SingularityNET — AGI of the People, by the People and for the People (and the Robots!)
AI is now Big Business and Big Science, and it’s getting smarter every year. But whose values is AI serving? As it provides more and more economic value to companies and consumers, can we ensure that AI works toward the common good?
This is the raison d’etre of SingularityNET — a project that was conceived last year and got off the ground this year, by myself and a bunch of friends who are AI researchers, futurists, roboticists and cryptocurrency hackers, spread across every continent save Antarctica. All of us on the SingularityNET team are joined together by a common goal of making AI more democratic and more broadly beneficial — and in the process giving it more general intelligence as well.
We are energized by the avenues the modern economy provides for radical transformational change. Certain technologies seem to rise almost out of nowhere, and suddenly become prominent or even dominant factors in the global economy. Web search, e-commerce, smartphones, ride-sharing, self-driving cars…. We are pushing to create the next entry in this list, with the potential to be the most transformative of them all: a decentralized, open market for AIs, enabling AI modules created by developers around the world to network together.
As many readers will know, I have been working on various aspects of AI for decades — and in particular have spent a lot of time endeavoring to push the AI research world toward “Artificial General Intelligence.” In 2011 I relocated from Washington DC, where I was doing AGI research and creating AI tools for US government agencies, to Hong Kong where I teamed up with my friend David Hanson (a world class artist and roboticist) to create the world’s most realistic and emotionally expressive humanoid robot, Sophia. Sophia has become a first class robot media star, appearing on the Tonight Show and garnering a billion+ views online.
David and I joined forces based on a shared vision to create a global AI mind cloud that would live in the Internet and supply general intelligence to billions of robots, devices and software programs helping people around the world. And this AI mind cloud should be free and open source, so it can benefit everyone, and include contributions from everyone — including people in the developing world, for instance in Ethiopia where in 2013 I co-founded the country’s first AI and robotics development shop, iCog Labs.
AI in 2017: Triumphs, Limitations Risks
The AI scene into which the SingularityNET team is injecting itself is an unpredecentedly exciting one. But it also has many complications that aren’t apparent to the non-technical observer.
First there’s the distinction between narrow AI and AGI — Artificial General Intelligence — which I’ve emphasized time and time again in my talks and writings. Yes — AI is pervading every area of industry, and becoming a major funding priority for governments everywhere. But today’s AI still has profound limitations relative to science fictional visions: it basically consists of an array of AI programs carrying out useful narrow functions for specific applications. These narrow programs are not effectively integrated into overall AI systems with general-purpose intelligence like that of humans.
It seems clear that the next big step in the evolution of AI is going to be the transition from AI to AGI. But exactly how this transition will unfold, industry and research experts disagree. Google Deep Mind, founded by neuroscientist Demis Hassabis along with my former employee Shane Legg and Mustafa Suleyman, is aiming to create AGI via emulating the human brain at an appropriate level of abstraction. Deep Mind numbered under 100 staff when Google purchased them for $500M in 2014, and now are many times that size; arguably they are the largest and best funded group on the planet focused toward Artificial General Intelligence.
The OpenCog open-source AGI platform I co-founded in 2008 takes a more inclusive approach, including other AI approaches such as symbolic logic and evolutionary learning alongside brain models like neural networks. SingularityNET represents an even broader and more inclusive approach: don’t rely on any one AI algorithm, but create an open market in which various AI algorithms such as neural networks and OpenCog can cooperate together, forming new patterns of emergent intelligence based on their interactions with customers and with each other.
There are also major questions regarding the risks and rewards of AI. Elon Musk and others have highlighted the long-term risks of superhuman AI annihilating humanity. Others have focused on nearer-term economic and social risks. In a world where nearly all AI startups are rapidly gobbled up by mega-corporations, and an increasing number of businesses are replacing human staff with AI and robots, will AI be used for the common good, or will it serve as a tool to increase corporate domination and exacerbate economic inequality?
My view tends to be that, if we want to militate toward beneficial AI outcomes, we should be proactive and create applications that deploy AI for human good. Toward this end I’ve been working on applying AI to the genomics of aging and disease, to eldercare and education, and to helping people meditate and achieve deeper states of consciousness. And there are plenty of other beneficial AI projects besides my own out there. But all in all they are a drop in the bucket compared to applications of AI to military defense/offense, banking or advertising.
These factors frame the key challenge faced by the SingularityNET team:In a world where the bulk of AI funding lives in government agencies, venture capitalists and megacorporations, how to get the needed brainpower, programming-hours, momentum and hardware behind an open-source initiative to bring about artificial general intelligence by the people, of the people and for the people?
AI Meets the Blockchain
In 2016 David Hanson and I realized that cryptocurrency might be a large part of the answer to making our vision of a democratic, beneficial AGI mind cloud a reality.
Both David and I had been following the crypto space with excitement for a while. Even back before Ethereum made smart contracts a practical tool, it had been evident to us that blockchain and associated technology could provide a powerful, flexible platform for creating a decentralized, globally distributed AI mind. At the Artificial General Intelligence 2015 conference in Berlin we had a session on AGI, blockchain and the emerging global brain; and in 2016 I co-organized a workshop on similar themes at the Global Brain Institute in Brussels.
What became clearer and clearer to us as 2016 progressed was that crypto token sale, as a funding mechanism, is fundamentally more democratic than typical routes like venture capital or government grants. It allows one to leverage the decentralized “crowdfunding” spirit that Kickstarter had in the early days before it became so heavily used as a marketing tool for VC-funded projects — but with tremendously more flexibility than Kickstarter-like sites or anything else around. Smart contracts make a technical reality out of the insight that — as Charles Stross dramatized in his novel Accelerando — companies are just another form of software, and relations between stakeholders can be much more customizable than traditional business has ever recognized.
Then David introduced me to AI/blockchain pioneer John Clippinger, and John introduced me to the 24 year old Italian blockchain guru Simone’ Giacomelli and his colleagues, with whom he’d worked on the early stages of the Swytch token. It didn’t take long for me, Simone’ and David to figure out how to bring blockchain, AI and robotics together to design a decentralized platform for an AI mind cloud.
In summer 2017, Simone’ brought in a modest amount of private funding from some of his colleagues in the crypto space, and our team began building the SingularityNET software, and working the crypto conference circuit building momentum for their Initial Coin Offering. The vision: an open platform in which any AI developer on the planet can install their software, enabling it to reach any AI user on the planet. A platform in which AIs carrying out specific functions can work together seamlessly, helping diverse users in various ways and coordinating together into AI networks of increasingly general intelligence.
If SingularityNET realizes even a fraction of its ambition, it will be a formidable competitor to current corporate cloud-AI providers like IBM Bluemix, Amazon Web Services, Microsoft Azure and Google Cloud. No company, no matter how large or smart, can provide as much AI cleverness as an energized, decentralized community of AI developers in every country around the world. Already there is a tremendous amount of powerful open source AI code in GitHub and similar repositories, but it’s not easily accessible to customers. Putting AI code in GitHub makes it accessible to AI developers with sufficient time and expertise; putting AI code in the SingularityNET makes it available for customers around the world to use, and for interaction with other AIs in complex multi-AI networks with their own emergent intelligence. SingularityNET has the potential to profit tremendously from the now-universal corporate need for online AI services, to leverage the usage patterns of customers to drive the emergence of general intelligence, to direct the profit thus generated to apply AI for global good.
Building a distributed blockchain based AI mind cloud is no small task, but I’m confident our team is up to it. Our AI team’s skills have been honed via building the OpenCog open-source general intelligence platform and the Hanson robots, and doing AI contracts for a long list of clients including Cisco, Huawei, DARPA and IARPA. Our blockchain team has built up their expertise via working on Swytch, Heavycoin, Eurocoin and a bunch of other blockchain systems and tokens.
All in all, the trickiest obstacles we face are probably not technical ones. Looking at it objectively, I have to see our choice to frame the project as a cryptocurrency-based “decentralized autonomous organization” as gutsy and risky, as well as incredibly promising. The cryptocurrency markets are tremendously volatile, and regulators around the world are beginning to take notice, most strikingly in the case of China’s recent decision to ban cryptocurrency exchange and Bitcoin mining altogether.
Some observers foresee the “ICO window” slamming shut before long, due to antsy regulators and defensive, threatened banks and VCs. Some see this recent spate of regulatory noises as the beginning of the end for the crypto boom. But I think this is very shortsighted.
My view is more like that of Hexayurt Capital founder Vinay Gupta, whose presentation I enjoyed recently at the September 2017 Blockchain Forum in London. Vinay was pushing the emerging crypto economy as the solution to modern society’s impotence at funding technologically risky Big Science and Big Engineering projects. “How can we fund the Singularity?” he asked. He proposed ICOs as the answer.
The global reserves of Bitcoin and Ethereum wealth are both deep and steadily increasing, and the true believers like Vinay Gupta view blockchain not merely as an avenue to make cryptocurrency profits, but more fundamentally as a way to build a new world economy outside the control of governments and multinational corporations. This vision of cryptocurrency as politically revolutionary fits in perfectly with the SingularityNET team’s goal of creating decentralized general intelligence for universal benefit. As we endeavor to bring their dream to reality, we will be navigating a technological, business, economic, political and regulatory landscape that is both more complex and more rapidly shifting than anything humanity has ever seen. Creating SingularityNET is going to be a hell of a ride — and a big success in this endeavor will, I believe, significantly increase the odds of a positive Singularity.