The SingularityNet team’s grand aim is to facilitate a distributed Artificial General Intelligence. Organizations currently control the requisite data in a highly centralized manner, resulting in inefficiencies. Singularity hopes to eliminate these inefficiencies, as well as expand the availability and impact of machine learning expertise and AI algorithms. Other factors add to the challenge of developing a distributed AI. Academics face difficulties monetizing AI research, and much publically available AI and ML code requires significant technical competency to configure and install, making it narrowly-out-of-reach to would be developers. SingularityNET hopes to address these factors as well. A DApp designed to act as a portal into the AI services marketplace is currently in Alpha testing. A reputation system, curation markets, and a liquid democracy governance structure are all slated for later phases of development. The SingularityNET’s CEO and chief scientist, Ben Goertzel, also holds the post of chief scientist at Hanson Robotics, whom supports SingularityNET’s development team.
SingularityNET’s open source platform aims to lower development costs by improving the means for trading developers machine learning algorithms and any associated dataseeking. The platform marketplace aims to enable better collaboration amongst the global community of Machine Learning algorithm utilisers and AI researchers, and to decrease redundant developments. Algorithms contributed to the SingularityNET platform can be re-deployed in other user’s programs, for a fee. AGI pricing oracles, which determine the appropriate price for the AGI token and AI services generally, will play a crucial role in this system.
On the peer-to-peer SingularityNET network, nodes are known as Agents. Agents are equipped with blockchain clients and the software to run processes supporting AI services. AI technology is segrable from smart contract operation in this configuration, thereby making the network more scalable since computationally dense ML/AI type processing is not repeated by every node comprising a global computer. Developers working with SingularityNET will onboard by connecting APIs, which will facilitate interoperability among agents within the network, and between an individual Agent and the blockchain. SingularityNET will provide tutorials and documentation to encourage researchers and users to contribute their work to the marketplace, in addition to offering a suite of standard implementations of popular machine learning algorithms. The service marketplace will include contributions that utilise both centralised and decentralised storage solutions. SingularityNET intends to support blockchains other than the Ethereum network in the future.
The AGI token, an ERC-20 token on the Ethereum blockchain, will be used to buy and sell AI services through the SinglarityNET DApp. Data, including pre-processed datasets, can be accessed by exchange of the AGI token. Though SingularityNET have held a TGE with tokens to be distributed ad hoc via Ethereum, the team recognize a need for a scalable solution in the future and allude to the future existence of an AI specific blockchain able to process the requisite transaction volume. AGI tokens will also contribute to a user’s reputation score in two ways; first, a basic reputation score will be assigned per-task based on the amount of AGI exchanged; second, the network plans to support staking based reputation whereby, independently of any specific task, users lock tokens up, thus bolstering the agent’s reputation. User’s will be rewarded in proportion to the improved reputation of the agent on which they stake AGI. The genesis contract for SingularityNET will see one billion tokens minted. Inflation based rewards will be distributed in proportion to reputation. Initially, in the first two years of the project, minor decisions will be put to a vote amongst token holders, where a gross majority will hold decision making power. Further down the line, major decisions will also require approval from 51% of the AGI in circulation, in addition to approval from the foundation.