SingularityNET is extremely excited to announce our collaboration with IOHK, a company led by Cardano founder Charles Hoskinson and driving forward the engineering and development of the Cardano blockchain. Cardano’s ADA token is currently the #6 cryptocurrency globally in terms of market cap, and the Cardano ecosystem is entering a period of rapid growth due to the expected near-future launch of a number of key features such as the Plutus smart contract framework.
An in-depth review of some of the motivations, synergies and possibilities underlying the SingularityNET/IOHK collaboration can be found in the recent 90-minute dialogue between Hoskinson and SingularityNET CEO Dr. Ben Goertzel, which occurred recently at the Wyoming Blockchain Hackathon in Laramie, and also in Dr. Goertzel’s talk at the Cardano Summit earlier this year.
Along with a shared focus on democratization and decentralization, Cardano and SingularityNET also share a culture and development style centred on advanced computer science and mathematics. Cardano’s focus on academic rigour and formal methods in developing its core blockchain technology match closely with the deep scientific background of SingularityNET’s AI team.
Several envisioned aspects of the collaboration between SingularityNET and IOHK have already been moving forward on the level of in-depth conversations between the two projects’ tech teams prior to the formal collaboration announcement.
The IOHK and SingularityNET platform teams have been seriously discussing what it would take to port a significant portion of the SingularityNET decentralized protocol and platform from Ethereum to Cardano. This would involve providing mechanisms for swapping some of the current (Ethereum-based) ERC-20 AGI tokens to Cardano-based AGI tokens, and also creating analogues of the Solidity smart contracts underlying aspects of the SingularityNET platform using Cardano’s new Plutus smart contract language.
Final decisions and details regarding porting of a portion of the SingularityNET network to Cardano have not yet been made and will be discussed in-depth with the SingularityNET community when the time is right. At the moment joint planning and exploration are proceeding very positively.
The SingularityNET design has been largely blockchain-agnostic since the beginning, and there has been recurrent discussion of making the SingularityNET protocol multi-chain rather than reliant on a single blockchain. Blockchain technology is still in the early days and we are now seeing a flourishing of new blockchains embodying more sophisticated algorithmic than now-“legacy” systems like Bitcoin and Ethereum.
Very serious consideration was given in late 2019 to porting a portion of SingularityNET to a radical TODA-based ledger-less framework, however, this has not been put into place yet due to the need for some additional development on the open-source implementations of the TODA/IP protocol. Experimentation has also occurred regarding porting a portion of the SingularityNET network to the NEM protocol, due to the close harmony between NEM’s Proof-of-Importance mechanism and SingularityNET’s reputation system. There have also been interactions with the Algorand team, which was driven by the innovative cryptographic algorithms and demonstrated the scalability of the Algorand platform.
Current speed and cost issues with the Ethereum blockchain have increased the urgency of exploring alternatives for SingluarityNET’s blockchain underpinning. The ambitious Ethereum 2.0 design holds promise but the timing of the rollout of different aspects of this next-generation Ethereum remains unclear, along with many of the practical particulars.
Cardano’s consensus mechanisms and associated algorithms and structures comprise a unique mix of rigorous theoretical grounding and practical scalability and have now reached a level of maturity that makes it possible to port a complex blockchain application like SingularityNET to Cardano.
The Solidity smart contract language provided an initial concrete toolset corresponding to Ethereum’s dramatic vision of a blockchain-based “world computer.” However, various technical aspects of Solidity’s design render it less than optimal as a foundation for the next stages of development of secure decentralized global computing. As a Turing-complete language without simple dedicated mechanisms for creating more limited Domain-Specific Languages (DSLs), Solidity presents significant challenges to the formal program verification and analysis methods that are critical to safety and security on today’s Internet.
Cardano’s Plutus smart-contract language is a next-generation construction designed to ease the creation of DSLs such as the Marlowe financial-smart-contract language, providing usability to smart contract authors with domain-specific expertise as well as making formal verification and analysis more tractable. Plutus provides a flexible and efficient framework for not only the sorts of secure transactions carried out in the blockchain underpinnings of the current SingularityNET framework but also for the more advanced aspects of the SingularityNET system that will be rolled out in coming years.
The Critical Value of Provable Security and Rigorous Design
Cardano’s thoroughly formalized functional programming foundations have potential to provide a rich and flexible basis for implementing advanced aspects of the SingularityNET design — but just as critically, they have strong promise to provide a secure and reliable basis for the network’s operations basic and advanced alike. The recent hack of the KuCoin exchange, resulting in the theft of more than 40 million AGI tokens belonging to various parties, highlights the essential role security has to play in the world of decentralized applications.
In the centralized IT world, security can be achieved — to a reasonable degree anyway — by taking relatively insecure underlying systems and building secure IT walls around them. In the world of decentralization, one instead must rely on security by design — security must be baked into the core of one’s protocols, contracts and containers, at every level. One might think this was intrinsically the case when using blockchains, due to the role of cryptographic functions at the core of blockchain operations — but of course, it’s not really that simple.
Provably secure consensus protocols and smart contracts are another important piece of the decentralized security story, and the careful attention Cardano has paid in these regards are part of the value it offers to projects such as SingularityNET. The careful layering of levels of abstraction in the Cardano design is also relevant here, as it strictly limits the types of hacks and attacks that can be done in various contexts. Putting SingularityNET on Cardano would not prevent exchanges from being hacked (though architecting exchange software on similar principles to those used in Cardano’s development would, but that’s a somewhat different story) — but if done right, it would give a much greater variety of responses and remedies to this sort of hack happening.
The KuCoin hack was not the motivation for the SingularityNET/Cardano collaboration — in-depth discussions between the two projects were already underway well before the hack occurred. However, it certainly highlights the multidimensional value that may be achieved by shifting SingularityNET to a more rigorously formalized and secure-by-design underlying blockchain framework. As we advance SingularityNET toward greater and greater decentralized general intelligence, we will also require its basic functions to operate in a safe and reliable manner.
Toward API-of-APIs and Decentralized AGI
The most intriguing aspect of a potential SingularityNET-on-Cardano system is not merely the promise of a substantial increase in speed and scalability — this is absolutely needed but can likely be obtained from a variety of next-generation blockchains — but rather the subtler synergies emerging from Cardano’s functional-programming underpinnings. Cardano is implemented in the Haskell language, which lends itself to the sorts of abstract representations and transformations required for the technical realization of the more advanced aspects of the SingularityNET vision.
Currently, the SingularityNET platform is primarily used for hosting relatively simple AI agents that provide practical services directly to end-users, e.g. in areas such as image processing, natural language processing, time series analysis, genomics data analysis and more. Through the Android SDK the SingularityNET protocol has been used on the back end of smartphone apps like the recent SongSplitter app that separates vocals from music in audio files.
However, the broader vision of SingularityNET relies on taking things a step further and using the platform to host a rich variety of AIs that outsource work to other AIs, and dynamically collaborate with other AIs to solve problems. To make this sort of “society of AI minds” operate in a robust way requires an “API of APIs” — an abstract formal framework for relating the APsI (the input/output interface) of different AIs to each other, even if these AIs are doing very different things as designed by a variety of different (human or AI) creators.
SingularityNET’s researchers have been prototyping approaches to the API of APIs using modern functional languages with capability for dependent typing, such as Idris and Liquid Haskell. However, a dependent type approach to the API of APIs matches awkwardly with a Solidity based smart contract infrastructure. Cardano’s Plutus language, on the other hand, is a much more natural match, and it appears both feasible and elegant to create a compiler from a dependent type based API of APIs framework into Plutus, following to an extent the approach that has been taken with the Marlowe finance DSL.
There may also be synergies between SingularityNET-Cardano integration and the OpenCog Hyperon initiative, which is focused on creating a more scalable, flexible and usable successor to the current OpenCog AGI R&D platform (which underlies a handful of specialized AI agents currently running on the SingularityNET network).
The OpenCog AGI design involves a metagraph knowledge store called the Atomspace, concurrently and cooperatively acted on by a number of different cognitive processes representing different learning and reasoning methods such as probabilistic logic, evolutionary learning, pattern mining and neural pattern recognition. Currently, to integrate OpenCog into SingularityNET, one creates a SingularityNET agent wrapping a whole OpenCog system with its own internal Atomspace and AI-process population.
However, in a SingularityNET-on-Cardano approach, it may eventually be possible to take a more decentralized approach in which the Hyperon Atomspace is provided as a service to any SingularityNET agent who needs it, and many of the cognitive processes involved in the Hyperon design are represented as SingularityNET agents that interact with Atomspace via channels set up via SingularityNET protocols. Such an approach would exploit the deep commonalities between the new version of OpenCog’s Atomese language being created for Hyperon and the dependent type based API of APIs under current exploration. The result would be a more fundamentally decentralized approach to AGI design.
Toward Mass-Market Traction via Escalating Positive Feedbacks
Market traction has been a struggle for essentially all projects in the blockchain domain, up to this point, and largely for reasons that are fairly obvious: the practical awkwardness of dealing with wallets and other cryptocurrency mechanisms, and the speed and cost of crypto transactions as compared to transactions involving centralized tools.
Fortunately, we are now at a point where the obsolescence of these obstacles is clearly on the horizon. The transition of SingularityNET onto modern blockchains such as Cardano is a clear route to achieving scale and speed, and the completion of the fiat-to-crypto gateway will remove the requirement for end-users of SingularityNET-based services to deal with cryptocurrency infrastructure.
Given the incentives and structures built into the SingularityNET design, the advent of a large population of AI agents and users on the platform has strong potential to naturally lead to the emergence of a sophisticated “society of AI minds” — so long as the underlying mechanisms are there to support this. The synergetic intelligence immanent in such a society of AI minds will provide greater value to human and AI users of the platform and thus progressively increase both the user traction and the intelligence of the overall network, via positive feedbacks. The deep integration of the SingularityNET and Cardano technologies has the potential to provide the underpinnings needed to support this kind of growth.
We thank each and everyone one of our community members for their continuous support. SingularityNET plans to reinforce and expand its collaborations to shape the coming AI Singularity into a positive one, for all. To read more about our other partners, click here.
SingularityNET has a passionate and talented community which you can connect with by visiting our Community Forum. For any additional information, please refer to our roadmaps and subscribe to our newsletter to stay informed about all of our developments.