H. Zenil, “On the Limits of Self-Improving in LLMs and Why AGI, ASI and the Singularity Are Not Near Without Symbolic Model Synthesis”, SI, vol. 2, no. 4, Jan. 2026.