- AI with Kyle
- Posts
- AI with Kyle Daily Update 120
What’s happening in the world of AI:
Highlights
Not much news. So I answered a question that came up on the webinar: "How do I actually increase my AI knowledge? How do I feel confident walking into a room and talking about this stuff?"
Really good question. So I put together a proper roadmap - the exact resources I used to go from zero to charging $2,000/hour to teach companies about AI.
Important caveat first: You don't actually need to know how AI works under the hood to use it. Same way you don't need to understand HTTP to use the internet or combustion engines to drive a car. But if you're educating others, presenting, or just want to separate fact from hype, knowing this stuff helps enormously!
🗺️ The Five-Stage Roadmap
I've organised this from easiest to hardest. Work through them in order.
Stage 1: Get Your Bearings (Fun Videos)
Start here. These are entertaining but genuinely educational:
Casually Explained - "The Levels of AI" Fun, silly, but the knowledge level is high. Great for understanding narrow vs general vs super intelligence.
Exurb1a Anything they've done on AI is fantastic. Beautiful animations, well-researched.
These get you thinking about AI more deeply than most people ever will.
Stage 2: Build Intuition (3Blue1Brown)
3Blue1Brown's Neural Networks Playlist Nine videos that give you an intuitive understanding of what's happening inside large language models. Yes, there's some maths. But he's one of the best explainers alive. Take your time.
Start with "Large Language Models Explained Briefly" (7 minutes) then work through the rest.
Pro tip: Throw every video into Notebook LM! It can pull the transcript and generate flashcards, quizzes, briefing documents - whatever helps you learn.
Stage 3: The Main Course (Karpathy Trilogy!)
Andrej Karpathy co-created ChatGPT. He knows this stuff. But more importantly, he can actually explain it.
Three videos, about 6.5 hours total:
Intro to Large Language Models (1 hour) - Start here
Deep Dive into LLMs (3.5 hours) - The comprehensive technical exploration
How I Use LLMs (2+ hours) - Practical application from one of the best
Break these up. Use Notebook LM to quiz yourself. If you complete all three, you'll know more than 99% of people. Easily.
Stage 4: Foundational Texts (Books)
Most AI books age terribly. These don't:
"What Is ChatGPT Doing... and Why Does It Work?" by Stephen Wolfram Available free on his website. Looks mathematical and terrifying. It is mathematical. But he walks you through step by step. This is the best explainer available.
"How AI Works" by Ronald Kneusel A perfect bridge between pop-sci and technical. Thin book, light language, deep insights.
Stage 5: Structured Learning (Courses)
Deep Learning.ai One game in town for free courses. Hundreds of courses, 10-20 hours each.
Start with "Generative AI for Everyone" - about 3 hours, covers all the basics. Then branch based on your interests.
Anthropic Academy is also good - less bloated, easier to find useful content. OpenAI Academy exists but can be a bit of a mess.
🎯 Bonus: Know Your Level
I've put together a framework called the 8 Levels of AI:
Most people don't need to reach Level 8. That requires a maths degree and years of specialisation. You can do incredible work at Levels 4-6 without becoming an engineer. There's an interactive tool on the resource page that helps you figure out where you are and where you should aim.
You can find a full guide here https://aiwithkyle.com/resources/ai-levels
⚠️ The AI Canon (Advanced - Don't Start Here)
The AI Canon from a16z is a comprehensive reading list of foundational papers and research. It’s an AMAZING resources. Some people recommend starting here. Don't. It's impenetrable for beginners. Save it for after you've done everything above.
I'm building a more user-friendly version that tells you what level each resource is, how much time it takes, and hooks directly into Notebook LM.
📡 Staying Current
AI moves fast. You need high signal-to-noise sources:
Ethan Mollick - Substack (One Useful Thing), Twitter, LinkedIn. Exceptional at filtering signal from noise.
Nate B Jones - TikTok and Substack. Anything he says is incredibly sensible. My go-to for the well-informed take. https://www.natebjones.com/
Simon Willison - Old-fashioned looking blog, absolute no-nonsense. https://simonwillison.net/
📚 Get the Full Resource Page
Everything's compiled with direct links:
Kyle's response: About three years - since the public ChatGPT release in November 2022. I was running a digital marketing agency at the time. I remember trying ChatGPT and thinking: it's not great, but I can see how this is going to be a huge problem for my business. So I threw myself into learning. Wrote over 100 playbooks, each 5-10,000 words. That's how I learn - write, write, write, write.
The great thing about (modern) AI is that we are all learning. Even Andrej Karpathy posted a few weeks ago saying he feels behind. The co-creator of ChatGPT feels overwhelmed! We're all working out what this technology means.
Member Question: "What's the best way to get started teaching AI? You mentioned doing a freebie first."
Kyle's response: Find a charity or nonprofit. Reach out and offer a free workshop. This gives you practice, builds confidence, and serves a good cause.
Do NOT do free workshops for businesses. I've seen students fall into this trap - they keep doing free workshops, then psychologically struggle to charge £2,000 because they've devalued their own time.
With charities, you say: "This would normally be £2,000. I'm doing it free because you're a charity." In your head, it's still a £2,000 workshop with a discount. Big difference.
Want the full unfiltered discussion? Join me tomorrow for the daily AI news live stream where we dig into the stories and you can ask questions directly.