Eugene Asahara, Soft Coded Logic
Putting the "I" Back into BI
TL;DR AI does not need to become superintelligent, reach the singularity, whatever, to create serious risks. The deeper problem is that no intelligence operates with perfect information. Better models can reason faster and make better inferences, but they are limited in conjuring facts that are missing, hidden, stale, deceptive, or unknowable. That has been obvious … Continue reading Even ASI Cannot Escape Imperfect Information
Abstract In The Assemblage of AI architecture, System 2 is modeled after the deliberate reasoning processes described in human cognition. But rather than focusing on psychology, this article examines what System 2 produces inside an AGI system. The outputs are not merely answers to questions but structured artifacts such as stories, plans, procedures, and models. … Continue reading The Products of System 2 – Prolog in the LLM Era – Spring Break 2026 Special
Situation Over the past few months there has been a trend to move from referring to "large language models" (LLM) to just "language models". The reason for that is the recognition that small language models (SLM) have clear advantages under certain situations. So "language model" is inclusive of both. And of course, as soon as … Continue reading Long Live LLMs! The Central Knowledge System of Analogy
