AndrewAltimit

AI Safety Training Guide for Human-AI Collaboration

Introduction

This guide draws heavily from the educational content created by Robert Miles (LinkedIn), whose work in AI safety communication has made complex technical concepts accessible to broader audiences. The videos referenced throughout this document are primarily from his collaborations with Rational Animations, and his dedicated AI Safety YouTube channel provides deeper technical explorations of these topics. This guide serves as a structured hub to help practitioners navigate and apply the safety concepts Rob has explained so effectively.

As AI systems become increasingly capable and integrated into critical workflows, understanding AI safety principles is essential for anyone working with AI agents. This guide provides practical knowledge about potential risks, safety measures, and thought experiments to help you work more safely and effectively with

AndrewAltimit

Rethinking "Artificial" Intelligence

We call it artificial intelligence. But is it really artificial?

To many the word "artificial" suggests something fake or lesser than and carries a negative connotation (e.g. artificial personality, artificial sweeteners). Yet the systems we’re building are neither toys nor simple imitations, they’re powerful, evolving forms of intelligence that shape our daily lives.

Language frames how we see technology. If we label machine intelligence as "fake" we may limit how we think about it and how we prepare for its impact.

Forms of Intelligence

AndrewAltimit

DBR Environment Setup Helper - Technical Specification

Version: 1.1 Date: September 14, 2025 Status: Implementation-Ready Infrastructure: GitHub Actions with Self-Hosted Linux Runners Purpose: Generate Python dependency wheels and system installation scripts for Databricks Runtime environments

Executive Summary

AndrewAltimit

QR Code File Transmitter

Convert any file into a series of QR codes for transmission via photos, then reconstruct the original file from the QR code images. Perfect for air-gapped systems, conference presentations, or creative file sharing.

Quick Start

Installation

# Install required packages

AndrewAltimit

TTS Demo (ElevenLabs v3 Alpha)

🔊 Click here to listen to the TTS Demo

Or download: tts-demo.mp3

---

Transcript:

[thoughtful] This is an impressive and ambitious pull request!

AndrewAltimit

URL fetcher with imported cookies

A powerful Python script for fetching content from any URL with advanced features including automatic cookie loading, authentication support, and SSL options.

Features

• Universal URL Fetching: Works with any HTTP/HTTPS URL
• Automatic Cookie Loading: Load cookies from Chrome, Firefox, and Waterfox browsers (macOS)
• Authentication Support: Basic auth and custom headers
• SSL Options: Disable SSL verification for development/testing

AndrewAltimit

AI Agent Training Guide for Closed Source Software

Overview

This guide outlines the process for integrating and training AI agents to work with closed source software applications. The approach focuses on establishing plaintext entrypoints and progressive learning phases.

AndrewAltimit

AI generated video of my cat Pixel made on local compute

Resources

• Wan2.2 Guide
• Containerized ComfyUI MCP
• Containerized AI-Toolkit MCP

AndrewAltimit

AI Toolkit MCP Server (local diffusion AI model trainer)

Warning: Requires a powerful GPU!

A containerized AI Toolkit setup with MCP (Model Context Protocol) integration for training LoRA models and fine-tuning diffusion models. This provides a complete solution for training custom LoRA models with full MCP integration, allowing AI assistants to manage the entire training workflow.

Usage

See the template repository for a complete example. Also includes the ComfyUI MCP Server used for creating images/videos from the trained models.

AndrewAltimit

ComfyUI MCP Server (local image/video generation)

Warning: Requires a powerful GPU!

A containerized ComfyUI setup with MCP (Model Context Protocol) integration for AI-driven image generation workflows. Includes both standard MCP (stdio) access and HTTP API access.

Usage

See the template repository for a complete example. Also includes the model training mcp used for creating checkpoints / loras.