Also available in Portuguese, Indonesian, French, Italian, Spanish, Arabic, Turkish, Russian

Over the last few years, we have been pushing the boundaries of decentralized AI, developing a comprehensive suite of innovative tools and products designed to develop and utilize AI services in ways that prioritize decentralization, safety, scalability, and benefits to all — developing the SingularityNET decentralized AI Platform.

The core goal of our Platform remains the same as it was in 2017 when we founded SingularityNET as the first truly decentralized AI network: To create a foundation suitable for running AGI systems with general intelligence at the human level and beyond in a secure, efficient, easily usable and fully decentralized way, without any central owners or controllers. Along the way, as our AI systems gradually move toward full AGI capabilities, the platform must also provide a high-quality decentralized infrastructure for AI applications serving diverse vertical markets.

One thing this mandate means is that — unlike most of the more recent entrants in the decentralized AI space — the Platform cannot be specialized to any particular class of AI algorithms or data types, nor any particular vertical market application. If it is going to serve as the decentralized infrastructure for the global economy as the world enters the AGI phase, it must be far more generic and flexible than that.

The purpose of this Platform progress overview is to detail the achievements made during 2023 and outline our Platform development roadmap for 2024. While the progress ongoing is detailed and heterogeneous there are three key thrusts involved:

• The Internet of Knowledge — a unique approach to extending the applicability of decentralized AI in the immediate term, and paving the way for the emergence of decentralized AGI;
• Integrations between SingularityNET Platform and other decentralized networks: Hypercycle, NuNet, Cardano, Dfinity (ICP), and others … part of the process of moving toward a next-gen cross-chain decentralized AI ecosystem;
• Scalability and Usability improvements, including large new features in basic areas like hosting and billing.

If the brief summary given here whets your appetite for more, you may find a full-length report of our Platform progress and plans at SingularityNET AI Platform 2024 Roadmap.

The Internet of Knowledge — A Distributed and Decentralized Metaframework for the Next Phase of AI Evolution

The Internet of Knowledge is a decentralized network, built on the SingularityNET platform, in which Knowledge Nodes (graph-based knowledge stores) and Model Nodes (AI algorithms) interact to create powerful AI solutions. This framework aims to make AI development more accessible and efficient for users of various technical levels — both in the context of building immediate-term practical AI applications, and of using SingularityNET to work toward advanced AGI with maximal efficiency and reliability.

Knowledge Nodes offer a flexible, dynamically updated knowledge representation along with a storage and a query engine for it. Each knowledge node may have its own knowledge representation, should store knowledge and query it. Users configure these nodes declaratively, specifying the data format and structure. This simplifies knowledge graph creation and management. Model Nodes leverage the knowledge stored within Knowledge Nodes for tasks like training and inference. SingularityNET supports a variety of AI algorithms, including symbolic, neural, and hybrid approaches. By leveraging blockchain technology for deploying these nodes, SingularityNET aims to become the foundational knowledge layer of the AI internet era, promoting decentralized interaction and the advancement of beneficial AI.

To streamline AI deployment, we are developing a scalable infrastructure tailored to AI-specific needs. It focuses on optimizing GPU usage, distributed processing, and autoscaling. AI-DSL, a specialized language for describing AI service interfaces, will simplify development and promote the creation of self-assembling AI workflows.

Finally, this framework of Knowledge and Model nodes works naturally with the use of OpenCog Hyperon to enact neural-symbolic AI, merging powerful Large Language Models (LLMs) with symbolic reasoning techniques. This comprises a step along the path to AGI and also enhances immediate applied AI solutions via enhancing problem-solving capabilities and improving knowledge integration.

The combined goal of all these aspects is to create a decentralized AI Metaframework that enables a collaborative, evolving system where knowledge and AI capabilities are not siloed but shared across a wide array of participants, catalyzing synergetic emergence of higher and higher levels of intelligence.

Zero-Code AI Model Training

To make our new Internet of Knowledge functionalities as broadly as accessible as possible, we have added new Platform features allowing service providers to easily train and deploy custom AI models, even without coding experience. This is made possible through a zero-code training option where providers can activate AI model training based on their specific datasets.

This is why we will be introducing the concept of Training-On-Platform (TOP). As a pilot, we have equipped three highly sought-after domain-focused implementations with this new capability:

• Speech synthesis service with training support based on the provided unique voice sample;
• Image generation service with training support based on the provided image collection;
• Textual generative conversational AI service based on LLM fine-tuning using custom data.

One of our key priorities is to showcase the TOP feature and its advanced functionalities through top-tier, pre-trained AI services available in our marketplace. This makes TOP accessible to diverse audiences, including developers with deep technical expertise and AI designers and customizers with much less background.

SingularityNET Platform Assistant: Decentralized AI for Guiding Use of Decentralized AI

An advance as radical as the Internet of Knowledge requires a cutting-edge user interface methodology, which in the current era is best supplied by AI itself.

This is provided by our new SingularityNET Platform Assistant, which was initially conceptualized as a chatbot for answering service-related questions, and helping users sort through the increasing number of services on the SingularityNET marketplace to find what they need, but has expanded to encompass a wider range of functions.

The Assistant’s development will follow a phased approach. The initial focus is on creating a chatbot capable of answering service-specific and general platform questions. It will leverage existing documentation to provide seamless user support. Medium-term goals include expanded functions, such as:

• Onboarding support: Guiding new users through the Platform’s features and operations;
• Automated code generation: Streamlining specific tasks by automatically generating code for AI models wrapping, data processing, etc.;
• Integrated service calls: Enabling users to interact with Platform services directly within the chatbot interface.

The Assistant’s technical foundation harnesses the Internet of Knowledge itself along with the MeTTa-MoTTo which has emerged from the OpenCog Hyperon initiative, which provides interoperability of LLMs with Knowledge Graphs and reasoning.

Integrations: Toward a Robust Cross-Chain Decentralized AI Ecosystem

One of the beautiful things about the modern decentralized software ecosystem is that no one project has to do everything for itself. Integrations across networks and blockchains are more and more tractable to construct and have the potential to multiplicatively improve all the parties in the integrated system. We are proceeding aggressively with the integration of the SingularityNET Platform with other decentralized networks, including HyperCycle and NuNet which originated as part of the SingularityNET ecosystem, and external networks such as Cardano and Dfinity.

ICP Integration and decentralized AI Marketplace deployment

We have partnered with Dfinity to improve the decentralized AI infrastructure on SingularityNET and Internet Computer Protocol (ICP). Complementing our partnerships with other entities such as Input Output Global (IOHK) from the Cardano ecosystem, and our work with HyperCycle to create our own unique Layer 0++ blockchain framework, this Dfinity collaboration exemplifies our commitment to a cross-chain approach to decentralized AI.

A major part of this initiative is the development of a decentralized AI Marketplace on the ICP, facilitating easier access to AI services. This marketplace will enable users to test, purchase, and explore AI services, moving us towards complete decentralization and a wider, cross-chain distribution of AI services and development activities.

To simplify AI service integration, we plan to introduce a universal template for AI service interfaces, supporting functionalities like web3 authorization and payment services. We’re also considering pre-built templates for common AI services to ease interface development.

Exploring the potential of ICP for hosting AI services, we’re conducting feasibility studies on scalability, resource distribution, and accessibility. This includes testing AI models and services in ICP canisters to understand hosting capabilities (and its limitations).

Accelerating Progress on Cardano Integration

In late 2023, we launched AGIX token staking on the Cardano blockchain, culminating an extensive development effort throughout much of last year.

Leveraging the Cardano knowledge we gained during the staking implementation, we have also worked out a detailed design for the full port of the SingularityNET protocol marketplace to the Cardano chain, and expect to progress with this during 2024, bringing the powerful decentralization, robustness and security of the Cardano network to the universe of SingularityNET AI agents.

HyperCycle: Steps Toward a Fully AI-Customized Decentralized Software Stack

HyperCycle is creating the infrastructure for a revolutionary concept: the Internet of AI. This network will enable AI agents to collaborate seamlessly, tackling increasingly complex problems through microtransactions for the exchange of microservices. We see immense potential in how HyperCycle can transform our decentralized AI Platform. One key area is the creation of an AI marketplace where humans and machines can transact freely, advancing machine intelligence and opening pathways to true AGI. This is why we are actively exploring ways to integrate HyperCycle into our Platform.

Initial research focuses on model hosting, where HyperCycle’s decentralized framework would streamline service registration, hosting, and utilization of AI models. However, HyperCycle’s real strength lies in its embedded blockchain, allowing for trustless value exchange through secure financial transactions and token conversion. This would revolutionize billing on the platform by providing new billing options. Additionally, integrating HyperCycle at the core level can lead to lower fees, faster processing, and increased security for the platform as a whole.

Currently, our team is researching the optimal architecture and interaction model to achieve the perfect balance between speed, security, and user experience. This research paves the way for an even more exciting future: the potential to create smart contracts within the HyperCycle network in MeTTa (Meta Type Talk), the native language of our OpenCog Hyperon AGI system. This development direction is poised to bridge the gap between AI and blockchain development, opening doors to a new era of AI-powered smart contracts (aka smart contracts that are actually smart)!

NuNet AI Model Hosting for the SingularityNET Platform

NuNet is building a globally decentralized computing framework that combines the latent computing power of independently owned compute devices across the globe into a dynamic marketplace of compute resources. This approach transcends limitations like physical location or device size, empowering users to discover and utilize the precise amount of computing power they need whenever they need it.

A key collaboration between NuNet and SingularityNET focuses on integrating the SingularityNET AI Platform with NuNet’s decentralized ecosystem. This integration will focus on seamless AI service hosting on NuNet, offering several benefits:

• Simplified Hosting: Service providers will easily host their AI models on NuNet.
• Dynamic Scaling: Ability to adjust resources based on AI service load, model training speed, and other parameters.
• Streamlined Onboarding: Users won’t need to worry about resource allocation, as the platform will handle it automatically. This makes it simple for providers to launch, train, and publish models.
• Efficient Billing: A simplified internal billing system reduces the complexity of payments, especially for scenarios with custom model training on unique datasets.

Scalability and Usability Improvements, Smoothing the Path to a Decentralized AI Future

One of the advantages SingularityNET derives from its 6+ years of history is that we have had plenty of time to receive user feedback regarding which aspects of our Platform work fantastically for the community and which ones need improvement. Based on this we are driving forward a series of fundamental improvements in the usability and scalability of our Platform, tied together with our Internet of Knowledge initiative and cross-network integrations.

Improving the Onboarding Experience

The SingularityNET decentralized AI Platform is doing some unique and complex things, encompassing a wide range of AI services and technologies. For this reason, as it’s gained capabilities over the years, the user experience from the standpoint of service developers has become increasingly involved. Simplifying the process of understanding and getting started with the platform has been one of our top priorities recently.

Recent usability improvements include:

• Development of a Text User Interface (TUI): We created a user-friendly interface to make the developer experience on the platform more accessible and dev-friendly, especially for those with limited system administration knowledge.
• Improved Technical Documentation: The documentation and guidelines for the Developer Portals have been revamped so they are clearer and more user-friendly, allowing for a much smoother onboarding process.
• Streamlining the Onboarding Process and Publisher Experience: We made enhancements in the test network, front-end functionalities, user experience, and documentation to ensure a more seamless experience for service providers.
• Facilitated Service Development and Deployment Automation: Implementing automation and event-driven solutions for AI model deployment, ultimately simplifying the service wrapping process, reducing costs, and allowing for efficient scaling of AI models.

Improving the Key Components: CLI, SDK, Daemon

Upgrading our decentralized AI Platform’s user experience relies on enhancing three of its most critical components:

• The Daemon: the adapter that a service can use to interface with the SingularityNET platform.
• CLI: the primary tool for interacting with the platform’s smart contracts, managing deployed services, and managing funds.
• SDK: a tool for customers of AI services to integrate and make calls to services.

These three components serve as the gateway for service providers to deploy and manage AI services and for clients to integrate these services seamlessly (as well as for end-users to access the services directly).

Right now, we are working on:

• Separating the CLI and SDK into distinct repositories and refactoring the SDK to increase flexibility by segregating components based on application context.
• Improving the general Platform architecture and developing multi-platform tools to make interactions with the Platform more intuitive.
• Automating and streamlining Daemon, CLI, and SDK setup and reducing configuration complexity.

The point of all this is to ensure the platform’s smooth operation, security, and performance, which are critical factors to support a simple, accessible, and user-friendly onboarding process.