Skip to main content
skills deep dive

Deep reading when everything summarises

AI will summarise anything for you. The summary is not the reading. Here is what the reading is for.

Updated Reviewed by Senwitt Editorial Team

What does daily deep-reading practice look like when AI summarises everything?

One full article a day read without an AI summary, without skimming, without parallel tabs. The cognitive act being practised is sustained attention on a single piece of text, which is the substrate that supports comprehension, judgment, and the ability to disagree with what you read. AI summaries are useful — and they are not the reading.

The Reading Skill in Senwitt sits at an unusual intersection in 2026. AI summarisation is now in every browser, every reading app, every meeting tool. Articles, papers, contracts, emails, books — there is a summary one click away from almost anything you might be expected to read. The productivity case is real and the tools are not going away.

The narrower question is the one this post addresses. What does daily deep-reading practice actually look like in 2026, and what does the published evidence say about why it is worth keeping on the calendar?

Why this matters — the published evidence

The 2025 EDUCAUSE Review piece on the "productivity paradox of AI on students' thinking" names the pattern at the knowledge-work level. AI-mediated reading produces better artefacts and weaker thinking habits in parallel. The mechanism is the same as in the wider cognitive-offloading literature: cognitive acts that get fewer reps weaken.

The 2025 PsyPost coverage of recent cognitive-offloading research (PsyPost) reports the same direction with reference to AI-mediated reading specifically. Heavy reliance on AI summaries was associated with lower self-reported engagement with the underlying material and with weaker recall of the content. The findings are survey-based and limit applies, but the direction is consistent.

The 2025 Phys.org coverage of AI and critical thinking (Phys.org) reports on workplace-side findings — heavier AI use was associated with lower self-reported critical-thinking engagement on knowledge tasks. The reading-side framing is one specific case of the broader pattern.

The foundational reference is the older cognitive-offloading literature (Risko & Gilbert, 2016, Trends in Cognitive Sciences). The mechanism is straightforward: cognitive acts get reps; the ones that get fewer reps weaken. AI summarisation substitutes for the cognitive act of sustained attention on a piece of text, and the substitution pattern is what the literature would predict.

The deliberate-practice literature (Ericsson, Krampe & Tesch-Römer, 1993) is the constructive frame. Daily, effortful, on-purpose engagement with the specific skill maintains the skill. For reading, the cognitive act is sustained attention on a single piece of text.

What deep reading is and is not

The phrase "deep reading" has been used loosely enough that it is worth being specific. The cognitive act in question is sustained attention on a single piece of text, in a single sitting, without parallel tabs, without skimming, and without an AI summary alongside.

It is not speed reading. It is not skimming for the argument and moving on. It is not the AI-summary-plus-headline pattern most knowledge workers use to keep up with the volume.

It is the act of sitting with a piece of writing, following its argument, noticing what it does and does not say, registering when you disagree, and finishing it. The artefact at the end of the act is comprehension and judgment, not a summary you can pass on.

Comprehension and judgment are what AI summaries cannot produce. They can produce a summary; they cannot produce your comprehension of the material. The cognitive act has to happen in you for the comprehension to be yours.

A daily deep-reading practice

The routine below takes fifteen to thirty minutes a day depending on the piece. None of it requires special equipment. All of it requires the discipline of reading one thing without an AI summary alongside.

1. Pick one piece a day to read fully. An article, a chapter, a paper, a long-form essay. One. Not the inbox-clearing version where you skim five things. The act is sustained attention on a single piece of text.

2. No AI summary alongside. Not before, not during. After is fine — comparing what you took away to what the AI summary says can be useful — but the originating act has to be your own reading.

3. No parallel tabs. Sustained attention is the cognitive act being practised. Tab-switching is its enemy. Close everything else for the duration of the piece.

4. Note where you disagree. The cognitive act of disagreement is the part of reading that AI summaries flatten most. When you read something, register the places where you push back, even silently. The act of pushing back is the rep that builds critical reading.

5. Summarise it yourself, a day later. Without going back to the source. The retrieval is the rep that consolidates the reading into memory. AI summaries skip this step entirely.

6. Read one source unmediated for every AI-summarised one. A useful ratio. AI summaries are productive for the bulk of your reading load; the unmediated minimum keeps the muscle warm.

What this is not

A few honest disclaimers.

This routine does not claim that AI summarisation causes general cognitive decline or any clinical condition. The published evidence does not support that claim. The narrower claim — that sustained attention on text is a cognitive act, that the act gets fewer reps when AI substitutes for it, and that acts which get fewer reps weaken — is what the literature supports.

It is not a case for refusing AI summarisation. The productivity case is real and the tools are not going away. For the bulk of your reading load — meeting notes, long email threads, dense reports — AI summaries are productive. The recommendation is to keep one piece a day on the unmediated side.

It does not promise that daily deep-reading practice transfers to general intelligence. The deliberate-practice literature is consistent on transfer specificity. The practice maintains the sustained-attention muscle. That is the scope.

It is not a substitute for the lifestyle factors that have the strongest evidence base for cognitive health. Sleep, physical activity, social engagement, and managing health conditions are the load-bearing variables; the deep-reading practice sits alongside them.

How Senwitt fits

Senwitt's daily Set includes a Reading Skill rep most days. The reps are short — comprehension exercises, focused-attention prompts, disagreement-noticing tasks — designed to maintain the sustained-attention muscle. The deliberate-practice frame is taken from the Ericsson 1993 paper.

The research/cognitive-offloading page covers the foundational research. The research/ai-overreliance page covers the workplace-side pattern.

A small observation on what changes after a few weeks of running this routine. The first week is the hardest because the unmediated reading session feels disproportionately effortful — the muscle of sitting with text for thirty minutes without skimming or tab-switching has, for many knowledge workers, been quietly weakening since the smartphone era and is now under further pressure from AI summarisation. The act feels harder than it should. By the third or fourth week, the friction tends to drop noticeably. The piece you sit with stops feeling like an obligation and starts feeling like the part of the day where you actually get to think. That subjective shift is not a small reward — it is the most reliable signal that the daily practice is doing structural work. The AI summary still does the bulk of the volume; the unmediated piece does the work of keeping the underlying surface available, and the working week feels different when it is back in regular rotation.

From Senwitt · advertisement

The text above is editorial. What follows is a promotional message from Senwitt, the maker of this site. Senwitt is a brain-exercise app and is not a medical product. Read the full disclaimer in the footer.

Sources

  1. 1.The Paradox of AI Assistance: Better Results, Worse Thinking EDUCAUSE Review, 2025.
  2. 2.AI tools may weaken critical thinking skills by encouraging cognitive offloading, study suggests PsyPost, 2025.
  3. 3.Increased AI use linked to eroding critical thinking skills Phys.org, 2025.
  4. 4.Cognitive Offloading Trends in Cognitive Sciences 20(9):676–688 (DOI 10.1016/j.tics.2016.07.002), 2016.
  5. 5.The Role of Deliberate Practice in the Acquisition of Expert Performance Psychological Review 100(3):363–406, 1993.
Get the app

Take this argument with you. Daily practice in the app.

Download on the App StoreGet it on Google Play

Free download. Super Senwitt available in-app.

We use cookies to make the site work, measure aggregate usage, and (if you opt in) attribute organic app installs. You can accept all, reject all, or customize.

See our cookie policy and privacy policy.