2017. Despite a Google spokesperson saying that the company’s technology is not being used in any sort of combat capacity, some company employees who learned of this development were reportedly not too happy about the company being involved with a military project that amounts to surveillance via drone. This comes after a meeting back in July, where Google executive Milo Medin, who is a member of a federal committee called the Defense Innovation Board, said that the government is essentially wasting valuable data that could be used to train AI programs. Presumably, this new drone program is meant to help determine what data is safe to discard, and what should be kept for training. What that training is for, exactly, was not touched upon.

Google’s founding motto of “Don’t Be Evil” and its new modus operandi of “Do the right thing” could both be argued to be violated by working with any kind of military organization. Ethics, however, can be subjective, and Google has likely taken all possibilities into consideration. It is worth noting that other tech companies have not been nearly as shy about working with the government in the past, and even the military. Though it sounds complacent, even if TensorFlow winds up being used for “evil” purposes, it could be said that Google is merely profiting from a situation that would happen with or without its involvement. Given Google’s well-known political leanings, it is also quite likely that this program could allow the company to help ensure that AI technology is not used for “evil” by inserting its own technology and policing its use.