The SingularityNET Foundation is currently shifting gears from the early phase of its growth cycle, focused on building software, essential partnerships and establishing brand reach, to what is being conceived as Phase Two of the project’s development.
The following is a breakdown of some of the most pertinent questions asked surrounding SingularityNET Phase Two.
There are aspects of the SingularityNET Phase Two plan (specifically the launching of new AGI-ADA tokens that would expand the total token supply beyond the current 1 billion AGI tokens) that will require a token-holder vote for enaction – The proposal will be put to a vote during the period Feb 3-5 2021 and is linked here.
Q: Do you really think you’re going to be able to make serious headway toward decentralized Artificial General Intelligence during the 5 year period that your SingularityNET Phase Two proposal suggests?
A: Absolutely yes. This is the right time to be doing AGI – which is why Microsoft, Google, Facebook and others are putting billions of dollars into it. They know an AGI breakthrough is coming and they don’t want to be left behind. But our aim with SingularityNET Phase Two is to make sure they are in fact left behind. Due to our work over the past few years, have the software platform needed for democratic and decentralized AGI – and we have an awful lot of the ingredients needed to build both powerful AGI and a lot of useful proto-AGI and narrow-AI applications on this platform. We “merely” need to put all the pieces together (OK and improve the scalability and usability of some of the pieces along the way). Which of course will be a lot of work. But we have the core ideas and designs needed to get to decentralized Artificial General Intelligence – and while there can always be hidden rocks when doing revolutionary work, it seems very clear that we are poised to make some true breakthroughs in the next few years, building on our prior work and the general astounding recent progress in the AI community overall.
Q: Why is the additional token issuance you’re proposing as part of SingularityNET Phase Two necessary? Why weren’t the needs addressed in the Phase Two proposal foreseen when the original SingularityNET project was planned out and the original whitepaper was written?
A: There are three main answers to this question: 1) The crypto winter we’ve just experienced, 2) The inherent infeasibility of projecting the growth course of any advanced technology startup, and 3) The global pandemic.
The original financial planning for the project did not account for the crypto winter that started in mid-2018 and lasted through much of 2020. The crypto winter had two main impacts on the project.
The first impact of the crypto winter as a result of the collapse of ETH prices during this period. The funds (mostly Ethereum) contributed to the SingularityNET Foundation in exchange for AGI tokens at the time of the initial Token Generation Event (Dec 2017) were nominally worth around $36M USD at that time. But at the times they were actually converted to fiat in order to cover development expenses, the ETH and BTC to USD exchange rates were much lower, resulting in a total of around $15M rather than $36M yielded to the Foundation (as noted by SingularityNET Foundation CEO Dr Ben Goertzel in a 2019 AMA).
The second impact of the crypto winter was due to the way the altcoin market as a whole collapsed concurrently with ETH and BTC price during this period, resulting in the AGI token falling far below the price at the time of the TGE. This made it far less viable than envisioned to use AGI tokens as incentives for development — meaning that some planned community-development initiatives just didn’t get done, and others got done via direct funding from the foundation. It also meant that using AGI tokens to compensate developers on a subcontract basis to the Foundation, while still viable, went through the Foundation’s reserve tokens faster than planned.
The second answer to the question is that no technology startup, ever — and especially no tech startup in a fast-moving, cutting-edge technology area — has correctly foreseen years in advance the evolution of the markets and ecosystems relevant to its business, nor the technical challenges it would face. A basic, low-level technical protocol like TCP/IP can often be specified in advance in a robust way that is unlikely to need significant changes in the short term after its detailed design, implementation and launch. Something as complex and multifaceted as SingularityNET cannot be.
The crypto ecosystem and the AI ecosystem have both developed in ways that nobody foresaw in 2017, and the reaction of the AI and blockchain community to the SingularityNET beta platform and the other associated AI tools we’ve developed has taught us many things we could not have understood at that point. The result of these developments and this learning is the focus on SingularityNET Layer 0 and Layer 2 that you see in the SingularityNET Phase Two development plan.
Finally, the global pandemic has had a complex impact on the SingularityNet ecosystem. Arguably it may have helped lead to the current resurgence in the cryptocurrency markets, which is positive for SingularityNET. However, it has also greatly strengthened the Big Tech companies that are our main competitors — and has in many ways made large enterprises more conservative in their technology experimentation, with the exception of innovations urgently needed specifically to adapt to the pandemic in the immediate term. In these and other ways it has made the market conditions more difficult for SingularityNET and other blockchain platforms, slowing the pace of adoption and delaying the onset of the dynamic wherein increasing adoption leads to increasing overall network value.
The good news is, we have developed a great deal of valuable software, we have curated a fantastic team and a vibrant community, and we have fundamentally changed the conversation regarding AI and centralization. And with SingularityNET Phase Two we are now poised to proceed full speed ahead toward the project’s original vision, building on all we have accomplished and learned.
Q: Wouldn’t the SingularityNET Phase Two proposal, if accepted and executed, decrease the value of the AGI tokens I hold? Because after 5 years when the new 1B tokens have all been issued, my tokens will then be half as much of the total AGI token supply as they are now.
A: You shouldn’t think about AGI tokens as if they were a security. Think about AGI more as if it’s the currency of a country, as a first approximation. When the US Treasury issues more dollars to stimulate the US economy, this doesn’t actually cause inflation in a predictable way – actually, according to Modern Monetary Theory, it will sometimes cause each dollar to be worth more relative to foreign currencies via stimulating domestic economic growth. And of course, growth stimulation is precisely what SingularityNET Phase Two is about.
The question to ask is whether, comparing the scenario where the proposal is approved with the scenario where it isn’t, the SingularityNET network in 5 years from now will be worth more than 2x as much in the “approved” scenario as in the “not approved” one.
One point to think about here is: If the additional development work to be enabled by the Phase Two proposal propels the network to large-scale adoption – or to a breakthrough in AGI technology (or both) — then the corresponding increase in value will be massively more than 2x.
Comparing the scenarios where SingularityNET Phase Two proceeds with those where it doesn’t, we are strongly convinced the expected value of the overall SingularityNET network 1, 2, 3, 4, 5 years from now and beyond will be significantly more than 2x greater in the former scenarios. The speed with which the Phase Two plans will allow the Foundation and community to build on its Phase One achievements will put the network in a far stronger position.
With the technologies we’ve built and the understanding we’ve gained, we are in a far better position now than in 2017; however, Big Tech has gained its own momentum during the pandemic and is not slowing down anytime soon. The Phase Two initiatives will allow SingularityNET to genuinely compete in a much more compelling way.
The main point is – What we are talking about here is not just an incremental improvement, but rather a dramatic coordinated increase in both the utilization and the capabilities of the network. As a visual metaphor, it’s as if in Phase One we have built a modest-sized platform and rocket, as well as a large number of sophisticated parts designed to be fitted into a larger rocket. The modest-sized rocket can indeed take you to space, but the ladder to climb in is a bit awkward to use, and it can’t fit that many people at once without slowing down. We have also spread the news far and wide about what our big rocket will eventually be able to do. In Phase Two we are assembling the fancy parts into the big rocket, making the underlying platform stronger so it can hold up the immense new vehicle, and adding an easy-to-use, wheelchair-compatible on-ramp as they have in the airport. The difference between the metaphor and the reality is, a rocket just gets you into space, whereas the impact of a beneficial democratic, decentralized AGI may be far, far greater.
Q: How do we know this will be the last batch of tokens the core SingularityNET team wants to generate? How do we know you won’t come back to us next year — or 5 years from now — asking to authorize the creation of another billion tokens, or whatever?
A: We (the core SingularityNET team) believe we know what we’re doing at this stage far better than we did in 2017."]
We believe the SingularityNET Phase Two token issuance will quite likely be sufficient to propel the SingularityNET network to massive adoption, which then will shift the SingularityNET Foundation into a dramatically cash-flow-positive situation and propel the network emphatically toward its broader goals. While acknowledging that in such highly uncertain, rapidly developing times and markets and technology spaces, any plans may need to be adapted based on future circumstances, our plan is that future token offerings in the SingularityNET ecosystem will occur as offerings of SingularityNET Layer 2 Assets.
Q: The Cardano port makes a lot of sense, but what happened to the SingularityNET/TODA partnership? I thought you guys were working with TODA to create a custom ledgerless blockchain specifically for SingularityNET?
A: TL;DR – we still intend to explore the integration of SingularityNET with TODA/IP but are currently oriented to do this in a SingularityNET-on-Cardano context.
The issues currently posed for SingularityNET via Ethereum’s slow speed and high cost will be remedied for a great number of use-cases by the creation of a Cardano-based platform leveraging the AGI-ADA token and the ADA network. However, there are also some important SingularityNET algorithm categories for which Cardano’s consensus mechanisms as currently implemented or envisioned on the current Cardano roadmap are not sufficiently streamlined, and more radical methods will be needed – but these more radical methods can, it seems, be implemented within Cardano’s framework.
Most critically, one of the original motivations of SingularityNET was to support the decentralized implementation of Multi-Agent System (MAS) AI algorithms, in which large populations of simple agents interact according to their individual goals and understandings but the crux of the intelligence lies in the system rather than the individual agent. For example,
MAS algorithms of this nature are used in finance to simulate markets (the agents are simulated market participants), and MAS simulations of the crypto markets will be highly valuable for the SingularityDAO project along with many other applications of SingularityNET in the traditional and crypto finance spheres.
The OpenCog Hyperon system, a new version of the OpenCog AGI system currently being developed with a view toward close interoperation with SingularityNET platform, incorporates a computational-creativity algorithm called Cogistry, which involves small software “codelets” whose purpose is to transform other codelets into yet other codelets , forming a sort of primordial soup of self-rewriting code assemblies out of which radical novelty may form
SingularityNET Foundation has in 2020 been using a novel MAS framework to simulate the spread of COVID-19 and evaluate the implications of various COVID management policies; and before that, in 2018-19, it developed a MAS simulation framework for studying reputation systems in decentralized networks.
The latter SingularityNET MAS simulations have been implemented so far as standalone systems, rather than having each simulated agent be an agent acting on the SingularityNET network — and this is precisely for reasons of cost and speed. Of course, one could work around this by making an MAS run entirely on a side-chain, but in this case one is losing many of the benefits of being on SingularityNET —if the agents in the MAS need to access AI agents running on the broader SingularityNET, they then need to leave the side-chain, encountering cost and scalability issues again.
The TODA/IP protocol provides an elegant solution here by enabling decentralized blockchain-based multi-agent coordination without requiring a distributed or replicated ledger. TODA “rings” (subnetworks) can include distributed ledgers when necessary, but can also be constructed to rely on pure ledgers, fully decentralized (“sharded all the way down”) consensus mechanisms.
It is possible to implement TODA/IP in such a way that a TODA-based network of agents can interact with a Cardano-based network as if it were a side-chain. In this way, one can effectively use a Cardano-sidechain-on-TODA to run SingularityNET Multi-Agent Systems that interoperate flexibly with other SingularityNET agents running on the SingularityNET-on-Cardano platform. While algorithmically subtle, this can be achieved without writing a large amount of code due to the simplicity and elegance of the TODA algorithms and the abstractions provided by the Cardano framework.
In sum, this aspect of SingularityNET Phase Two will add a level of sophistication to SingularityNet Layer 0 that goes far beyond anything else out there in the decentralized or centralized software spaces, and in a way directly relevant to multiple core SingularityNET goals, including Artificial General Intelligence itself along with SingularityDAO and MAS for reputation management and social and medical simulation.
Interweaving TODA, Cardano and SingularityNET in this way involve much subtler customization of blockchain infrastructure than was envisioned in the original SingularityNET whitepaper, but our exploratory prototyping work in the intervening years has made it clear that this is a bullet which must be bitten. We don’t need our own blockchain for SingularityNET, but for some critical sorts of AI, we do need our own customizations and combinations of existing leading blockchain technologies.
Q: The original SingularityNET whitepaper set a portion of funds aside for “benefit” oriented projects (the Beneficial Reserve). Why isn’t something like this incorporated in SingularityNET Phase Two? Have you given up on social benefit?
A: Social benefit will be one of the major criteria the community will be encouraged to use in rating and selecting projects to receive tokens from the DEEP Funding pool. What we found during Phase One was that it was very hard in many cases to differentiate benefit projects from other projects, because so many decentralized AI projects building on SingularityNET provide benefit to humanity in some form or another. For instance, is AI for clinical trial analytics a benefit project? (one would suppose so, it’s curing disease). What about SingularityDAO, which is aiming to create a better financial ecosystem for tokenized startups, including those from the developed world and underrepresented founders. Is Hyperon a benefit project because it’s open-source and aimed at beneficial AGI? In the end, it seemed what we really want is a bias toward funding projects that are beneficial in a diversity of senses, rather than a division into buckets labelled Beneficial vs. non-Beneficial.
Q: What is the point of having SingularityNET Foundation spin-off separate companies (like Rejuve or NuNet) for some of the Layer 2 projects, rather than just doing the projects within SingularityNET Foundation?
A: SingularityNET Foundation has too many things on its plate – it needs to focus on building the platform and coordinating the ecosystem and getting traction for the platform and the various tools built on it. Each of these SL2 projects being spun off is big, complex and multifaceted enough that it should logically be its own company or foundation in itself.
Of course one could make such projects semi-autonomous divisions of SingularityNET rather than fully separate entities. However, having a separate entity for such a project has a couple of big advantages.
It allows senior staff to be brought in and given founders’ token allocations (or share allocations, in the case where the entity is for-profit) that are significant percentage-wise – something that isn’t so easy to do when bringing new staff into a project like SingularityNET that’s been around awhile. This makes it easier to recruit senior staff of the needed quality – very important as recruiting for various key functions is a major challenge in today’s competitive market.
In the case where the new separate entity is for-profit, the separate entity also enables the raising of investment from an external angel or VC investors who are oriented toward seed rounds of new enterprises.
Q: Why do SL2 projects like SingularityDAO, Rejuve, NuNet, Xccelerando etc. need their own tokens? Why don’t they just use the AGI token?
A: AGI token is designed specifically to be a payment token for AI services. SingularityDAO Dynasets represent portfolios, and SingularityDAO governance tokens have to be limited to those who have earned or bought specifically the right to vote in SingularityDAO governance (we want SingularityDAO to be governed by folks who are engaged with that particular project, not necessarily by everyone who is engaged with SingularityNET in any way – the latter would not be optimal for actual attentive governance). Rejuve token is designed to be a reward token for contributing longevity-related personal data. NuNet token is tied to processing power. Etc.
Part of the beauty of smart contract based cryptographic tokens is that you can script them to serve specific sorts of functions. So that’s what we’re doing.
All these SL2 projects will make use of AGI tokens for their AI services, and where appropriate they’ll also pay a small SingularityNET fee drawn from each transaction done w/ one of their tokens – where the fee will be in AGI tokens and will be used for reputation system and network simulation modelling development, ensuring that the SL2 projects will continue to have the world’s best reputation and network simulation tech behind them.
Q: How will the move to AGI-ADA affect staking? Will I be able to stake using these new tokens and will we see rewards increase as they are so small at the moment and the cost of gas is so high?
A: Yes, you can expect significantly higher staking rewards on AGI-ADA tokens.
Q: You reported you would be self-sufficient by this time last year if this is true why do you need to generate more funds for Phase 2?
The combination of the COVID-19 global crisis and an ongoing bear crypto market (until quite recently) threw our projections off, basically.
Q: Will current token holders be able to swap 1:1 or even better gain any bonus for swapping?
A: AGI-ADA tokens will be 1-1 swappable for AGI ERC-20 tokens (and vice versa). There is no swapping bonus, but staking rewards will be significantly higher on the AGI-ADA side, and most importantly as we build out a version of the platform on Cardano further it will be achieving a variety of capabilities that we can’t easily do using ETH. The AI-DSL being the first example.
Q: If I agree to expand the token supply for SingularityNET, I want more say in how it is run.
A: See the Governance section of the SingularityNET Phase Two document. Our plan has always been to shift to more fully democratic governance in Year 4 after the original TGE, and following through on this is a big part of SingularityNET Phase Two. We have also learned more about how democratic governance can and should work in this sort of network, during the last few years, which is reflected in the Governance section of the document and the ways the proposal there differs from the original whitepaper.
The main points are giving the Supervisory Council (which is elected by the token-holders) more of a role (including auditing regular financial reports and summarizing them for the community), and dispensing a substantial percentage (30%) of newly released tokens each month based on a community voting procedure.
Q: With all the talk of regulations tightening and large projects like XRP coming under fire, aren’t you scared you may breach regulation? How sure are you that Cardano won’t fall under the same scrutiny by the regulators?
A: AGI is a utility token, ADA is a utility token. XRP/Ripple was always a notably centralized project, with more of the vibe of a traditional banking project than a decentralized crypto project w/ the spirit of libertarian cypherpunk anarchy and so forth.
Q: If we vote against minting new tokens will the foundation be able to continue?
A: Yes we are continuing to pursue the mission no matter what, but we will not be able to progress nearly as fast unless Phase Two is ratified. And being slowed down would not be a good thing, because the centralized competition is speeding up not slowing down. We can beat them but it will be more likely to do so if we’re not needlessly resource-starved as we push forward.
Q: Given that big tech now has the power to deplatform any competitor i.e. Parler, what is to stop Google / Amazon / Facebook from coordinating against SingularityNET?
A: We don’t need to rely on any of their services. Hosting on AWS or Azure or Google Cloud is convenient but there are loads of other hosting providers that can run SingularityNET agent networks, albeit with less slick APIs and infrastructures that require a bit more configuration and effort (which however we know how to do). And of course, a big part of the SingularityNET Phase Two thrust is to create SingularityNET infrastructure-as-a-service, which once mature will allow you to effectively run large-scale complex SingularityNET agent networks on any server farm, via leveraging appropriately configured OSS software.
Q: Will you merge with Cardano completely and be the AI arm of Cardano? Or will you be a true partner? Given Cardano are much larger won’t they be in the driving seat?
A: A full merger w/ Cardano has not been the nature of our discussion with IOHK up to this point. There is indeed a great deal of synergy between SingularityNET and IOHK/Cardano – but it’s also important to understand that two decentralized networks can operate and coordinate together with more subtlety than two typical commercial companies.
As for “leadership” in the partnership – IOHK is a fantastic group of people but they don’t have nearly as much AI expertise as we do in SingularityNET Foundation and the close SingularityNET community, so it appears likely in the partnership SingularityNET will have a leadership role as regards AI and IOHK will have a leadership role as regards blockchain platform tech – nothing too mysterious here.
Q: How will this drive adoption? You’ve been talking about how you’ll work on creating adoption, but the market is filled with demo services that don’t yet have a huge amount of traction.
One thing we’ve realized in the last couple years is that the best way to get massive traction will be to create application-specific decentralized networks (SL2 networks) on top of the platform, which appeals to specific audiences and apply SingularityNET AI in specific ways appropriate to those audiences. In this way, the SingularityNET platform becomes more of a meta-platform – an underlayer for various other platforms providing more customized agent populations for particular classes of users.
Appendix 2 of the SingularityNET Phase Two document summarizes some of the specific reasoning behind the token-use estimates we’ve presented in the graphs in the main body of the document, for the various initial SL2 projects.
Q: How will you prevent the market from crashing due to flooding it with a billion new tokens?
The new tokens will be released gradually, in equal tranches month by month over 5 years. This gives plenty of time for the growth rate of the value provided by the network to vastly exceed the growth rate of the token supply, month by month.
Q: How will adding more tokens to the pool affect your internal tokenomic structure? Have you run any simulations of this?
What will complicate the dynamics of the AGI token ecosystem most dramatically in the next few years will be the launch of multiple SL2 networks that will use the AGI token to a dramatic degree – along with SingularityDAO that will specifically deal with trading and market-making for both AGI token and SL2 Assets along with other tokens. We haven’t yet tried to simulate all this in detail but will do so as 2021 progresses using our existing multi-agent simulation framework.