Random Topics with Jack

Join host Jack on Random Topics with Jack for a clear, beginner-friendly tour of artificial intelligence — what it is, why we built it, and how it actually works. In this episode Jack walks listeners through the core idea that underpins modern AI: pattern recognition used to make predictions. He explains the AI family tree, including machine learning, deep learning and neural networks, reinforcement learning, evolutionary AI, and the generative AI branch that creates text, images, music and video.

The episode breaks down how large language models (LLMs) like ChatGPT function at a fundamental level: tokens, context windows, next-token prediction, pre-training, fine-tuning, and reinforcement learning from human feedback (RLHF). Jack emphasizes the practical limits of current systems — they don’t think or feel, they predict — and previews why that leads to both powerful capabilities and problems like hallucinations.

Jack also pulls back the curtain on the physical infrastructure behind AI: data centers, GPUs, cooling, electricity demand, battery energy storage, and the environmental and community trade-offs involved in locating and powering facilities. The episode reviews who’s building the AI ecosystem — from broad-stack players like OpenAI and Google to specialized companies such as Anthropic, MidJourney, Runway, Suno and others — and explains how products often combine shared, open-source and proprietary models.

Key takeaways include an accessible mental map of terms (LLM, generative AI, tokens, pre-training, fine-tuning, RLHF), an understanding that AI is an ecosystem rather than a single tool or company, and a reminder that behind every AI answer is real hardware, energy use, and design choices. Jack closes by setting up the next episode, which will dig into the ethical, legal and societal implications of deploying AI widely.