ZENIL, Hector. On the Limits of Self-Improving in LLMs and Why AGI, ASI and the Singularity Are Not Near Without Symbolic Model Synthesis. SuperIntelligence - Robotics - Safety & Alignment, [S. l.], v. 2, n. 4, 2026. DOI: 10.70777/si.v2i4.17159. Disponível em: https://s-rsa.com/index.php/agi/article/view/17159. Acesso em: 19 may. 2026.