[{"data":1,"prerenderedAt":1527},["ShallowReactive",2],{"navigation":3,"\u002Fnews\u002Finsights\u002Fthe-part-you-meant-to-keep":357,"\u002Fnews\u002Finsights\u002Fthe-part-you-meant-to-keep-surround":870},[4,8,17,21,25,29,33,345,349,353],{"title":5,"path":6,"stem":7},"About","\u002Fabout","about",{"title":9,"path":10,"stem":11,"children":12},"Authentication","\u002Fauth","auth",[13],{"title":14,"path":15,"stem":16},"Email Confirmation","\u002Fauth\u002Fconfirmation","auth\u002Fconfirmation",{"title":18,"path":19,"stem":20},"Industry Briefs","\u002Fcase-studies","case-studies",{"title":22,"path":23,"stem":24},"Contact","\u002Fcontact","contact",{"title":26,"path":27,"stem":28},"Thinkata | Production AI Engineering","\u002F","index",{"title":30,"path":31,"stem":32},"Insights","\u002Finsights","insights",{"title":34,"path":35,"stem":36,"children":37},"News","\u002Fnews","news",[38,41,65],{"title":39,"path":35,"stem":40},"News & Insights","news\u002Findex",{"title":18,"path":42,"stem":43,"children":44},"\u002Fnews\u002Fcase-studies","news\u002Fcase-studies",[45,49,53,57,61],{"title":46,"path":47,"stem":48},"Building Secure and Scalable AI Infrastructure: Integrating with Existing Systems through Modern Cloud Frameworks","\u002Fnews\u002Fcase-studies\u002Fcloud-infrastructure-ai","news\u002Fcase-studies\u002Fcloud-infrastructure-ai",{"title":50,"path":51,"stem":52},"Making Sense of Financial Regulations: How AI Teams Can Tackle Complex Documents","\u002Fnews\u002Fcase-studies\u002Ffinancial-regulations","news\u002Fcase-studies\u002Ffinancial-regulations",{"title":54,"path":55,"stem":56},"AI-Powered Transformations in Healthcare","\u002Fnews\u002Fcase-studies\u002Fhealth-care","news\u002Fcase-studies\u002Fhealth-care",{"title":58,"path":59,"stem":60},"Generative AI in Upstream Natural Gas: Shell's Exploration Initiative","\u002Fnews\u002Fcase-studies\u002Foil-gas","news\u002Fcase-studies\u002Foil-gas",{"title":62,"path":63,"stem":64},"Optimizing Manufacturing with AI-Driven Multi-Agent Systems","\u002Fnews\u002Fcase-studies\u002Fsupply-chain-optimization","news\u002Fcase-studies\u002Fsupply-chain-optimization",{"title":30,"path":66,"stem":67,"children":68},"\u002Fnews\u002Finsights","news\u002Finsights",[69,73,77,81,85,89,93,97,101,105,109,113,117,121,125,129,133,137,141,145,149,153,157,161,165,169,173,177,181,185,189,193,197,201,205,209,213,217,221,225,229,233,237,241,245,249,253,257,261,265,269,273,277,281,285,289,293,297,301,305,309,313,317,321,325,329,333,337,341],{"title":70,"path":71,"stem":72},"Nothing Gets Deleted, Just Blurred","\u002Fnews\u002Finsights\u002Fadaptive-context-memory","news\u002Finsights\u002Fadaptive-context-memory",{"title":74,"path":75,"stem":76},"The Capability-Reliability Split in Agent Systems","\u002Fnews\u002Finsights\u002Fagent-capability-reliability-split","news\u002Finsights\u002Fagent-capability-reliability-split",{"title":78,"path":79,"stem":80},"When Code Gets Cheap, Judgment Gets Expensive","\u002Fnews\u002Finsights\u002Fagentic-coding-engineer-skills","news\u002Finsights\u002Fagentic-coding-engineer-skills",{"title":82,"path":83,"stem":84},"The Rise of AI Agents in Cyberattacks: Latest Research and Threats","\u002Fnews\u002Finsights\u002Fai-agent-cyber-threats","news\u002Finsights\u002Fai-agent-cyber-threats",{"title":86,"path":87,"stem":88},"Winning the Bid Is Not the Same as Knowing the Price","\u002Fnews\u002Finsights\u002Fai-agent-economics","news\u002Finsights\u002Fai-agent-economics",{"title":90,"path":91,"stem":92},"The Smart Enterprise AI Stack: Why Teams of AI Agents Beat Solo Models Consistently","\u002Fnews\u002Finsights\u002Fai-architecture","news\u002Finsights\u002Fai-architecture",{"title":94,"path":95,"stem":96},"When Seeing Everything Becomes the Only Option","\u002Fnews\u002Finsights\u002Fai-comprehensive-observability","news\u002Finsights\u002Fai-comprehensive-observability",{"title":98,"path":99,"stem":100},"The Data Infrastructure AI-Native Systems Can't Ignore","\u002Fnews\u002Finsights\u002Fai-data-layer","news\u002Finsights\u002Fai-data-layer",{"title":102,"path":103,"stem":104},"Enterprise AI Triage Systems: Intelligent Automation for Large-Scale Operations","\u002Fnews\u002Finsights\u002Fai-enterprise-triage","news\u002Finsights\u002Fai-enterprise-triage",{"title":106,"path":107,"stem":108},"When Oversight Becomes Infrastructure","\u002Fnews\u002Finsights\u002Fai-governed-autonomy","news\u002Finsights\u002Fai-governed-autonomy",{"title":110,"path":111,"stem":112},"Designing for Graceful Failure in Compound AI Systems","\u002Fnews\u002Finsights\u002Fai-graceful-failure","news\u002Finsights\u002Fai-graceful-failure",{"title":114,"path":115,"stem":116},"Intelligent Composability: Building AI Systems Like Orchestra, Not Soloists","\u002Fnews\u002Finsights\u002Fai-intelligent-composability","news\u002Finsights\u002Fai-intelligent-composability",{"title":118,"path":119,"stem":120},"Building the Plane While Flying It — Migrating from Monolith to AI-Native Without Stopping","\u002Fnews\u002Finsights\u002Fai-migration-path","news\u002Finsights\u002Fai-migration-path",{"title":122,"path":123,"stem":124},"Stability Through Continuous Adaptation","\u002Fnews\u002Finsights\u002Fai-native-overview","news\u002Finsights\u002Fai-native-overview",{"title":126,"path":127,"stem":128},"Provable Stability: Mathematical Guarantees for Adaptive AI Systems","\u002Fnews\u002Finsights\u002Fai-provable-stability","news\u002Finsights\u002Fai-provable-stability",{"title":130,"path":131,"stem":132},"How Temperature Tuning Makes or Breaks Reinforcement Learning","\u002Fnews\u002Finsights\u002Fai-soft-actor-critic-entropy-collapse","news\u002Finsights\u002Fai-soft-actor-critic-entropy-collapse",{"title":134,"path":135,"stem":136},"Testing What Can't Be Predicted","\u002Fnews\u002Finsights\u002Fai-systems-testing","news\u002Finsights\u002Fai-systems-testing",{"title":138,"path":139,"stem":140},"Delegating the Job Is Not the Same as Delegating the Rules","\u002Fnews\u002Finsights\u002Fauthorization-drift-multi-agent","news\u002Finsights\u002Fauthorization-drift-multi-agent",{"title":142,"path":143,"stem":144},"Closing the Loop: How Human Corrections Can Make AI Systems Smarter Over Time","\u002Fnews\u002Finsights\u002Fclosing-the-loop","news\u002Finsights\u002Fclosing-the-loop",{"title":146,"path":147,"stem":148},"A Panel Can Be Safer Than Any Member of It","\u002Fnews\u002Finsights\u002Fcoalitional-alignment-panel-review","news\u002Finsights\u002Fcoalitional-alignment-panel-review",{"title":150,"path":151,"stem":152},"Multi-Path Reasoning: Collaborative and Competitive Approaches in AI","\u002Fnews\u002Finsights\u002Fcollaborative-competitive-agents","news\u002Finsights\u002Fcollaborative-competitive-agents",{"title":154,"path":155,"stem":156},"Why Challenges Supercharge Smarts for Humans and AI","\u002Fnews\u002Finsights\u002Fcompetition-improves-ai","news\u002Finsights\u002Fcompetition-improves-ai",{"title":158,"path":159,"stem":160},"Context is Infrastructure, Not Instructions","\u002Fnews\u002Finsights\u002Fcontext-is-infrastructure","news\u002Finsights\u002Fcontext-is-infrastructure",{"title":162,"path":163,"stem":164},"Context is the New Code","\u002Fnews\u002Finsights\u002Fcontext-is-new-code","news\u002Finsights\u002Fcontext-is-new-code",{"title":166,"path":167,"stem":168},"Continuous Thought Machines","\u002Fnews\u002Finsights\u002Fcontinuous-thought-machines","news\u002Finsights\u002Fcontinuous-thought-machines",{"title":170,"path":171,"stem":172},"Don't Vibe, Architect","\u002Fnews\u002Finsights\u002Fdont-vibe-architect","news\u002Finsights\u002Fdont-vibe-architect",{"title":174,"path":175,"stem":176},"The Edge of the Underdefined","\u002Fnews\u002Finsights\u002Fedge-of-the-underdefined","news\u002Finsights\u002Fedge-of-the-underdefined",{"title":178,"path":179,"stem":180},"Experts All the Way Down","\u002Fnews\u002Finsights\u002Fexperts-all-the-way","news\u002Finsights\u002Fexperts-all-the-way",{"title":182,"path":183,"stem":184},"A Multi-Tier Safety Architecture for Critical Applications","\u002Fnews\u002Finsights\u002Ffour-tier-architecture","news\u002Finsights\u002Ffour-tier-architecture",{"title":186,"path":187,"stem":188},"Green Dashboard, Unhappy Users","\u002Fnews\u002Finsights\u002Fgreen-dashboard-unhappy-users","news\u002Finsights\u002Fgreen-dashboard-unhappy-users",{"title":190,"path":191,"stem":192},"Hybrid Autoregressive Residual Tokens","\u002Fnews\u002Finsights\u002Fhart-model","news\u002Finsights\u002Fhart-model",{"title":194,"path":195,"stem":196},"Hierarchical Reasoning in Artificial Intelligence","\u002Fnews\u002Finsights\u002Fhierarchical-approaches","news\u002Finsights\u002Fhierarchical-approaches",{"title":198,"path":199,"stem":200},"Taking Turns Instead of Splitting the Difference","\u002Fnews\u002Finsights\u002Finterleaved-distillation-rl","news\u002Finsights\u002Finterleaved-distillation-rl",{"title":202,"path":203,"stem":204},"Latent Diffusion for Language Generation: A Comprehensive Overview","\u002Fnews\u002Finsights\u002Flatent-diffusion-for-language","news\u002Finsights\u002Flatent-diffusion-for-language",{"title":206,"path":207,"stem":208},"Breaking Language Barriers: How AI Can Translate Without Examples","\u002Fnews\u002Finsights\u002Flearning-languages","news\u002Finsights\u002Flearning-languages",{"title":210,"path":211,"stem":212},"The Emergence of AI Deception: How Large Language Models Have Learned to Strategically Mislead Users","\u002Fnews\u002Finsights\u002Fllm-deception","news\u002Finsights\u002Fllm-deception",{"title":214,"path":215,"stem":216},"Grading on a Shared Curve","\u002Fnews\u002Finsights\u002Fllm-judge-correlated-errors","news\u002Finsights\u002Fllm-judge-correlated-errors",{"title":218,"path":219,"stem":220},"Synergizing Specialized Reasoning and General Capabilities in AI","\u002Fnews\u002Finsights\u002Fllm-reasoning-advances","news\u002Finsights\u002Fllm-reasoning-advances",{"title":222,"path":223,"stem":224},"The Expensive Default","\u002Fnews\u002Finsights\u002Fllm-routing-cost-quality","news\u002Finsights\u002Fllm-routing-cost-quality",{"title":226,"path":227,"stem":228},"The AI That Rewrites Itself: MIT's Breakthrough in Self-Adapting Language Models","\u002Fnews\u002Finsights\u002Fllm-seal","news\u002Finsights\u002Fllm-seal",{"title":230,"path":231,"stem":232},"Forgetting on Request","\u002Fnews\u002Finsights\u002Fmachine-unlearning-removal-suppression","news\u002Finsights\u002Fmachine-unlearning-removal-suppression",{"title":234,"path":235,"stem":236},"The Map and the Walk","\u002Fnews\u002Finsights\u002Fmap-and-walk","news\u002Finsights\u002Fmap-and-walk",{"title":238,"path":239,"stem":240},"Metacognitive Reinforcement Learning for Self-Improving AI Systems","\u002Fnews\u002Finsights\u002Fmetacognitive-reinforcement-learning","news\u002Finsights\u002Fmetacognitive-reinforcement-learning",{"title":242,"path":243,"stem":244},"Revolutionary Advancements in Mixture of Experts (MoE) Architectures","\u002Fnews\u002Finsights\u002Fmixture-of-experts","news\u002Finsights\u002Fmixture-of-experts",{"title":246,"path":247,"stem":248},"One Model, Many Customers, and the Leak Nobody Tests For","\u002Fnews\u002Finsights\u002Fmulti-tenant-ai-isolation","news\u002Finsights\u002Fmulti-tenant-ai-isolation",{"title":250,"path":251,"stem":252},"Balancing Neural Plasticity and Stability","\u002Fnews\u002Finsights\u002Fneural-plasticity","news\u002Finsights\u002Fneural-plasticity",{"title":254,"path":255,"stem":256},"Offline RL and the Data Flywheel","\u002Fnews\u002Finsights\u002Foffline-rl-data-flywheel","news\u002Finsights\u002Foffline-rl-data-flywheel",{"title":258,"path":259,"stem":260},"Orchestration Without a Signal From the Room","\u002Fnews\u002Finsights\u002Fplanner-in-the-loop","news\u002Finsights\u002Fplanner-in-the-loop",{"title":262,"path":263,"stem":264},"Second-Guessing Has a Price","\u002Fnews\u002Finsights\u002Freasoning-budget-allocation","news\u002Finsights\u002Freasoning-budget-allocation",{"title":266,"path":267,"stem":268},"Reasoning You Can Check","\u002Fnews\u002Finsights\u002Freasoning-you-can-check","news\u002Finsights\u002Freasoning-you-can-check",{"title":270,"path":271,"stem":272},"When Optimization Optimizes Itself","\u002Fnews\u002Finsights\u002Frecursive-goodhart","news\u002Finsights\u002Frecursive-goodhart",{"title":274,"path":275,"stem":276},"Reward Design as Architecture","\u002Fnews\u002Finsights\u002Freward-design-as-architecture","news\u002Finsights\u002Freward-design-as-architecture",{"title":278,"path":279,"stem":280},"When Success Has No Author: The Temporal Credit Assignment Problem","\u002Fnews\u002Finsights\u002Frl-credit-assignment-problem","news\u002Finsights\u002Frl-credit-assignment-problem",{"title":282,"path":283,"stem":284},"Beyond Entropy Collapse: When Exploration Succeeds but Learning Fails","\u002Fnews\u002Finsights\u002Frl-optimization-gaps","news\u002Finsights\u002Frl-optimization-gaps",{"title":286,"path":287,"stem":288},"The Body Was Supposed to Be the Easy Part","\u002Fnews\u002Finsights\u002Frobot-embodiment-interface","news\u002Finsights\u002Frobot-embodiment-interface",{"title":290,"path":291,"stem":292},"The Path to Practical Confidential Computing for AI Systems","\u002Fnews\u002Finsights\u002Fsecure-ai-architectures","news\u002Finsights\u002Fsecure-ai-architectures",{"title":294,"path":295,"stem":296},"When an AI Edits Its Own Memory","\u002Fnews\u002Finsights\u002Fself-editing-agent-memory","news\u002Finsights\u002Fself-editing-agent-memory",{"title":298,"path":299,"stem":300},"The Guard at the Front Door","\u002Fnews\u002Finsights\u002Fshallow-safety-alignment-refusal","news\u002Finsights\u002Fshallow-safety-alignment-refusal",{"title":302,"path":303,"stem":304},"Guess First, Check Later","\u002Fnews\u002Finsights\u002Fspeculative-execution-pattern","news\u002Finsights\u002Fspeculative-execution-pattern",{"title":306,"path":307,"stem":308},"Spiking Neural Networks for Energy-Efficient AI","\u002Fnews\u002Finsights\u002Fspiking-neural-networks","news\u002Finsights\u002Fspiking-neural-networks",{"title":310,"path":311,"stem":312},"When Replay Is Not an Option: Streaming Q-Learning and SARSA Get a Second Look","\u002Fnews\u002Finsights\u002Fstreaming-q-learning-revival","news\u002Finsights\u002Fstreaming-q-learning-revival",{"title":314,"path":315,"stem":316},"The Turn as the Unit of Quality","\u002Fnews\u002Finsights\u002Fstructured-iteration-quality","news\u002Finsights\u002Fstructured-iteration-quality",{"title":318,"path":319,"stem":320},"AI Speech Translation: Breaking Down Language Barriers","\u002Fnews\u002Finsights\u002Fsts-performance-advances","news\u002Finsights\u002Fsts-performance-advances",{"title":322,"path":323,"stem":324},"Test-Time Training Layers: The Next Evolution in Transformer Architecture","\u002Fnews\u002Finsights\u002Ftest-time-training-layers","news\u002Finsights\u002Ftest-time-training-layers",{"title":326,"path":327,"stem":328},"The Part You Meant to Keep","\u002Fnews\u002Finsights\u002Fthe-part-you-meant-to-keep","news\u002Finsights\u002Fthe-part-you-meant-to-keep",{"title":330,"path":331,"stem":332},"Predicting the Next Second: Tile Coding as a Real-Time Prediction Engine","\u002Fnews\u002Finsights\u002Ftile-coding-real-time-prediction","news\u002Finsights\u002Ftile-coding-real-time-prediction",{"title":334,"path":335,"stem":336},"Breakthrough: Large Language Models Pass the Turing Test","\u002Fnews\u002Finsights\u002Fturing-tests","news\u002Finsights\u002Fturing-tests",{"title":338,"path":339,"stem":340},"Algorithms Used in Autonomous Fighter Jet Flight Research","\u002Fnews\u002Finsights\u002Fvenom-f16-flight-autonomy","news\u002Finsights\u002Fvenom-f16-flight-autonomy",{"title":342,"path":343,"stem":344},"Training in a World That Does Not Exist Yet","\u002Fnews\u002Finsights\u002Fworld-models-as-infrastructure","news\u002Finsights\u002Fworld-models-as-infrastructure",{"title":346,"path":347,"stem":348},"Privacy Policy","\u002Fprivacy","privacy",{"title":350,"path":351,"stem":352},"Research","\u002Fresearch","research",{"title":354,"path":355,"stem":356},"Terms of Service","\u002Fterms","terms",{"id":358,"title":326,"body":359,"date":850,"description":851,"extension":852,"image":853,"meta":855,"navigation":867,"path":327,"seo":868,"stem":328,"__hash__":869},"insights\u002Fnews\u002Finsights\u002Fthe-part-you-meant-to-keep.md",{"type":360,"value":361,"toc":838},"minimark",[362,381,385,397,401,411,414,424,428,438,448,452,460,470,473,569,579,585,589,599,607,617,623,627,630,702,705,709,715,723,726],[363,364,367,368,367,374],"div",{"className":365},[366],"page-title","\n  ",[369,370,326],"h1",{"className":371,"id":373},[372],"page-title__main","the-part-you-meant-to-keep",[375,376,380],"h2",{"className":377,"id":379},[378],"page-title__sub","is-there-a-point-where-ai-help-turns-into-harm-and-could-a-scale-find-it","Is There a Point Where AI Help Turns Into Harm, and Could a Scale Find It?",[382,383,384],"p",{},"Few people ask whether a letter went through spell-check. The small bottle of correction fluid that once sat beside the typewriter went away with the carbon paper, and the red underline that replaced it stopped being a topic of conversation long ago. A cover letter written with a chatbot still draws suspicion, and detectors and watermarks exist to answer whether a model was involved. As models settle into ordinary revision, a different question may matter more. Did the help cost anything, and at what point?",[382,386,387,388,396],{},"Psychologists call this cognitive offloading, using a physical action to make a task less demanding on the mind, like setting a phone reminder instead of memorizing a date ",[389,390,391],"sup",{},[392,393,395],"a",{"href":394},"#source-1","[1]",". Usually that is a fair trade. So where does a convenience start to become a loss? Is there a line, and could anyone measure it?",[375,398,400],{"id":399},"the-cost-arrives-later","The Cost Arrives Later",[363,402,404,405],{"style":403},"width: 100%; padding: 2%;","\n    ",[406,407],"img",{"src":408,"alt":409,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1793393444608-3886cf95b354?w=1200&auto=format&fit=crop","A lifter on a bench press with a spotter standing behind the bar","width: 100%; height: auto;",[382,412,413],{},"The spotter's hands rest at the hips until the bar stalls. A nudge on the last repetition finishes the set, and the lifter's chest still did the work. Lifting from the first repetition would move the same weight along the same path, and nothing about the bar would show the difference. AI help has a similar measurement problem. The output can look fine either way, and any cost tends to appear once the hands come away.",[382,415,416,417,423],{},"How quickly can that happen? Across three randomized experiments, people practiced fractions or reading comprehension in sessions of about a quarter of an hour, some with a chatbot beside them, then took a short test alone. The chatbot helped during practice. On the test, those in the first experiment who had used it solved 57 percent of problems, compared with 74 percent for those who had not, and they gave up more often ",[389,418,419],{},[392,420,422],{"href":421},"#source-2","[2]",". The gap was smaller in a replication, and testing came right after practice, so the authors leave open whether the effect builds or fades with daily use.",[375,425,427],{"id":426},"counting-requests-misses-it","Counting Requests Misses It",[382,429,430,431,437],{},"The obvious dial is quantity. Could some number of requests per task, or hours per week, mark the point of trouble? One small experiment points elsewhere. In it, 124 adults worked through logic puzzles, most with a simulated assistant whose price per request varied. Once starting ability was accounted for, request counts said almost nothing about skill gained. What tracked it was solo share, the fraction of time a person worked alone on a problem before asking ",[389,432,433],{},[392,434,436],{"href":435},"#source-3","[3]",". Scores while assisted also overstated how people later did on their own, which suggests that a session going well is a poor guide.",[382,439,440,441,447],{},"A larger experiment on fraction practice gives the nearest thing to a dose curve. Each extra 10 percentage points of problems on which someone requested the full answer came with 32 percent lower odds of getting a later test question right ",[389,442,443],{},[392,444,446],{"href":445},"#source-4","[4]",". That analysis is correlational, and it assumes a smooth trend, so a sharper break could be hiding inside it. Access alone produced no detectable harm in that study, perhaps because the assistant gave full answers only when explicitly asked. No study found in this research set out to look for a break point. For now a ramp seems at least as plausible as a cliff, which is inconvenient for anyone hoping to draw one line.",[375,449,451],{"id":450},"five-rungs-drawn-in-pencil","Five Rungs, Drawn in Pencil",[382,453,454,455,459],{},"If quantity is a weak guide, depth looks stronger. In the replication of those short experiments, people who said they mostly asked for hints scored about the same as people with no chatbot, while those who mostly asked for answers fell behind ",[389,456,457],{},[392,458,422],{"href":421},". Those groups chose their own habits, so the comparison only suggests a cause.",[382,461,462,463,469],{},"One recent writing study offers a vocabulary for depth. It separates four layers, surface (grammar and local wording), structural (outlines and organization), idea (claims and examples) and reasoning (evidence interpretation and argumentative logic), with coders rating each from 1, where the writer did the work, to 5, where the tool did it and little changed. Over an eight-week course, heavier offloading at every layer went with weaker thinking on later no-AI tasks, and the cost of open access ran most faintly through the surface layer and most strongly through reasoning ",[389,464,465],{},[392,466,468],{"href":467},"#source-5","[5]",". Open-access writers also sent nearly twice as many prompts per task, yet far fewer of their revisions were writer-led, so more turns may not mean more involvement. Six classes without random assignment can only suggest so much, and the author says the framework still needs independent validation.",[382,471,472],{},"Adding memory at the bottom gives a rough ladder. It was assembled for this article, and no validation stands behind it.",[474,475,476,495],"table",{},[477,478,479],"thead",{},[480,481,482,486,489,492],"tr",{},[483,484,485],"th",{},"Rung",[483,487,488],{},"What moves to the tool",[483,490,491],{},"Usually helps when",[483,493,494],{},"May start to cost when",[496,497,498,513,527,541,555],"tbody",{},[480,499,500,504,507,510],{},[501,502,503],"td",{},"1. Storing",[501,505,506],{},"Dates, lists, references",[501,508,509],{},"The facts can be looked up again",[501,511,512],{},"Recall is the skill being built",[480,514,515,518,521,524],{},[501,516,517],{},"2. Polishing",[501,519,520],{},"Spelling, grammar, local wording",[501,522,523],{},"The meaning is already settled",[501,525,526],{},"Checking stops, or edits drift into meaning",[480,528,529,532,535,538],{},[501,530,531],{},"3. Arranging",[501,533,534],{},"Outline and order",[501,536,537],{},"The person knows the material",[501,539,540],{},"The person is still learning how the pieces fit",[480,542,543,546,549,552],{},[501,544,545],{},"4. Supplying",[501,547,548],{},"Claims, examples, solutions",[501,550,551],{},"The task is routine",[501,553,554],{},"Working it out was the point",[480,556,557,560,563,566],{},[501,558,559],{},"5. Deciding",[501,561,562],{},"Interpretation and judgment",[501,564,565],{},"Stakes are low and the call is easy to check",[501,567,568],{},"The person will answer for the conclusion later",[382,570,571,572,578],{},"Even the second rung has a price. In a 2005 experiment, people with stronger verbal skills left nearly twice as many unflagged errors behind when a spelling and grammar checker was switched on ",[389,573,574],{},[392,575,577],{"href":576},"#source-6","[6]",".",[382,580,581,582,584],{},"Read across the rows and the threshold stops being a place on the ladder. One working definition might be that help turns to harm at the rung where the tool starts doing something the person wanted, or needed, to still be able to do alone. A status email can sit at rung five with little lost. A junior engineer at rung four on debugging may be giving away what the job was supposed to build, the worry behind ",[392,583,78],{"href":79},". Same ladder, opposite verdicts, so a useful scale would need to ask about intent as well as behavior.",[375,586,588],{"id":587},"the-closest-thing-to-a-number","The Closest Thing to a Number",[382,590,591,592,598],{},"Could the rungs be measured automatically? The most direct attempt found here is the Offloading Score. It reads screen recordings and keystrokes, has a language model rebuild how an average person would finish each AI-assisted step without the tool, and reports the fraction of steps saved, between 0 and 1. Among 40 freelance developers, a one-hour limit produced more offloading than a four-hour limit, 0.45 against 0.31, while self-reported mental load barely moved ",[389,593,594],{},[392,595,597],{"href":596},"#source-7","[7]",". People may not feel the shift as it happens.",[382,600,601,602,606],{},"The authors also sketch an actual line, an overreliance threshold of 0.33 tied to how well developers could recall their own system. They call it illustrative and say real thresholds should be set empirically. The link to recall was weak overall and stronger only after setting aside about 30 percent of users who combined moderate to high reliance with high recall, apparently using the tool to learn ",[389,603,604],{},[392,605,597],{"href":596},". Those users could cross that line with no visible harm, and they may be the most interesting people in the study.",[382,608,609,610,616],{},"Questionnaires take the other route. A free self-check called PAUSE sorts readings into minimal, mild, moderate and strong bands, and its author is candid that the cut points were chosen for legibility, with no reliability or validity evidence behind them ",[389,611,612],{},[392,613,615],{"href":614},"#source-8","[8]",". Levels are easy to print. Knowing what they mean is the hard part.",[382,618,619,620,622],{},"From a machine learning standpoint, this resembles choosing an operating point on a classifier. A score can come from logs, but the cut depends on what each kind of mistake costs, and those costs differ between an expense note and a thesis. The ground truth, what someone can do later with the tool gone, is slow and expensive to collect, and several studies above tested within minutes of practice. Then comes the feedback loop. Once anyone is judged on the number, prompts will likely get padded, the pattern ",[392,621,270],{"href":271}," traced in systems that learn to satisfy their own metrics.",[375,624,626],{"id":625},"right-use-wrong-use","Right Use, Wrong Use",[382,628,629],{},"Until a validated scale exists, the ladder can still sort everyday choices. Right use leaves a person able to do what they wanted to keep. Wrong use gives that ability away, often with little sign at the time. The labels depend on that goal, so the same request can land in either column. These pairings follow from the evidence above, and no study has tested them as a set.",[474,631,632,645],{},[477,633,634],{},[480,635,636,639,642],{},[483,637,638],{},"Task",[483,640,641],{},"Right use",[483,643,644],{},"Wrong use",[496,646,647,658,669,680,691],{},[480,648,649,652,655],{},[501,650,651],{},"Grammar and wording",[501,653,654],{},"Polishing a draft with settled meaning, then checking the edits",[501,656,657],{},"Accepting every change unread, even when the meaning shifts",[480,659,660,663,666],{},[501,661,662],{},"Learning something new",[501,664,665],{},"Trying first, then asking for a hint",[501,667,668],{},"Asking for the full answer before trying",[480,670,671,674,677],{},[501,672,673],{},"Organizing work",[501,675,676],{},"Outlining familiar material",[501,678,679],{},"Letting the tool arrange material still being learned",[480,681,682,685,688],{},[501,683,684],{},"Ideas and arguments",[501,686,687],{},"Getting help on routine work that can be checked",[501,689,690],{},"Submitting an argument the writer could not explain",[480,692,693,696,699],{},[501,694,695],{},"Decisions",[501,697,698],{},"Handing off low-stakes calls that are easy to check",[501,700,701],{},"Handing off a conclusion the person must answer for later",[382,703,704],{},"A rough test fits in one question. After the session, could the person explain, defend or reproduce the parts of the work they meant to keep? A yes suggests the tool supported them. A no may be fine for a status email, as long as the trade was made on purpose.",[375,706,708],{"id":707},"a-bubble-between-two-lines","A Bubble Between Two Lines",[363,710,404,711],{"style":403},[406,712],{"src":713,"alt":714,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1786341594766-22826f989d0b?w=1200&auto=format&fit=crop","A yellow spirit level vial with its bubble between two black lines",[382,716,717,718,722],{},"A bubble between two lines straightens nothing. It tells the person holding the level where things stand, and the rest is up to them. That may be the more realistic job for a scale, and there is some early support for it. In the larger fraction experiment, some participants read a short note between problems describing what they had just done and what it meant for the practice they got, with no grades, points or instructions. Their odds of requesting a full answer fell by about half and they did better on the unaided test, while a points reward for effort changed neither ",[389,719,720],{},[392,721,446],{"href":445},". The margins of error were wide, and it was one brief session with an immediate test, so durability remains unknown. It also hints at something awkward for anyone building a scale. The note seems to have worked without a score.",[382,724,725],{},"Spell-check seems to have stopped being a question once people had decided, mostly without discussing it, what they no longer needed to do themselves. That decision has not been made for reasoning, and it may not get made the same way for everyone. A scale might never find one point where help becomes harm. Could it still show a person where they stood on a given afternoon, close enough to change what they hand over next time?",[363,727,367,731,367,734],{"className":728},[729,730],"references","mt-8",[375,732,733],{"id":729},"References",[735,736,404,742,404,760,404,772,404,784,404,794,404,806,404,818,404,828,367],"ol",{"className":737},[738,739,740,741],"list-decimal","list-inside","space-y-2","mt-4",[743,744,746,747,751,752],"li",{"id":745},"source-1","E. F. Risko and S. J. Gilbert, \"Cognitive Offloading,\" ",[748,749,750],"em",{},"Trends in Cognitive Sciences",", vol. 20, no. 9, pp. 676–688, 2016. DOI: ",[392,753,759],{"href":754,"target":755,"className":756},"https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.tics.2016.07.002","_blank",[757,758],"text-blue-600","underline","[Online]",[743,761,763,764,767,768],{"id":762},"source-2","G. Liu et al., \"AI Assistance Reduces Persistence and Hurts Independent Performance,\" ",[748,765,766],{},"arXiv",", 2026, ",[392,769,759],{"href":770,"target":755,"className":771},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2604.04721",[757,758],[743,773,775,776,779,780],{"id":774},"source-3","S. Wu et al., \"How AI Assistance Affects Human Skill Development: A Study of Learning with Logic Puzzles,\" in ",[748,777,778],{},"Proc. ACM Conference on Human-AI Complementarity and Alignment (HCOMP 2026)",", 2026. DOI: ",[392,781,759],{"href":782,"target":755,"className":783},"https:\u002F\u002Fdoi.org\u002F10.1145\u002F3834580.3838741",[757,758],[743,785,787,788,767,790],{"id":786},"source-4","S. Maier et al., \"Designing Against Deskilling: Metacognitive Feedback Reduces Cognitive Offloading to LLM Assistants,\" ",[748,789,766],{},[392,791,759],{"href":792,"target":755,"className":793},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.20143",[757,758],[743,795,797,798,801,802],{"id":796},"source-5","X. Chen, \"Layer-sensitive cognitive offloading in generative AI-assisted writing: supported performance and independent no-AI outcomes,\" ",[748,799,800],{},"Frontiers in Psychology",", vol. 17, art. 1906199, 2026. DOI: ",[392,803,759],{"href":804,"target":755,"className":805},"https:\u002F\u002Fdoi.org\u002F10.3389\u002Ffpsyg.2026.1906199",[757,758],[743,807,809,810,813,814],{"id":808},"source-6","D. F. Galletta et al., \"Does Spell-Checking Software Need a Warning Label?,\" ",[748,811,812],{},"Communications of the ACM",", vol. 48, no. 7, pp. 82–86, 2005. DOI: ",[392,815,759],{"href":816,"target":755,"className":817},"https:\u002F\u002Fdoi.org\u002F10.1145\u002F1070838.1070841",[757,758],[743,819,821,822,767,824],{"id":820},"source-7","V. Padmakumar et al., \"Offloading Score: Measuring AI Reliance Through Counterfactual Workflows,\" ",[748,823,766],{},[392,825,759],{"href":826,"target":755,"className":827},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2605.29392",[757,758],[743,829,831,832,767,834],{"id":830},"source-8","M. U. Alam, \"PAUSE: A Privacy-Preserving Self-Reflection Tool for AI-Associated Cognitive Offloading,\" ",[748,833,766],{},[392,835,759],{"href":836,"target":755,"className":837},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2609.13155",[757,758],{"title":839,"searchDepth":840,"depth":840,"links":841},"",2,[842,843,844,845,846,847,848,849],{"id":379,"depth":840,"text":380},{"id":399,"depth":840,"text":400},{"id":426,"depth":840,"text":427},{"id":450,"depth":840,"text":451},{"id":587,"depth":840,"text":588},{"id":625,"depth":840,"text":626},{"id":707,"depth":840,"text":708},{"id":729,"depth":840,"text":733},"2026-11-02","AI help tends to look good in the moment and cost something later. The evidence so far offers no single cutoff, and the line seems to move with what a person wants to keep. Could a cognitive offloading scale find it?","md",{"src":854},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1530686350401-7de25243dd89?w=1200&auto=format&fit=crop",{"authors":856,"badge":862,"source":864},[857],{"avatar":858,"name":860,"to":861},{"src":859},"\u002Fimg\u002Fmark_avatar.png","Mark Williams","#",{"label":863},"Cognitive Offloading",{"name":865,"url":866},"Thinkata Research","https:\u002F\u002Fthinkata.com",true,{"title":326,"description":851},"I2CiH4E-x_25GqPZ7G3zu0bgjJ6xoQJ8ikXT2zhnVQ4",[871,1172],{"id":872,"title":330,"body":873,"date":1160,"description":1161,"extension":852,"image":1162,"meta":1163,"navigation":867,"path":331,"seo":1170,"stem":332,"__hash__":1171,"_path":331},"insights\u002Fnews\u002Finsights\u002Ftile-coding-real-time-prediction.md",{"type":360,"value":874,"toc":1146},[875,888,894,902,910,914,917,924,931,939,943,951,954,976,980,988,996,1013,1017,1024,1037,1041,1048,1061],[363,876,367,878,367,883],{"className":877},[366],[369,879,882],{"className":880,"id":881},[372],"predicting-the-next-second","Predicting the Next Second",[375,884,887],{"className":885,"id":886},[378],"tile-coding-as-a-real-time-prediction-engine","Tile Coding as a Real-Time Prediction Engine",[363,889,404,890],{"style":403},[406,891],{"src":892,"alt":893,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1531259960912-746d0c872ee6?w=1200&auto=format&fit=crop","A cyclist in motion blur on a city street",[382,895,896,897,901],{},"A cyclist weaving through traffic is not reacting to the street as it is. The body is already braced for the pothole a second ahead, already leaning into a turn that has not started yet. Psychologists call this constant, small-scale forecasting nexting, and a group of reinforcement learning researchers borrowed the word to describe a robot that learned to do something similar ",[389,898,899],{},[392,900,395],{"href":394},". Reinforcement learning is the branch of machine learning in which software improves through trial and error, and one of its quieter strengths turns out to be learning to predict.",[382,903,904,905,909],{},"What made the robot practical was not a large neural network. It was tile coding, a simple technique with roots in an older control method known as CMAC ",[389,906,907],{},[392,908,422],{"href":421},", and one that most students meet briefly before moving on to deep learning. Looking at where tile coding has actually been put to work suggests it deserves a second look, less as a way to make machines act and more as a way to help them anticipate.",[375,911,913],{"id":912},"overlapping-grids","Overlapping Grids",[382,915,916],{},"Imagine laying a sheet of graph paper over a map. Any location now falls in one square, and everything learned about that square applies to the whole of it. That is a coarse way to describe a place, since two points just either side of a line end up in different squares and share nothing.",[382,918,919,920,578],{},"Now add a second sheet, shifted a little, and a third shifted a little more. A location falls in one square on every sheet. Two nearby locations share most of their squares, two distant ones share none, and the overlap between sheets lets the system tell apart points much finer than any single square. That is tile coding. Each sheet is a tiling, each square a tile, and the set of tiles a reading lands in becomes its description ",[389,921,922],{},[392,923,436],{"href":435},[382,925,926,927,578],{},"The same trick works for any measurement, not just position on a map. A motor's current, a joint's angle or a room's temperature can all be described this way. Richard Sutton's reference software for the technique, called tiles3, is short enough to fit on a single page ",[389,928,929],{},[392,930,422],{"href":421},[382,932,933,934,938],{},"From an engineering point of view, the appeal is predictability. Every reading lights up the same small number of tiles, so each learning step does the same amount of work, and the software caps how much memory the tiles can use ",[389,935,936],{},[392,937,422],{"href":421},". Learning about one situation also leaves unrelated situations untouched, which matters a great deal for a system that has to keep learning while it runs.",[375,940,942],{"id":941},"asking-questions-about-the-future","Asking Questions About the Future",[382,944,945,946,950],{},"The second idea is about what to predict. Reinforcement learning usually estimates how much reward lies ahead. The nexting research kept that machinery but pointed it at ordinary sensor readings instead, so the robot could learn how bright its surroundings were about to get, or how hard a motor would be working over the next few seconds ",[389,947,948],{},[392,949,395],{"href":394},". A single setting controls how far ahead each question looks, from a fraction of a second to several seconds.",[382,952,953],{},"Because every question reads from the same tiles, asking one more costs very little. There is no new model to build, only one more small set of numbers to keep up to date.",[363,955,404,957,404,961,404,967],{"style":956},"width: 100%; margin: 20px 0;",[406,958],{"src":959,"alt":960,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1678225867994-e7a5b071ebfd?w=1200&auto=format&fit=crop","A small hobby robot with exposed wiring and an ultrasonic sensor",[962,963,966],"h3",{"style":964,"id":965},"margin: 1rem 0 0.5rem 0;","thousands-of-forecasts-on-a-laptop","Thousands of Forecasts on a Laptop",[382,968,970,971,975],{"style":969},"margin: 0;","Every wire on a hobby robot like this one carries a signal that could be forecast, battery level, wheel effort, the distance reported by the sensor up front. The research robot in the nexting study had several dozen such signals. Its learning system made more than two thousand forecasts about them at once, from an ordinary laptop, and still finished each round of updates comfortably inside the robot's tenth-of-a-second rhythm ",[389,972,973],{},[392,974,395],{"href":394},". Most of the learning was done within about half an hour of driving around its pen.",[375,977,979],{"id":978},"what-the-robot-learned","What the Robot Learned",[382,981,982,983,987],{},"Speed alone would not make the result interesting. The researchers also compared the robot's forecasts, learned on the fly, against the best forecasts that could be worked out afterward with all the data in hand, and the live versions held up well ",[389,984,985],{},[392,986,395],{"href":394},". A single set of tiles, built once, was enough to forecast light, motion and motor signals across every time horizon tested. Nobody had to design separate inputs for each kind of signal.",[382,989,990,991,995],{},"One small detail in the paper hints at a practical use. The robot's motors had overheating protection and would stop to cool down every so often ",[389,992,993],{},[392,994,395],{"href":394},". In the experiment that was just a nuisance. For anyone responsible for motors, pumps or fans, a forecast of the next few seconds of effort and heat, learned from nothing but the machine's own sensors, starts to look like an early warning.",[363,997,404,998,404,1002,404,1006],{"style":956},[406,999],{"src":1000,"alt":1001,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1581090464777-f3220bbe1b8b?w=1200&h=700&auto=format&fit=crop","A robotic prosthetic hand raised against a teal background",[962,1003,1005],{"style":964,"id":1004},"anticipating-the-next-joint","Anticipating the Next Joint",[382,1007,1008,1009,578],{"style":969},"A powered hand like this one has more joints than its wearer has muscle signals to drive them. Many prosthetic arms controlled by muscle signals solve this with a switch that steps control from elbow to wrist to hand in a fixed order, an approach widely regarded as slow and difficult to use ",[389,1010,1011],{},[392,1012,446],{"href":445},[375,1014,1016],{"id":1015},"predictions-that-help-a-person","Predictions That Help a Person",[382,1018,1019,1020,578],{},"Patrick Pilarski and colleagues at the University of Alberta asked whether the same kind of forecasting could reorder that list as a person used the arm, offering first whichever joint the person was most likely to want next. Their system tile coded the arm's position and other signals, learned its forecasts in real time on a laptop, and on a standard skills test beat even the best fixed order, one identified afterward from the same session, cutting the expected wait between joints by more than 14 percent ",[389,1021,1022],{},[392,1023,446],{"href":445},[382,1025,1026,1027,1031,1032,1036],{},"Later work moved closer to everyday use. Real-time prediction learning of this kind was demonstrated with an amputee using a training robot arm ",[389,1028,1029],{},[392,1030,468],{"href":467},", and a follow-up case series found that adaptive switching reduced both how often and how long people spent switching, for an amputee and for able-bodied participants ",[389,1033,1034],{},[392,1035,577],{"href":576},". These are small studies, and clinicians are far better placed to judge what they mean for people who wear prostheses every day. From a machine learning perspective, the notable part is that the system learned from each person's own signals while the arm was in use, with no separate training phase beforehand.",[375,1038,1040],{"id":1039},"what-this-suggests-for-builders","What This Suggests for Builders",[382,1042,1043,1044,578],{},"For anyone building software for physical equipment, this work points to a pattern that seems underused. A handful of sensor readings, overlapping grids, and a set of simple forecasting questions can give a machine a running sense of what is about to happen, on hardware no bigger than a laptop and at a cost that is known in advance. The tiles3 manual covers most of the practical setup, such as choosing how wide each tile should be and which measurements to combine ",[389,1045,1046],{},[392,1047,422],{"href":421},[382,1049,1050,1051,1055,1056,1060],{},"The catch is that the number of tiles explodes as more measurements are combined. The nexting robot, for instance, tiled most of its sensors on their own or in small pairs rather than all together ",[389,1052,1053],{},[392,1054,395],{"href":394},". Deciding which measurements belong together is where knowledge of the equipment enters the design. For richer, more complex data, one research team working on laser welding first used a neural network to boil the raw data down to a few key features, then handed those to the same kind of lightweight forecasting ",[389,1057,1058],{},[392,1059,597],{"href":596},". Whether that pairing, a learned summary in front of a cheap and predictable forecaster, becomes a common recipe for small devices is still an open question. It looks like a reasonable bet for machines that need to keep learning from their own sensors without ever stopping to retrain.",[363,1062,367,1064,367,1066],{"className":1063},[729,730],[375,1065,733],{"id":729},[735,1067,404,1069,404,1080,404,1091,404,1102,404,1113,404,1124,404,1135,367],{"className":1068},[738,739,740,741],[743,1070,1071,1072,1075,1076],{"id":745},"J. Modayil et al., \"Multi-timescale Nexting in a Reinforcement Learning Robot,\" in ",[748,1073,1074],{},"From Animals to Animats 12 (SAB 2012), Lecture Notes in Computer Science",", vol. 7426, Springer, 2012, pp. 299–309. DOI: ",[392,1077,759],{"href":1078,"target":755,"className":1079},"https:\u002F\u002Fdoi.org\u002F10.1007\u002F978-3-642-33093-3_30",[757,758],[743,1081,1082,1083,1086,1087],{"id":762},"R. S. Sutton, \"Tile Coding Software, Reference Manual, Version 3.0,\" ",[748,1084,1085],{},"Reinforcement Learning and Artificial Intelligence Lab, University of Alberta",", accessed Nov. 6, 2026, ",[392,1088,759],{"href":1089,"target":755,"className":1090},"http:\u002F\u002Fincompleteideas.net\u002Ftiles\u002Ftiles3.html",[757,758],[743,1092,1093,1094,1097,1098],{"id":774},"R. S. Sutton and A. G. Barto, ",[748,1095,1096],{},"Reinforcement Learning: An Introduction",", 2nd ed., Cambridge, MA, USA, MIT Press, 2018, sec. 9.5.4, p. 217. ",[392,1099,759],{"href":1100,"target":755,"className":1101},"http:\u002F\u002Fincompleteideas.net\u002Fbook\u002FRLbook2018.pdf#page=239",[757,758],[743,1103,1104,1105,1108,1109],{"id":786},"P. M. Pilarski et al., \"Dynamic switching and real-time machine learning for improved human control of assistive biomedical robots,\" in ",[748,1106,1107],{},"Proc. 4th IEEE RAS\u002FEMBS International Conference on Biomedical Robotics and Biomechatronics (BioRob'12)",", Rome, Italy, 2012, pp. 296–302. DOI: ",[392,1110,759],{"href":1111,"target":755,"className":1112},"https:\u002F\u002Fdoi.org\u002F10.1109\u002FBioRob.2012.6290309",[757,758],[743,1114,1115,1116,1119,1120],{"id":796},"P. M. Pilarski et al., \"Adaptive artificial limbs: a real-time approach to prediction and anticipation,\" ",[748,1117,1118],{},"IEEE Robotics & Automation Magazine",", vol. 20, no. 1, pp. 53–64, 2013. DOI: ",[392,1121,759],{"href":1122,"target":755,"className":1123},"https:\u002F\u002Fdoi.org\u002F10.1109\u002FMRA.2012.2229948",[757,758],[743,1125,1126,1127,1130,1131],{"id":808},"A. L. Edwards et al., \"Application of real-time machine learning to myoelectric prosthesis control: A case series in adaptive switching,\" ",[748,1128,1129],{},"Prosthetics and Orthotics International",", vol. 40, no. 5, pp. 573–581, 2016. DOI: ",[392,1132,759],{"href":1133,"target":755,"className":1134},"https:\u002F\u002Fdoi.org\u002F10.1177\u002F0309364615605373",[757,758],[743,1136,1137,1138,1141,1142],{"id":820},"J. Günther et al., \"Intelligent laser welding through representation, prediction, and control learning: An architecture with deep neural networks and reinforcement learning,\" ",[748,1139,1140],{},"Mechatronics",", vol. 34, pp. 1–11, 2016. DOI: ",[392,1143,759],{"href":1144,"target":755,"className":1145},"https:\u002F\u002Fdoi.org\u002F10.1016\u002Fj.mechatronics.2015.09.004",[757,758],{"title":839,"searchDepth":840,"depth":840,"links":1147},[1148,1149,1150,1154,1157,1158,1159],{"id":886,"depth":840,"text":887},{"id":912,"depth":840,"text":913},{"id":941,"depth":840,"text":942,"children":1151},[1152],{"id":965,"depth":1153,"text":966},3,{"id":978,"depth":840,"text":979,"children":1155},[1156],{"id":1004,"depth":1153,"text":1005},{"id":1015,"depth":840,"text":1016},{"id":1039,"depth":840,"text":1040},{"id":729,"depth":840,"text":733},"2026-11-06","Tile coding is usually taught as a classroom exercise in reinforcement learning. Its most practical use may be prediction, helping a robot or a prosthetic arm learn to anticipate what happens next while it runs, on modest hardware and with no separate training phase.",{"src":892},{"authors":1164,"badge":1167,"source":1169},[1165],{"avatar":1166,"name":860,"to":861},{"src":859},{"label":1168},"Reinforcement Learning",{"name":865,"url":866},{"title":330,"description":1161},"aBKbRiZDaa0itNJxwsVAVsbUsxF-1fiWgFR524wtojw",{"id":1173,"title":78,"body":1174,"date":1515,"description":1516,"extension":852,"image":1517,"meta":1518,"navigation":867,"path":79,"seo":1525,"stem":80,"__hash__":1526,"_path":79},"insights\u002Fnews\u002Finsights\u002Fagentic-coding-engineer-skills.md",{"type":360,"value":1175,"toc":1502},[1176,1188,1194,1197,1214,1217,1226,1230,1238,1243,1251,1254,1272,1276,1284,1292,1295,1299,1305,1318,1326,1336,1346,1350,1363,1371,1379,1383,1386,1389,1394],[363,1177,367,1179,367,1183],{"className":1178},[366],[369,1180,78],{"className":1181,"id":1182},[372],"when-code-gets-cheap-judgment-gets-expensive",[375,1184,1187],{"className":1185,"id":1186},[378],"agentic-coding-speeds-up-the-typing-the-evidence-suggests-the-scarce-work-moves-to-specifying-checking-and-maintaining","Agentic Coding Speeds Up the Typing. The Evidence Suggests the Scarce Work Moves to Specifying, Checking and Maintaining.",[363,1189,404,1190],{"style":403},[406,1191],{"src":1192,"alt":1193,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1515879218367-8466d910aaa4?w=1200&auto=format&fit=crop","Close view of Python source code on a dark editor screen",[382,1195,1196],{},"Ask a room of engineers how much faster AI coding tools make them and the answers tend to come back as multipliers, sometimes two, sometimes ten for the right kind of task. From the keyboard those numbers feel accurate. A function that used to take twenty minutes of typing now appears in a few seconds, and that change is hard to miss.",[382,1198,1199,1200,1204,1205,1209,1210,578],{},"The controlled evidence is more uneven. In an early lab experiment, developers given an AI pair programmer, a tool that suggests code as the person types, finished a small web server task about 56 percent faster than a control group ",[389,1201,1202],{},[392,1203,395],{"href":394},". Field trials at three large companies, with several thousand developers doing their ordinary work, found a smaller and noisier gain in completed tasks, and the less experienced developers benefited most ",[389,1206,1207],{},[392,1208,422],{"href":421},". A randomized trial with experienced open-source maintainers, working in large repositories they knew well, pointed the other way. With AI tools allowed, their tasks took about 19 percent longer, even though the same developers believed afterwards that the tools had sped them up ",[389,1211,1212],{},[392,1213,436],{"href":435},[382,1215,1216],{},"One way to reconcile those results is to stop treating productivity as a single dial. Writing code is one station in a longer line. Before it comes deciding what to build and describing it precisely enough for someone, or something, else to build it. After it comes checking what came back and living with it for years. AI tools have made the writing station far cheaper. So far they have done much less for the others. When one station on an assembly line speeds up, the line does not run ten times faster. The queue moves to the next slowest station. For software, the evidence suggests that station is human judgment, and the skills gaining value are the ones that work there.",[382,1218,1219,1220,1222,1223,1225],{},"That framing picks up where The Meta-Engineer, an earlier Thinkata series, left off. ",[392,1221,170],{"href":171}," looked at how professional developers plan, delegate and validate agent work, and ",[392,1224,174],{"href":175}," asked which parts of engineering stay human as agents improve. This piece comes at the same shift from the measurement side, asking what the productivity numbers actually capture and what they leave out.",[375,1227,1229],{"id":1228},"from-autocomplete-to-delegation","From Autocomplete to Delegation",[382,1231,1232,1233,1237],{},"The tools have changed shape, and that matters for where the queue forms. Early assistants completed a line or a block. Agentic coding tools, systems that can read a repository, edit several files, run the tests and retry on failure without a person typing each step, take a task description and return a finished change. One measurement effort tracks the length of task an AI system can complete on its own, expressed as how long the same task takes a skilled human. That horizon has been doubling roughly every seven months since 2019 ",[389,1234,1235],{},[392,1236,446],{"href":445},". As it grows, more of the human's day shifts from typing alongside the tool to handing it a job and judging the result.",[382,1239,1240,1242],{},[392,1241,170],{"href":171}," described two ways developers work with agents, acceleration on familiar tasks where the goal is speed, and exploration of unfamiliar territory where the goal is learning. That split matters for the productivity numbers. Large speedups live in acceleration, because the person already holds the answer and only needs it typed. Exploration depends on being able to tell which of several plausible options is actually right, and that check runs at human speed.",[382,1244,1245,1246,1250],{},"One study logged what programmers spent their time on during assisted sessions. Verifying suggestions took more of the session than writing new functionality did, and checking plus editing the tool's output together took up a large share of the total ",[389,1247,1248],{},[392,1249,468],{"href":467},". From a systems perspective that is close to what a pipeline model predicts. Generation got cheap, but every generated line still passes through a reviewer, and the reviewer still runs at human speed.",[382,1252,1253],{},"This also offers one reading of the open-source slowdown. A small, self-contained lab task is nearly pure acceleration, because the spec is clear and the context fits in a paragraph. A mature repository is full of unwritten conventions, old decisions and edge cases that live in the maintainers' heads. In that setting the time saved on typing can go straight back into explaining context to the tool and checking whether it was respected.",[363,1255,404,1256,404,1260,404,1264],{"style":956},[406,1257],{"src":1258,"alt":1259,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1581092160562-40aa08e78837?w=1200&auto=format&fit=crop","Overhead view of an engineer marking up technical drawings with a pen",[962,1261,1263],{"style":964,"id":1262},"the-pen-does-more-of-the-work","The Pen Does More of the Work",[382,1265,1266,1267,1271],{"style":969},"A stack of technical drawings becomes something buildable only after someone has gone over each sheet, circling the dimension that does not match and the hole that will not line up. Agent-written pull requests seem to follow the same path. Across several hundred pull requests produced with an agentic coding tool in open-source projects, most were eventually merged, yet close to half of the merged ones needed human revisions first, most often for bugs, documentation and project-specific conventions ",[389,1268,1269],{},[392,1270,577],{"href":576},". The agent drafted the change. Someone who knew the project still had to go over it.",[375,1273,1275],{"id":1274},"the-bill-arrives-later","The Bill Arrives Later",[382,1277,1278,1279,1283],{},"What happens when review does not keep pace? A difference-in-differences study, a method that compares projects adopting a tool with similar projects that did not, before and after adoption, followed open-source projects that started using an AI-first code editor ",[389,1280,1281],{},[392,1282,597],{"href":596},". Development velocity jumped after adoption and then faded. Static analysis warnings, automated flags for likely bugs and risky patterns, rose and stayed high, and so did measures of code complexity. The authors' modeling points to that accumulated complexity as a major factor in the later slowdown. In effect, the early speed was borrowed against future maintenance.",[382,1285,1286,1287,1291],{},"Security shows a similar pattern with a human twist. In a controlled experiment, participants with an AI assistant wrote less secure code than those without one, yet were more likely to believe their code was secure ",[389,1288,1289],{},[392,1290,615],{"href":614},". Participants who trusted the assistant less, and who rephrased and adjusted their prompts more actively, introduced fewer vulnerabilities. What separated them looks like calibrated distrust, a sense of when output that looks finished has not really been checked.",[382,1293,1294],{},"Both findings sit on the same gap the open-source trial exposed, the distance between how productive AI-assisted work feels and what it delivers once someone has to own it. That gap is plausibly where engineering skill now earns most of its value.",[375,1296,1298],{"id":1297},"skills-gaining-weight","Skills Gaining Weight",[382,1300,1301,1302,1304],{},"If the scarce work has moved toward specification and verification, the skills that matter are likely to move with it. None of them is new, and ",[392,1303,170],{"href":171}," found experienced developers already working this way, planning before they delegate and validating everything that comes back. What seems to be changing is their share of the job, and how early in a career they are needed.",[363,1306,404,1307,404,1311,404,1315],{"style":956},[406,1308],{"src":1309,"alt":1310,"style":410},"https:\u002F\u002Fimages.unsplash.com\u002Fphoto-1434030216411-0b793f4b4173?w=1200&auto=format&fit=crop","A hand writing notes with a pen on paper beside a coffee cup",[962,1312,1314],{"style":964,"id":1313},"a-task-written-down-is-a-task-that-can-be-checked","A Task Written Down Is a Task That Can Be Checked",[382,1316,1317],{"style":969},"A pen on a blank page forces choices that a quick request in conversation can skip. Specification, describing a change precisely enough that someone without the surrounding context could carry it out, has long been a senior engineering skill. With agents it becomes an everyday one. The quality of the task description puts a ceiling on the quality of the change, and a spec that names its acceptance tests also hands the reviewer something concrete to check against.",[382,1319,1320,1321,1325],{},"Decomposition sits right beside it. Agents do well on tasks with clear boundaries, and the length of task they can handle, while growing, is still finite ",[389,1322,1323],{},[392,1324,446],{"href":445},". Splitting a feature into pieces that each fit inside that horizon, and that can each be verified on their own, is a design skill closer to architecture than to typing.",[382,1327,1328,1329,1335],{},"A less obvious skill is shaping the environment the agent works in. Work on SWE-agent, a research system that lets a language model fix real bugs in real repositories, found that the design of the agent-computer interface, the set of commands the agent uses to search, view and edit files and to run tests, had a large effect on how often it succeeded ",[389,1330,1331],{},[392,1332,1334],{"href":1333},"#source-9","[9]",". The same underlying model did markedly better with tools built around its limitations. For a working team, the equivalent is the test suite, the lint rules, the written project instructions and the fast feedback loops that let an agent catch its own mistakes before a person has to. Engineers who build that scaffolding raise the ceiling on every task anyone delegates afterwards.",[382,1337,1338,1339,1341,1342,578],{},"Then there is review, which now covers a larger share of the work and asks something slightly different of the reviewer. An agent does not get tired or cut corners on purpose, but it also has no memory of which shortcuts the team agreed to avoid. The reviewer has to hold the architecture in mind and ask whether a change that passes its tests also fits the system. ",[392,1340,174],{"href":175}," argued that this kind of taste lives in the fit between code and human intentions, which is part of why it resists automation. The complexity findings above suggest it is also easy to lose under delivery pressure ",[389,1343,1344],{},[392,1345,597],{"href":596},[375,1347,1349],{"id":1348},"where-judgment-comes-from","Where Judgment Comes From",[382,1351,1352,1353,1355,1356,1362],{},"A harder question follows about how that judgment gets built in the first place. ",[392,1354,174],{"href":175}," raised the practitioner worry about junior engineers shipping code nobody on the team understands, and asked how judgment develops if the routine work that used to train it goes away. Controlled evidence on that question has started to appear. A randomized experiment on developers learning an unfamiliar programming library found that those working with AI assistance scored worse afterwards on conceptual understanding, code reading and debugging, without a significant gain in speed on average ",[389,1357,1358],{},[392,1359,1361],{"href":1360},"#source-10","[10]",". Participants who handed the coding over entirely saw some speedup but learned the least.",[382,1364,1365,1366,1370],{},"The same study found that the cost was not spread evenly. Several interaction patterns preserved learning, including asking only conceptual questions, and generating code and then asking the assistant to explain it ",[389,1367,1368],{},[392,1369,1361],{"href":1360},". What those patterns share is that the person kept doing the thinking and used the tool to check or extend it.",[382,1372,1373,1374,1378],{},"Set that beside the field trials, where less experienced developers gained the most output from AI tools ",[389,1375,1376],{},[392,1377,422],{"href":421},", and a tension appears. The engineers with the most to gain in short-term output may also be the ones whose long-term skill formation is most exposed. A team pushing for multiplied output from its newest engineers could find, a few years on, that it has fewer people able to do the review and architecture work the agents depend on.",[375,1380,1382],{"id":1381},"looking-ahead","Looking Ahead",[382,1384,1385],{},"Across these studies the multiplier looks real at the station where code is typed, and much smaller, sometimes negative, once the work includes understanding a large system, verifying the result and maintaining it afterwards. That suggests a practical way to read any 10x claim, which is to ask which station was measured and over what time span.",[382,1387,1388],{},"For individual engineers, the evidence points toward skills that resemble what strong senior engineers already practice, clear written specs, careful decomposition, investment in tests and tooling, skeptical review and a feel for architecture, now exercised at higher volume and much earlier in a career. For teams, it suggests measuring outcomes over a longer horizon than a sprint, since the velocity gains and the complexity costs arrive on different schedules.",[382,1390,1391,1393],{},[392,1392,174],{"href":175}," closed on the idea that engineers move to wherever meaning is still underdefined. The productivity evidence adds a practical footnote. Working at that edge seems to depend on judgment built through close contact with code, and the early results on learning-preserving interaction patterns hint that the same tools can either support that contact or erode it, depending on how they are used. Whether that holds outside a single experiment, and across a whole career, is a question the field has only begun to study.",[363,1395,367,1397,367,1399],{"className":1396},[729,730],[375,1398,733],{"id":729},[735,1400,404,1402,404,1412,404,1423,404,1433,404,1442,404,1452,404,1461,404,1470,404,1481,404,1492,367],{"className":1401},[738,739,740,741],[743,1403,1404,1405,1407,1408],{"id":745},"S. Peng et al., \"The Impact of AI on Developer Productivity: Evidence from GitHub Copilot,\" ",[748,1406,766],{},", 2023, ",[392,1409,759],{"href":1410,"target":755,"className":1411},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2302.06590",[757,758],[743,1413,1414,1415,1418,1419],{"id":762},"Z. Cui et al., \"The Effects of Generative AI on High-Skilled Work: Evidence from Three Field Experiments with Software Developers,\" ",[748,1416,1417],{},"Management Science",", Articles in Advance. DOI: ",[392,1420,759],{"href":1421,"target":755,"className":1422},"https:\u002F\u002Fdoi.org\u002F10.1287\u002Fmnsc.2025.00535",[757,758],[743,1424,1425,1426,1428,1429],{"id":774},"J. Becker et al., \"Measuring the Impact of Early-2025 AI on Experienced Open-Source Developer Productivity,\" ",[748,1427,766],{},", 2025, ",[392,1430,759],{"href":1431,"target":755,"className":1432},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2507.09089",[757,758],[743,1434,1435,1436,1428,1438],{"id":786},"T. Kwa et al., \"Measuring AI Ability to Complete Long Software Tasks,\" ",[748,1437,766],{},[392,1439,759],{"href":1440,"target":755,"className":1441},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2503.14499",[757,758],[743,1443,1444,1445,1447,1448],{"id":796},"H. Mozannar et al., \"Reading Between the Lines: Modeling User Behavior and Costs in AI-Assisted Programming,\" ",[748,1446,766],{},", 2022, ",[392,1449,759],{"href":1450,"target":755,"className":1451},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2210.14306",[757,758],[743,1453,1454,1455,1428,1457],{"id":808},"M. Watanabe et al., \"On the Use of Agentic Coding: An Empirical Study of Pull Requests on GitHub,\" ",[748,1456,766],{},[392,1458,759],{"href":1459,"target":755,"className":1460},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2509.14745",[757,758],[743,1462,1463,1464,1428,1466],{"id":820},"H. He et al., \"Speed at the Cost of Quality: How Cursor AI Increases Short-Term Velocity and Long-Term Complexity in Open-Source Projects,\" ",[748,1465,766],{},[392,1467,759],{"href":1468,"target":755,"className":1469},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2511.04427",[757,758],[743,1471,1472,1473,1476,1477],{"id":830},"N. Perry et al., \"Do Users Write More Insecure Code with AI Assistants?,\" in ",[748,1474,1475],{},"Proc. ACM SIGSAC Conference on Computer and Communications Security (CCS'23)",", 2023, pp. 2785–2799. arXiv: ",[392,1478,759],{"href":1479,"target":755,"className":1480},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2211.03622",[757,758],[743,1482,1484,1485,1487,1488],{"id":1483},"source-9","J. Yang et al., \"SWE-agent: Agent-Computer Interfaces Enable Automated Software Engineering,\" ",[748,1486,766],{},", 2024, ",[392,1489,759],{"href":1490,"target":755,"className":1491},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2405.15793",[757,758],[743,1493,1495,1496,767,1498],{"id":1494},"source-10","J. H. Shen and A. Tamkin, \"How AI Impacts Skill Formation,\" ",[748,1497,766],{},[392,1499,759],{"href":1500,"target":755,"className":1501},"https:\u002F\u002Farxiv.org\u002Fabs\u002F2601.20245",[757,758],{"title":839,"searchDepth":840,"depth":840,"links":1503},[1504,1505,1508,1509,1512,1513,1514],{"id":1186,"depth":840,"text":1187},{"id":1228,"depth":840,"text":1229,"children":1506},[1507],{"id":1262,"depth":1153,"text":1263},{"id":1274,"depth":840,"text":1275},{"id":1297,"depth":840,"text":1298,"children":1510},[1511],{"id":1313,"depth":1153,"text":1314},{"id":1348,"depth":840,"text":1349},{"id":1381,"depth":840,"text":1382},{"id":729,"depth":840,"text":733},"2026-10-26","Agentic coding tools have made writing code far cheaper. The controlled evidence suggests the bottleneck moves rather than disappears, toward specifying, checking and maintaining, and the engineering skills worth building move with it.",{"src":1192},{"authors":1519,"badge":1522,"source":1524},[1520],{"avatar":1521,"name":860,"to":861},{"src":859},{"label":1523},"Agentic Coding",{"name":865,"url":866},{"title":78,"description":1516},"L3CVCVdkwDh7-6NBw3H3d1n_1Qw-G2IrTMxzPRjwBu0",1794600584680]