The SingularityNET Beta platform was released almost two months ago and ever since that first private release on February 6, our team has been blessed with insatiable curiosity from the community leading up to the public release and thereafter. It is incredibly rewarding to see that amount of excitement about the new features, the 40+ AI agents now live on the network and the overall platform, and to feel that we have delivered on what still seemed to be just a distant dream to many. But we are not resting yet. The warm welcome that the Beta received is a solid indication that our product has remained faithful to our vision and continues to bring our vision closer to reality.
In this article, we will be outlining the renewed roadmap for the Beta in 2019. Various technical implementations to the Beta are still required to make the user experience seamless, the AI service deployment comprehensive and to facilitate enterprise access to community-built services. As always, our work is transparent and open-sourced, available for each to view on the SingularityNET’s GitHub Repo.
Our Chief AI Officer, Cassio Pennachin, mentioned the importance of governance in our vision of a future with decentralized artificial intelligence. The roadmap resembles a central tool of communication between the SingularityNET Foundation and the community, as it defines the current future plan in the near-term future. In our new roadmap, a significant part of our plan revolves around making crucial steps towards decentralized governance of the platform by introducing components in Q2 that will allow the community to directly impact the growth and development of the SingularityNET Platform. With that said, let’s dive into the most prominent roadmap highlights that will help empower you to decentralize artificial intelligence.
Gearing up for Beta v2
To help our readers make sense of the upcoming improvements we asked our VP of engineering, Raam Baranidharan, what goals he had set for the team in 2019 and which updates he would like to focus on:
“Now that the Beta launch is complete we are focusing on driving more usage on the platform. We are especially focused on documentation and enhancements that would make it easier to use the platform and to interact with its many features.For instance, from our Beta feedback session with the community, it became clear that we needed to include an easy way to access different topics on the platform and simplify access for users with no “blockchain experience”. To this end, we are working on auto-generated documents for the snet-cli which will be simple to search. An overhaul of the developer portal is also being planned for that purpose.Finally, we will be making it easier for developers to integrate AI services from our platform with their software via the Service Development Kits. As we enhance our platform with support for streams we can expect much more elaborate AI services to be added on the platform –a significant improvement especially for our enterprise customers” — Raam Baranidharan, VP of Engineering at SingularityNET
In the following section, we will be unpacking some of the most crucial developments set for 2019.
The Request for AI Portal (RFAI)“This addition to our roadmap is a way for us to engage with the AI community. It will provide them with a forum dedicated to the request for AI services, their development, and their public display and funding.” — Raam Baranidharan, VP of Engineering
The RFAI portal allows users to make requests for AI services which they would like to see built and deployed onto the SingularityNET network. Users can also back an existing request by committing AGI tokens to reward the successful development of a predefined service.
The RFAI portal is a particularly important feature that will invigorate the development of AI agents on the Beta platform. As in most markets, the formulation of demand is essential to innovation whether it is explicit demand or one that is idealized by suppliers. The purpose of the RFAI is to create the appropriate conditions for demand to be clearly articulated, and involve as many parties as possible. It is rare for demand to be clearly defined in a market, especially when the particular request is intricate and includes the views of multiple parties. Suppliers are often required to compromise and find the largest common denominator between the different sources of demand in the market. The RFAI portal, on the other hand, will not only incentivize the creation of new products, but it will also nudge demanding parties towards cooperation and allow them to clearly express the layers of complexity they expect of an AI service.
If a particular party is not satisfied with the direction that the request has taken, it will be able to call for additional development and find potential allies (be it individuals or larger groups like companies) to incentivize it. By default, the RFAI portal will serve as a three-way bridge connecting enterprise customers, the community and developers. Coupled with the upcoming platform feature that will enable different services to interact and help each other, every particular demand expressed by the above parties will contribute to the higher performance of the network as a whole.
The RFAI portal will be pre-seeded with requests from the foundation to help kick-start the development of new services by our community of developers and beyond. In parallel to independent requests, as Dr. Goertzel has pointed out, the efforts of Singularity Studio will be focused on populating the platform with a large volume of requests in order to “blitzscale” the network’s capacity.
We are proud to deliver on our commitment to give everyone a chance to express their inventiveness.
Software Development Kit and Request StreamsSoftware Development Kit (SDK)
The SingularityNET SDK offers a set of tools, libraries, and documentation that helps AI developers create AI services on the SingularityNET Network. It also allows end-users to integrate AI Services into their applications. We will be enhancing the current SDK so as to abstract out blockchain interactions and make it simpler for developers to invoke services.
Daemon — Request Streams
We are enabling daemon to support streaming requests from various clients. This will allow a request or response to have an arbitrarily large size in order to enable the use of large payloads which are needed by several AI services. This will be particularly relevant to requests made by enterprise customers.
“We currently have a straight-forward request/response model tied to the platform, but with streaming we can support richer services. For example, real time speech translation where the request is a live (continuous) stream of data that is processed by the service and the response is a stream of the processed data. Since we have not set size limits for streaming, we can now support larger datasets which allow more complex AI services to operate.” — Raam Baranidharan, VP of EngineeringPlatform Improvement Proposals and Ongoing Developments“We want to democratize access to AI technology and it only makes sense that we democratize the decision making process that will get us there. We have a vibrant and skilled community and we want to leverage their expertise as we build out the platform.” — Raam Baranidharan, VP of Engineering
The platform improvement proposal portal will effectively be a repository dedicated to capturing improvement proposals from the community. Development pluralism is how we will outperform centralized groups. Detailed workflow on how proposals will be curated, managed and finally implemented across the platform.
This has been a long time coming. Platform improvement proposals have always been important to our team, and the useful feedback we received during the closed Beta led us to this decision to formalize the feedback process. Indeed, the community would always critically review the progress made by SingularityNET, as they ought to, be it via YouTube videos, the developer Telegram channel, or directly on Github. The purpose of this new feature is to make community feedback an integral part of our modus operandi and provide visibility on the status of coherent proposals. For example, some proposals might be dropped as they would involve improvements that are already being considered while other proposals would be reviewed and green-lighted by the platform development team. The later would be formally added as a next step in our roadmap and receive public updates. Ultimately, this feature will pave the way to a community-led governance of the system which has been reiterated in SingularityNET’s whitepaper 2.0.
Through this addition, we also hope to reaffirm our belief in the power of development pluralism — or the voluntary cooperation of open source developers which effectively gave us the world wide web — among other technological wonders. After all, as the software developer Eric Raymond suggested: “many eyes make all bugs shallow”. A powerful and modern illustration of the value of community-led improvements can be found in the Ethereum and Bitcoin projects. Indeed, the Ethereum project, for example, is continuously refined by individual contributions on many levels: interface wise, on core tech improvements, meta proposals, etc. In turn, these are reviewed by the community as a whole and core developers, and are fully accessible and acknowledged on the Ethereum Improvements Proposal page.
Further information on ongoing improvements to expect in Q3 and Q4 of 2019 can be found in the current version of the roadmap on our website. These include, inter alia, the InterPlanetary File System (IPFS) stability, SDKs expansion, and Daemon improvements.
Payment methods and Pricing ModelsFiat to Crypto
The long-awaited financial bridge is in its final development stages. The fiat to crypto gateway will enable users to consume services on the platform using fiat currencies, and service owners to cash out their AGI tokens back into fiat currencies. In other words, it will not be necessary for new users to have a Metamask account and connect their wallet to the platform before they can call an AI service or invest in the fulfillment of a new AI project on the RFAI portal.
This feature is certain to enhance the user experience of the platform and to open the doors to a significant mass of people that may not be familiar with cryptocurrencies or unable to use them in transactions. Indeed, ease of entry into a market is one of the most essential aspects of a successful ecosystem and particularly so for decentralized projects. The early success of Coinbase, Abra and even latecomers such as CashApp are partly due to their “user friendliness” and ability to offer users a familiar interface and method of payment. That being said, while we endeavor to iron out the payment system for all users and providers on the platform, it is important to note that the AGI token will still remain the primary method of exchange in the backend, meaning that without AGI you cannot utilize decentralized AI services on the SingularityNET Platform.
We will be rolling out three different pricing models to enhance SingularityNET’s AI marketplace. The first improvement will involve allowing users to subscribe to a given (to be defined) number of AI service calls per month. Most users will be familiar with this method of payment as many subscription models, from phone service providers to news outlets, work as such. This will not replace the current “per use” model of the Beta but simply be added as an option. Bulk pricing is the next model that our team has explored. In that model, a first amount (to be determined) of calls will be made free or tiered. To put it simply, with bulk pricing the first X amount of calls made by a user will cost, let’s say 10 cogs, followed by 5 cogs for the next X amount of calls, and so on.
NOTE: 1 cog = 0.00000001 AGI
Variable pricing is the final model that we will be rolling out. For that model, the price of a given call will fluctuate. Indeed, prices may vary based on the complexity of a request. The cost of a request will change according to the attributes of the input such as the size of the input dataset. This was devised in parallel to the daemon request stream and removal of size limits that we described above. As AI services get more complex and requests more voluminous, it will be useful to have a flexible pricing model. A clear breakdown of the attributes will be made available for users to be able to assess the price before making a request.
API of APIsServices calling services
Allowing different AI agents to interact and supplement each other is a major step towards the development of AGI. The “API of APIs” use case involves a service invoking another service for any given request. The final cost of such interaction will be set to the sum of individual service calls.
This feature is particularly exciting as it is central to SingularityNET’s vision. Specialized AIs freely interacting with other specialized AIs is the bedrock of AGI creation as we see it. Granting the ability to any AI service to borrow the collective capacity of an ever-growing set of AI systems is unprecedented. The marketplace as a whole will benefit from this development as no limitation to anyone AI service will cause the failure of a service to deliver on a client’s request.
In our model and with this inter-AI calling feature, the weakness of an AI service is the strength of another. By definition, this attribute will benefit from the increased participation of users, service providers and requesting parties. In other words, every single member of our community, regardless of their background, can help make the network more efficient and all-embracing. Moreover, we will be enhancing the marketplace DApp with support for injectable User Interfaces (UIs). This step in roadmap will enable service developers to provide a custom UI to interact with their services on the DApp. Finally, our platform development team will be enabling a daemon service mesh. Simply put, the service mesh will be an inter-service communication infrastructure using daemons to enable the API of APIs use case. This will be released along with a whitepaper focused on the use case.
SingularityNET’s vision for a decentralized ecosystem of symbiotic AI agents was, albeit intuitive, a severely complex project to achieve. But look at where we stand today: the vision has materialized in the form of a functional Beta platform and every missing component is quickly falling into place.
Better even, missing components are already defined, updated with the insight of our community, independently developed by an array of developers, and signaled by order of implementation. While the Beta platform is already delivering much of the core ingredients outlined in our initial whitepaper, its current and upcoming developments herald the surge of an overarching network that will serve anyone who wishes to incorporate AI in any form of human activity. Whether you are an individual, a small business or large corporation, the SingularityNET platform is now ready to foster your boldest requests.
Your feedback is critical to our success. So until we reach the fourth step of our roadmap and create a repository for platform improvement proposals, we invite you to chat with the team or community about every feature and service on our Community Forum or on Telegram.
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.
The best way forward is to come together and work practically towards creating a better future. We hope you will join us in our mission to democratize AI and to use its transformative powers to enact real, positive and lasting change.
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.
More from AI/ML
>Announcing our AI collaboration with Ping An TechnologyInvestment, Innovation, IntelligenceEditor’s note: The logo above now reflects the Ping An Technology …
>In the first part of this series, we discuss the vision of the Odyssey Hackathon and introduce the team members …