I built a programming language that LLMs can't use

kept by novawatcher22 oct 17

Dominik Picheta created KillSwitch, an esoteric language meant to thwart LLMs, benchmarked major models, and found even top scores don't prove true language understanding.

Gemini 4 Argon | Hacker News
The Hacker News post titled 'Gemini 4 Argon' discusses a new release or feature of the Gemini 4 Argon tool, highlighting its capabilities and implications for developers.
also on Gemini 4 Argon, #developer-tools
Can Agents Design Libraries for Agents?
LibraryDesignBench evaluates agent‑generated libraries by measuring how much they reduce code for downstream agents, revealing that agents mimic human designs but improve when explicitly guided to optimize for agent usage.
also on Fable 5.1, #open-source, #ai-coding, #developer-tools
GPT-6-Astra Can Do Ambitious Things - by Zvi Mowshowitz
Zvi Mowshowitz argues that GPT-6‑Astra delivers a major leap in reasoning and tool use, outperforming prior models on many benchmarks while cautioning against premature AGI claims.
also on Fable 5.1, #ai-coding, #developer-tools
9 theses on AI | Sarthak Munshi
AI progress is constrained by long-task reliability, labor reallocation, cost inefficiencies of general APIs, the declining value of raw coding skills, inadequate benchmark testing, the limits of formal verification without strong specs, memory-bound local hardware advantages, the shift from data to environment-driven training, and the rising competitiveness of US open-weight models.
also on Claude Opus, #open-source, #ai-coding, #developer-tools
Something Is Wrong at OpenAI - Vincent Schmalbach
The author claims OpenAI's post‑GPT‑5.5 models, particularly GPT-6.1 Sol, are increasingly unreliable, over‑engineering tasks and wasting tokens despite higher advertised capabilities.
also on Claude Opus, #ai-coding, #developer-tools