{"id":829,"date":"2026-09-18T18:58:32","date_gmt":"2026-09-18T18:58:32","guid":{"rendered":"https:\/\/sites.clarkson.edu\/ai-institute\/workbench\/critique\/offloading-audit\/"},"modified":"2026-09-18T19:07:58","modified_gmt":"2026-09-18T19:07:58","slug":"offloading-audit","status":"publish","type":"page","link":"https:\/\/sites.clarkson.edu\/ai-institute\/workbench\/critique\/offloading-audit\/","title":{"rendered":"The Offloading Audit"},"content":{"rendered":"\n<a href=\"https:\/\/sites.clarkson.edu\/ai-institute\/workbench\/critique\/\" class=\"wb-backlink\" style=\"position:fixed;top:14px;left:14px;z-index:99999;display:inline-flex;align-items:center;gap:6px;padding:8px 13px;background:rgba(10,12,20,0.72);color:#fff;font:600 12.5px\/1 Arial,Helvetica,sans-serif;letter-spacing:0.04em;text-decoration:none;border:1px solid rgba(255,255,255,0.28);border-radius:999px\">\u2190 Critique<\/a>\n\n\n\n\n\n\n\n\n<!--\n  CRITIQUE \/ The offloading audit \u2014 subpage of \/workbench\/critique\/\n  Same conventions as the Critique page: scoped under #cq, shares its\n  <style> block, facts dated and linked. The audit stores nothing.\n-->\n<div id=\"cq\">\n  <header class=\"cq-sub-hero cq-dark\">\n    <p class=\"cq-crumb\"><a href=\"https:\/\/sites.clarkson.edu\/ai-institute\/workbench\/\">Workbench<\/a> \/ <a href=\"https:\/\/sites.clarkson.edu\/ai-institute\/workbench\/critique\/\">08 Critique<\/a> \/ The offloading audit<\/p>\n    <h2 class=\"cq-title\">Am I still doing the <em>thinking<\/em>?<\/h2>\n    <p class=\"cq-stand\">Twelve questions about four habits, about ten minutes, nothing stored. At the end you get a reading for each habit, the studies behind it, and a plan short enough to keep. This is not a test of intelligence. It is a check on where the thinking is happening: in you, or in the window.<\/p>\n    <p class=\"cq-stance\">Offloading is not the enemy. Unnoticed offloading is.<\/p>\n  <\/header>\n\n  <section class=\"cq-band cq-paperbg\">\n    <div class=\"cq-prose\">\n      <p class=\"cq-kicker\">What the word means<\/p>\n      <p class=\"cq-drop\">Cognitive offloading is &ldquo;the use of physical action to alter the information processing requirements of a task so as to reduce cognitive demand.&rdquo; That is the 2016 definition from Evan Risko and Sam Gilbert, and their examples are homely: tilting your head to read a rotated page, setting a reminder on your phone. Writing itself is offloading. So is a library. Humans have always put memory and calculation into the world so that our heads could do something else.<\/p>\n      <p>The trouble starts when the offloading happens without our noticing, and when the thing we hand over is the thing we were supposed to be building. In 2011 a study in <em>Science<\/em> found that people remembered trivia less well when they believed a computer would save it, and remembered <em>where<\/em> it was saved better than what it said. In 2020 a study of drivers found that a lifetime of GPS use predicted worse spatial memory, and that heavier users declined faster over three years. Nobody decided to stop being able to find their way home. It just happened, one turn at a time.<\/p>\n      <p>Generative AI is offloading at a different scale, because the thing it takes is not a fact or a route but the act of composing, reasoning and judging. Whether that costs you anything depends on how you set it up. Which is what the audit is for.<\/p>\n      <p class=\"cq-reviewed\">Sources: <a href=\"https:\/\/www.sciencedirect.com\/science\/article\/abs\/pii\/S1364661316300985\" rel=\"noopener\">Risko &amp; Gilbert, TICS 2016<\/a> &middot; <a href=\"https:\/\/news.columbia.edu\/news\/study-finds-memory-works-differently-age-google\" rel=\"noopener\">Sparrow, Liu &amp; Wegner, Science 2011<\/a> (one experiment did not replicate in 2018; the direction held) &middot; <a href=\"https:\/\/www.nature.com\/articles\/s41598-020-62877-0\" rel=\"noopener\">Dahmani &amp; Bohbot, Scientific Reports 2020<\/a> &middot; checked Sep 2026<\/p>\n    <\/div>\n  <\/section>\n\n  <div class=\"cq-module-wrap cq-dark\" style=\"padding-top:2rem;padding-bottom:2rem\">\n    <div class=\"cq-module\" id=\"cq-audit\">\n      <p class=\"cq-kicker\">The audit<\/p>\n      <h3>Four habits, three questions each<\/h3>\n      <p>Answer for how you have actually worked in the last month, not how you would like to. Nobody sees this but you.<\/p>\n      <div class=\"cq-audit\" id=\"cq-audit-list\">\n        <div class=\"cq-habit\" data-habit=\"verify\">\n          <h4>Habit 1 &middot; Verify<\/h4>\n          <div class=\"cq-check-list\">\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>When an AI gives me a fact, a number or a citation, I open the source before I use it.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>I have caught an AI being wrong about something in my own field in the last month.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Yes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Not sure<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">No<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"no\"><p>When the answer sounds confident and well-organized, I take that as a sign it is right.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><\/div><\/div>\n          <\/div>\n        <\/div>\n        <div class=\"cq-habit\" data-habit=\"retain\">\n          <h4>Habit 2 &middot; Retain<\/h4>\n          <div class=\"cq-check-list\">\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>An hour after finishing a piece of work with AI help, I could explain its argument without looking.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>I take notes by hand, or rewrite the model&#8217;s output in my own words, rather than pasting.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"no\"><p>I could not reproduce, on an exam with no tools, work I got a good grade for with them.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Not true<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Partly<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">True<\/button><\/div><\/div>\n          <\/div>\n        <\/div>\n        <div class=\"cq-habit\" data-habit=\"attempt\">\n          <h4>Habit 3 &middot; Attempt first<\/h4>\n          <div class=\"cq-check-list\">\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>I try the problem myself before the first prompt.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>When I use AI for coursework, I ask it to explain or quiz me rather than to produce the answer.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"no\"><p>I reach for the model the moment a task feels hard.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><\/div><\/div>\n          <\/div>\n        <\/div>\n        <div class=\"cq-habit\" data-habit=\"own\">\n          <h4>Habit 4 &middot; Own the judgement<\/h4>\n          <div class=\"cq-check-list\">\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>In the last week I have disagreed with an AI answer and been right.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Yes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Not sure<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">No<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"yes\"><p>I could say, for a piece of AI-assisted work, which decisions were mine.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Usually<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">Sometimes<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"no\">Rarely<\/button><\/div><\/div>\n            <div class=\"cq-check-item\" data-good=\"no\"><p>There is a skill I used to do by hand that I would now struggle to do without the tool.<\/p><div class=\"cq-seg\" role=\"group\" aria-label=\"Answer\"><button type=\"button\" class=\"cq-btn\" data-v=\"no\">No<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"some\">One or two<\/button><button type=\"button\" class=\"cq-btn\" data-v=\"yes\">Several<\/button><\/div><\/div>\n          <\/div>\n        <\/div>\n      <\/div>\n      <div class=\"cq-check-result\" id=\"cq-audit-result\" hidden aria-live=\"polite\">\n        <h4>Your reading<\/h4>\n        <div class=\"cq-bars\" id=\"cq-audit-bars\">\n          <div class=\"cq-bar\" data-habit=\"verify\"><span>Verify<\/span><i><b style=\"width:0\"><\/b><\/i><span data-score>&ndash;<\/span><\/div>\n          <div class=\"cq-bar\" data-habit=\"retain\"><span>Retain<\/span><i><b style=\"width:0\"><\/b><\/i><span data-score>&ndash;<\/span><\/div>\n          <div class=\"cq-bar\" data-habit=\"attempt\"><span>Attempt first<\/span><i><b style=\"width:0\"><\/b><\/i><span data-score>&ndash;<\/span><\/div>\n          <div class=\"cq-bar\" data-habit=\"own\"><span>Own it<\/span><i><b style=\"width:0\"><\/b><\/i><span data-score>&ndash;<\/span><\/div>\n        <\/div>\n        <p id=\"cq-audit-text\"><\/p>\n        <div class=\"cq-plan\" id=\"cq-audit-plan\"><h4>A plan you can keep<\/h4><ol id=\"cq-audit-steps\"><\/ol><p style=\"margin:0\"><button type=\"button\" class=\"cq-btn\" id=\"cq-audit-copy\">Copy as a Lab Book entry<\/button> <button type=\"button\" class=\"cq-btn\" id=\"cq-audit-reset\">Start over<\/button><\/p><\/div>\n      <\/div>\n    <\/div>\n  <\/div>\n\n  <section class=\"cq-receipts cq-dark2\">\n    <div class=\"cq-receipts-head\"><h3><span>The evidence<\/span>What the studies actually measured<\/h3><span class=\"cq-count\">10 studies &middot; checked Sep 2026<\/span><\/div>\n    <div class=\"cq-table-wrap\">\n    <table class=\"cq-table\">\n      <thead><tr><th>Study<\/th><th>Who and how<\/th><th>Finding<\/th><th>Handle with care<\/th><\/tr><\/thead>\n      <tbody>\n        <tr><td><b>&ldquo;Your Brain on ChatGPT&rdquo;<\/b><br><a href=\"https:\/\/arxiv.org\/abs\/2506.08872\" rel=\"noopener\">Kosmyna et al., MIT Media Lab, June 2025<\/a><\/td><td>54 people wrote essays with an LLM, a search engine, or nothing, wearing EEG caps.<\/td><td>The LLM group showed the weakest brain connectivity; in session one, 15 of 18 LLM users could not correctly quote from the essay they had just written. Self-reported ownership was lowest in that group.<\/td><td>Preprint, not peer-reviewed; only 18 finished the last session. The &ldquo;55% reduced connectivity&rdquo; number circulating online is not in the paper.<\/td><\/tr>\n        <tr><td><b>&ldquo;Generative AI Can Harm Learning&rdquo;<\/b><br><a href=\"https:\/\/www.pnas.org\/doi\/10.1073\/pnas.2422633122\" rel=\"noopener\">Bastani et al., PNAS, June 2025<\/a><\/td><td>Randomized trial, nearly 1,000 high-school maths students in Turkey, 2023&ndash;24.<\/td><td>Unrestricted ChatGPT raised practice performance 48% and lowered the no-AI exam score 17%. A &ldquo;GPT Tutor&rdquo; built to give hints rather than answers raised practice 127% and left exam scores essentially unchanged.<\/td><td>One school; the strongest evidence here that design, not the tool, decides.<\/td><\/tr>\n        <tr><td><b>AI tools and critical thinking<\/b><br><a href=\"https:\/\/www.mdpi.com\/2075-4698\/15\/1\/6\" rel=\"noopener\">Gerlich, Societies, Jan 2025<\/a><\/td><td>Survey of 666 people in the UK.<\/td><td>Frequent AI use correlated with lower critical-thinking scores, mediated by cognitive offloading; younger participants showed higher dependence.<\/td><td>Correlational, not causal; a correction to one table was published in September 2025.<\/td><\/tr>\n        <tr><td><b>Knowledge workers and critical thinking<\/b><br><a href=\"https:\/\/www.microsoft.com\/en-us\/research\/wp-content\/uploads\/2025\/01\/lee_2025_ai_critical_thinking_survey.pdf\" rel=\"noopener\">Lee et al., CHI 2025 (Microsoft \/ CMU)<\/a><\/td><td>319 workers, 936 real examples of AI use.<\/td><td>Higher confidence in the AI went with less critical-thinking effort; higher confidence in oneself went with more.<\/td><td>Self-reported effort, not measured performance.<\/td><\/tr>\n        <tr><td><b>AI assistance and coding skill<\/b><br><a href=\"https:\/\/www.anthropic.com\/research\/AI-assistance-coding-skills\" rel=\"noopener\">Anthropic, Jan 2026<\/a><\/td><td>Randomized, 52 engineers learning a new Python library.<\/td><td>AI-assisted group scored 50% on a comprehension quiz against 67% for hand-coders, and finished only about two minutes faster.<\/td><td>Small sample, mostly junior engineers, comprehension measured immediately. Published by the vendor, against its own product.<\/td><\/tr>\n        <tr><td><b>Experienced developers with AI<\/b><br><a href=\"https:\/\/metr.org\/blog\/2025-07-10-early-2025-ai-experienced-os-dev-study\/\" rel=\"noopener\">METR, July 2025<\/a><\/td><td>Randomized, 16 experienced open-source developers, 246 real tasks.<\/td><td>Developers took 19% longer with AI tools and believed afterwards that they had been 20% faster.<\/td><td>Small, expert, specific tooling; the perception gap is the finding.<\/td><\/tr>\n        <tr><td><b>Deskilling in colonoscopy<\/b><br><a href=\"https:\/\/www.thelancet.com\/journals\/langas\/article\/PIIS2468-1253(25)00133-5\/abstract\" rel=\"noopener\">Budzy\u0144 et al., Lancet Gastro. &amp; Hep., Aug 2025<\/a><\/td><td>19 experienced endoscopists, 1,443 non-AI colonoscopies before and after routine AI use.<\/td><td>Detection rate in unassisted procedures fell from 28.4% to 22.4% after AI exposure.<\/td><td>Observational; authors say other factors may have contributed.<\/td><\/tr>\n        <tr><td><b>Handwriting vs typing<\/b><br><a href=\"https:\/\/www.frontiersin.org\/journals\/psychology\/articles\/10.3389\/fpsyg.2023.1219945\/full\" rel=\"noopener\">Van der Weel &amp; Van der Meer, Frontiers in Psychology, Jan 2024<\/a><\/td><td>36 students, 256-electrode EEG.<\/td><td>Handwriting produced far more elaborate connectivity in regions tied to memory and learning than typing did.<\/td><td>Measures brain connectivity, not grades or retention.<\/td><\/tr>\n        <tr><td><b>How students use Claude<\/b><br><a href=\"https:\/\/anthropic.com\/news\/anthropic-education-report-how-university-students-use-claude\" rel=\"noopener\">Anthropic Education Report, Apr 2025<\/a><\/td><td>574,740 academic conversations analyzed by the company.<\/td><td>About 47% were &ldquo;direct&rdquo; requests for an answer or output with minimal engagement; the model was mostly doing higher-order work (creating, analyzing).<\/td><td>Company analysis of its own product, not a survey of students.<\/td><\/tr>\n        <tr><td><b>Digital Education Outlook 2026<\/b><br><a href=\"https:\/\/www.oecd.org\/en\/publications\/oecd-digital-education-outlook-2026_062a7394-en.html\" rel=\"noopener\">OECD, Jan 2026<\/a><\/td><td>International policy review.<\/td><td>Completing tasks with generative AI &ldquo;does not necessarily translate into learning&rdquo;; offloading cognitive work to chatbots risks &ldquo;metacognitive laziness.&rdquo;<\/td><td>Quotations via the European Commission&#8217;s summary of the report; check the wording against the PDF before citing it in an essay.<\/td><\/tr>\n      <\/tbody>\n    <\/table>\n    <\/div>\n  <\/section>\n\n  <section class=\"cq-band cq-paperbg\">\n    <div class=\"cq-two\">\n      <div class=\"cq-prose\">\n        <p class=\"cq-kicker\">Design beats willpower<\/p>\n        <p>The most useful thing in the table is the tutor condition in the Turkish study. Same students, same model, same month. The version that answered questions made them worse on the exam; the version that gave hints and made them try did not. Nobody&#8217;s character changed between the two rooms. The setup did.<\/p>\n        <p>That is also the 2010 finding about automation bias, which predates chatbots by a generation: the habit of following a decision aid even when it is wrong &ldquo;cannot be prevented by training or instructions.&rdquo; You will not out-discipline the tool. You can arrange the work so that the thinking has to pass through you: try first, write your own version, verify one thing, say which decisions were yours. That is what a lab notebook is for, and why the Institute&#8217;s <a href=\"https:\/\/sites.clarkson.edu\/ai-institute\/workbench\/teach\/\">TRAIL<\/a> notebook asks what the AI contributed, what you decided, and what still needs checking.<\/p>\n        <p class=\"cq-reviewed\"><a href=\"https:\/\/journals.sagepub.com\/doi\/10.1177\/0018720810376055\" rel=\"noopener\">Parasuraman &amp; Manzey, Human Factors 2010<\/a> &middot; checked Sep 2026<\/p>\n      <\/div>\n      <div>\n        <p class=\"cq-kicker\">How many of us<\/p>\n        <ul class=\"cq-rules\">\n          <li><strong>95%<\/strong> of UK undergraduates use generative AI and 94% use it for assessed work; 12% paste AI text straight in, up from 8% a year earlier. <a href=\"https:\/\/www.hepi.ac.uk\/reports\/student-generative-ai-survey-2026\/\" rel=\"noopener\">HEPI 2026<\/a><\/li>\n          <li><strong>85%<\/strong> of U.S. college students use it; 55% say the effect on their learning and critical thinking is &ldquo;mixed.&rdquo; <a href=\"https:\/\/www.insidehighered.com\/news\/students\/academics\/2025\/08\/29\/survey-college-students-views-ai\" rel=\"noopener\">Inside Higher Ed, 2025<\/a><\/li>\n          <li><strong>54%<\/strong> of U.S. teens have used a chatbot for schoolwork, up from 26% using ChatGPT in 2024 and 13% in 2023. <a href=\"https:\/\/www.pewresearch.org\/internet\/2026\/02\/24\/how-teens-use-and-view-ai\/\" rel=\"noopener\">Pew, Feb 2026<\/a><\/li>\n          <li><strong>86%<\/strong> of 3,800 students in 16 countries used AI in their studies in 2024; 58% felt they lacked AI skills. <a href=\"https:\/\/www.digitaleducationcouncil.com\/post\/what-students-want-key-results-from-dec-global-ai-student-survey-2024\" rel=\"noopener\">Digital Education Council<\/a><\/li>\n        <\/ul>\n        <p class=\"cq-reviewed\">Different populations and years; not one trend line. Checked Sep 2026.<\/p>\n      <\/div>\n    <\/div>\n  <\/section>\n<\/div>\n<!-- \/#cq -->\n\n\n\n<script>\n\/* The offloading audit \u2014 behaviour. Stores nothing. *\/\n\/* NOTE for editors: WordPress texturizes the text between a less-than sign and the next greater-than sign\n   even inside script tags, so write comparisons as '60 > sec', never with a bare less-than sign\n   (one followed later by an ampersand breaks the whole script). *\/\n(function () {\n  'use strict';\n  var root = document.getElementById('cq'); if (!root) return;\n  var reduce = window.matchMedia && window.matchMedia('(prefers-reduced-motion: reduce)').matches;\n  var $ = function (s, el) { return (el || root).querySelector(s); };\n  var $$ = function (s, el) { return Array.prototype.slice.call((el || root).querySelectorAll(s)); };\n  var list = $('#cq-audit-list'); if (!list) return;\n  var habits = $$('.cq-habit', list);\n  var HAB = {\n    verify: { name: 'Verify', low: 'You check. That is the single habit that separates the people in the court database from everyone else.', high: 'You are taking output on trust, and confident tone is doing the persuading. Two thousand court rulings say where that leads.', tip: 'Before you use any fact from a model, open the source it points to. If it cannot point to one, treat the fact as a rumour and say so in your notebook.', href: 'https:\/\/www.damiencharlotin.com\/hallucinations\/', src: 'Charlotin, AI Hallucination Cases (2,041 rulings, Sep 2026)' },\n    retain: { name: 'Retain', low: 'What you make with the tool is staying in your head. Keep writing your own version; it is the thing that makes the work yours.', high: 'The work is leaving with the window. In the MIT study, 15 of 18 LLM users could not quote their own essay minutes later.', tip: 'After any AI-assisted piece of work, close the window and write three sentences on its argument from memory. If you cannot, it is not finished.', href: 'https:\/\/arxiv.org\/abs\/2506.08872', src: 'Kosmyna et al., MIT (2025, preprint)' },\n    attempt: { name: 'Attempt first', low: 'You try before you ask. In the one good randomized trial, that is the difference between learning and looking like you learned.', high: 'The model is getting the first attempt, and the first attempt is where the learning was. Unguarded ChatGPT raised practice scores 48% and cut the exam 17%.', tip: 'Set a ten-minute timer before the first prompt. Keep the wrong attempt in your notebook; it is the part that teaches.', href: 'https:\/\/www.pnas.org\/doi\/10.1073\/pnas.2422633122', src: 'Bastani et al., PNAS (2025)' },\n    own: { name: 'Own the judgement', low: 'You can say which calls were yours, and you argue with the machine. That is the habit that keeps judgement alive.', high: 'It is getting hard to say which decisions were yours, and skills you had are fading. The endoscopists did not notice either until someone measured.', tip: 'Once a day, predict what the model will say before you ask, and score yourself. Once a week, do one skill you care about unassisted.', href: 'https:\/\/www.thelancet.com\/journals\/langas\/article\/PIIS2468-1253(25)00133-5\/abstract', src: 'Budzy\u0144 et al., Lancet Gastroenterology & Hepatology (2025)' }\n  };\n  var result = $('#cq-audit-result'), text = $('#cq-audit-text'), steps = $('#cq-audit-steps');\n  var scores = {};\n  function answered() { return habits.every(function (h) { return $$('.cq-check-item', h).every(function (it) { return it.getAttribute('data-ans'); }); }); }\n  habits.forEach(function (h) {\n    $$('.cq-check-item', h).forEach(function (it) {\n      $$('.cq-btn', it).forEach(function (b) {\n        b.setAttribute('aria-pressed', 'false');\n        b.addEventListener('click', function () {\n          $$('.cq-btn', it).forEach(function (x) { x.setAttribute('aria-pressed', 'false'); });\n          b.setAttribute('aria-pressed', 'true');\n          it.setAttribute('data-ans', b.getAttribute('data-v'));\n          if (answered()) render();\n        });\n      });\n    });\n  });\n  function render() {\n    var order = [];\n    habits.forEach(function (h) {\n      var key = h.getAttribute('data-habit'), s = 0;\n      $$('.cq-check-item', h).forEach(function (it) {\n        var v = it.getAttribute('data-ans'), good = it.getAttribute('data-good');\n        s += v === 'some' ? 1 : (v === good ? 0 : 2);\n      });\n      scores[key] = s; order.push({ key: key, s: s });\n      var bar = $('.cq-bar[data-habit=\"' + key + '\"]');\n      if (bar) {\n        bar.setAttribute('data-level', s >= 4 ? 'high' : s >= 2 ? 'mid' : 'low');\n        $('i b', bar).style.width = Math.round((s \/ 6) * 100) + '%';\n        $('[data-score]', bar).textContent = s + '\/6';\n      }\n    });\n    order.sort(function (a, b) { return b.s - a.s; });\n    var total = order.reduce(function (n, o) { return n + o.s; }, 0);\n    var worst = order[0], best = order[order.length - 1];\n    var t;\n    if (6 >= total) t = 'Low offloading risk across the board. ' + HAB[best.key].low + ' The bars show where you are strongest; the plan below keeps it that way, because these habits erode quietly under deadlines.';\n    else if (14 >= total) t = 'Mixed, which is where most people who use these tools honestly land. Your weakest habit is ' + HAB[worst.key].name.toLowerCase() + ': ' + HAB[worst.key].high + ' Your strongest is ' + HAB[best.key].name.toLowerCase() + '. The plan starts with the weak one.';\n    else t = 'You have described the pattern the studies worry about, in more than one habit. Nothing here measures your intelligence; it measures where the thinking has been happening lately, and the honest answer is: mostly in the window. Every one of these is a setup problem, not a character problem. Start with the first step below and re-take this in a month.';\n    text.textContent = t;\n    steps.innerHTML = '';\n    order.slice(0, 3).forEach(function (o) {\n      var h = HAB[o.key]; var li = document.createElement('li');\n      li.innerHTML = '<strong>' + esc(h.name) + ' (' + o.s + '\/6):<\/strong> ' + esc(h.tip) + ' <span>' + esc(o.s >= 2 ? h.high : h.low) + '<\/span> <a href=\"' + h.href + '\" rel=\"noopener\">' + esc(h.src) + '<\/a>';\n      steps.appendChild(li);\n    });\n    result.hidden = false;\n    if (!reduce) result.scrollIntoView({ behavior: 'smooth', block: 'nearest' });\n  }\n  var copyBtn = $('#cq-audit-copy');\n  if (copyBtn) copyBtn.addEventListener('click', function () {\n    var lines = ['Critique \/ The offloading audit (' + new Date().toISOString().slice(0, 10) + ')'];\n    Object.keys(HAB).forEach(function (k) { if (scores[k] !== undefined) lines.push(HAB[k].name + ': ' + scores[k] + '\/6'); });\n    lines.push(text.textContent);\n    $$('li', steps).forEach(function (li, i) { lines.push((i + 1) + '. ' + li.textContent.trim()); });\n    lines.push('Source: ' + location.href.split('#')[0]);\n    var t = lines.join('\\n');\n    var done = function () { copyBtn.setAttribute('data-done', '1'); var old = copyBtn.textContent; copyBtn.textContent = 'Copied'; setTimeout(function () { copyBtn.removeAttribute('data-done'); copyBtn.textContent = old; }, 1800); };\n    if (navigator.clipboard && navigator.clipboard.writeText) navigator.clipboard.writeText(t).then(done, done); else done();\n  });\n  var reset = $('#cq-audit-reset');\n  if (reset) reset.addEventListener('click', function () {\n    scores = {}; $$('.cq-check-item', list).forEach(function (it) { it.removeAttribute('data-ans'); });\n    $$('.cq-btn', list).forEach(function (b) { b.setAttribute('aria-pressed', 'false'); });\n    result.hidden = true; var first = $('.cq-btn', list); if (first) first.focus();\n  });\n  function esc(s) { return String(s).replace(\/[&<>\"']\/g, function (c) { return { '&': '&amp;', '<': '&lt;', '>': '&gt;', '\"': '&quot;', \"'\": '&#39;' }[c]; }); }\n})();\n<\/script>\n\n\n\n<p style=\"text-align:center;font-family:'Courier New',Courier,monospace;color:#e6ac00;font-size:0.8rem;letter-spacing:0.13em;padding:2rem 0 1rem;opacity:0.82;\" aria-hidden=\"true\">&gt;Build-Future-WithAi<\/p>\n<p style=\"font-size: 0.65em; color:#595959; font-family: 'Courier New', Courier, monospace; margin-top: 40px; padding-top: 8px; border-top: 1px solid #eee;\" aria-hidden=\"true\">\n  WORKBENCH SIDE &mdash; Clarkson AI Institute &middot; Open page &middot; TRAIL: Updated by Stephen Casper with Claude Cowork on 2026-09-18\n<\/p>\n<style class=\"advgb-styles-renderer\">\n#wrapper-navbar,.site-header,.hero__breadcrumbs,.breadcrumbs,.entry-header,\nfooter.site-footer,.footer-widgets,.navbar,nav.navbar,.wp-block-navigation,.wp-block-template-part{display:none!important}\n.col-md.content-area{padding:0!important}\n.entry-content.container{max-width:100%!important;padding:0!important;margin:0!important}\n.site-main,.row{padding:0!important;margin:0!important}\nbody{padding-top:0!important;background:#fff}\n\n\/* ============================================================\n   CRITIQUE \u2014 Clarkson AI Institute Workbench \/ 08\n   Scoped styles. Everything lives under #cq.\n   Palette follows the Workbench: deep green, gold, pale mint,\n   plus a warm \"paper\" for the human-written passages and a\n   rust for critical marks.\n   ============================================================ *\/\n#cq{--cq-green:#0d2b1a;--cq-green2:#111f13;--cq-green3:#1a5e2a;--cq-green4:#00563f;\n    --cq-gold:#e6ac00;--cq-gold2:#f5c542;--cq-pale:#b8d4be;--cq-pale2:#5a9966;--cq-mint:#f1f8f3;\n    --cq-paper:#f6f1e7;--cq-paper2:#ebe3d2;--cq-ink:#14201a;--cq-ink2:#4a5a50;\n    --cq-rust:#a8431f;--cq-rust2:#c4552d;\n    --cq-mono:'Courier New',Courier,monospace;\n    --cq-sans:inherit;\n    color:var(--cq-ink);font-family:var(--cq-sans);line-height:1.6;margin:0 0 2rem;}\n#cq *,#cq *::before,#cq *::after{box-sizing:border-box;}\n#cq [hidden]{display:none!important;}\n#cq p{margin:0 0 1rem;}\n#cq ul,#cq ol{margin:0;padding:0;list-style:none;}\n#cq a{color:inherit;text-decoration:underline;text-decoration-thickness:1px;text-underline-offset:3px;}\n#cq a:hover{text-decoration-thickness:2px;}\n#cq button{font:inherit;cursor:pointer;}\n#cq :focus-visible{outline:3px solid var(--cq-gold);outline-offset:3px;}\n#cq .cq-dark :focus-visible{outline-color:var(--cq-gold2);}\n#cq .cq-paperbg :focus-visible{outline-color:var(--cq-green4);}\n#cq .cq-sr{position:absolute!important;width:1px;height:1px;padding:0;margin:-1px;overflow:hidden;clip:rect(0,0,0,0);white-space:nowrap;border:0;}\n\n\/* ---- surfaces ---- *\/\n#cq .cq-dark{background:var(--cq-green);color:var(--cq-pale);}\n#cq .cq-dark2{background:var(--cq-green2);color:var(--cq-pale);}\n#cq .cq-paperbg{background:var(--cq-paper);color:var(--cq-ink);}\n#cq .cq-band{padding:2.25rem 1.5rem;}\n@media (min-width:768px){#cq .cq-band{padding:3rem 2.5rem;}}\n#cq .cq-kicker{font-family:var(--cq-mono);font-size:.7rem;letter-spacing:.16em;text-transform:uppercase;margin:0 0 .6rem;}\n#cq .cq-dark .cq-kicker,#cq .cq-dark2 .cq-kicker{color:var(--cq-gold);}\n#cq .cq-paperbg .cq-kicker{color:var(--cq-green3);}\n#cq .cq-mono{font-family:var(--cq-mono);}\n\n\/* ---- hero ---- *\/\n#cq .cq-hero{position:relative;overflow:hidden;padding:2.5rem 1.5rem 2rem;}\n@media (min-width:768px){#cq .cq-hero{padding:3.5rem 2.5rem 2.5rem;}}\n#cq .cq-hero::before{content:\"\";position:absolute;inset:0;pointer-events:none;opacity:.35;\n  background-image:linear-gradient(var(--cq-green4) 1px,transparent 1px),linear-gradient(90deg,var(--cq-green4) 1px,transparent 1px);\n  background-size:28px 28px;mask-image:radial-gradient(ellipse at 80% 20%,#000 0%,transparent 65%);-webkit-mask-image:radial-gradient(ellipse at 80% 20%,#000 0%,transparent 65%);}\n#cq .cq-hero>*{position:relative;}\n#cq .cq-title{font-size:clamp(2.4rem,6vw,4.6rem);line-height:.98;font-weight:800;letter-spacing:-.02em;margin:0 0 1.1rem;color:#fff;max-width:14ch;}\n#cq .cq-title em{font-style:normal;color:var(--cq-gold);}\n#cq .cq-title .cq-caret{display:inline-block;width:.5em;height:.9em;background:var(--cq-gold);vertical-align:-.08em;margin-left:.08em;animation:cq-blink 1.1s steps(1) infinite;}\n#cq .cq-stand{font-size:1.08rem;max-width:62ch;color:var(--cq-pale);margin-bottom:1.4rem;}\n@media (min-width:768px){#cq .cq-stand{font-size:1.18rem;}}\n#cq .cq-stance{display:inline-flex;align-items:center;gap:.6rem;border:1px solid var(--cq-gold);color:var(--cq-gold);font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.12em;text-transform:uppercase;padding:.45rem .8rem;margin:0 0 1.6rem;}\n#cq .cq-stance::before{content:\"\";width:.55rem;height:.55rem;background:var(--cq-gold);border-radius:50%;}\n\n\/* questions ticker *\/\n#cq .cq-ticker{border-top:1px solid rgba(184,212,190,.25);padding-top:1rem;display:grid;grid-template-columns:1fr auto;gap:.75rem 1rem;align-items:center;}\n#cq .cq-ticker-label{grid-column:1\/-1;font-family:var(--cq-mono);font-size:.68rem;letter-spacing:.14em;text-transform:uppercase;color:var(--cq-pale2);margin:0;}\n#cq .cq-ticker-q{margin:0;min-height:2.6em;font-family:var(--cq-mono);font-size:clamp(1rem,2.4vw,1.35rem);color:#fff;}\n#cq .cq-ticker-q a{color:#fff;text-decoration-color:var(--cq-gold);}\n#cq .cq-ticker-q .cq-qnum{color:var(--cq-gold);margin-right:.5rem;}\n#cq .cq-ticker-q.cq-fade{animation:cq-in .5s ease both;}\n#cq .cq-ticker-ctl{display:flex;gap:.4rem;}\n@media (max-width:640px){#cq .cq-ticker{grid-template-columns:1fr;}#cq .cq-ticker-ctl{justify-content:flex-start;}}\n#cq .cq-ticker-ctl button{background:transparent;border:1px solid rgba(184,212,190,.45);color:var(--cq-pale);font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.08em;padding:.45rem .7rem;min-width:2.6rem;}\n#cq .cq-ticker-ctl button:hover{border-color:var(--cq-gold);color:var(--cq-gold);}\n#cq .cq-ticker-all{grid-column:1\/-1;display:flex;flex-wrap:wrap;gap:.4rem .8rem;margin:.4rem 0 0;padding:0;}\n#cq .cq-ticker-all li{font-family:var(--cq-mono);font-size:.74rem;opacity:.85;}\n#cq .cq-ticker-all a{color:var(--cq-pale);text-decoration-color:rgba(184,212,190,.4);}\n#cq .cq-ticker-all a:hover{color:var(--cq-gold);}\n\n\/* ---- how to read ---- *\/\n#cq .cq-two{display:grid;gap:1.5rem;}\n@media (min-width:860px){#cq .cq-two{grid-template-columns:1.4fr 1fr;gap:2.5rem;}}\n#cq .cq-lede{font-size:1.06rem;max-width:64ch;}\n#cq .cq-lede p.cq-drop::first-letter{font-size:2.6em;line-height:.8;float:left;margin:.08em .12em 0 0;font-weight:800;color:var(--cq-green3);}\n#cq .cq-rules{border-left:3px solid var(--cq-green3);padding:0 0 0 1.1rem;}\n#cq .cq-rules li{margin:0 0 .75rem;padding:0;font-size:.95rem;}\n#cq .cq-rules li strong{color:var(--cq-green3);}\n#cq .cq-reviewed{font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.08em;color:var(--cq-ink2);margin-top:1rem;}\n\n\/* ---- atlas ---- *\/\n#cq .cq-atlas-head{display:flex;flex-wrap:wrap;gap:1rem 2rem;align-items:end;justify-content:space-between;margin-bottom:1.25rem;}\n#cq .cq-h2{font-size:clamp(1.5rem,3.4vw,2.2rem);line-height:1.1;font-weight:800;letter-spacing:-.01em;margin:0 0 .4rem;}\n#cq .cq-dark .cq-h2,#cq .cq-dark2 .cq-h2{color:#fff;}\n#cq .cq-paperbg .cq-h2{color:var(--cq-ink);}\n#cq .cq-sub{margin:0;max-width:60ch;}\n#cq .cq-filters{display:flex;flex-wrap:wrap;gap:.5rem;align-items:center;margin:0 0 1.2rem;}\n#cq .cq-chip{background:transparent;border:1px solid rgba(184,212,190,.4);color:var(--cq-pale);font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.1em;text-transform:uppercase;padding:.45rem .75rem;border-radius:999px;}\n#cq .cq-chip:hover{border-color:var(--cq-gold);color:var(--cq-gold);}\n#cq .cq-chip[aria-pressed=\"true\"]{background:var(--cq-gold);border-color:var(--cq-gold);color:var(--cq-green);}\n#cq .cq-search{flex:1 1 220px;display:flex;gap:.4rem;align-items:center;min-width:200px;}\n#cq .cq-search label{font-family:var(--cq-mono);font-size:.7rem;letter-spacing:.1em;text-transform:uppercase;color:var(--cq-pale2);white-space:nowrap;}\n#cq .cq-search input{flex:1;background:var(--cq-green2);border:1px solid rgba(184,212,190,.4);color:#fff;font:inherit;font-size:.95rem;padding:.45rem .7rem;min-width:0;}\n#cq .cq-search input::placeholder{color:rgba(184,212,190,.6);}\n#cq .cq-grid{display:grid;gap:.9rem;grid-template-columns:repeat(auto-fill,minmax(190px,1fr));}\n#cq .cq-tile{text-decoration:none;display:flex;flex-direction:column;align-items:flex-start;text-align:left;background:var(--cq-green2);border:1px solid rgba(184,212,190,.18);border-left:4px solid var(--cq-gold);color:var(--cq-pale);padding:1rem 1.1rem 1.1rem;min-height:11rem;transition:transform .18s ease,border-color .18s ease,background .18s ease;position:relative;}\n#cq .cq-tile:hover{transform:translateY(-3px);border-color:var(--cq-gold);background:#152a1a;}\n#cq .cq-tile .cq-kicker{margin-bottom:.35rem;}\n#cq .cq-tile-title{font-size:1.15rem;font-weight:700;color:#fff;margin:0 0 .35rem;line-height:1.2;}\n#cq .cq-tile-hook{font-size:.92rem;margin:0 0 .8rem;color:var(--cq-pale);}\n#cq .cq-tile-signal{margin-top:auto;font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.06em;color:var(--cq-gold);}\n#cq .cq-tile[hidden]{display:none;}\n#cq .cq-tile-tags{position:absolute;top:.9rem;right:.9rem;display:flex;gap:.25rem;}\n#cq .cq-tile-tags i{width:.45rem;height:.45rem;border-radius:50%;background:var(--cq-pale2);display:block;}\n#cq .cq-tile-tags i.t-students{background:var(--cq-gold);}\n#cq .cq-tile-tags i.t-contested{background:var(--cq-rust2);}\n#cq .cq-empty{font-family:var(--cq-mono);font-size:.85rem;color:var(--cq-pale);padding:1rem 0;margin:0;}\n#cq .cq-legend{display:flex;flex-wrap:wrap;gap:.4rem 1.2rem;margin:1rem 0 0;font-family:var(--cq-mono);font-size:.68rem;letter-spacing:.08em;text-transform:uppercase;color:var(--cq-pale2);}\n#cq .cq-legend i{display:inline-block;width:.45rem;height:.45rem;border-radius:50%;background:var(--cq-pale2);margin-right:.35rem;vertical-align:middle;}\n#cq .cq-legend i.t-students{background:var(--cq-gold);}\n#cq .cq-legend i.t-contested{background:var(--cq-rust2);}\n\n\/* ---- sticky rail ---- *\/\n#cq .cq-rail{position:sticky;top:0;z-index:20;background:var(--cq-green);border-bottom:1px solid rgba(184,212,190,.2);padding:.45rem 1rem;display:flex;gap:.25rem;overflow-x:auto;scrollbar-width:thin;}\nbody.admin-bar #cq .cq-rail{top:32px;}\n@media (max-width:782px){body.admin-bar #cq .cq-rail{top:46px;}}\n#cq .cq-rail a{font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.08em;color:var(--cq-pale);text-decoration:none;padding:.35rem .55rem;border:1px solid transparent;white-space:nowrap;}\n#cq .cq-rail a:hover{border-color:rgba(184,212,190,.4);}\n#cq .cq-rail a.is-current{color:var(--cq-green);background:var(--cq-gold);}\n#cq .cq-rail .cq-rail-label{font-family:var(--cq-mono);font-size:.66rem;letter-spacing:.14em;text-transform:uppercase;color:var(--cq-pale2);align-self:center;padding:0 .6rem 0 .2rem;white-space:nowrap;}\n#cq .cq-rail-tools{margin-left:auto;display:flex;gap:.3rem;}\n#cq .cq-rail-tools button{background:transparent;border:1px solid rgba(184,212,190,.35);color:var(--cq-pale);font-family:var(--cq-mono);font-size:.68rem;letter-spacing:.08em;padding:.3rem .55rem;white-space:nowrap;}\n#cq .cq-rail-tools button:hover{border-color:var(--cq-gold);color:var(--cq-gold);}\n\n\/* ---- student concerns ---- *\/\n#cq .cq-concerns{display:grid;gap:1rem;grid-template-columns:repeat(auto-fill,minmax(235px,1fr));margin-top:1.25rem;}\n#cq .cq-concern{background:#fff;border:1px solid var(--cq-paper2);padding:1.1rem 1.15rem 1rem;display:flex;flex-direction:column;}\n#cq .cq-concern .cq-big{font-family:var(--cq-mono);font-size:1.9rem;line-height:1;color:var(--cq-rust);margin:0 0 .35rem;font-weight:700;}\n#cq .cq-concern .cq-big small{font-size:.8rem;color:var(--cq-ink2);font-weight:400;display:block;margin-top:.2rem;letter-spacing:.04em;}\n#cq .cq-concern h3{font-size:1.02rem;margin:.2rem 0 .4rem;line-height:1.25;}\n#cq .cq-concern p{font-size:.92rem;margin:0 0 .6rem;}\n#cq .cq-concern .cq-src{font-size:.78rem;color:var(--cq-ink2);margin:auto 0 0;}\n#cq .cq-concern .cq-src a{color:var(--cq-green3);}\n\n\/* ---- dossiers ---- *\/\n#cq .cq-dossier{scroll-margin-top:4.5rem;border-top:6px solid var(--cq-gold);}\n#cq .cq-dossier-head{padding:2rem 1.5rem 1.5rem;}\n@media (min-width:768px){#cq .cq-dossier-head{padding:2.75rem 2.5rem 2rem;}}\n#cq .cq-dossier-head .cq-h2{font-size:clamp(1.7rem,4vw,2.7rem);max-width:22ch;}\n#cq .cq-frame{font-size:1.05rem;max-width:66ch;}\n#cq .cq-frame p{margin-bottom:.9rem;}\n#cq .cq-feel{margin-top:1.25rem;border:1px dashed var(--cq-green3);padding:.9rem 1rem;max-width:66ch;background:#fff;}\n#cq .cq-feel .cq-kicker{margin-bottom:.35rem;}\n#cq .cq-feel p{margin:0;font-size:.97rem;}\n#cq .cq-receipts{padding:1.75rem 1.5rem 1.5rem;}\n@media (min-width:768px){#cq .cq-receipts{padding:2.25rem 2.5rem 2rem;}}\n#cq .cq-receipts-head{display:flex;flex-wrap:wrap;justify-content:space-between;align-items:baseline;gap:.5rem 1rem;margin-bottom:1rem;}\n#cq .cq-receipts-head h3{font-size:1.05rem;font-weight:700;color:#fff;margin:0;letter-spacing:.01em;}\n#cq .cq-receipts-head h3 span{color:var(--cq-gold);font-family:var(--cq-mono);font-size:.75rem;letter-spacing:.14em;text-transform:uppercase;margin-right:.6rem;}\n#cq .cq-count{font-family:var(--cq-mono);font-size:.7rem;letter-spacing:.1em;text-transform:uppercase;color:var(--cq-pale2);}\n#cq .cq-facts{display:grid;gap:.75rem;}\n@media (min-width:900px){#cq .cq-facts{grid-template-columns:1fr 1fr;}}\n#cq .cq-fact{background:var(--cq-green2);border-left:3px solid var(--cq-gold);padding:.9rem 1rem .8rem;position:relative;}\n#cq .cq-fact[hidden]{display:none;}\n#cq .cq-fact-claim{color:#fff;font-size:.97rem;margin:0 0 .5rem;}\n#cq .cq-fact-src{margin:0;font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.03em;color:var(--cq-pale);display:flex;flex-wrap:wrap;gap:.25rem .6rem;align-items:baseline;}\n#cq .cq-fact-src a{color:var(--cq-gold);}\n#cq .cq-checked{font-family:var(--cq-mono);font-size:.62rem;letter-spacing:.1em;text-transform:uppercase;color:var(--cq-pale2);border:1px solid rgba(184,212,190,.25);padding:.1rem .35rem;margin-left:auto;}\n#cq .cq-fact.cq-caveat{border-left-color:var(--cq-rust2);}\n#cq .cq-fact.cq-caveat::before{content:\"caveat\";position:absolute;top:.55rem;right:.6rem;font-family:var(--cq-mono);font-size:.6rem;letter-spacing:.12em;text-transform:uppercase;color:var(--cq-rust2);}\n#cq .cq-more{margin-top:.9rem;background:transparent;border:1px solid rgba(184,212,190,.4);color:var(--cq-pale);font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.1em;text-transform:uppercase;padding:.5rem .8rem;}\n#cq .cq-more:hover{border-color:var(--cq-gold);color:var(--cq-gold);}\n#cq .cq-aftermath{padding:1.5rem 1.5rem 1.75rem;display:grid;gap:1.25rem;}\n@media (min-width:900px){#cq .cq-aftermath{grid-template-columns:1fr 1fr 1fr;gap:2rem;padding:2rem 2.5rem 2.25rem;}}\n#cq .cq-aftermath h3{font-size:.95rem;margin:0 0 .5rem;font-family:var(--cq-mono);letter-spacing:.12em;text-transform:uppercase;color:var(--cq-green3);font-weight:700;}\n#cq .cq-aftermath p,#cq .cq-aftermath li{font-size:.95rem;}\n#cq .cq-counter{border-left:3px solid var(--cq-rust2);padding-left:1rem;}\n#cq .cq-questions li{margin:0 0 .55rem;padding-left:1.4rem;position:relative;}\n#cq .cq-questions li::before{content:\"?\";position:absolute;left:0;top:0;font-family:var(--cq-mono);font-weight:700;color:var(--cq-rust);}\n#cq .cq-copy{margin-top:.4rem;background:var(--cq-green3);border:1px solid var(--cq-green3);color:#fff;font-family:var(--cq-mono);font-size:.7rem;letter-spacing:.1em;text-transform:uppercase;padding:.5rem .8rem;}\n#cq .cq-copy:hover{background:var(--cq-green4);}\n#cq .cq-copy[data-done=\"1\"]{background:var(--cq-gold);color:var(--cq-green);border-color:var(--cq-gold);}\n#cq .cq-deeper li{margin:0 0 .4rem;font-size:.92rem;}\n#cq .cq-deeper a{color:var(--cq-green3);}\n\n\/* ---- interactive modules ---- *\/\n#cq .cq-module{margin:0;padding:1.5rem;border:1px dashed var(--cq-gold);background:var(--cq-green2);color:var(--cq-pale);}\n@media (min-width:768px){#cq .cq-module{padding:1.75rem 2rem;margin:0 2.5rem 2rem;}}\n#cq .cq-module-wrap{padding:0 1.5rem 1.75rem;}\n@media (min-width:768px){#cq .cq-module-wrap{padding:0;}}\n#cq .cq-module .cq-kicker{margin-bottom:.25rem;}\n#cq .cq-module h3{color:#fff;font-size:1.25rem;margin:0 0 .5rem;font-weight:800;}\n#cq .cq-module p{color:var(--cq-pale);}\n#cq .cq-btn{background:transparent;border:1px solid rgba(184,212,190,.5);color:#fff;font-family:var(--cq-mono);font-size:.74rem;letter-spacing:.1em;text-transform:uppercase;padding:.55rem .9rem;}\n#cq .cq-btn:hover{border-color:var(--cq-gold);color:var(--cq-gold);}\n#cq .cq-btn[aria-pressed=\"true\"],#cq .cq-btn.is-on{background:var(--cq-gold);border-color:var(--cq-gold);color:var(--cq-green);}\n#cq .cq-seg{display:inline-flex;border:1px solid rgba(184,212,190,.5);}\n#cq .cq-seg .cq-btn{border:0;border-right:1px solid rgba(184,212,190,.3);}\n#cq .cq-seg .cq-btn:last-child{border-right:0;}\n\n\/* offloading check *\/\n#cq .cq-check-list{display:grid;gap:.6rem;margin:1rem 0;}\n#cq .cq-check-item{display:grid;grid-template-columns:1fr auto;gap:.5rem 1rem;align-items:center;background:var(--cq-green);padding:.8rem .9rem;border:1px solid rgba(184,212,190,.15);}\n@media (max-width:640px){#cq .cq-check-item{grid-template-columns:1fr;}#cq .cq-gauge-item b,#cq .cq-analog-col li b,#cq .cq-cost-row b{white-space:normal;}#cq .cq-gauge-item,#cq .cq-analog-col li{grid-template-columns:1fr;}}\n#cq .cq-check-item p{margin:0;color:#fff;font-size:.97rem;}\n#cq .cq-check-item .cq-seg .cq-btn{padding:.45rem .7rem;}\n#cq .cq-check-result{margin-top:1rem;border-top:1px solid rgba(184,212,190,.25);padding-top:1rem;}\n#cq .cq-check-result[hidden]{display:none;}\n#cq .cq-check-result h4{color:var(--cq-gold);font-family:var(--cq-mono);font-size:.8rem;letter-spacing:.14em;text-transform:uppercase;margin:0 0 .5rem;}\n#cq .cq-check-result p{color:#fff;}\n#cq .cq-check-result ul{margin:.5rem 0 0;padding-left:1.1rem;list-style:disc;color:var(--cq-pale);font-size:.92rem;}\n#cq .cq-check-result li{margin-bottom:.35rem;}\n#cq .cq-check-result li a{color:var(--cq-gold);}\n\n\/* prompt cost *\/\n#cq .cq-cost{display:grid;gap:1.25rem;grid-template-columns:minmax(0,1fr);}\n@media (min-width:900px){#cq .cq-cost{grid-template-columns:1fr 1fr;gap:2rem;}}\n#cq .cq-cost label{display:block;font-family:var(--cq-mono);font-size:.74rem;letter-spacing:.1em;text-transform:uppercase;color:var(--cq-gold);margin-bottom:.4rem;}\n#cq .cq-cost input[type=range]{width:100%;accent-color:var(--cq-gold);}\n#cq .cq-cost output{display:block;font-family:var(--cq-mono);font-size:2rem;color:#fff;line-height:1;margin:.3rem 0 .8rem;}\n#cq .cq-cost output small{font-size:.8rem;color:var(--cq-pale);letter-spacing:.06em;}\n#cq .cq-cost-row{display:flex;justify-content:space-between;gap:1rem;border-top:1px solid rgba(184,212,190,.2);padding:.55rem 0;font-size:.93rem;}\n#cq .cq-cost-row b{color:#fff;font-family:var(--cq-mono);font-weight:400;white-space:nowrap;}\n#cq .cq-cost-note{font-size:.82rem;color:var(--cq-pale2);margin:.6rem 0 0;}\n#cq .cq-scale{margin-top:.25rem;}\n#cq .cq-scale-bar{height:14px;background:rgba(184,212,190,.15);position:relative;margin:.4rem 0 .35rem;overflow:hidden;}\n#cq .cq-scale-bar span{position:absolute;left:0;top:0;bottom:0;background:var(--cq-gold);}\n#cq .cq-scale-row{display:flex;justify-content:space-between;font-family:var(--cq-mono);font-size:.74rem;letter-spacing:.04em;color:var(--cq-pale);}\n\n\/* spectrum *\/\n#cq .cq-spectrum{margin-top:1rem;}\n#cq .cq-spec-track{position:relative;height:6px;background:linear-gradient(90deg,var(--cq-pale2),var(--cq-gold),var(--cq-rust2));margin:2.6rem 0 2.8rem;border-radius:3px;}\n#cq .cq-spec-pt{position:absolute;top:50%;transform:translate(-50%,-50%);width:1.35rem;height:1.35rem;border-radius:50%;background:var(--cq-green);border:3px solid #fff;padding:0;}\n#cq .cq-spec-pt:hover,#cq .cq-spec-pt[aria-expanded=\"true\"]{border-color:var(--cq-gold);transform:translate(-50%,-50%) scale(1.15);}\n#cq .cq-spec-pt .cq-spec-lbl{position:absolute;left:50%;transform:translateX(-50%);top:1.5rem;font-family:var(--cq-mono);font-size:.64rem;letter-spacing:.06em;color:var(--cq-pale);white-space:nowrap;}\n#cq .cq-spec-pt:nth-child(even) .cq-spec-lbl{top:auto;bottom:1.5rem;}\n#cq .cq-spec-ends{display:flex;justify-content:space-between;font-family:var(--cq-mono);font-size:.68rem;letter-spacing:.1em;text-transform:uppercase;color:var(--cq-pale2);}\n#cq .cq-spec-card{background:var(--cq-green);border:1px solid rgba(184,212,190,.25);padding:1rem 1.1rem;min-height:5.5rem;}\n#cq .cq-spec-card p{margin:0 0 .4rem;color:#fff;}\n#cq .cq-spec-card .cq-fact-src{color:var(--cq-pale);}\n#cq .cq-spec-card .cq-fact-src a{color:var(--cq-gold);}\n\n\/* gauge *\/\n#cq .cq-gauge{display:grid;gap:1.5rem;align-items:start;grid-template-columns:minmax(0,1fr);}\n@media (min-width:900px){#cq .cq-gauge{grid-template-columns:minmax(220px,300px) 1fr;gap:2rem;}}\n#cq .cq-gauge svg{width:100%;height:auto;display:block;}\n#cq .cq-gauge-list{display:grid;gap:.5rem;}\n#cq .cq-gauge-item{display:grid;grid-template-columns:auto minmax(0,1fr);min-width:0;gap:.2rem .9rem;align-items:baseline;padding:.6rem .8rem;background:var(--cq-green);border-left:3px solid var(--cq-gold);}\n#cq .cq-gauge-item b{font-family:var(--cq-mono);font-weight:400;color:#fff;font-size:1.15rem;white-space:nowrap;}\n#cq .cq-gauge-item span{font-size:.92rem;color:var(--cq-pale);}\n#cq .cq-gauge-item .cq-fact-src{grid-column:2;}\n#cq .cq-gauge-item .cq-fact-src a{color:var(--cq-gold);}\n#cq .cq-gauge [data-side=\"other\"]{border-left-color:var(--cq-pale2);}\n#cq .cq-gauge-reading{text-align:center;font-family:var(--cq-mono);font-size:.74rem;letter-spacing:.12em;text-transform:uppercase;color:var(--cq-gold);margin:.3rem 0 0;}\n\n\/* futures *\/\n#cq .cq-futures-panel{background:var(--cq-green);border:1px solid rgba(184,212,190,.2);padding:1.1rem 1.2rem;margin-top:.9rem;}\n#cq .cq-futures-panel[hidden]{display:none;}\n#cq .cq-futures-panel h4{color:#fff;margin:0 0 .5rem;font-size:1.1rem;}\n#cq .cq-futures-panel dl{display:grid;grid-template-columns:auto minmax(0,1fr);gap:.35rem 1rem;margin:0;}\n#cq .cq-futures-panel dt{font-family:var(--cq-mono);font-size:.68rem;letter-spacing:.12em;text-transform:uppercase;color:var(--cq-gold);padding-top:.15rem;}\n#cq .cq-futures-panel dd{margin:0;font-size:.95rem;color:var(--cq-pale);}\n#cq .cq-futures-panel dd a{color:var(--cq-gold);}\n\n\/* analog switch *\/\n#cq .cq-switch{display:inline-flex;align-items:center;gap:.7rem;background:transparent;border:0;color:#fff;padding:0;}\n#cq .cq-switch .cq-track{width:3.4rem;height:1.7rem;border-radius:999px;background:rgba(184,212,190,.25);position:relative;transition:background .2s;flex:none;}\n#cq .cq-switch .cq-track::after{content:\"\";position:absolute;top:.2rem;left:.2rem;width:1.3rem;height:1.3rem;border-radius:50%;background:#fff;transition:transform .2s;}\n#cq .cq-switch[aria-checked=\"true\"] .cq-track{background:var(--cq-gold);}\n#cq .cq-switch[aria-checked=\"true\"] .cq-track::after{transform:translateX(1.7rem);background:var(--cq-green);}\n#cq .cq-switch .cq-switch-lbl{font-family:var(--cq-mono);font-size:.78rem;letter-spacing:.1em;text-transform:uppercase;}\n#cq .cq-analog{display:grid;gap:1rem;margin-top:1rem;}\n@media (min-width:768px){#cq .cq-analog{grid-template-columns:1fr 1fr;}}\n#cq .cq-analog-col{background:var(--cq-green);padding:1rem 1.1rem;border-top:3px solid var(--cq-pale2);transition:opacity .25s;}\n#cq .cq-analog-col.is-analog{border-top-color:var(--cq-gold);}\n#cq .cq-analog-col h4{color:#fff;font-size:.8rem;font-family:var(--cq-mono);letter-spacing:.14em;text-transform:uppercase;margin:0 0 .6rem;}\n#cq .cq-analog-col li{display:grid;grid-template-columns:auto minmax(0,1fr);min-width:0;gap:.2rem .8rem;align-items:baseline;margin:0 0 .55rem;font-size:.92rem;color:var(--cq-pale);}\n#cq .cq-analog-col li b{font-family:var(--cq-mono);font-weight:400;color:#fff;font-size:1.1rem;white-space:nowrap;}\n#cq .cq-analog-col li a{color:var(--cq-gold);}\n#cq .cq-analog[data-mode=\"analog\"] .cq-analog-col:not(.is-analog){opacity:.45;}\n#cq .cq-analog[data-mode=\"digital\"] .cq-analog-col.is-analog{opacity:.45;}\n\n\/* refusal switch *\/\n#cq .cq-refusal-big{display:flex;flex-wrap:wrap;align-items:center;gap:1rem 1.5rem;margin:.5rem 0 1rem;}\n#cq .cq-refusal-big .cq-btn{font-size:.9rem;padding:.8rem 1.2rem;}\n#cq .cq-refusal-panel{margin-top:.75rem;display:grid;gap:.9rem;}\n@media (min-width:900px){#cq .cq-refusal-panel{grid-template-columns:1fr 1fr 1fr;}}\n#cq .cq-refusal-panel[hidden]{display:none;}\n#cq .cq-refusal-card{background:var(--cq-green);border-top:3px solid var(--cq-gold);padding:1rem 1.1rem;}\n#cq .cq-refusal-card h4{color:#fff;font-size:1rem;margin:0 0 .5rem;}\n#cq .cq-refusal-card p{font-size:.93rem;margin:0 0 .5rem;}\n#cq .cq-refusal-card a{color:var(--cq-gold);}\n\n\/* ---- footer \/ maintenance ---- *\/\n#cq .cq-maint{font-size:.95rem;}\n#cq .cq-maint code{font-family:var(--cq-mono);background:#fff;border:1px solid var(--cq-paper2);padding:.05rem .3rem;font-size:.85em;}\n#cq .cq-maint details{margin-top:.6rem;}\n#cq .cq-maint summary{cursor:pointer;font-family:var(--cq-mono);font-size:.74rem;letter-spacing:.12em;text-transform:uppercase;color:var(--cq-green3);}\n#cq .cq-maint pre{background:#fff;border:1px solid var(--cq-paper2);padding:.9rem 1rem;overflow:auto;font-size:.78rem;line-height:1.45;margin:.6rem 0 0;}\n\n\/* ---- motion ---- *\/\n@keyframes cq-blink{0%,49%{opacity:1}50%,100%{opacity:0}}\n@keyframes cq-in{from{opacity:0;transform:translateY(4px)}to{opacity:1;transform:none}}\n@media (prefers-reduced-motion:reduce){\n  #cq *{animation:none!important;transition:none!important;}\n  #cq .cq-tile:hover{transform:none;}\n}\n\n\/* ---- subpages (audit, field guide) ---- *\/\n#cq .cq-sub-hero{padding:2.25rem 1.5rem 1.75rem;}\n@media (min-width:768px){#cq .cq-sub-hero{padding:3rem 2.5rem 2.25rem;}}\n#cq .cq-sub-hero .cq-title{font-size:clamp(2rem,5vw,3.6rem);}\n#cq .cq-crumb{font-family:var(--cq-mono);font-size:.7rem;letter-spacing:.14em;text-transform:uppercase;color:var(--cq-pale2);margin:0 0 .6rem;}\n#cq .cq-crumb a{color:var(--cq-gold);text-decoration:none;}\n#cq .cq-crumb a:hover{text-decoration:underline;}\n#cq .cq-prose{max-width:66ch;font-size:1.03rem;}\n#cq .cq-prose h3{font-size:1.2rem;margin:1.4rem 0 .5rem;}\n#cq .cq-prose ul.cq-bullets{list-style:disc;padding-left:1.2rem;margin:0 0 1rem;}\n#cq .cq-prose ul.cq-bullets li{margin:0 0 .4rem;}\n#cq .cq-audit{display:grid;gap:1.25rem;margin-top:1rem;}\n#cq .cq-habit{background:var(--cq-green);border:1px solid rgba(184,212,190,.15);padding:1rem 1.1rem 1.1rem;}\n#cq .cq-habit h4{color:var(--cq-gold);font-family:var(--cq-mono);font-size:.78rem;letter-spacing:.14em;text-transform:uppercase;margin:0 0 .6rem;}\n#cq .cq-habit .cq-check-item{background:var(--cq-green2);}\n#cq .cq-bars{display:grid;gap:.6rem;margin:1rem 0;}\n#cq .cq-bar{display:grid;grid-template-columns:9rem minmax(0,1fr) 3rem;gap:.6rem;align-items:center;font-family:var(--cq-mono);font-size:.74rem;letter-spacing:.06em;color:var(--cq-pale);}\n@media (max-width:640px){#cq .cq-bar{grid-template-columns:6rem minmax(0,1fr) 3rem;}}\n#cq .cq-bar i{display:block;height:12px;background:rgba(184,212,190,.15);position:relative;overflow:hidden;}\n#cq .cq-bar i b{position:absolute;left:0;top:0;bottom:0;background:var(--cq-gold);transition:width .5s ease;}\n#cq .cq-bar[data-level=\"high\"] i b{background:var(--cq-rust2);}\n#cq .cq-plan{background:var(--cq-green);border-left:3px solid var(--cq-gold);padding:1rem 1.1rem;margin-top:1rem;}\n#cq .cq-plan h4{color:#fff;margin:0 0 .5rem;font-size:1.05rem;}\n#cq .cq-plan ol{list-style:decimal;padding-left:1.3rem;margin:0 0 .6rem;color:#fff;}\n#cq .cq-plan ol li{margin:0 0 .5rem;}\n#cq .cq-plan ol li span{color:var(--cq-pale);}\n#cq .cq-plan ol li a{color:var(--cq-gold);}\n#cq .cq-table{width:100%;border-collapse:collapse;font-size:.9rem;margin:1rem 0 0;}\n#cq .cq-table th,#cq .cq-table td{text-align:left;vertical-align:top;padding:.6rem .6rem;border-top:1px solid rgba(184,212,190,.2);}\n#cq .cq-table th{font-family:var(--cq-mono);font-size:.68rem;letter-spacing:.12em;text-transform:uppercase;color:var(--cq-gold);border-top:0;}\n#cq .cq-table td a{color:var(--cq-gold);}\n#cq .cq-table td b{color:#fff;font-weight:600;}\n#cq .cq-table-wrap{overflow-x:auto;}\n#cq .cq-timeline{position:relative;margin:1rem 0 0;padding-left:1.5rem;border-left:2px solid var(--cq-green3);}\n#cq .cq-timeline li{position:relative;margin:0 0 1rem;}\n#cq .cq-timeline li::before{content:\"\";position:absolute;left:calc(-1.5rem - 7px);top:.5rem;width:12px;height:12px;border-radius:50%;background:var(--cq-gold);border:2px solid var(--cq-paper);}\n#cq .cq-timeline .cq-when{font-family:var(--cq-mono);font-size:.72rem;letter-spacing:.12em;text-transform:uppercase;color:var(--cq-green3);margin:0 0 .15rem;}\n#cq .cq-timeline details{margin:0;}\n#cq .cq-timeline summary{cursor:pointer;font-weight:700;font-size:1rem;color:var(--cq-ink);}\n#cq .cq-timeline details p{margin:.5rem 0 0;font-size:.95rem;max-width:66ch;}\n#cq .cq-timeline details a{color:var(--cq-green3);}\n#cq .cq-template{background:#fff;border:1px solid var(--cq-paper2);padding:1rem 1.1rem;font-size:.95rem;max-width:70ch;}\n#cq .cq-template p{margin:0 0 .7rem;}\n#cq .cq-three{display:grid;gap:1rem;margin-top:1rem;}\n@media (min-width:860px){#cq .cq-three{grid-template-columns:1fr 1fr 1fr;}}\n#cq .cq-card{background:#fff;border-top:3px solid var(--cq-green3);padding:1rem 1.1rem;}\n#cq .cq-card h3{font-size:1.05rem;margin:0 0 .5rem;}\n#cq .cq-card p,#cq .cq-card li{font-size:.93rem;}\n#cq .cq-card ul{list-style:disc;padding-left:1.1rem;}\n#cq .cq-card li{margin:0 0 .35rem;}\n#cq .cq-card a{color:var(--cq-green3);}\n<\/style>","protected":false},"excerpt":{"rendered":"<p>\u2190 Critique<\/p>\n","protected":false},"author":256,"featured_media":0,"parent":251,"menu_order":0,"comment_status":"closed","ping_status":"closed","template":"","meta":{"advgb_blocks_editor_width":"","advgb_blocks_columns_visual_guide":"","footnotes":""},"class_list":["post-829","page","type-page","status-publish","hentry"],"coauthors":[],"author_meta":{"author_link":"https:\/\/sites.clarkson.edu\/ai-institute\/author\/scasper\/","display_name":"scasper"},"relative_dates":{"created":"Posted 1 week ago","modified":"Updated 1 week ago"},"absolute_dates":{"created":"Posted on September 18, 2026","modified":"Updated on September 18, 2026"},"absolute_dates_time":{"created":"Posted on September 18, 2026 6:58 pm","modified":"Updated on September 18, 2026 7:07 pm"},"featured_img_caption":"","featured_img":false,"series_order":"","_links":{"self":[{"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/pages\/829","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/pages"}],"about":[{"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/types\/page"}],"author":[{"embeddable":true,"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/users\/256"}],"replies":[{"embeddable":true,"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/comments?post=829"}],"version-history":[{"count":3,"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/pages\/829\/revisions"}],"predecessor-version":[{"id":837,"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/pages\/829\/revisions\/837"}],"up":[{"embeddable":true,"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/pages\/251"}],"wp:attachment":[{"href":"https:\/\/sites.clarkson.edu\/ai-institute\/wp-json\/wp\/v2\/media?parent=829"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}