AI Before-and-After Learning Loop
Use AI to apply expert teachings, compare revisions, and expose the lesson
- Difficulty
- Easy
- Time to result
- ~days to results
- Steps
- 5
- Confidence
- 94%
The loop begins with a real artifact, such as an article or landing page, rather than a purely theoretical lesson. The learner then asks an AI assistant to examine a chosen expert's teachings, revise the artifact using those teachings, and identify where each principle improved the result. Comparing the original with the revision makes otherwise abstract ideas visible in context. The AI's explanation supplies the causal bridge between a principle and its practical effect, placing the human in the role of active evaluator rather than passive reader. Repeating the loop across several artifacts helps the learner distinguish reusable principles from changes that only fit one example.
Origin
Extracted from Marketing Against The Grain during Kieran Flanagan's discussion of using Bing AI to revise an article according to an author's teachings.
Core principles
- 01Learn through visible application rather than abstract theory
- 02Use expert teachings as explicit revision criteria
- 03Compare before and after versions to reveal mechanisms
- 04Ask AI to explain where and how each principle changed the work
How to run it
- 1
Create the baseline
Write or select an initial artifact that can be evaluated and improved. Preserve this version so the later comparison remains clear.
Pro tip Use work that matters to you; genuine context makes the lesson easier to retain.
- 2
Choose the teaching source
Identify an expert, methodology, or body of teachings relevant to the artifact's goal.
Pro tip Choose a source with specific practices rather than a vague personal style.
Watch out The exercise degrades if the AI has incomplete or inaccurate knowledge of the source.
- 3
Apply the teachings
Ask the AI to study the selected teachings and revise the baseline by applying them directly.
Pro tip State the artifact's audience, objective, and constraints in the prompt.
Watch out Do not accept stylistic imitation as proof that substantive principles were applied.
- 4
Expose the changes
Require the AI to highlight where it used each teaching and explain how the change improved the artifact.
Pro tip Request a side-by-side or change-by-change comparison.
- 5
Extract reusable lessons
Review the before-and-after versions and convert the strongest improvements into rules you can apply without AI.
Pro tip Test each inferred rule on a second artifact to see whether it generalizes.
Watch out Do not treat every AI-generated change as a valid lesson.
In the wild
A writer asks Bing AI to produce an article, study a particular author's teachings, rewrite the article using those teachings, and highlight where the teachings made it better. The writer studies the original, the revision, and the explanations together.
→ The writer sees how abstract advice changes a concrete article and learns through comparison.
A marketer creates baseline landing-page copy, asks an AI assistant to apply a respected copywriter's practices, and requests annotations connecting each revision to a specific practice.
→ The marketer receives an improved draft and a practical lesson in applying copywriting principles.
Common mistakes
Skipping the baseline
Without an original version, the learner cannot see what changed or isolate the effect of the teachings.
Accepting unexplained revisions
A polished rewrite alone produces an artifact, not a transferable lesson. Require the AI to connect changes to principles.
Trusting source knowledge blindly
Verify that the AI accurately represents the expert before learning from its revisions.
Is it for you?
Best for
People who learn most effectively by seeing principles applied to real work.
Not ideal for
High-stakes work where the AI cannot reliably access or represent the selected expert's teachings.
From the transcript
“Go read the teachings of this author and come back to me and write the article and highlight where you use those teachings to make…”
“That is actually how I learn. That is the only way I can learn, is like I can see the before and after and see…”
“That's like something taking the information, making something better and actually teaching you how it made it better and now it is put in the…”
From the episode
Bing A.I. Search is Way More Advanced Than We Realized…