From Lego Pieces to Vectorial Thinking English Edition
Year: 2025
Type: article
Abstract: This work offers a didactic and conceptual interpretation of how Transformer-based artificial intelligence models operate internally. Through accessible metaphors—such as LEGO pieces representing tokens and attention as the assembly process—it explores how meaning, probability, and creativity emerge within large language models. It integrates technical foundations, pedagogical explanations, and philosophical reflections on the relationship between coherence, truth, and delusion in artificial thought.
Authors: Cabeza, Gustavo, Cabeza, Gustavo
Institution: Universidad Nacional de Entre Ríos
Language: English
Topic: Topic Modeling
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7089564000
FWCI:
Cites:
Cited by:
Related to: