Google Paid Microworkers to Train AI Algorithms for the Controversial Project Maven, But Workers Say They were not Aware of Google’s Involvement

A new revelation from The Intercept has revealed that Google hired thousands of microworkers on the Figure Eight platform for building a military use artificial intelligence program for the Pentagon.

Figure Eight is a crowdsourcing platform, which pays gig economy workers $1 per hour for performing short tasks such as labelling images and identifying objects on different images. Google in collaboration with Pentagon worked on the controversial Project Marven where Google helped the intelligence agency for developing an AI-based software to identify humans and objects in the thousands of hours-long drone footage. When Google’s involvement in the project was discovered, the company received backlash from its own employees which accused the tech giant for violating its own policy of ‘Don’t be Evil’. Their concerns centered on the fact that their services were being used for developing war technology without their knowledge.

Google-parent Alphabet announced in June 2018 that it will not be renewing its contract with the Pentagon’s Defense Department after nearly 3000 Google employees signed on a petition protesting against the company’s involvement in the Project Maven. The recent revelation that Google used a crowdsourcing platform Figure Eight employing thousands of microworkers who helped identify objects in CAPTCHA like images. The platform helped train its AI algorithms to distinguish between surrounding objects and human targets.

The Intercept reported that the people who were involved in the crowdsourced project were absolutely ignorant of the actors who will benefit from the project as is the case with the projects on the Figure Eight platform. An Account executive at Figure Eight, Will Pleskow confirmed his company’s role in the Project Maven in an interview in September 2018. He added that their customers are provided with an anonymity option where they can choose to remain anonymous meaning that the contributors which performed the data labelling job for Google were kept in the dark about whether they are working for a regular machine learning project or a military venture.

The Intercept report has started new round of questions on ethical issues on the use of AI for warfare technologies. Although Google’s employees demanded that Google should swear off to never get involved in warfare technology projects in future, Google only announced that it won’t be renewing license without any promises to never collaborate with military. Boston College’s sociology professor Juliet Schor said that the workers have a right to know which projects they are working on and who will benefit from it especially in projects where political and moral judgment is required. She added that the right to information is the basic dimension of democracy.

Author: Sudip Saha

Sudip Saha is one of the leading tech consultants in APAC, having served at key positions in leading IT consulting firms. He has extensively written about the commercial viability and impact of next-generation technologies, most notably AI, IoT, and Big Data. Sudip’s forte lies in offering a nuanced analysis on the key developments in the tech landscape. A reputed thought-leader, his views have been published in leading publications, including CIO, ZD Net, Economic Times, and The Economist