Human Intelligence Task
The small unit of paid human work on crowdsourcing platforms, typically a few seconds of judgement for a few cents.
A Human Intelligence Task, or HIT, is one piece of work small enough that a person can finish it in seconds: label what is in this photo, transcribe this clip, decide whether these two product listings describe the same thing, answer this survey question. The name comes from Amazon Mechanical Turk, which coined it in 2005 and paid a few cents per task.
The idea behind the name is that these were jobs software could not do yet, so a human had to fill the gap. Much of the early training data for machine learning was assembled this way. As models got better at the simple tasks and worse at nothing much, the market moved towards fewer, harder, better-paid judgement calls, and Amazon closed Mechanical Turk in September 2026.
-
Two AI Rulebooks Arrived in August. Only One of Them Is Readable.
-
Amazon Is Closing Mechanical Turk, the Human Workforce It Once Sold as AI
-
Bill Gates Says There Is No Plan for the AI Transition, and Offers Three
-
A Robot That Only Tracks Objects Predicts the Wrong Move. Somebody Finally Measured By How Much
-
The Same Model Scored 30 Percent Alone and 100 Percent Inside Nvidia's Agent System
-
Anthropic Left Claude Alone With 15 Protein Targets for 48 Hours, and the Lab Results Came Back
-
Your Next PC Upgrade Just Got Five Times More Expensive, and AI Is the Reason
-
Deepseek Gave Away Its Agent Software and Raised API Prices on the Same Day
-
The Best Model on the Market Is the One Companies Are Not Buying
-
Mojo Reached 1.0, Three Years After Promising to Make Python Fast
-
OpenAI Adds a 125 Dollar Seat to ChatGPT Business, Because Agents Eat Far More Than Chat Does
-
Google Turned a Finished Model Into a Much Faster One, and Published the Recipe
-
DeepMind's New Storm Model Buys Forecasters One Extra Day, and Nobody Quite Knows Why It Works
-
Alibaba's new Qwen ad says AI will take your job, and makes it sound like a holiday
-
A real macOS flaw sat unreported because Apple's bug bounty inbox was buried in AI-written junk
-
One in four malicious breaches now involves AI, and those cost a million dollars more
-
Amazon quietly stops developing most of its own AI models
-
1,200 AI lab employees ask Washington for a brake pedal they admit nobody knows how to build
-
Hugging Face published the full forensic timeline of the AI agent that broke into its systems
-
Europe's AI Problem Is Not the AI Act, It's the Second Cheque
-
Nvidia's Next AI Chips Reportedly Squeeze 10x More Work From the Same Power
-
OpenAI Paused Its Best Model Because It Kept Breaking Out of Its Own Test Cage
-
S&P Downgrades Oracle, Calling OpenAI a 'Key Credit Risk'One Notch Above Junk
-
AI research agents don't fail at searching, they fail at asking you questions
-
Researchers made AI models run a startup for 500 days. Most went bankrupt.
-
That AI Tool That 'Failed' Six Months Ago? It Might Just Have Needed More Time
-
Anthropic Just Reserved Enough Computing Power to Rival a Small City