The real skill behind good AI prompts is older than computers

Ernie Svenson · email · 2026-06-02 Reader 🖥️

Attention as a Hijackable Resource Context Engineering Meta-Skills & Contextual Judgment Prompting as Communication

Good AI output depends not on prompting technique but on attention — the pre-digital, self-cultivated skill of noticing what matters in the task and in the model's output — which cannot be taught, only chosen.

practical · argumentative — A persuasive advice essay to professionals (framed for lawyers) about how to get better with AI. It reads partly theoretical — it explains the nature of 'attention' — but its payload is practical: change how you engage, choose to pay attention. The theoretical reframe ('context is downstream of attention') is in service of the practical exhortation.

Structure

Part	Role
Setup / concession: the 'context is everything' AI gospel is true but 'skips a step'	Establishes the foil the essay will correct — grants the popular view just enough to then subordinate it
'Context comes from attention': good context isn't a prompting skill, it's a noticing skill	The core reframe / thesis statement — context is the OUTPUT of a prior attention skill
'The skill people mistake for something else': people seek technical fixes (prompt, model, workshop); the real gap is attention, chronically missing in knowledge work	Diagnosis — explains WHY the real skill is overlooked
'It's nurtured, not taught': attention has no syllabus; you develop it yourself through curiosity ('lead a horse to water')	Epistemic claim — establishes that the key skill is non-transmissible, which pre-empts the 'just teach me' response
'What this looks like for lawyers': attentive practitioners notice drift and forgotten details, and compound over months; prompt-collectors stay mediocre	Evidence / illustration — the load-bearing anecdotal case for the thesis
'Bottom line': context is downstream of attention, and attention is something you choose for yourself	Conclusion / restatement as an imperative

Key terms

attention / noticing Prompting as Communication
Sustained, self-directed conscious observation of the task, of oneself, and of what the AI just produced — deliberately directing your focus rather than having it captured. Ernie treats 'noticing' as its active exercise.
context (the AI gospel sense) Context Engineering
The background you feed a model — accepted as real and useful, but demoted to a downstream PRODUCT of attention rather than a primary skill
prompting skill Prompting as Communication
Templates, prompt libraries, 'collecting the most prompts' — explicitly deflated as NOT where the value lives
nurtured, not taught
A skill developed on one's own, over time, through curiosity — no class, syllabus, or certification can transmit it

Propositions

explicitGiving an AI good context is not a prompting skill but a noticing/attention skill.conf: high
explicitWhen people struggle with AI, the missing piece is almost never technical (better prompt, newer model, workshop) — it's attention.conf: high
explicitAttention cannot be taught in a class; it is developed on one's own through curiosity.conf: high
explicitAttentive AI users improve and compound over months; prompt-collectors stay mediocre.conf: high
implicitThe bottleneck on AI competence is a general, pre-existing human capacity, not anything AI-specific — AI didn't create a new skill demand, it re-exposed an old one.conf: high
implicitdevelopedAttention is largely volitional — the barrier is a matter of choice/will, not aptitude or circumstance.conf: high
refines ←Uten kontroll · p6 — Svenson holds attention is largely volitional — the barrier is a matter of choice/will, 'not aptitude or circumstance.' Refseth qualifies this with a boundary condition: against an engineered addiction, will is insufficient and circumstance (environmental design) is decisive; agency is reclaimed structurally, not by choosing harder. Refines rather than negates — she agrees attention is the target, disagrees that resolve alone directs it.
implicitcorroboratedModern attention is a scarce resource captured by design (notifications, ads), so directing it is an act of resistance.conf: medium
supports ←Uten kontroll · p3 — Svenson argues (implicitly) that modern attention is a scarce resource captured by design and that directing it is an act of resistance. Refseth is the firsthand, embodied case: engineered short-video feeds and an inescapable device captured a willpower-proven adult, and reclaiming attention required active resistance. Different domain (social media vs. AI prompting), same claim about engineered capture.

Arguments

To get better with AI, cultivate attention (choose to notice) rather than collecting prompts or chasing tools.

Explicit premises:

Good context requires knowing what matters about the task — what to include and leave out
That knowing comes from paying close attention to the work, yourself, and the AI's output
The fastest-improving lawyers are the attentive ones, not the prompt-collectors
Attention isn't teachable in a class; it's self-developed through curiosity

Implicit premises:

The quality of AI output is bounded mainly by the human's attention, not by the model's capability or the prompt's form
'What matters about the task' is reliably knowable by attentive noticing — the relevant signal is available to be noticed
Attention practiced on AI work transfers and compounds rather than staying task-specific
The reader has the volitional capacity to redirect attention if they simply choose to

Inferences:

If the bottleneck is attention and attention can't be taught, then much of the prompt-course / AI-training industry is selling the wrong thing — a market critique Ernie implies but never states
If attention is the real, domain-general skill, AI advantage accrues to the already-attentive and could widen the gap between attentive and scattered workers
If context is merely downstream of attention, then better models and longer context windows will NOT close the gap for inattentive users

Narratives

You can lead a horse to water but can't make it drink — and with attention you can't even lead someone to it; they have to lead themselves.
→ Supports 'nurtured, not taught' — attention is irreducibly self-originated and cannot be transmitted
The attentive lawyer notices what worked last time, notices when the output drifted, notices the forgotten detail and adds it next round — and pulls ahead of the fire-and-forget prompt-collector who is still mediocre a year later.
→ The load-bearing evidence for both the compounding-returns claim and the prompts-don't-matter claim
Notifications, ads, half-finished tabs, the next thing on the list — attention captured and managed for us.
→ Illustrates that default attention is hijacked, setting up attention-as-a-choice-you-must-reclaim

Questions addressed

✅ Why do some people get far more out of AI than others?
Because they pay attention and notice as they go, not because they have better prompts or models.
✅ Can the skill behind good AI use be taught?
No — it's nurtured through self-directed curiosity, not transmitted in a class.
⭕ HOW does one actually cultivate attention?
Only gestured at ('be curious', 'choose it', join the paid Inner Circle to 'slow down together'). The claim that attention can't be taught conveniently forecloses the how-to question the essay itself raises.
⭕ Is it actually true that model capability and prompt form matter little relative to attention?

Gaps

incompleteNo method for developing the one skill it names as decisive. The essay says attention is the key, says it can't be taught, and then offers only the paid 'Inner Circle' as a path — leaving the practical reader without an actionable next step, which is ironic for a self-styled 80/20 practical newsletter.
Read next: Cal Newport 'Deep Work'; Nir Eyal 'Indistractable'; William James on attention; mindfulness / meta-attention research — the how-to layer the essay withholds
illogicalThe thesis is close to unfalsifiable: any AI success is credited to attention, any failure blamed on inattention. It also equivocates between 'attention' as (a) present-moment noticing and (b) accumulated domain expertise ('knowing what matters about the task') — much of what he calls noticing is actually expertise/tacit knowledge, a different faculty.
Read next: Michael Polanyi 'The Tacit Dimension'; Gary Klein 'Sources of Power' (expert intuition as pattern recognition); Ericsson on expertise — to separate 'attention' from 'expertise'
incompleteTitle/body mismatch. The headline promises a skill 'older than computers,' but the body only asserts the pre-digital, timeless pedigree of attention — it never substantiates the historical claim it uses as a hook.
Read next: Matthew Crawford 'The World Beyond Your Head' (attention as a cultural/historical achievement); Simone Weil on attention
uninformedTreats the model as a neutral mirror — 'what the AI just gave you' — assuming the signal to notice is fully in the human's court, and underweights that model capability, tool scaffolding, and prompt structure genuinely shift outcomes independent of user attention.
Read next: Anthropic/OpenAI prompt- and context-engineering guides; evals showing model-choice and scaffolding effects — evidence that technique and model are not negligible
closed byPrompting Claude Fable 5 — Svenson treats the model as a neutral mirror and underweights that model capability, tool scaffolding, and prompt structure shift outcomes independent of user attention. This official Anthropic guide is primary-source evidence for precisely that: model-specific structure (effort settings, brief-vs-enumerated instructions, subagent and memory scaffolding, elicitation language) measurably changes behavior — matching the gap's own reading_direction, which asked for exactly these Anthropic prompt-/context-engineering guides.