Select Page

AI Work powered by Blockchain

AIWORK is a blockchain network that facilitates a marketplace of crowdsourced AI Human Experts to help create, verify and validate AI data sets that make AI smarter

 

White Paper (English)White Paper ( 한국어)

How AIWORK Works

The AIWORK community will provide an open, distributed and crowd-sourced community for transcription and translation. Content owners and distributors in need of transcription and translation of their content will come into the open marketplace to transcribe and translate their content through AIWORK’s AI machine transcription and translation with distributed crowd-sourced computing resources combined with a crowd-sourced open community of transcribers and translators, all incentivized using the AIWORK Token.

Proof of Intelligence (POI)

AIWORK has developed the Proof of Intelligence (POI) blockchain protocol to verify and validate AI data sets that are created thru a crowdsourced network of human experts

AI JOB REQUESTORS

AI Job Requesters submit AI data jobs to the AIWORK Human Expert Network and pay AIWORK Tokens in exchange for the AI data work performed by AI Human Experts and validated by AI Validators. Therefore, one can think of AI Job Requestors as the Buyers in the AIWORK artificial intelligence ecosystem providing jobs and tasks that are fulfilled by AI Human Experts and validated by AI Validators.

AI HUMAN EXPERTS

AI Human Experts perform simple micro tasks to verify that the output from AI. AI Human Experts also provide additional metadata that cannot be easily generated by AI. Relevant results can be fed into the AI network to teach and improve it. AI Human Experts gain AIWORK token in exchange for well performed tasks, which must first be verified and validated by AI Validators.

AI VALIDATORS

AI Validators validate the work of AI Human Experts, staking AIWORK Token for the right to perform the validation tasks in accordance to consensus. AI Validator review the results submitted from AI Human Experts. An AI Validator may lose a portion of the staked tokens if they do not perform the tasks in accordance to consensus. AI Validator decides the next validator via psuedo-random algorithm.

Why AIWORK?

AIWORK is a decentralized, open-source blockchain protocol and ecosystem built on a consensus network of Artificial Intelligence (AI) computing resources and a community of human experts, used to generate normalized and enhanced metadata for video content. Developers can build Decentralized Applications (DApps) on top of the AIWORK protocol. Platform participants can either run AI nodes or provide expertise and be rewarded with AIWORK Tokens. 

GLOBAL NETWORK OF AI HUMAN EXPERTS

AIWORK brings together a global network of crowdsourced AI Human Experts to help generate, verify and validate AI data sets. You can think of AIWORK as the “Uber” of AI.

/

AI Match Verification and Correction

AI is not perfect, and since faces and objects can look similar, there will always be misidentifications or “false positives.” Humans, while limited by speed and dataset, are still better at comparing and identifying faces and objects. Therefore, a hybrid approach of combining AI with human verification and correction is the most optimal way of achieving AI computer vision at scale and accuracy. AIWORK will provide AI match verification and correction through the AIWORK open community. Requesters such as AI technology providers and content owners, can use AIWORK to request AI match verification and correction on their AI generated datasets. They would then pay the AIWORK open community for this match verification and correction using AIWORK Tokens.

AI Decentralized Distributed Computing Cloud

Although the power of AI is efficient and scalable, it still requires lots of computing power to process, recognition, match and tag content. One growing pain of AI is the lack of computing resources a company can have and utilize without jeopardizing cost, efficiency and scalability. In addition, the compute demand may fluctuate based on the “spikes and valleys” of videos uploaded. A decentralized and distributed AI computing cloud using crowd-sourced computing cycles is a good solution to handle the fluctuations in demand while maintaining optimal costs. This decentralized AI Compute Network has the added benefit of lower environmental impact by utilizing pre-existing and under-utilized computers around the world.

Better Content Metadata for Better Search and Discoverability

The growth of online video has grown exponentially over the last few years.  Yet, currently, video metadata is very inconsistent, incomplete and subjective. Normalized and standardized metadata generated by AIWORK will help individuals query better matching results which will benefit online video platforms providing them with more consistent, completed, normalized and standardized metadata. This will allow their content to be easily discoverable and searchable by subscribers. Online video platforms could use AIWORK in multiple ways to add or enhance their metadata. They would receive detailed scene-accurate metadata and tags for each item, paying AIWORK in tokens for the AI computation and human work.

CONTENT GRAPH

The consumption of video content has increased dramatically over the last few years due to new platforms like YouTube, Facebook, and Twitch. However, it is difficult for these platforms to manage, review and validate all the videos uploaded therefore, leading to underlying issues of content safety. Through AIWORK, videos like above can be easily detected and flagged as inappropriate by the power of AI and human experts. AIWORK’s ContentGraph score would label rejected and with a low score of safety. The scene level detection and metadata, will clearly define the inappropriate scenes for children. Meanwhile, Service Providers could use AIWORK and ContentGraph to offer content safety filters which viewers could use to search for safe and appropriate content. Contributors to the AIWORK platform, those who are providing computing resources or expertise, would be rewarded with tokens.

Roadmap

Versions

Estimated Date

Milestone

Oct 2020

Design of AIWORK architecture

0.5

July 2021

Alpha testing of AIWORK AI client

0.6

Sept 2021

Alpha testing of AIWORK Marketplace

0.7

Oct 2021

Alpha testing of the validator and AI nodes, and task allocation

0.8

Nov 2021

Release of Minimum Viable Product (MVP), including Testnet

0.9

Dec 2021

Release of beta with workflow support for human experts to verify and validate

1.0

Mar 2022

Release of V1.0 with Mainnet and workflow support for human experts to correct transcription and translation

2.0

Oct 2022

Release of V2.0 with support for Validator nodes and human experts to contribute in training the AI model