Use Cases
Any ML practitioner can have their data labeled through HUMAN Protocol, as exemplified by the Impact Projects (see below). One significant benefit of using HUMAN is the access it offers to global workpools. Many comparative platforms do not have this. If AI products are designed to interpret the world, they must understand its different cultures. More detailed datasets means more fair and appropriate products to serve a complex world, and helps to mitigate the bias created through narrow datasets.
Another significant benefit is HUMAN Protocol’s ‘clearing rate’. This is where the Protocol, in fact, is not a competitor to existing solutions, but rather a complementary service. The Protocol can support jobs coming in via, for example, scale.api, and then distribute the work across the platforms offering the lowest fees. For example, it could send 25% of the work to scale.ai, 25% to Mturk, 25% to the HUMAN App’s CAPTCHA system, and 25% to CVAT. The user would see it all as a Web2 request over the scale API via the Job Launcher.
In this instance, an ML user can launch a request with the same code and same experience of existing vendors, but behind the scenes the Job Launcher can get them the best deal automatically. These different services would all fulfill the same smart contract.
Pyth Network, a next-generation oracle solution that aims to bring valuable financial market data to the general public, has outlined in its whitepaper a blueprint to use HUMAN Protocol’s question and answer system to provide information to pricing services. This could be very useful in the cases in which trust is required – and impartial, crowd-sourced, on-chain data can solve disputes.
This is exemplified by Hummingbot, an automated market making solution that rewards individuals for providing liquidity, which is planning to operate as a Requester on the Protocol, seeking liquidity providers. The HUMAN Protocol oracles assesses the liquidity on exchanges, track contributions of liquidity, and pay out rewards based on the requirements established by the Requester.
Factored cognition is the decomposition of a piece of work into many microtasks. It is a useful tool in machine learning. Greater minutiae (detail) in responses helps machines recognize subtle differences, and develop a more complex and comprehensive understanding.
Because the Protocol is automated, Requesters can input a generic job which can be decomposed into many arbitrary subsections. The Protocol is designed to decompose a single medical record into images, words, and numbers, and send each out to different applications/Job Exchanges. This provides a map of how the Protocol can facilitate the completion of increasingly complex work.
All parties can benefit from the opportunities supported by the Protocol. On the Requesters’ side, software takes care of the sourcing, management, quality evaluation, and payment of Workers. Workers on the other side are given new opportunities to work and earn; existing knowledge based gig workers can maximize their time by choosing how and when they contribute.
HUMAN is built to enable anyone to publish a job and have it completed by a distributed workforce; in principle, almost any kind of job can be published on the Protocol.
The global Q&A functionalities enabled by the Protocol allow companies to test product-market fit. With the right application, a Requester will be able to ask “Which company logo do you prefer?” or “Which shoe do you prefer?"
Last modified 4mo ago