Towards Explainable and Language-Agnostic LLMs

Explore a groundbreaking approach to overcoming Large Language Models' limitations in natural language processing and artificial intelligence in this 36-minute conference talk by Walid S. Saba. Discover a novel solution integrating symbolic representations with LLMs' empirical power through bottom-up reverse engineering of language. Learn about creating interpretable, language-agnostic models that bridge the gap between symbolic AI and LLMs. Gain insights into advancements in machine learning training, data engineering, and explainable AI. Understand how this innovative approach could redefine AI interaction, making it more accessible and understandable across languages and cultures.

- Introduction
- Agenda
- What Happened in AI/NLU
- LLMs: Bottom-up reverse engineering of language
- Limitations of LLMs
- Symbolic bottom-up reverse engineering of language
- Discovering the Universal? ontology of NL