Writing Hub
AI governance essays, reasoning systems notes, experiment logs, and technical writing across BioAI and engineering practice.
Project Topics
Beyond M15: Why STEM BIO-AI Started Acting More Like a Governance Report in v1.8.x
STEM BIO-AI v1.8.x moved beyond M15 integration by turning its audit output into a clearer governance report with bounded scores, traceability, and release integrity.
AI Can Write the Code. It Still Cannot Place the Stone.
AI can now write code, patch files, and finish releases. But a real case from an AI-assisted release shows that the harder human work may be deciding what the system should expose, which output belongs to which reader, and how agent-generated work remains inspectable after the code is written.
We Made a High-Formality, Fake Physics Slop Artifact - QSOT (Quantum State Over Time) Compiler
A post-mortem on QSOT Compiler v1.2.3, a high-formality AI-generated scientific software artifact that looked rigorous but failed core reproducibility and claim-validation checks.
The Quality Author: Taste as the Last Bottleneck in AI Development
On where craftsmanship went, why verification gaps appear in its absence, and the one practice AI cannot automate for you.
When the Memory Gate Met a Real Archive: What 90 Experiments Taught Us About Cheap LLM Slop
How to enforce data integrity against AI-generated slop using MICA. Explore a 11-step session-start validator that locks rules, playbooks, and contracts in code before code is ever touched.
The Two Problems No One Talks About in AI Agent Coding Pipelines
AI agent coding pipelines fail not because models are weak, but because verification is structurally broken. This article identifies four empirically documented failure mechanisms — agreement bias, latent entanglement, echoing, and right-for-wrong-reasons — and proposes a concrete architecture: hash-chained audit records, hybrid recurrence scoring, dynamic context budgets, and evidence-first review across three independent axes. Covers multi-agent pipeline design, agentic code review, blueprint indexing, and P0–P4 governance gates.
Stanford. Princeton. A bioRxiv Paper. So Why Did Nobody Ask Where the Data Goes?
BioClaw processes EHR data. Its primary showcase channel is WhatsApp. We audited the repository: 60/100, Tier 2 Caution. Here is what the bioRxiv paper says that the README does not.
The README Was a Protocol. The Entrypoint Was Still Optional.
README-as-Protocol solved explicit invocation at the schema level. It did not solve entry control at the workflow level. This version adds the missing hierarchy: natural, guided, and forced activation.
STEM-BIO-AI Audit Report: yorkeccak/bio
When a README Claim Meets a Deterministic Scanner
Beyond Repo Scanning: How AIRI Expanded the Risk Vocabulary in STEM BIO-AI 1.7.x
How STEM BIO-AI uses the MIT AI Risk Repository as a governed local risk-vocabulary layer without replacing deterministic repository scanning
Building a Deterministic Governance Kernel: Separating Custody from Truth
CGF separates domain truth from custody mechanics, turning AI governance from Markdown/YAML policy language into deterministic, inspectable artifacts.
Each /slop Is a Calibration Signal — AI-SLOP Detector v3.6.0 and the Claude Code Skill
Every /slop invocation records to a project-scoped history. After 10 re-scanned files, bounded self-calibration adjusts detection weights for your codebase. Here is the mechanism, the data, and what actually shipped in v3.6.0.
Showing page 1 of 5 · 53 matching posts