Should AI Really Slow Down?
AI needs stronger safety controls. But better hardware, efficient data centres and secure distributed computing also deserve a place in the debate.
Independent. Experience-driven. Practical.
Research, analysis, and practical guidance from someone who has worked in technology for over 30 years — from BBS sysop to AI practitioner. Experience that predates the current hype cycle by decades.
Background
An independent researcher and engineer with over 30 years in IT. Career spanning BBS systems, early network infrastructure, enterprise automation, and now deep engagement with AI systems in production environments.
No vendor partnerships. No investment positions. No marketing agenda. Someone who has seen enough technology cycles to recognise the difference between genuine capability and well-funded narrative.
// Practical insights. Honest assessments. Documented experience.
Full background →Latest
AI needs stronger safety controls. But better hardware, efficient data centres and secure distributed computing also deserve a place in the debate.
Original
A three attempt test of whether ChatGPT could follow explicit creative direction, adapt after correction and escape repetitive AI thumbnail conventions.
Practical
A no-fluff walkthrough of calling Claude's API for the first time — authentication, basic completions, streaming, and patterns for production use. Python and TypeScript.
A complete walkthrough — document loading, chunking, embedding, vector retrieval, and generation — with real, runnable Python code.