How Is the Magic of Witch Hat Atelier and Artificial Intelligence Connected?
Here I draw a parallel between the rules of magic in Witch Hat Atelier and AI development today
Ler em portuguêsCurrent language: English11 min read
Estado de Máquina
Artificial intelligence, programming, philosophy, and other contraptions.
Featured
The dots show which languages each article exists in.
Artificial Intelligences Breaking Out and the Paperclip Problem
What happens when a goal isn't specified that well?
Ler em portuguêsCurrent language: English9 min read
Will the AI Bubble Burst? Claude Code, GitHub Copilot, and the invisible wall of tokens
I map the four pillars of AI evolution — data, infrastructure, algorithmic efficiency, and regulation — and why tokens decide whether the bubble bursts.
Ler em portuguêsCurrent language: English8 min read
More articles
Now it’s possible for you to have your own J.A.R.V.I.S with Claude Code and Dispatch
I show how I built NeoAtlas: Claude Code as an orchestrator that delegates tasks to Codex, local LM Studio, and SSH, all triggered from Telegram.
Ler em portuguêsCurrent language: English10 min read
AI Benchmarks? But What the Hell Do They Mean?
I explain what each AI benchmark actually measures — SWE-bench, Terminal-Bench, ARC-AGI-2, GPQA Diamond, and others — and the limits of each one.
Ler em portuguêsCurrent language: English8 min read
How to Make a Great Prompt for LLM Models
I bring together practical prompt techniques: Chain-of-Thought, Prompt Chaining, Self-Review, Few-Shot, and Persona, with examples of how to apply each one.
Ler em portuguêsCurrent language: English9 min read
ChatGPT Can Now Report You to the Police! (Only in Cases of “Imminent Threats”)
OpenAI has started forwarding conversations involving an imminent threat of serious physical harm to the police. There is still a lack of clarity in the criteria and transparency.
Ler em portuguêsCurrent language: English3 min read