Google’s termination of a multi-million-dollar contract with Appen, an Australian company that uses global contractors to train artificial intelligence, exposes the precarious employment conditions of these “ghost workers.”
The Alphabet Union, as a result, claimed that Google’s decision to sever a multi-million-dollar contract with an Australian company that assists in artificial intelligence training through contractors worldwide exposes the precarious working conditions of the invisible “ghost workers” who develop artificial intelligence tools.
Appen informed the Australian Stock Exchange on Monday, as part of a strategic review process, that Google had informed it on Saturday that it would be ending its relationship with Appen as of March 19, 2024.
Precarious conditions of AI ‘ghost workers’ revealed by Google termination of Appen contract, union says https://t.co/XNtvifgrf8
— Toby Walsh (@TobyWalsh) January 23, 2024
Ghost workers at Appen
Appen is a global employer with 1 million skilled contractors spread across 170 countries and more than 235 languages spoken.
They are known as “ghost workers,” the unseen human labor involved in training systems people use daily. They work by labeling images, text, audio, and other data to improve AI systems used by major tech companies like Google.
At Appen, employees work from home, manually reviewing content to train the AI that facilitates public searches and voice recognition apps like Google Assistant.
According to the company, its Google revenue for the 2023 fiscal year was $82.8 million (A$125 million).
Google's Breakup with Appen Leaves "Ghost Workers" Stranded: The Impact on AI Training Subcontractors – Ai Me Mate |#AI #ChatGPT #DALLE3–https://t.co/VNMJt1RGnR
— AiMeMate (@aimemate) January 23, 2024
Additionally, the company said the news is unexpected and disappointing, particularly considering the progress made against Appen’s transformation and performance in November and December 2023. However, the company added that it was focused on cost management, business turnaround, and the delivery of “high-quality AI data for its customers.”
As a result, the news caused a 40% decline in the company’s share price.
Significantly, the company said that to guarantee that their vendor operations are as efficient as possible, a Google spokesperson stated that the decision to terminate the contract” was made as part of their ongoing effort to evaluate and adjust many of their supplier partnerships across Alphabet.”
Google said, however, that the quality rating work that Appen contractors completed will be transferred to new suppliers. This means that Appen is just one of many contractors impacted by the company’s review to identify cost savings and process efficiencies throughout the board, not just AI work.
Alphabet Workers’ Union counters the move
The Alphabet Workers’ Union, the union representing Google employees in the US, however, countered that the move would have a “devastating impact” on subcontracted workers. It will, in addition, serve as a wake-up call for tech sector employees regarding the impact of AI on working people, given that Google accounts for one-third of Appen’s business revenues.
According to Toni Allen, the executive board secretary of the Alphabet Workers Union-Communications Workers of America, in tandem with subcontractors for Google, they have been a canary in the AI coalmine, calling out the precarious labor conditions they face as human workers standing between large language models and their end users.
Wired, in October, revealed that Appen contractors were paid as little as 2.2 US cents per task to train AI systems for Google and other businesses. One worker in Colombia told Wired that her monthly pay was about $US280, which is slightly less than the country’s $US285 minimum wage.
Millions of Workers Are Training AI Models for Pennies https://t.co/mLhOVeMYef pic.twitter.com/UtoJT6nMKR
— Reg Saddler (@zaibatsu) October 16, 2023
The business, however, restated earlier claims that it guarantees contractors receive more money than the minimum wage in their work area. Rates, however, differ based on the nature of the project, its complexity, and the level of experience needed.
and then