Amazon to shut down Mechanical Turk, ending the era of 'artificial artificial intelligence'

Amazon is shutting down Mechanical Turk (MTurk), the crowdworking platform that pioneered the digital gig economy by outsourcing complex tasks to human workers. The service will permanently close on September 30, 2026.

Launched in 2005, Mechanical Turk allowed businesses to post "Human Intelligence Tasks" (HITs)—small, discrete jobs that were too complex for computers to handle at the time. These tasks ranged from data labeling and audio or video transcription to conducting academic surveys. The platform's name was a nod to an 18th-century hoax involving a chess-playing robot that appeared autonomous but was actually operated by a human hidden inside.

The rise and fall of the digital crowd

For two decades, MTurk served as a precursor to modern freelance marketplaces like Upwork and Fiverr. It operated on a model where workers were paid small amounts, often just cents, to complete menial digital chores. This system provided the essential human intuition needed to train early machine learning models, effectively acting as the manual labor behind the curtain of early automation.

However, the platform's utility declined as the technology it helped build evolved. In recent years, MTurk struggled with a surge of AI bots posing as human participants to claim payments. This erosion of data quality led many academic researchers, who had long relied on the platform for behavioral studies, to abandon the service in favor of more secure alternatives.

The shift to true automation

The closure symbolizes a fundamental transition in the tech industry: the move from "human-in-the-loop" systems to true artificial intelligence. Jeff Bezos famously described the platform as "artificial artificial intelligence," acknowledging that the "intelligence" was merely a human masquerading as a machine.

Today, the tasks that once required human intuition—such as categorizing images or summarizing text—are performed more efficiently and at a lower cost by Large Language Models (LLMs) and other generative AI. The crowd-labor model for simple data processing has been rendered obsolete by the very technology it helped create.

What remains

While the platform will vanish by late 2026, the legacy of MTurk persists in the way modern AI is trained. The massive datasets created by millions of HITs provided the foundation for the current AI boom. As the service winds down, the industry will likely see a further consolidation of data labeling into specialized, high-quality RLHF (Reinforcement Learning from Human Feedback) pipelines rather than the open, low-cost crowd model MTurk championed.