344
Productivity & Workflow355
Automation & Workflow224
Software Development251
Marketing & Growth192
AI Infrastructure & MLOps174
Writing & Content Creation203
Data & Analytics141
Design & Creative170
Photography & Imaging156
Customer Support131
Sales & Outreach125
Voice & Speech135
Education & Learning131
Operations & Admin87
As AI tools improve, some human online tasks are fading, while new paid work grows in factories and expert review jobs to train AI.
In short: As AI gets better at routine online tasks, older “crowd work” is shrinking, and human training work is shifting to the physical world and to expert checking.
Amazon says its Amazon Mechanical Turk service will be closed to new customers from July 30, 2026. Mechanical Turk is a marketplace where businesses pay many people small amounts to do small tasks online, like typing what they hear in an audio clip.
For years, this kind of work helped fill gaps where computers were not good enough. It also helped train AI by doing “data annotation,” which means labeling examples so a model can learn (like putting name tags on photos so a computer learns what a cat looks like).
The Financial Times argues that large language models, which are AI systems that write and summarize text, can now do many of the simple computer-based tasks that once went to these workers. But the need for humans has not disappeared. Instead, it is moving to new places, including factory work where people may wear cameras to record actions for training robots, and higher-paid expert roles where professionals are paid to grade AI outputs for errors.
There is also a newer problem. Some people doing paid online tasks now use AI to answer faster, which can make it harder for platforms to tell if responses come from real people. One company, Prolific, has tested checks like tracking cursor movement and using trick questions to spot AI “agents” (software that tries to act like a person).
Expect more pressure to prove when “human data” is really human, especially for surveys and research. At the same time, more training work may move into real-world settings and into jobs that require specific expertise.
Source: Financial Times