Formation · Reference Card

Month 09 — AI & Attention

Texts

Carr — The ShallowsA chapter every few days. The brain rewired by the medium it thinks in.

Newport — Deep WorkArgument chapters once; Rules as a weekly checklist you actually apply.

Weizenbaum — Computer Power and Human ReasonSlow down for the closing arguments — the ones that still land now.

Core distinction

Augmenting — the tool widens search, drafts a start, flags an option; your judgment still decides.

Outsourcing — the tool's output becomes real (sent, submitted, decided) without your undivided read.

Danger isn't slow thinking anymore — it's thinking that becomes fast, broad, and shallow because the tools make that cheap.

Key questions per text

Supporting ideas: Borges ("Funes the Memorious," "The Library of Babel") — total recall or total information isn't wisdom. Nielsen ("Augmenting Long-term Memory") — memory systems should free attention for synthesis, not replace judgment.

Sob medida — anxiety metabolized as construction

The pattern: when something uncontrollable looms (PMF, income, fatherhood timing), you build something controllable instead — avoidance that photographs like virtue. AI supercharges it: instant response, an audience, endless novelty, none of the real stakes.

The counter-move: one line before opening the machine outside a sanctioned block — "o que estou evitando decidir ou sentir agora?" — then decide anything small, real, first.

The tell: artifact-production about the system rising · a new framework adopted in under a week · mirror-sessions outnumbering executed days · escalation language.

The month practice

Write your personal rules for AI use without cognitive outsourcing.

  1. Four domains: research/writing, clinical decision support, patient-facing comms, personal knowledge systems.
  2. Per domain: one line on what the tool may do alone; one line on what always needs your read first.
  3. Name your weekly deep-work blocks — device and agent-free — for writing, video review, family.
  4. Write the one thing no model ever decides alone, full stop.
  5. Save to learning-records/ai-use-rules.md, dated. Revise, don't silently overwrite.
Maxim

Let the machine be fast. Keep the judgment slow enough to be yours.

Full rationale: SYLLABUS.md · shelf: RESOURCES.md · lesson: 0009-ai-and-attention.html