Episode Overview

In this insightful episode of AI for Lifelong Learners, host Tom Parish sits down with Preston McCauley, author, educator, and software developer who wrote "Generative AI for Everyone: A Practical Guidebook."

Preston brings his unique perspective, gained from teaching, curriculum development, and hands-on AI system building, to share practical insights about mastering artificial intelligence.

He reveals how he built an AI system to teach himself AI and discusses his approach to staying ahead of technology trends by "living five years in the future." The conversation covers essential topics for anyone looking to effectively utilize AI tools, including Preston's CLEAR methodology for prompting, the current state of various AI models such as ChatGPT-5 and Claude, understanding AI agents, and the growing importance of open-source models. Preston demystifies complex AI concepts and provides practical frameworks that help transform AI from a simple chat tool into a true collaborative partner, making this episode essential listening for both beginners and experienced AI users.

Book Snapshot

Generative AI for Everyone by Preston McCauley is the ultimate guide to understanding and harnessing the potential of artificial intelligence. Whether you're new to AI or an experienced professional, this book equips you with the tools to revolutionize your work, learn, and create in an AI-driven world.

This guide's core is prompt engineering, a critical skill for effectively communicating with AI systems like ChatGPT and other advanced models.

Learn how to:

• Craft effective prompts by mastering key elements like context, clarity, constraints, and adaptation.

• Refine your prompts iteratively to achieve precise, high-quality outputs.

• Use proven frameworks such as CLEAR AI (focusing on 15 essential elements) and FOCUS AI (creating reusable templates for various applications).

With practical examples, including a creative take on Goldilocks and the Three Bears, you'll gain hands-on experience in constructing and perfecting prompts.

What You'll Learn in This Episode

[00:33] - Preston's "Living Five Years in the Future" Philosophy

• Preston's work mantra of staying ahead of mainstream technology adoption

• Building his first AI system specifically designed to teach him about AI

• The importance of immersing yourself in new technology, like learning a new language

• How this approach helped him anticipate the evolution of GPT and early AI systems

[03:10] - Reverse Engineering the Learning Process

• Creating an AI-powered curriculum and syllabus for learning AI

• Breaking down complex concepts into teachable structures

• Learning faster through AI assistance than traditional methods

• The importance of understanding things in a way you can teach them

[05:12] - Writing "Generative AI for Everyone"

• The book's intentionally timeless design and methodology

• Using AI assistants as a team to edit and review the book

• Applying the Goldilocks principle ("getting it just right") to AI interactions

• Creating content that flows conversationally rather than lecturing at readers

[07:41] - Who Benefits from AI Training

• Designing content for diverse demographics beyond tech professionals (gardeners, story writers, etc.)

• Breaking through the "AI is just chat" mindset

• Moving from using AI as a tool to collaborating with it as a partner

• The visible "gasp" moment when people grasp AI's collaborative potential

[10:41] - Common Barriers to AI Adoption

• Trust and privacy concerns are holding people back

• Fear of "breaking" something when experimenting with AI

• The challenge of sycophantic AI responses that always agree

• Understanding AI inference (like Wheel of Fortune) vs. factual truth

[14:19] - Comparing Current AI Models

• Claude 4.1 excels at deep written thought

• ChatGPT-5 offers a good balance but requires proper prompting structure

• Perplexity combines search engine functionality with AI

• Gemini 2.5 for text generation

• The reality that businesses don't need "supercomputer" level AI for most tasks

[18:48] - ChatGPT-5 Deep Dive

• Managing expectations vs. reality for the new model

• The thinking mode and when to use it vs. faster responses

• Cost and energy challenges of training advanced models

• Why technical users may be disappointed while everyday users find it sufficient

[21:57] - Cross-Platform Prompting Strategy

• Running the same prompt across multiple LLMs for comparison

• Understanding that different models provide different perspectives

• All LLMs use similar training datasets but produce varied outputs

• The encyclopedia analogy - different sources tell stories differently

[24:18] - Identifying AI-Generated Content

• The telltale word "unlock" frequently appears in AI content

• Common patterns across different AI models

• Techniques for getting deeper, less generic responses

• The importance of reflection and going beyond first-level results

[27:16] - The CLEAR Methodology Framework

• Clarity: Setting the AI on the right path (GPS route entry)

• Limits: Defining boundaries and constraints (routes along the way)

• Examples: Showing what good looks like, not just telling

• Adaptation: Adjusting when obstacles arise (like construction on I-635)

• Reflection: Both AI and human reviewing the output

• Why this framework is more important than ever with GPT-5's thinking model

[31:28] - Practical Prompting Tips

• Adding confidence rankings (1-5 scale) to AI responses

• Using "Does this make sense?" as a quick reflection check

• Asking AI for clarification questions before proceeding

• Building nested inference and establishing references

• Not going more than 3-5 requests deep to maintain context

[35:44] - Understanding AI Agents

• Agents as specialized team members with specific roles (like a marketing team)

• The difference between task-based agents and truly agentic systems

• The importance of providing GPS-like structure to prevent wandering

• Achieving 50% workflow automation with human review • CrewAI as an example of truly agentic systems

[40:36] - Open Source and Local Models

• LM Studio, Ollama, and GPT4ALL for running models locally

• Hardware requirements: 13 billion parameters or less for responsive performance on Mac M1 Pro

• Privacy benefits of running models locally

• ChatGPT OS 20B is the current best open-source model size

• Memory optimization advances allowing larger models on consumer hardware

[49:02] - Beyond Fine-Tuning

• Why fine-tuning isn't always necessary anymore

• Alternative techniques like advanced RAG (Retrieval-Augmented Generation)

• Cost-effective approaches to domain-specific knowledge

• Building medical AI models with minimal training using Unsloth

[53:07] - Preston's Future Projects

• New intelligent AI website with six AI personalities

• MELD framework (Model Engagement Language Directive) - open-sourced ver 1

• Upcoming book: "Generative AI for Everyone: Images"

• New frameworks for complex image and brand structures

Resources

* Book: "Generative AI for Everyone: A Practical Guidebook" by Preston McCauley

* Reference: The Goldilocks principle applied to AI interactions

* Newsletter: The White Box by Ignacio (Nacho)

* Organizations: OpenAI, Anthropic, Google, Hugging Face

* Tools Mentioned: ChatGPT-5, Claude 4.1, Perplexity, Grok, Gemini, LM Studio, Ollama, GPT4ALL, CrewAI, Keras, Unsloth, Google Colab

Connect

* LinkedIn: Preston McCauley

* Website: clearsightdesigns.com

* Book (physical copy): books.by/clearai

* Amazon: "Generative AI for Everyone: A Practical Guidebook"

Enjoyed the episode? Share these notes and help more learners discover AI insights!

Get full access to AI for Lifelong Learners at aiforlifelonglearners.substack.com/subscribe