Brian Jones

Case-Based Moral Reasoning Without an LLM: Jaccard Similarity as Ethical Judgment

How do you build an ethics system for an autonomous AI agent that runs 24/7, evaluates hundreds of actions per day, and can't afford an LLM call for every moral decision? You turn ethics into set comparison. This post describes how we built a casuistic moral engine

Deduplicating AI Thoughts at Scale: MinHash Fingerprinting in Practice

When your AI agent generates thousands of thoughts per hour, many of them start to look the same. Not identical — that would be easy to catch — but structurally similar: the same concepts rearranged, the same patterns with slightly different phrasing. We needed a way to detect this repetition in real