The reading room · Digna Legi
The right kind of AI sceptic
/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. The evidence is argument-driven and anecdotal; it does not establish broader empirical prevalence of the described patterns.
Scores reflect one reader’s profile, not an objective quality rating. Best is a separate personal selection.
How scoring works →This brief · about 3 min with detail
Why read this
AI scepticism and optimism both fail when they become identities protected from firsthand use, specific claims, and contrary evidence.
AI brief · Checked against source text
The main idea
The author argues that AI debate splits less usefully into believers and sceptics than into grounded and ungrounded positions. A defensible view comes from sustained use, specific claims, and willingness to name what would change your mind; a brittle view borrows slogans, treats predictions as facts, or dismisses inconvenient evidence to protect identity.
Go a little deeper
Earned scepticism is specific
The author’s key distinction is between scepticism reached after use and scepticism absorbed from a group. The former can say where AI fails, for which work, after what attempts; the latter leans on shorthand like dismissive memes. Specific scepticism is harder to refute because it has conditions, evidence, and boundaries rather than a tribal signal.
Hype creates its own backlash
The mirror failure is enthusiasm that treats every capability as proof of sweeping transformation. The author argues this is more organizationally dangerous when leaders control budgets, issue mandates, or make public claims without evaluation. Failed promises then become evidence for disengaged sceptics, making serious adoption harder even when the tools are genuinely useful in narrower settings.
Separate capability from prediction
A useful mental filter is to split observable abilities from forecasts. A model producing working code from a prompt is something a person can test directly; claims about replacing most engineers within a fixed time frame are speculative. Collapsing those categories lets both camps argue lazily: enthusiasts overgeneralize from demos, while sceptics dismiss real capabilities because they reject grand predictions.
Teams need experiments, not loyalty tests
For managers, the author recommends creating conditions for honest exploration rather than mandating belief. The useful conversation is whether a tool helped a team ship faster or discover problems, not whether someone “believes in AI.” This also requires acknowledging that contexts differ: greenfield web applications and old enterprise systems may yield very different results.
A case from the article
An AI friction log
The author describes feeding a video recording of someone using an app into Claude Code, splitting it into screenshots, and running UX analysis on each frame. The result was a quick friction log that surfaced several issues he says he would not have noticed. The example illustrates the article’s standard for belief: try a concrete workflow, inspect the output, and share both useful and useless experiments.
How the case is made
The case is made through workplace observation, personal tool use, named public examples, and a practical taxonomy of belief-forming habits.
Where the idea has limits
The argument establishes how to hold an AI view responsibly; it does not settle whether AI will transform software work, because the author explicitly treats that as context-dependent and evidence-sensitive.
A question to take away · from Digna Legi
What belief about a fast-moving tool are you defending as belonging rather than testing as a claim?
What the original adds
The source includes more practical prompts for teams, the author's tool-use progression from Copilot to Claude Code, and examples of hype-driven backlash such as AI mandates and overbroad replacement claims.
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.