Amazon has announced it will shut down its Mechanical Turk platform on September 30, 2026. This 21-year-old service has connected a vast workforce with simple digital tasks known as Human Intelligence Tasks, which include labeling data, transcribing audio, and answering surveys. The platform, launched in 2005, was originally created to tackle jobs that were straightforward for humans but difficult for computers, a concept Amazon founder Jeff Bezos famously termed “artificial artificial intelligence.”

Mechanical Turk was named after an 18th-century chess-playing hoax machine that relied on a hidden human operator, reflecting the platform’s reliance on human workers for complex tasks that AI couldn’t handle at the time. At its height, the platform supported over 500,000 workers worldwide, known as turkers. Amazon used Mechanical Turk to support its sprawling online store and later positioned it as a data annotation tool for its AI development services, including the Amazon SageMaker platform.

Despite its initial success, Mechanical Turk saw a decline in recent years as artificial intelligence progressed and new competitors, like Scale AI, Mercor, and Prolific, entered the market. These companies have attracted many data workers, partly because Amazon invested less in updating Mechanical Turk. Additionally, a 2023 study found that a significant portion of workers on the platform used AI tools themselves to complete tasks, raising questions about data quality. Amazon had already stopped accepting new customers on Mechanical Turk last month, signaling the impending closure.

For many workers, Mechanical Turk offered flexible, remote income similar to gig economy jobs, providing vital earnings for individuals such as Krista Pawloski, a long-time turker and advocate for data workers. The platform’s shutdown leaves some companies scrambling to find alternatives, as insurance and travel firms have relied on it for data-driven tasks. As Mechanical Turk closes, the labor market for digital data work is expected to shift further toward newer, more integrated AI training platforms.