Skip to content

Latest commit

 

History

History
45 lines (23 loc) · 2.77 KB

README.md

File metadata and controls

45 lines (23 loc) · 2.77 KB
Raw

EchoCortex: AI Inner Dialogue Framework

EchoCortex aims to make Language Learning Models (LLMs) such as ChatGPT more human-like by introducing an 'inner dialogue'. This is a structure that mirrors the cognitive processes in human beings, enabling an AI to reason, plan, and make decisions effectively.

Key Components

The framework consists of six main elements:

  1. Sensory Organ: Analogous to human senses, this component receives input data. For ChatGPT, this is the user's prompt. In a more complex system, it could include different input devices.

  2. Stimulus: The sensory input from the Sensory Organ. For instance, the prompt in the case of ChatGPT.

  3. Actuator: This is how the AI influences the world. For example, the response generated by ChatGPT.

  4. Action: The specific output from the Actuator. This could be the generated reply from ChatGPT or a command to move a robot.

  5. Inner Dialogue: This component allows the AI to propose, evaluate, and select from different potential actions, enabling the AI to 'think before it speaks'.

  6. Monitoring Function: This supervisory function ensures that the AI is not stuck in a loop and its decisions do not violate certain ethical or axiomatic constraints.

Chain-of-Thought and Tree-of-Thought Reasoning

Chain-of-Thought (CoT) and Tree-of-Thought (ToT) are strategies that inform this framework. CoT is a linear approach that tackles a problem step-by-step, while ToT explores multiple solutions in parallel. The Inner Dialogue serves to conduct these reasonings internally.

Human Memory Emulation

Different types of memory mechanisms can be introduced to the AI system to mimic human cognition:

  1. Semantic Memory: Provides a reservoir of factual knowledge for the AI to draw upon during the Inner Dialogue.

  2. Episodic Memory: Allows the AI to remember past experiences and incorporate them into its decision-making process.

  3. Working Memory: Acts as a short-term memory that holds the 'thread' of thought as the AI works through a problem.

Towards Greater Autonomy

The framework also aims to increase AI autonomy. Currently, models like ChatGPT react to user prompts but are otherwise idle. This framework introduces an ongoing Inner Dialogue, allowing AI to 'think' continuously, respond to asynchronous stimuli, and wait based on specific criteria. This gives the AI the ability to consider problems deeply and to act autonomously.

Getting Started

For developers interested in contributing to the development of this framework, you can start by cloning the repository and examining the available codebase. Look at the issue tracker for current tasks and features in need of assistance.

License

This project is licensed under the terms of the MIT license. For more information, see the LICENSE file.