The Pentagon is using Google’s TensorFlow on Project Maven drones

Google’s TensorFlow AI framework is now being employed by the US Department of Defense’s Project Maven. The machine learning system is being used to analyze and detect subjects of interest in the footage provided by American drones around the globe, in order to relieve human military staff of their tasks related to analysis and reviews.

This controversial collaboration with Google comes as part of the Department of Defense’s $7.4 billion invested in AI data analysis, in their efforts to fight against the Islamic State. However, Google’s assistance to Project Maven with both cloud and AI services has been highly controversial internally and externally. According to Gizmodo, several employees have been outraged.

Even though Google hasn’t been the only company to provide assistance to the Department of Defense, with both Amazon and Microsoft providing cloud services to military missions for a while, it is the use of proprietary artificial intelligence that continuously concerns workers, sparking internal debates at the company.

TensorFlow AI and Algorithmic Warfare

Google Brain developed TensorFlow as an open source software for the programming of a wide variety of tasks through machine learning neural networks that can be easily run on CPUs. This framework enables a higher analysis capability lightening heavy data full tasks for personnel, through a higher image recognition, API models, and cloud services.

Through this technology implemented in drones, it can easily extract the most useful footage collected in order to spare tragic mistakes. It has been stated that this technology may help tell the difference between civilians and military camps or hideouts, which is the main goal of Algorithmic warfare.

Google thinks AI can save lives in armed conflicts

Google, on the other hand, has delivered calm responses claiming that it is “naturally concerning” to adopt machine learning technologies in warfare. Internal debates at the company are not only recurrent among employees but also on the high office as Google will continue to establish safeguard policies for military AI employment.

The company has also stated that the implementation of advanced technology might prove effective on warfare on the safety aspect, as object recognition technology may help save lives by correcting miscalculations and assumptions made by human personnel.

A Google spokesperson said “This specific project is a pilot with the Department of Defense, to provide open source TensorFlow APIs that can assist in object recognition on unclassified data. The technology flags images for human review, and is for non-offensive uses only.”

Source: Center for a New American Security