Summary

AI-DSL Technical Report (February to May 2021)

Nil Geisweiller, Kabir Veitas, Eman Shemsu Asfaw, Samuel Roberti, Matthew Ikl´e, Ben Goertzel

August 10, 2021

Retreived from

https://github.com/singnet/ai-dsl/blob/master/doc/technical-reports/2021/ai-dsl-techrep-2021-05_may.pdf

Abstract

This technical report summarizes the work completed between February and August 2021, aimed at realizing the concept of an AI-DSL (AI Domain-Specific Language) initially introduced in a blog post on the SingularityNET website in December 2020. The AI-DSL serves as a language for AI agents operating on a decentralized platform to exchange their pertinent attributes, thereby facilitating diverse forms of agent interaction. As a vital component of a larger initiative focused on establishing a system that supports autonomous interoperability between AI services across a decentralized blockchain-based network, this report offers a comprehensive overview of the achievements made in the prototyping stage thus far, along with plans for forthcoming implementation and experimentation.

Chapter 1

Introduction

One big part of the SingularityNET vision is the creation of a decentralized “society of minds” in which multiple AI agents can cooperate and collaborate to solve problems together – so that the SingularityNET network as a whole becomes a complex self-organizing system in which the intelligence of the whole greatly exceeds the intelligence of the parts. 

The current SingularityNET marketplace comprises mainly agents that carry out AI functions for customers on their own, without referencing other agents on the platform in their own back-end operations. There is no obstacle to connecting multiple agents on the current network into complex assemblages, however, the platform also does not provide any particular tools designed to make this easier and more natural.

The SingularityNET whitepaper alludes to a meta-API or API-of-APIs according to which AI agents can describe to each other various aspects of their operation, to make it easier for agents to make automated decisions about which other AIs to connect to for what reasons. However, the whitepaper does not describe in any detail how this API-of-APIs would be designed or implemented. 

This technical gap is filled by the concept of an AI-DSL or “AI DomainSpecific Language”, by which is meant not a language for the implementation of the nitty-gritty internal operations of AI algorithms or data structures, but rather a language that AI algorithms can use to describe their properties to each other, so as to allow agent interactions to be automatically determined in an appropriately knowledgeable way.

Much as the SingularityNET platform could be used to host any sort of software agents, not just AI-based agents, similarly the AI-DSL could be used to allow any sorts of software processes to describe their properties to other software processes. However, both SingularityNET and the AI-DSL are being initially tuned for particular effectiveness in the AI context, in ways that do not constrain their potential for more general application. 

The high level concept of the AI-DSL, and the first steps toward technical specification of the idea, were presented in a December 2020 blog post, AI-DSL [23]. Since that time considerable work has been done toward an initial technical 3 implementation of the AI-DSL. This document describes much of that work, and then briefly discusses what still needs to be done to move from these initial achievements toward an AI-DSL that can be broadly and efficiently deployed in the SingularityNET platform.

Full paper: https://groupkos.com/dev/images/Ai-dsl-techrep-2021-05_may.pdf