SingularityNET’s Ben Goertzel on Building AI For All And Turning the Metaverse Into a Training Ground for AGI

Ben Goertzel is a pioneer in artificial intelligence and robotics. A leading voice in the Singularity-Transhuman community, Goertzel is also the CEO and founder of SingularityNET, a project combining artificial intelligence and blockchain to democratize access to AI. In this The Metaverse Insider exclusive, Goertzel offers insights on SingularityNet, the decentralization and democratization of AI and AGI, the Metaverse and the convergence of technologies and more!

The Metaverse Insider: What drove the creation of SingularityNet?
Ben Goertzel: I founded SingularityNet because I wanted to make sure that, when the big breakthrough from today’s narrow AI to tomorrow’s AGI (Artificial General Intelligence) occurs, it will be possible to then deploy this AGI in a decentralized and democratic way, so that it doesn’t end up controlled by some elite group like a government agency or a big company or even some clever individual hacker who believes he knows what’s best for everyone.

Since we don’t have AGI yet though, to mature such a platform the sensible course is to deploy it first for narrow AIs in various vertical markets — and then use the platform to connect narrow AIs with early stage AGIs and aim for the emergence of AGI within one’s decentralized platform via a sort of heavily engineering guided emergence.

TMI: What role did access to AI and decentralization play in its creation and development?
Ben: What triggered me to actually create SNet was the availability of Ethereum smart contracts.  Which have quite profound shortcomings and are basically obsoleted by new developments like Cardano’s Plutus smart contracts now — but they got the ball rolling in terms of making it practical to build complex “decentralized world computer” applications.

TMI: What’s the ultimate goal for your community?
Ben: I think most of the SingularityNET community is driven by a desire to see beneficial AGI occur — and a healthy mistrust of big tech and big government and other centralized authorities.  There is a lot of greed and speculative fervor and such in the blockchain world, but there is also a strong thread of fundamental faith in the power of decentralized self-organizing communities to make better — smarter and more ethical — decisions than typical top-down hierarchical organizations.

TMI: What would you list as major successes so far?
Ben: Given the primitive state of blockchain technology, just creating a decentralized AI platform that actually works and lets you run a bunch of AI agents on separate machines located anywhere on the Internet, coordinating and cooperating via the blockchain without any centralized controller or database, has been a significant technical endeavor and a major engineering success for the SNet platform team.

In terms of applications, I guess the two things that are deepest from the AI view are the Grace humanoid eldercare robot, which uses Snet platform based AI processes on the back end, and the work we’ve done using ML and OpenCog based machine reasoning for genomic data analysis.   OTOH the most downloaded app using SNet platform has I
believe been the Song Splitter, which is also interesting to me as a spare-time musician — and in fact I’m collaborating with some local friends from the Seattle area music scene on creating a SingularityNet-based music AI and tokenization project called Jam Galaxy.  So we may see a lot more happening on sNet platform in the music space.

“If it weren’t a David Vs. Goliath type battle it wouldn’t be any fun, right?”

TMI: What are some of the challenges-limitations facing the community?

The primitive nature of blockchain tech is a challenge — it’s slow, it’s expensive, it’s hard to use.   I think all these problems are solvable within the next few years — and indeed our own Hypercycle project aims to create a custom sidechain for Cardano that will be a be able to address all these issues.  But they are challenges at the
moment.

The other big challenge ties into this — it’s the super high quality and slickness of the toolsets created by centralized Big Tech.  These are competitors with trillion dollar market caps and huge teams of brilliant engineers and marketing gurus and so forth.   On the other hand we have on our side the creativity that comes from being an open, decentralized, somewhat anarchic cabal of crazy decentralized AI enthusiasts out to save the world!

If it weren’t a David Vs. Goliath type battle it wouldn’t be any fun, right?

TMI: What are some of the big opportunities for the community?

Ben: There are loads of vertical market specific opportunities.  DeFi needs AI just as badly as traditional finance, so there’s a huge market opportunity for decentralized AI.   Every human being needs to own their own medical records and control the application of AI to understand their bodies and minds — there’s another big application, right?   Everyone is now fed up with the centralized tyranny of social media companies — even the CEOs of these companies can’t usefully
defend themselves anymore — and there’s a huge decentralized-AI opportunity there…

Also the fact that SingularityNET happens to employ most of the team behind the OpenCog Hyperon AGI project — an open source initiative rebuilding the OpenCog AGI framework that’s been around since 2008 or 2001 depending how you measure it for greater scalability and usability.   Hyperon is quite different from the deep neural nets that dominate the commercial AI discussion today — it combines neural nets with symbolic reasoning and evolutionary learning and a bunch of other AI paradigms, in a common unified architecture inspired by the high level structure of human cognition.    If Hyperon proves key to the breakthrough from narrow Ai to AGI, then the close alliance between Hyperon and SingularityNET will play a pivotal role in enabling the rollout of decentralized AGI.   SingularityNET is spinning off a
project called TrueAGI aimed at enterprise deployments of Hyperon, and TrueAGI may play a major role in getting Hyperon and SNet platform working together in a way that meets the needs of industry customers
across vertical markets.

TMI: With all the chatter about the Metaverse, what’s your take on how SingularityNET fits in the Metaverse?

Ben: From a pure AGI R&D view the metaverse is an amazing opportunity — it’s basically the perfect venue for teaching young AGI systems.   A reasonably realistic metaverse with a lot of people controlling avatars interacting in a complex shared environment — this provides an incredible arena for young AGIs to learn about perception, action, space, time, language, social interaction and so forth without the cost and complexity of physical robotics.  There are some things you’ll only get from putting your AI in a physical robot — and we’re working on AGI for robotics together w. Hanson Robotics, e.g. in the Awakening Health project with the Grace humanoid robot — but there is also a lot your AGI systems can learn from a simulation world like one has in metaverses.

A lot of the interest in metaverses in the blockchain space is related to the role of payment tokens and NFTs in metaverses, of course.  This is all very interesting to SingularityNET from a business and customer acquisition viewpoint.   And they’re interesting from an AI view as well to the extent tokenomics can be used to help pull the humans and AIs involved in a metaverse into a sort of emergent non-linear-dynamical mind process — an emergent metatarsal brain.

In general I do buy the notion that processing and bandwitdh have finally reached the point to make widespread metaverse rollout finally feasible… just at it took a while for computer vision and NLP to hit the big time, but they did once the underlying hardware was cheap enough … so it is going for the metaverse, and so it will go for AGI
as well.

TMI: Do you have any misgivings about the Metaverse? Are there some threats we should watch for and try to mitigate?

Ben: Well I personally tend to spend my non work time in the outdoors by the beach or in the forest .. I have no desire to while away my off-work hours in a virtual space with a headset on.   I’m open to it in principle but in practice right now I really love nature and the richness of the perceptual surround and the physical exercise I get from moving around in it and the states of consciousness it puts me in.   I’m definitely a hard core tech geek but also a bit of a nature loving hippie.   OTOH I do spend an awful lot of my waking hours in front of the computer, and perhaps some of the activities I’m doing now in , say, Zoom calls or programming language IDEs could be done more efficiently and more enjoyably in the metaverse.  So I’d say my concern is that romping in the metaverse is used to complement romping in glorious nature, not to replace it.

At some point we may have a metaverse with the complexity and richness and human depth of physical nature — let’s all work to make it so — but until we’re there, let’s keep one foot in the Metaverse and one foot in the natural ecosystems we emerged from…

“I think the breakthrough from narrow AI to AGI will be the catalyst for all these other technologies converging and breaking through, because AGI that’s even a small integer multiple of human IQ will be able to quickly solve a lot of technology and science problems that we humans find difficult given the limitations of how our brains work.”

TMI: Just to give you a background, we started out with quantum… The Quantum Insider… then created DeepTech Insider and, now, The Metaverse Insider, so I personally feel  technological convergence is well underway. I see SingularityNET sitting at the nexus of all these technological trends — Blockchain, AI, Deep Tech, Longevity, Robotics, Metaverse, etc. — as I am someone who is (painfully) trying to navigate these technologies, could you give me some sense of how you see this all coming together? Do you have any near-term predictions, or even far-term predictions?

Ben: I’m a Singularitarian, I tend to think within the next couple decades all these technologies are going to come together in amazing ways we can’t currently foresee.  I think the breakthrough from narrow AI to AGI will be the catalyst for all these other technologies converging and breaking through, because AGI that’s even a small integer multiple of human IQ will be able to quickly solve a lot of technology and science problems that we humans find difficult given the limitations of how our brains work.

Sophia and Ben (Wikimedia Commons)

TMI: Finally… Will Sophia have an avatar?

Ben: There will be quite a few Sophia avatars running (and leaping and flying) around in the Sophiaverse metaverse we’re launching later this year, that’s for sure.   And a bunch of avatars representing Sophia’s robot brothers and sisters too.  All learning from humans, entertaining humans, working together to create new artworks some of which can be minted as NFTs.  It’s going to be pretty fun and interesting, keep an eye out at sophidverse.ai .. and it may just be
the venue in which the first real  AGI emerges, and on a decentralized platform no less … who knows?

Read More