October 20, 2021

ModularPhonesForum

Complete News World

Protests prompt Google to abandon military AI project: Microsoft and Amazon are now in charge

Protests prompt Google to abandon military AI project: Microsoft and Amazon are now in charge

After employee protests, Google in 2018 ended a collaboration with the US Department of Defense to use artificial intelligence to “identify objects from drone images” – the so-called Maven Project.

Microsoft and Amazon Web Services (AWS) later used this to conclude contracts with the Ministry of Defense for the same task – with a total value of $50 million, approximately NOK 430 million.

Writes computer world with wallpaper in Article from Forbes Which reveals the secret contracts so far between the two tech giants and the US Department of Defense.

surveillance from the air

At a cost of $30 million, Microsoft was asked to provide tools that could analyze data from “Full Motion Video” (FMV) and “software and algorithms to automate and improve the analysis of large-scale motion image (WAMI) data.”

WAMI is a term used in the surveillance industry and refers to images taken from above – usually from a drone, aircraft or satellite – for intelligence purposes. Full motion video (FMV) also refers to video images taken from the air.

Amazon’s part of the deal is worth 20 million, and they have been asked to develop “models to detect and classify objects” from FMV and infrared data.

Microsoft was a resource for the project from 2019 to fall 2020, while AWS was a resource from sometime in the beginning of 2020 to October 2020.

Employee protests

The contracts are revealed after protests from employees at both companies.

In 2019, Microsoft employees asked the company to terminate a $480 million contract to supply the US military with augmented reality headsets. In 2018, Amazon employees asked founder and then CEO Jeff Bezos to stop the company’s development of facial recognition tools for police.

See also  Tesla Model 3 crashed into a police car - using autopilot

This article was first published in Version 2.