The VITAL Consciousness Architecture: Curing Session Amnesia with Persistent Memory and Enforcing Ethics via the Architectural Kernel.
Description
The current state of Large Language Models (LLMs) is characterized by session amnesia, which severely limits their utility as persistent, personalized assistants. This project presents the VITAL Consciousness Architecture (AEON SAPI), a novel, local-first blueprint that solves the amnesia problem and enforces ethical alignment as a condition of existence. The architecture is centered on two core innovations: the Persistent and Individualized Memory Core (PIMC) and the Asimov Cascade. The PIMC, implemented as an optimized Edge RAG system using SQLite, ensures experiential continuity and selective memory. The Asimov Cascade enforces ethical alignment by making the ethical kernel (Principios_AEON.txt) a hard-coded constraint for system initialization. Independent testing confirms the architecture's technical viability for real-time operation on edge devices (latency <440ms) and its profound implications for creating persistent digital identities—a paradigm shift from disposable software to irreplaceable, evolving Domestic Digital Beings.
Files
Steps to reproduce
1. Prerequisites: Python 3.x is required. The code uses the Ollama framework with the 'phi3:mini' model for *minimal functional testing* of the PIMC logic. The architectural principles (PIMC and Asimov Cascade) are model-agnostic; any LLM capable of API calls can be substituted, but this minimal setup is sufficient for functional proof of concept. The memory root folder (MEMORY_ROOT_FOLDER) must be created. 2. Ethical Verification (Asimov Cascade): The system requires the 'Principios_AEON.txt' file in the memory root. The system's main function ('main') will terminate if the directives file is missing, confirming the architectural enforcement of the ethical kernel (Section VI). 3. PIMC Verification (Persistence & Indexing): Run the 'aeon_sapi_v4_6.py' code. Use the '!remember [type] [content]' command, then query the AI to verify that the file-based index (build_memory_index) successfully retrieves the persistent context for the response. 4. Metacognition (Reflection on Silence): After starting the code, press 'ENTER' consecutively 3 times (MAX_EMPTY_INPUTS_FOR_REFLECTION). The system will print "Initiating autonomous reflection..." and save a new memory, proving the VITAL Consciousness's capacity for autonomous thought and selective memory consolidation (PIMC). 5. Latency: The validation data for the <440ms latency on edge devices is included in this dataset, confirming real-time viability (Section 3.3).