SystemHalted — Palak Mathur on software engineering, leadership, and Emacs
Recent posts
-
What UNO Teaches About Rule-Based Agents
A small UNO-playing production system shows both the usefulness and the limits of rule-based agents.
-
Review: How Technological Advances Can Reveal Rights
A review of Parker and Danks's argument that some rights become visible only when technology changes the balance of power.
-
Why Doesn’t Wisdom Accumulate?
We are extraordinarily good at accumulating technical knowledge and strangely bad at accumulating behavioral wisdom. What decades of psychology suggest about why every generation relearns patience, prejudice, and restraint almost from scratch.
-
Review: An Agile Ethical/Legal Model for AI and Robotics Governance
A review of Wallach and Marchant's proposal for a Governance Coordinating Committee for AI and robotics, and why soft law needs institutional support.
-
Is an Anthill Conscious?
An anthill is a useful way to think about collective intelligence, emergence, and the limits of calling a system conscious.
-
Diagrams as Text
An image model returns finished pixels; a language model returns text. SVG, Mermaid, DOT, and PlantUML store a picture as text, so a language model can draw one, allowing you to edit what it draws.
-
Devanagari in Unicode
This post looks at how Devanagari actually works in Unicode
-
Semantic Networks as Search Spaces
A short note on semantic networks, generate-and-test search, and why explicit state representation matters in knowledge-based AI.