The Writing Skill in Senwitt is the one where the AI integration question is sharpest. Tools that write faster than you are in every browser tab, in every editor, in every email client. The productivity case is real and the tools are not going away.
The question this post addresses is narrow and specific: what does daily writing practice actually look like in 2026, and what does the published evidence say about why it matters?
Why the originating act is the load-bearing one
The 2025 MIT Media Lab preprint by Kosmyna et al. (arXiv 2506.08872) is the cleanest recent reference. It measured EEG, recall, and self-reported ownership across essay writers in three conditions — brain-only, search-engine-assisted, and LLM-assisted. The LLM-assisted group showed the weakest neural connectivity during composition, the lowest self-reported ownership of the essays they produced, and the worst recall of their own arguments afterward. The paper introduces the phrase "cognitive debt" for that gap.
It is a preprint. The Stanković 2026 commentary raises methodological objections about sample size and EEG interpretation. Both are preprints and neither is the final word. The careful read is that the directional finding holds even where specific numbers are contested, and it is consistent with the older cognitive-offloading literature.
The Conversation's 2023 piece on ChatGPT and student writing is the qualitative classroom-side reference. It reported the same direction early in the AI-in-classroom timeline: AI-assisted students produced more polished work and developed weaker motivation to think for themselves.
The foundational frame is the older cognitive-offloading literature (Risko & Gilbert, 2016, Trends in Cognitive Sciences). Cognitive acts get reps. The ones that get fewer reps tend to weaken. AI writing tools substitute for the originating cognitive act of putting an idea into your own words for the first time, which is the act that builds and maintains writing voice.
The deliberate-practice literature (Ericsson, Krampe & Tesch-Römer, 1993) is the constructive frame. Daily, effortful, on-purpose engagement with the specific cognitive act is what maintains the skill. Editing an AI draft is a different act from writing one. The repetition does not transfer.
What "daily writing practice" actually means
The phrase is overused, so it is worth being specific. Daily writing practice in this context means: a small volume of words, written by you, from a blank doc, without AI in the first-draft pipeline, on most days of the week. The volume is small on purpose. The unmediated existence of the act is what counts.
The act is not the same as journaling, although journaling can be a delivery mechanism. It is not the same as taking notes, although note-taking shares some surface features. It is the cognitive act of generating prose from a blank surface, in your own voice, for some purpose you have in mind.
The purpose can be modest. A journal entry. A long Slack message you would otherwise have AI-drafted. A response to an article. A paragraph of a piece you are working on. The artefact is not what matters. The act is.
The 200-word daily minimum
The Senwitt threshold for unmediated daily writing is two hundred words. The number is deliberately low. The reasoning is structural.
A higher threshold pushes the practice off busy days. A lower threshold (a sentence, a fragment) does not constitute a real cognitive act. Two hundred words is a paragraph or two — small enough to survive any working day, large enough to be a real piece of generated prose.
The daily-practice literature is consistent on the load-bearing variable being daily volume rather than weekly or monthly volume. The Ericsson 1993 paper is the canonical reference. The maintenance effect drops sharply when the practice gaps lengthen.
What gets written at the threshold matters less than people think. Working writers who maintain voice across the AI era often run their daily volume through whatever happens to be at hand — a journal, a working-document draft, a long message — rather than scheduling it as a dedicated session. The structural commitment is the daily existence, not the form.
A practice that holds up
The routine below is for someone who already writes professionally or who wants to maintain writing capability over time. It does not require special tools. It requires fifteen to twenty-five minutes a day.
1. Open a blank doc. Not a doc with notes. Not a doc with an outline. Not a doc with an AI draft. Blank.
2. Write at least two hundred words. Whatever is on your mind, whatever the working day is bringing, a paragraph of something you owe later in the week. The volume is the rep.
3. Read it aloud. The cheapest voice test. AI-assisted prose is syntactically correct and slightly off in rhythm. Reading aloud surfaces the patterns immediately. Your own first draft has rough edges; the rough edges are usually you.
4. Now AI joins. Structural critique, line edits, length adjustments, alternative phrasings. AI is excellent at all of these. Use it freely.
5. Verify the AI edits against your ear. Each suggestion. The cognitive act is deciding the edit fits, not seeing it. The verification step is what keeps the editorial judgment in you.
6. Weekly voice continuity check. Once a week, read something you wrote a fortnight ago. Can you still hear yourself in the sentences? If not, the AI integration has crept past the calibration line.
What this is not
A few honest disclaimers.
This routine does not promise voice immortality. Voice in writing changes with what you read, who you write for, and what tools you use. The routine protects the daily reps. It does not lock voice in amber.
It does not claim that AI writing tools cause lasting cognitive harm. The published evidence does not support that claim. The narrower claim — that the originating cognitive act of writing, like any cognitive act, weakens when it gets fewer reps — is what the evidence supports.
It is not an abstinence position. AI tools are useful for revision, structure, and the substantial work that happens after the first draft exists. Calibration is the recommended posture, not refusal.
How Senwitt fits
Senwitt's daily Set includes a Writing Skill rep on most days. The Set takes about seven minutes and the Writing rep is one mechanism for the 200-word minimum. It does not replace your actual writing work; it is the warm-up that keeps the muscle warm for the work that matters.
The research/your-brain-on-chatgpt page covers the Kosmyna preprint with the Stanković critique alongside. The research/cognitive-debt page covers the broader framing.
A useful test for whether the routine has settled. After about three or four weeks of running the 200-word daily minimum, the act of opening a blank doc and starting to type should feel less effortful than it did at the start. The friction that the AI workflow tends to install — the half-conscious reach for the autocomplete or the prompt box — gets smaller as the unmediated act gets more routine. That is the signal that the daily reps are doing structural work. If after a month the routine still feels like a chore that needs willpower every morning, the threshold is probably too high for the current week and the right move is to bring it down to a hundred words for a stretch, not to drop the practice. Voice work survives off small consistent reps, not off heroic occasional ones.
