The reading room · Digna Legi
The "I don't know, Claude wrote this" pandemic
/100
80–100: high value. 70–79: worth the time. Below 70: below the usual publication threshold.
Evidence-reviewed score based on available publisher text. Referenced external article and book claims are not evaluated beyond the supplied excerpts.
Scores reflect one reader’s profile, not an objective quality rating. Best is a separate personal selection.
How scoring works →This brief · about 2 min with detail
Why read this
Useful AI delegation turns dangerous when engineers stop owning decomposition, architectural judgment, review, and the reasoning behind generated code.
AI brief · Checked against source text
The main idea
The central distinction is not whether engineers use AI, but whether they still own the reasoning behind the result. AI planning and code generation become dangerous when a plausible plan lets someone skip decomposition, architectural judgment, and review until neither author nor reviewer can explain the change.
Some background helpful. Software review and AI coding workflows.
Go a little deeper
Ownership is the review standard
The author treats a pull request as an accountability claim: opening it means the engineer wrote it in the professional sense. The failure is not using Claude, but being unable to answer fundamental questions about architecture, tradeoffs, or file changes after submitting the work.
Vagueness transfers decisions
The mechanism is subtle: when the task is ambiguous and the engineer lacks enough domain understanding, AI planning fills the missing judgment with decisions that feel coherent. Momentum then rewards skimming, because the result works, but understanding has been displaced rather than accelerated.
Skill loss is selective
The author’s advice is not to preserve every old skill equally. He is willing to lose some manual implementation fluency, including syntax and CSS, but not the higher-level capacities needed to steer work: reading plans, analyzing tradeoffs, and understanding systems.
A case from the article
The Segment tracking task
In an unfamiliar internal agents repository, the author used Claude to plan and execute event tracking, then reviewed every file before committing. For each unclear change, he questioned the tool, revised bad explanations, and re-reviewed until he could discuss senior engineers’ comments directly.
How the case is made
The argument is made through workplace observation, a personal failure mode, one contrasting work example, and historical analogy.
Where the idea has limits
The author’s boundary is strongest for engineering changes where architecture, reviewability, and system understanding matter; it does not reject delegating syntax or manual coding.
A question to take away · from Digna Legi
Which parts of your work must remain explainable by you, even when generated elsewhere?
What the original adds
The source adds a concrete self-audit: distinguish skills you can afford to lose from skills required to direct the tool.
About this brief
AI-written, then separately checked for source support, useful detail and clarity. The author’s claims and our editorial question are kept separate. The original remains the author’s work. How we select and summarise →
How was this brief?
Rate this summary, separately from the author’s article.
Optional. Saved in this browser; shared only if you allow analytics.
How was the original article?
Rate the author’s original after reading it.
Optional. Saved in this browser; shared only if you allow analytics.
Digna legi. Worth reading.