Employees urge Google to stop developing machine vision for Pentagon military drones



“We believe that Google should not become part of the military business. Therefore, we ask the company to cancel the Maven project, as well as publish and apply a policy that clearly states that neither Google nor its contractors will ever build military technology, ”is the beginning of the text of a collective letter to an employer that is now massively signed by Google employees. The document has already been signed by more than 3,100 people , according to the NY Times . The editors got the text (pdf) .

As it became known a month ago, the search giant signed a contract with the US Department of Defense . As part of the Maven project, by the end of 2017, it was planned to “introduce advanced computer algorithms into government platforms for recognizing objects on a large number of moving or still images”. In fact, this is a machine vision system for military UAVs.

In the letter, the Maven project is briefly described as “a custom AI engine that uses Wide Area Motion Imagery data collected by US government unmanned aerial vehicles to detect vehicles and other objects, track their movements and transfer recognition results to the Department of Defense.”

Among the ordinary employees of Google discussion of the project is at least a month . The letter says that at first they tried to discuss the problem inside the company. The staff was addressed by Google Chairman of the Board of Directors Diane Greene. She assured that Google’s technology will not “control drones” and “will not be used to launch weapons”. As if these facts are enough to calm everyone down. But no. In fact, the technology can be used for a large number of other military applications, including combat UAVs that carry lethal weapons on board. And no one guarantees that the operator of the UAV will decide to use this weapon, having received data from Google. Ordinary civilian programmers, machine vision specialists and artificial intelligence do not want to take such responsibility.

In the fall of 2017, at the summit on artificial intelligence and global security systems, Eric Schmidt expressed concern that such advanced computer vision and AI systems might “not be used correctly by the military to kill people.” According to him, in the technology industry there is some concern about this, because, generally speaking, IT companies are not at all developing machine vision systems. Many scientists and researchers from universities and departments of AI do not assume that their development can be used in this way.

The Pentagon thinks otherwise. The Maven project was launched in April 2017, and last year the Department of Defense invested a total of $ 7.4 billion in areas related to Artificial Intelligence, wrote The Wall Street Journal . Now all the work on the purchase of hardware and software, as well as the introduction of AI systems is coordinated through the Maven project. The primary task is object recognition and analytical processing of a huge number of video frames taken by reconnaissance drones. Based on this information, decisions can be made to destroy targets.

“This plan will cause irreparable damage to the brand and Google’s ability to compete in the labor market,” the letter said. - Against the backdrop of growing fears of the emergence of a biased and armed AI, Google is trying so hard to maintain public confidence. By signing this contract, Google will join the ranks of companies such as Palantir, Raytheon and General Dynamics. The argument that other companies, such as Microsoft and Amazon, are also involved in the project does not reduce the risks to Google. Google’s unique story, its motto Do not Be Evil , and the direct impact on the lives of billions of people put it outside the brackets. ”

Employees believe that it is impossible to transfer to outsiders the moral responsibility for the use of the technologies created . Google’s stated values ​​make it clear that each user trusts the company. This can never be risked. This contract jeopardizes Google’s reputation and is in direct conflict with core values. "Creating this technology to assist the US government in military intelligence - and potentially fatal outcomes - is unacceptable," the letter said.

Thus, the thousands of employees who subscribe to the letter ask the management for:

  1. Immediately cancel the project.
  2. Develop, publish and apply a clear policy that states that neither Google nor its contractors will ever create war technologies.

In response to complaints from the public, Google has unsubscribed that the Pentagon is using "free software available to any Google Cloud user." The technology "is used to identify candidate images to be studied by man and is designed to save human lives and save people from having to do very tedious work."

According to available information, in the near future, Google, Amazon and Microsoft will argue for a multi-billion dollar contract for the provision of cloud services to the US Department of Defense.

Source: https://habr.com/ru/post/411323/


All Articles