AI Guides › Workbench

The AI Literacy Ladder: Three Rungs, One Source Each

By Nigel Guy · 6 min read

Most people who want to "understand AI" subscribe to six newsletters, bookmark forty videos and finish none of them. It feels like progress because the pile keeps growing, but a pile is not understanding. A few well-chosen sources, read in order and tested against something you actually do, will take you further than a feed ever will.

This kit is a short, checked list of free or cheap sources. Each is something we could confirm exists and is described accurately at the time of writing. Where we could not confirm something, we say so below rather than pad the list.

The kit at a glance

The ladder

Rung 1: Beginner, get the shape

Goal: be able to explain, in your own words, what a model is, what training means, and why chatbots get things wrong.

1. Take Elements of AI's introductory course at your own pace. It is built for non-specialists.
2. Watch AlphaGo or The Thinking Game on a quiet evening. Treat them as context for why people care, not as technical teaching.
3. Subscribe to one newsletter only. One Useful Thing is the gentlest start.

Rung 2: Intermediate, get the mechanism

Goal: know why the tools behave as they do, and use one daily.

1. Watch Karpathy's talk. It explains a chatbot as a next-word predictor and covers training, tools and jailbreaks. Pause and take notes.
2. Read one book: Co-Intelligence if you want practical habits, AI Snake Oil if you want a defence against hype.
3. Add The Batch for weekly news once you have enough context to sort it.
4. Pick one assistant and use it on real work. Claude's free plan is $0 and Pro is $20 a month billed monthly, or $17 a month on annual billing, which is about £15 and about £13 at time of writing. Check the £ price at checkout, as the pricing page shows only dollars.

Rung 3: Advanced, get the argument

Goal: hold an informed view on risk, policy and limits.

1. Read The Alignment Problem or Human Compatible, then argue with it. Read AI Snake Oil too if you took the other route earlier, so you meet both sides.
2. Add Import AI or Latent Space depending on whether you lean towards research and policy or towards building.
3. Start reading primary documents: model release notes and vendor documentation, not summaries of them.

Make it stick: the Teach-Back Test

After each source, write five sentences explaining it to a friend with no background, then name one thing you will try this week. If you cannot, you consumed it rather than learned it. Use this prompt to get an honest critique of your five sentences.

You are a patient tutor helping an adult beginner build AI literacy.

Context: I have just finished [SOURCE_NAME], which is a [FORMAT]. My level is [BEGINNER / INTERMEDIATE / ADVANCED].

My task: Below is a five-sentence explanation of what I learned, written for a friend with no background.

[YOUR_FIVE_SENTENCES]

What a good result looks like: you tell me what I got right, what I got wrong or oversimplified, and what I left out that matters.

Steps:
1. Check each sentence against what you know about the source. If you do not know the source well, say so and ask me for the key claims instead of guessing.
2. List errors first, then gaps, then what is solid.
3. Ask me two questions that test whether I understand the idea rather than the wording.
4. Suggest one small task I could do this week with [TOOL_OR_WORK_CONTEXT] to apply it.

Constraints: plain English, no jargon without a one-line definition, no invented facts or quotes. If you are not sure something is true, mark it as uncertain.

Before answering, re-read your feedback and delete anything you cannot defend.

Fill in the source, your level, your five sentences and a tool or work context you use.

How to choose

• Short on time: Elements of AI, Karpathy's talk and one newsletter.
• Worried about hype: AI Snake Oil before any product-led source.
• Want to build: Karpathy, then Latent Space.
• Want the ethics and policy side: The Alignment Problem, then Import AI.

What to skip

• A second newsletter before you have finished the first.
• Daily news feeds. Weekly is enough to see patterns.
• "Top 100 AI tools" lists. Learn one assistant properly.
• Books that promise to make you an expert in a weekend.
• Any source you have not finished but keep re-subscribing to.

Guardrails

• This list is a starting point, not the field. We confirmed each entry exists and what it is. We have not verified the claim that any of them is the best, and we cannot vouch for every episode, issue or chapter.
• Several sources come from people or organisations with a stake in AI: DeepMind made both documentaries available, and Import AI is written by an Anthropic co-founder. Read them as informed perspectives, not neutral verdicts.
• Prices change. Check the £ price at checkout before paying for anything.
• Not confirmed and so left out: specific YouTube channels beyond the Karpathy talk, and podcasts beyond Latent Space. Check any recommendation you find elsewhere before building a plan on it.
• Do not paste confidential work or personal data into an assistant while practising.

Sources

• Elements of AI
• One Useful Thing: About
• The Batch, DeepLearning.AI
• Import AI: About
• Latent Space: About
• Claude pricing
• Google DeepMind blog: The Thinking Game
• AI Snake Oil, Princeton research news
• Simon Willison on Karpathy's Intro to LLMs

All 751 AI guides · JulieMango plans from £17/mo