小公司想找到质量训练数据来构建AI算法时他们是处于严重劣势的，意识到这点之后，Chris Dawe和其创始团队成员Jesse Eisses和Lauren Verspeek决定要创建一个去中心化的模型来注释数据，其目的是帮助小公司启动自己的AI项目。就这样，Effect.AI诞生了，【它是】“一个开放、民主和去中心化的AI网络”。具体来说，团队意识到该网络的基础是全球性的微任务工作集合，它将执行机器学习任务，如数据注释、情感分析和完成调查——和亚马逊的Mechanical Turk(MTurk)一样，市场环境有利于AI解决方案和服务的交流，这也支持了Effect. AI的发展。Effect.AI是MTurk的区块链版本，它能让开发者构建AI算法并从中获益，同时它会提供一个分布式的“全球超级电脑”，为TEN协议构建的GPU重型深度学习框架提供支持，这个协议让是一种让任何人都能在网络上开发自己的AI服务的智能合约和算法堆栈。
因为区块链和AI这两种新兴技术在未来30年内都在全球经济发展上都有巨大的潜能，所以Dawe, Eissens和Verspeek能够发现这两种技术上的联系实在太正常了。尤其是，预计2030年，AI能够给全球市场带来13万亿美元的财富，而预计2027年，区块链相关的服务将占到全球GDP的10%。同时，只有区块链能够解决目前采用AI技术存在的几大难题：因为区块链网络带来去中心化的AI所有权和使用权，因此它能打开新的大型数据注释集池。同时，这一特点还能通过分布式计算机帮助支持实施AI所需的复杂的技术基础架构。这两者都是GAFA在人工智能开发方面的垄断优势。比如，因为在去中心化的微任务平台（所谓Effect Force）上没有中间人，所以其注册员工可以因为即刻从请求服务方获得报酬。反过来，相比寻求亚马逊、谷歌或者苹果的AI服务，这些服务寻求方将能通过更低廉的价格获得AI算法和解决方案。
在过去短短18个月内，Effect Newwork上已经执行了两百多万条数据注释，其中包括使用NLP的公司（如聊天机器人、翻译、情感分析）和计算机视觉（如汽车、图像标记）。Effect.AI团队已经和联合国及新加坡政府合作完成了几个试点项目，但它还有其它大胆的计划来获得全球影响。他们正在和联合国合作，在格鲁吉亚建立一个社会影响中心，其目的是“（通过招募当地员工到’Effect Force’平台上的微任务），给[发展中国家中]需要找工作的人提供就业机会，让他们能够自学并做一些有意义和影响的事情。”从Effect. AI一月的公报来看，这次合作其中一个目标是鼓励联合国“使用Effect Force来丰富和构建数据，以解决格鲁吉亚的紧迫问题，如灾害/洪水预测和救灾、可持续生活、以及其它联合国特别关注的该地区问题”。同样，Effect Force在为新加坡的信息通信媒体发展管理局（IMDA）提供地理测绘项目所需的劳动，这将需要“Effect Force员工从卫星图像中识别和分类地理细节”。因此，这些项目彰显了去中心化、基于区块链的AI系统有能力加快人道主义援助和政府的发展。
Since the legendary Go match between AlphaGo and Lee Sedol in 2016, when Google’s DeepMind product beat the 18-time world champion — marking the first time a computer beats a 9-dan professional, the world has been all the more captivated by the power of artificial intelligence. At once fascinating and formidable, machine learning has the potential to radically change human lives for the better, whether it be through aspects such as natural language processing, bioinformatics, or credit card fraud detection. For its many obvious benefits, however, AI technology has yet to see mass adoption by businesses across the board. In fact, only 21 percent of enterprises surveyed in a McKinsey report say that they have incorporated AI in parts of their business, mostly in areas such as service operations and product development. Among the top ranked reasons for this include a lack of AI talent and a lack of technological infrastructure to support AI implementation. So far, only the big players in the tech space have been able to develop solid go-to-market AI solutions, with the gamut not extending much further beyond Amazon’s Alexa, Google’s TensorFlow, and Apple’s Siri.
Realizing that small companies are placed at an acute disadvantage when it comes to finding quality training data to build AI algorithms, Chris Dawe and his founding team members Jesse Eisses and Lauren Verspeek decided to create a decentralized model for data annotations with the aim of helping companies kickstart their AI initiatives. With that, Effect.AI was born as “an open, democratic & decentralized network for artificial intelligence.” Specifically, the team recognized that the foundation of this network is a global microtasking workforce, one that will carry out machine learning tasks such as data annotation, sentiment analysis and survey completion — supported by a marketplace that facilitates the exchange of AI solutions and services, much like Amazon’s Mechanical Turk (MTurk). Offering a blockchain alternative to MTurk, Effect.AI helps developers build and monetize AI algorithms, all the while providing a distributed ‘global supercomputer’ that can power the GPU-heavy deep learning frameworks built on top of the TEN protocol, which is a tech stack of smart contracts and algorithms that allows anyone to build their own AI services on the network.
It is indeed timely that Dawe, Eissens and Verspeek have identified this natural synergy between blockchain and artificial intelligence, given that both forms of emerging technology possess huge potential to grow global economic activity in the next three decades. AI in particular is forecasted to deliver an additional USD 13 trillion to the world market by 2030, while blockchain-related services have been suggested to comprise 10 percent of global GDP by 2027. Blockchain is also uniquely positioned to address some of the biggest pain points in AI adoption at present: by allowing for the decentralized ownership of and access to AI, blockchain networks can open up new pools of large annotated datasets, as well as help power the complex technical infrastructure needed for AI implementation with distributed computation, both of which are aspects that have thus far given the likes of GAFA a monopolistic edge over AI development. For example, because there is no middleman involved in the decentralized microtasking platform (called ‘Effect Force’), its registered workers can receive instant payment from service requesters, who will in turn receive AI algorithms and solutions for a way lower price than if they were to engage Amazon, Google or Apple for AI services.
And over the span of just 18 months, more than 2 million data annotations have already been performed on the Effect Network, some of which are for companies working with NLP (e.g. chatbots, translations, sentiment analysis) and computer vision (e.g. automotive, image tagging). Having partnered with the United Nations and the Government of Singapore on several pilot projects, the Effect.AI team also has big, bold plans to make a global impact. In collaboration with the UN, they are now setting up a social impact hub in Georgia, which aims to “provide jobs to those [in developing countries] who need them (by enlisting local workers for the microtasking ‘Effect Force’ platform), allowing them to educate themselves and work on something that has meaning and impact”. According to Effect.AI’s January newsletter, one of the goals of this collaboration is to encourage the UN to “use Effect Force to enrich and structure data to solve pressing issues in Georgia, like disaster/flooding prediction and relief, sustainable living, and other matters the UN are closely monitoring in the region”. In a similar vein, ‘Effect Force’ is providing Singapore’s Info-communications Media Development Authority (IMDA) with the required labor for a geo-mapping project, which would entail “Effect Force workers identifying and categorizing geographic details from satellite images”. As such, these initiatives show that a decentralized, blockchain-based AI system has the power to accelerate developments in humanitarian aid and government.
On 21 February 2019, Effect.AI made a landmark decision to officially switch from the NEO protocol to EOSIO. Given the considerable costs that would have been involved in any strategic change, what motivated the team to nonetheless go ahead with the move? “It’s simple, building a network as complex as ours just does not work on NEO at the moment,” said Dawe. “The EOS blockchain, on the other hand, provides us with a more robust alternative, with its clearly scalable infrastructure, ability to easily iterate upon, speed, security, high TPS, and of course, its incredible community.”
Having participated in Block.one’s 2018 EOS London Hackathon as mentors, the founding team found the support to be overwhelming. “We felt the powerful strength and positivity of the EOS community [at the hackathon], and the experience solidified our decision to build on EOSIO,” added Dawe. “When we officially announced the migration of our project to EOSIO, the community response was unreal. We must have received over 200 messages from EOS developers and community members welcoming us and offering their support.” And as well they should, since the Effect.AI team is clearly working towards a worthy and innovative cause of making artificial intelligence more accessible to the world through blockchain infrastructure, whether it be for technologists, businesses or governments.
More information on Effect.AI available on https://effect.ai
Stay tuned to our EOSIO Spotlight series where we’ll highlight some of the truly exceptional projects being built on our platform. If you have a project you’d like to share with us, please email firstname.lastname@example.org.
-Developer Relations team
Block.one is a software company that is producing the EOSIO software as a free, open-source protocol. This software may, among other things, enable those who deploy it to launch a blockchain, or decentralized applications with various features. For more information, please visit https://github.com/eosio. Block.one does not provide financial support to anyone seeking to become a block producer on any version of the EOSIO platform that may be adopted or implemented.
Block.one will not be launching any of the initial public blockchains based on the EOSIO software. It will be the sole responsibility of third parties, the community, and/or those who wish to become block producers, to adopt and implement EOSIO in the manner they choose, with the features they choose, and/or providing the services they choose. Block.one does not guarantee that anyone will adopt or implement such features, or provide such services, or that the EOSIO software will be adopted and implemented in any way.
Block.one does not endorse any third party or its products or services, even if they are mentioned herein. Block.one is not responsible for any linked content or content provided by third parties, whether used directly or incorporated into this document.
Please note that the statements herein are an expression of Block.one’s vision, not a guarantee of anything. While we will try to make that vision come true, all aspects of it are subject to change in all respects at Block.one’s sole discretion. We call these “forward looking statements”, which includes statements in this document, other than statements of historical facts, such as statements regarding Block.one’s business strategy, plans, prospects, developments and objectives. These statements are only predictions and reflect Block.one’s current beliefs and expectations with respect to future events; they are based on assumptions and are subject to risk, uncertainties and change at any time.
We operate in a rapidly changing environment. New risks emerge from time to time. Given these risks and uncertainties, you are cautioned not to rely on these forward-looking statements. Actual results, performance or events may differ materially from what is predicted in the forward-looking statements. Some of the factors that could cause actual results, performance or events to differ materially from the forward-looking statements include, without limitation: market volatility; continued availability of capital, financing and personnel; product acceptance; the commercial success of any new products or technologies; competition; government regulation and laws; and general economic, market or business conditions.
All statements are valid only as of the date of first posting and Block.one is under no obligation to, and expressly disclaims any obligation to, update or alter any statements, whether as a result of new information, subsequent events or otherwise. Nothing herein constitutes technological, financial, investment, legal or other advice, either in general or with regard to any particular situation or implementation. Please consult with experts in appropriate areas before implementing or utilizing anything contained in this document.
The ideas and information expressed herein are solely those of the author and do not necessarily reflect the positions, views or advice of Block.one or any other employee of Block.one.