What is the five-rung AI learning ladder?

The five-rung AI learning ladder is a sequence for turning free AI material into working skill: get taught by the tool, rewire your inputs, apply it the same week, distil it, and teach it. Each rung has a piece of proof, such as a saved curriculum for your job or something built that did not exist last Monday. Rungs 1, 2 and 4 take information in. Rungs 3 and 5 produce output another person can inspect.

How does learning AI differ from keeping up with AI news?

Keeping up with AI news means consuming updates, which covers the input rungs of the ladder. Learning AI means applying the material to real work and being able to explain it under questioning. A simple test separates them. Describe a recent development in three sentences, then add one sentence on what you did with it. People who only follow the news can write the first three and stall on the fourth.

Why do most people stall when learning AI?

Most people stall at the application rung because it is the first rung that can fail in public. Watching tutorials, tuning a feed and saving notes are private and pleasant, and they produce the feeling of progress. Applying a technique to real work the same week risks a broken build or a worse report. The discomfort is where learning starts, which is why the pull back to one more video is so strong.

When should a team lead use the learning ladder with a team?

A team lead should use the ladder whenever AI training is being measured by course completions or hours watched. Those metrics only capture the input rungs. Replace the question 'what are you learning?' with 'what did you build with it last week?' and ask for a ten minute demonstration. The answers show which rung each person is on and give a far better read on real AI fluency than any completion rate.

What is the first step to learning AI without a technical background?

The first step is to ask a free chat model to teach you, with a precise description of who you are. State your role and background, ask it to explain things simply, give it a budget of 15 minutes a day and a target of intermediate level, and narrow the curriculum to examples from your own function. Save the result as a project so the context carries forward, then complete day one immediately.

The Five-Rung Ladder for Learning AI, and the Rung Where Most People Stall

The AI learning ladder is the order in which a working professional turns free material into usable skill: get taught, rewire your inputs, apply, distil, teach. Three of the five rungs take information in. Two put something out where another person can see it. Most of the freshers and early-career professionals I have talked to about learning AI have stopped at the first rung that requires output, because saving a video produces the same feeling as understanding it.

The follow-up question

In the weeks before I went live, I had a run of one-to-one conversations with freshers and people early in their careers. I asked each of them a version of the same question: what has changed in AI recently?

The answers were confident. A few people told me they had gone deep on exactly that period. Then I asked one follow-up question, usually something as plain as how the thing they mentioned actually worked, or where they had tried it. For the ones who claimed to have studied it closely, the confidence ended on that question. They had the headline and nothing underneath it.

I came away frustrated, and I wrote in my journal that evening that it was sad. Learning has never been cheaper. The free tier of ChatGPT will write you a curriculum. Many of the best builders in this field post what they are building on YouTube, for nothing. The World Economic Forum's Future of Jobs Report 2025 estimates that 39% of workers' core skills will change or become outdated between 2025 and 2030, and that 59 of every 100 workers will need training by the end of the decade. On the live session I said I think the real figure is above 50%. On those numbers, I think the gap between the people who keep learning and the people who perform it starts showing up in careers well before the decade is out.

The frustration faded once I checked my own record, because the honest version of the story includes me.

For roughly a year I had been telling people to learn in public, while sitting on a live session I had never run. I kept waiting for the lighting to be right, for the microphone settings to be tested, for a rehearsal I never scheduled. I was doing a quieter version of what those freshers were doing. I had the material and the audience. I was not putting anything out.

So on the morning of 27 July I went live on YouTube and X at the same time. No script, no rehearsal, bad light, a mic I had not checked. I opened by admitting I had delayed it for a year. Then I turned the complaint about freshers into the session itself, and I screen-shared the actual prompts, the actual feed and the actual code editor I learn with.

What I taught that morning was a five-rung ladder. Writing it down afterwards, I noticed something about its shape that I had not seen while I was using it, and that shape explains the follow-up question better than any theory about attention spans.

The Five-Rung Ladder

Five rungs, climbed in order. Each rung has a piece of evidence that proves you are standing on it. If you cannot produce the evidence, you are on the rung below.

  RUNG                      PROOF YOU ARE ON IT               DIRECTION
  5  Teach it               answered a question you            OUT
                            did not see coming
  4  Distil it              one page you could hand             in
                            to a colleague
  3  Apply it this week     something that did not exist        OUT
                            last Monday
  2  Rewire your inputs     your feed recommends builders       in
  1  Get taught by the tool a saved curriculum for your job     in

Rung 1: Get taught by the tool. Open any chat model, including a free one, and tell it exactly who you are. In the live session I said, by voice, that I was a commerce graduate with a finance background and no technology knowledge. Ask it to act as an experienced practitioner training a beginner, to explain things the way it would to a grandparent, and to build a step by step curriculum at 15 minutes a day to an intermediate level. Then narrow it to your actual function. An accountant and an HR lead need different first weeks. Save the result as a project so the context carries into every later chat. The proof for this rung is that saved project, narrowed to your job. The common stall here is accepting the generic plan and never narrowing it, which gives them a reading list with no connection to Monday's work.

Rung 2: Rewire your inputs. I deliberately trained my YouTube recommendations to serve people who build with AI and show their work, creators like Nate Herk and shows like My First Million. Anything I cannot watch goes to Watch Later. Podcasts go on the treadmill and in the car. One hard rule sits underneath it: never pair two active tasks. Walking plus a podcast works. Writing a report plus a tutorial fails, for me as much as anyone. The proof for rung 2 is that your feed now recommends builders without being asked. This is also the rung where the trap is set, because a well-tuned feed delivers a steady supply of the feeling of progress.

Rung 3: Apply it the same week. Whatever you learned on Monday has to change something you produce by Friday. Learn a prompting technique and your prompts that week must be visibly better. Learn a tool and you must build something with it. This is how I built the personal assistant I now use every day. I took trending GitHub repositories I found through creators on rung 2, learned to fork and customise them, and assembled the system over about six months. When I started, I did not know how to use git. The proof for rung 3 is an artefact that did not exist last Monday. In my experience, this is the rung where most people stop.

Rung 4: Distil it. Too much input produces paralysis. People freeze under a long Watch Later list and act on none of it. I load the relevant links and PDFs into NotebookLM and ask for flashcards, a short quiz, a one-page infographic or an audio overview I can play on a walk. The proof is a single page you could hand to a colleague. Distilling material you have never applied just produces a tidier version of the pile, which is why this rung sits above application.

Rung 5: Teach it. Explaining something to a live audience exposes every gap you were able to hide from yourself. The proof for rung 5 is specific: someone asked you a question you did not predict, and you answered it from experience. This was the rung I skipped for a year, and the reason was vanity about lighting.

Now look at the direction column. Rungs 1, 2 and 4 take material in. Rungs 3 and 5 put something out where another person can inspect it. The two output rungs are the only two that can fail in public, and they are the two that people skip.

That is what the follow-up question was measuring. A person on rung 2 can describe what happened in AI last week in detail, because they have consumed it. The follow-up asks whether they have used it, and nobody who stopped at rung 2 has a sentence ready for that. Performed awareness is rung 2 wearing the clothes of rung 5.

The ladder also explains why rung 3 feels expensive. Every input rung is private and pleasant. You choose the video, you control the pace, and nobody grades you. Rung 3 is the first time the material meets real work with real consequences. The thing you build might not run. The prompt might produce a worse report. The discomfort is the signal that the learning has started, which is also why the pull back to one more video is so strong.

What this means for a career

Two consequences follow for anyone managing their own development.

The first is about what to show. A manager, a client or an interviewer can only see the output rungs: the thing you built, and the way you answer a question you did not prepare for. If your plan for staying relevant is mostly rungs 1, 2 and 4, it is invisible to every person who decides your next role, however many hours it takes.

The second is about pace. The WEF figure of 39% is spread over five years, so the skill changes arrive gradually across that period. A ladder climbed once becomes out of date. What compounds is the loop: a new rung 1 curriculum whenever the tools shift, a feed that keeps surfacing the shift, and a rung 3 artefact every week. In my own case, the assistant I built from forked repositories is still not finished, because a system you use daily keeps showing you what it cannot yet do.

For team leads there is a third, quieter consequence. When you ask your team what they are learning, you will get rung 2 answers, because rung 2 is what people have to report. Ask them what they built with it last week, and ask to see it. A ten minute demonstration from one person tells you more than a slide of training hours from the whole team. Course completion rates only ever measure rungs 1 and 2.

Three things to do this week

Build a narrowed curriculum today. Paste this into a free chat model: "I am a [your role]. I have 15 minutes a day. Act as an experienced practitioner and build me a step by step AI curriculum to intermediate level, explained simply, using only examples from my function." Save it as a project. Do day one before you close the tab.
Run the follow-up test on yourself. Pick the last AI development you mentioned to someone. Write three sentences on how it works and one sentence on what you did with it. If the fourth sentence will not come, that item lives on your watch list, and you know which rung you are on.
Ship one small thing by Friday. Choose one task you do every week, use one AI technique you learned this month on it, and keep the result. Then spend ten minutes showing one colleague what you did. You will reach rung 3 and rung 5 in the same week, and their first question will show you what you have not yet understood.

What to read next

While you are here, the back catalogue has more on this:

Footer

Anees Merchant writes one essay every Tuesday about enterprise AI, agentic systems, and the human side of the work. He is the author of Merchants of AI, a TEDx speaker, and a doctoral researcher in Human-AI Communication at the Swiss School of Business and Management.

The newsletter version, with extra commentary, goes out separately on Mondays. Subscribe at /newsletter.

See you Tuesday.