Stay up to date with the HUMAN project through our bi-monthly newsletter.
Introduction to HUMAN Protocol
HUMAN Protocol creates a blockchain based, two sided marketplace. On one side, it enables requests to label large volumes of data. On the other, it allows for the workers who complete this work to be evaluated and remunerated.
Artificial Intelligence (AI) is becoming more powerful by the day. Whether it is a virtual assistant responding to our voice, the traffic predictions of GPS, facial recognition, or customer support, there is no doubting the reach and utility of machine learning (ML) in our daily lives.
In the growing conversation around AI, there is a common oversight: the human.
There are many things machines can do, but they do not learn by themselves. An AI cannot accurately distinguish a traffic light from a tree, or decipher messy handwritten notes, without being “taught” by data labeled by humans.
When a human completes a task via the Protocol, this creates a label to annotate the data. When we begin to label data, we are creating associations; it is these associations that ultimately power machine learning.
HUMAN Protocol is built on blockchain technology, which facilitates an open-source, secure transaction system for the compensation of human work. The nature of this work implies high-volume, low-value transactions, as well as the potential for one-to-many micropayments.
Because HUMAN Protocol is built on the blockchain, it can apply Smart Contracts to execute high-volume micropayments across the globe. The openly verifiable smart contracts hold all funds in escrow, so there is peace of mind for all parties involved.
The efficiency of HUMAN Protocol means we can pass cost savings onto the Requesters of work, charging them less than alternative, centralized marketplaces.
The Protocol allows arbitrary job types to be defined and published. These job types can be either simple, in which case they may be executed directly, or complex, in which case they may be factored into simpler tasks and then reassembled into the final result.
Factored Cognition simply means to disassemble a task into its component parts. For example, consider a request for human evaluation of an entire page of a mathematical textbook. Factored cognition can be applied to break the page into smaller coherent parts: the numbers, the formula, the pictures, the text. With these parts, the actors can now distribute these sub-jobs across the entire network of HUMAN exchanges publishing those job types, and therefore match specific work to specific agents based on real time cost, suitability, and reputation. The sub-results can then be combined.
The first application running on HUMAN Protocol is hCaptcha, an online security service that distinguishes humans from bots via simple questions, e.g. selecting which image squares contain a dog. hCaptcha uses HUMAN Protocol to assist in managing the evaluation and compensation of the end users who carry out this work, in order to compensate the web sites they visit when their answers are correct.
Requesters — the Requester submits a request for human evaluation work. Attached to the request is a bounty, which is a sum of HUMAN Tokens (HMT) held in escrow, to be released to the Agent when the task is complete.
Exchanges — the Exchange plays an intermediary matching role between the Requester and the Agents. The Protocol provides the Exchange with precise blocks of work, which then allows the Exchange to classify and bundle the tasks to create a match with the agents. The Exchange also manages the bidding for work.
Workers — the Workers perform the work as specified by the Requester. A Worker can be an individual, a website, or derive from labor pools.
By its nature, the blockchain can only retrieve information within its own network, or on-chain. To distinguish if a task is complete, it must have a reliable way of connecting to external information, off-chain. For example, when a human is asked to select which squares of an image contain a traffic light, this work happens off the Blockchain. The Blockchain therefore does not know whether the smart-contract has been fulfilled. Enter the Oracles.
Recording Oracle — these collect answers and provide a rolling evaluation of answer quality. They pass on their data to the Reputation Oracle.
Reputation Oracle — receive the answers, and evaluate the final quality of the answer. This feeds back into the network to trigger the release of bounties to the agents.
hCaptcha is only the first application on the Protocol. The Protocol is designed to allow arbitrary task types to be designed, published, and purchased, allowing large groups of workers (“labor pools”) to cooperate with requesters of work without requiring high trust on both sides.
The Foundation’s role is to help drive adoption of HUMAN Protocol technology and ecosystem, and we look forward to telling you more in the coming days.
The HUMAN Protocol Foundation makes no representation, warranty, or undertaking, express or implied, as to the accuracy, reliability, completeness, or reasonableness of the information contained here. Any assumptions, opinions, and estimations expressed constitute the HUMAN Protocol Foundation’s judgment as of the time of publishing and are subject to change without notice. Any projection contained within the information presented here is based on a number of assumptions, and there can be no guarantee that any projected outcomes will be achieved.