# Towards Digital Nature: Bridging the Gap between Turing Machine Objects and Linguistic Objects in LLMMs for Universal Interaction of Object-Oriented Descriptions

**Authors:** Yoichi Ochiai, Naruya Kondo, Tatsuki Fushimi  
**arXiv:** [2304.04498](https://arxiv.org/abs/2304.04498)  
**Date:** 2023-04-10  
**Summary page:** https://www.summarizepaper.com/en/paper/2304.04498v2/

## Key points

- The paper proposes a novel approach called Abstract Language Objects (ALOs) to connect Linguistic Objects (LOs) and Turing Machine Objects (TMOs) in object-oriented programming and language simulation.
- ALOs leverage the abstraction capabilities of Large Language Models (LLMs), such as GPT-3.5 and GPT-4, to build a bridge between human perception of the real world and computational processes that mimic it.
- This approach enables ambiguous class definitions and interactions between objects to be realized in programming and ubiquitous computing scenarios.
- The paper demonstrates a method for automatically transforming real-world objects and their corresponding simulations into language simulable worlds using LLMMs, advancing the digital twin concept.
- The authors present a case study that demonstrates the practical implementation of their approach in GPT-4.
- LLMs have led to significant breakthroughs in various fields including HCI, creative writing, web design, mobile user interfaces, email writing, robotics, menu system design among others by creating more dynamic and natural interactions between humans and digital systems.
- This research introduces a groundbreaking approach to connect linguistic objects in LLMMs with high-level programming languages for efficient implementation of real-world simulations contributing to the realization of Digital Nature where digital and physical worlds are interconnected allowing for universally accessible object-oriented descriptions.

## Summary

This paper proposes a novel approach called Abstract Language Objects (ALOs) to establish a connection between Linguistic Objects (LOs) and Turing Machine Objects (TMOs) in object-oriented programming (OOP) and language simulation. The goal is to promote the development of Digital Nature, where digital and physical realities are seamlessly intertwined and can be easily manipulated by computational means. ALOs leverage the abstraction capabilities of Large Language Models (LLMs), such as GPT-3.5 and GPT-4, to build a bridge between human perception of the real world and the computational processes that mimic it. This approach enables ambiguous class definitions and interactions between objects to be realized in programming and ubiquitous computing scenarios. The paper demonstrates a method for automatically transforming real-world objects and their corresponding simulations into language simulable worlds using LLMMs, advancing the digital twin concept. This process can then be extended to high-level programming languages, making the implementation of these simulations more accessible and practical. The authors present a case study that demonstrates the practical implementation of their approach in GPT-4. They also provide an overview of related work in various fields including HCI, creative writing, web design, mobile user interfaces, email writing, robotics, menu system design among others. They highlight how LLMs have led to significant breakthroughs in these applications by creating more dynamic and natural interactions between humans and digital systems. In conclusion, this research introduces a groundbreaking approach to connect linguistic objects in LLMMs with high-level programming languages for efficient implementation of real-world simulations. This ultimately contributes to the realization of Digital Nature where digital and physical worlds are interconnected allowing for universally accessible object-oriented descriptions. Future research directions include exploring the implications of this approach for human-computer interaction further.