AWS to Sunset Amazon Mechanical Turk After 21 Years as AI Takes Over Micro-TaskingAmazon Web Services (AWS) has officially announced plans to shut down Amazon Mechanical Turk (MTurk), its iconic crowdsourcing platform for digital micro-tasks, on September 30, 2026. The decision ends a pioneering 21-year run for the platform, following early signals last month when AWS quietly stopped accepting new customer registrations.
In a statement regarding the sunset, AWS explained that it continuously evaluates its portfolio of services and tools to align with evolving market demands. Mechanical Turk once boasted a global workforce of over 500,000 gig workers affectionately known as "Turkers" who completed human-intelligence tasks (HITs) such as image annotation, audio transcription, data tagging, and survey completion.
However, the rapid advancement of artificial intelligence and machine learning has rendered traditional human micro-tasking largely obsolete. AWS noted a steady decline in active workers on the platform as enterprise clients increasingly transitioned to automated AI models, alongside AWS's own suite of specialized, AI-driven data processing tools.
The origin of the name MTurk: Jeff Bezos, the founder of Amazon, coined the term Mechanical Turk for "artificial intelligence," naming it after an 18th-century chess robot that appeared to operate automatically but actually concealed a human chess master within. MTurk allowed software developers to embed human intelligence directly into APIs to handle tasks that computers in the mid-2000s could not solve.
In the early days of machine learning, platforms like MTurk were crucial for creating human-in-the-loop (HITL) data, providing labeled images and datasets to train early computer vision systems. Surprisingly, AI systems trained on human-labeled data eventually surpassed human speed and cost efficiency, allowing custom-built AI models to classify and label unstructured data in seconds at significantly lower cost.
The discontinuation of MTurk doesn't mean data labeling is dead, but rather evolved. Modern AI labs no longer rely on cheap, unverified crowdsourcing networks for basic labeling. Instead, cutting-edge AI development has shifted to specialized data management platforms utilizing specialized experts (e.g., software engineers, medical professionals). and linguists) to perform complex reinforcement learning from human feedback (RLHF) and fine-tuning.
Source: CNBC
AWS to Sunset Amazon Mechanical Turk After 21 Years as AI Takes Over Micro-TaskingAmazon Web Services (AWS) has officially announced plans to shut down Amazon Mechanical Turk (MTurk), its iconic crowdsourcing platform for digital micro-tasks, on September 30, 2026. The decision ends a pioneering 21-year run for the platform, following early signals last month when AWS quietly stopped accepting new customer registrations.
In a statement regarding the sunset, AWS explained that it continuously evaluates its portfolio of services and tools to align with evolving market demands. Mechanical Turk once boasted a global workforce of over 500,000 gig workers affectionately known as "Turkers" who completed human-intelligence tasks (HITs) such as image annotation, audio transcription, data tagging, and survey completion.
However, the rapid advancement of artificial intelligence and machine learning has rendered traditional human micro-tasking largely obsolete. AWS noted a steady decline in active workers on the platform as enterprise clients increasingly transitioned to automated AI models, alongside AWS's own suite of specialized, AI-driven data processing tools.
The origin of the name MTurk: Jeff Bezos, the founder of Amazon, coined the term Mechanical Turk for "artificial intelligence," naming it after an 18th-century chess robot that appeared to operate automatically but actually concealed a human chess master within. MTurk allowed software developers to embed human intelligence directly into APIs to handle tasks that computers in the mid-2000s could not solve.
In the early days of machine learning, platforms like MTurk were crucial for creating human-in-the-loop (HITL) data, providing labeled images and datasets to train early computer vision systems. Surprisingly, AI systems trained on human-labeled data eventually surpassed human speed and cost efficiency, allowing custom-built AI models to classify and label unstructured data in seconds at significantly lower cost.
The discontinuation of MTurk doesn't mean data labeling is dead, but rather evolved. Modern AI labs no longer rely on cheap, unverified crowdsourcing networks for basic labeling. Instead, cutting-edge AI development has shifted to specialized data management platforms utilizing specialized experts (e.g., software engineers, medical professionals). and linguists) to perform complex reinforcement learning from human feedback (RLHF) and fine-tuning.
Source: CNBC
Comments
Post a Comment