Replace Doing With Learning
Turn every AI-assisted project into a focused, compounding learning loop.
- Difficulty
- Moderate
- Time to result
- ~months to results
- Steps
- 5
- Confidence
- 92%
Replace Doing With Learning reframes an AI-assisted task as one repetition in a deliberate practice loop. Before beginning, the practitioner identifies the specific skill being developed, such as crafting a message, telling a story, or producing a video. AI is then used to reduce research time, generate alternatives, explain errors, and accelerate feedback—not to remove the decisions that exercise the skill. After completing the work, the practitioner reviews what improved, what failed, and what should change in the next repetition. Those lessons shape a slightly more demanding follow-up project. Over days, weeks, and months, the accumulated knowledge and judgment compound, turning AI from a task-completion shortcut into an engine for increasingly capable human performance.
Origin
Extracted from Marketing Against The Grain as the hosts argued that marketers should use AI to accelerate structured learning rather than outsource it.
Core principles
- 01Learning compounds more powerfully than isolated task completion.
- 02AI should shorten feedback loops rather than remove practice.
- 03Focused repetitions build transferable skill.
- 04Each project should make the next attempt better.
- 05Random output does not create cumulative capability.
How to run it
- 1
Replace the task with a learning target
Rewrite what you want to do as something you want to learn. Specify the underlying capability the project should strengthen.
Pro tip Use the sentence, “By completing this, I want to become better at…”
Watch out A vague goal such as “learn AI” cannot guide deliberate practice.
- 2
Choose a measurable repetition
Design a concrete project that exercises the target skill and can be completed soon enough to produce feedback. Define what a stronger attempt would look like.
Pro tip Keep the first repetition small enough to finish within days.
Watch out Oversized projects delay feedback and obscure which behavior caused the result.
- 3
Use AI to shorten the loop
Ask AI to accelerate research, propose alternatives, critique drafts, or explain mistakes. Retain the decisions that require you to exercise judgment.
Pro tip Request critiques before requesting rewrites so you still perform the revision.
Watch out If AI makes every decision, the task may finish while your skill remains unchanged.
- 4
Review the attempt
Compare the result with your success criteria and identify what worked, failed, or surprised you. Separate lessons about the craft from lessons about the AI tool.
Pro tip Capture one strength, one weakness, and one change for next time.
Watch out Engagement metrics alone may not reveal whether your underlying skill improved.
- 5
Increase the challenge
Repeat the process with a focused adjustment or slightly harder project. Carry the prior lesson explicitly into the next attempt.
Pro tip Change one major variable at a time so the lesson remains interpretable.
Watch out Jumping to unrelated projects interrupts compounding.
In the wild
A marketer replaces the goal “make five campaign posts” with “learn to open a campaign with a sharper customer tension.” They draft several openings, use AI to compare clarity and specificity, choose and revise one themselves, then record which wording generated qualified responses.
→ Each campaign becomes a structured repetition that improves the marketer's messaging judgment.
A creator makes one short video each week while focusing sequentially on hooks, pacing, transitions, and conclusions. AI speeds transcript analysis and offers critiques, but the creator makes the edits and reviews retention data after publication.
→ Production becomes faster while storytelling ability compounds across multiple projects.
Common mistakes
Counting outputs instead of lessons
A high volume of completed assets can conceal the fact that the same weaknesses are being repeated. Track the specific capability improved during each project.
Outsourcing the difficult decisions
Allowing AI to perform every judgment-heavy step removes the mental repetitions needed to develop the skill.
Practicing without focus
Randomly trying unrelated tasks and tools produces experiences that do not build coherently on one another.
Is it for you?
Best for
It is best for marketers, creators, and knowledge workers who want AI use to compound into stronger long-term capabilities.
Not ideal for
It is not ideal for routine tasks where learning is irrelevant and straightforward automation is the only objective.
From the transcript
“I would just urge everybody replace the word do with learn.”
“And AI allows you to learn way faster than ever before.”
“AI is a tool that helps you shorten the feedback loops and learn much much faster, but your core craft and skill is going to…”
From the episode
This Video Proves AI Generated Images are Dying