{"slug": "do-it-by-hand-intentional-practice-in-an-ai-world", "title": "Do it by hand: Intentional practice in an AI world", "summary": "Engineer and ProductMind co-founder warns that over-reliance on AI tools like Anthropic's Claude can atrophy core skills, advocating for regular unaided practice to maintain expertise and creative judgment.", "body_md": "# Do it by hand: Intentional practice in an AI world\n\n### Even if you have a lot of experience, overuse of AI will eventually atrophy your skills. If you value that, make it a habit to practice unaided from time to time.\n\n### A simple practice\n\nEvery Sunday morning, I create by hand. In a journal, on a computer, etc. I sometimes look at all the things I wrote pre-LLM to make sure I understand them, and that I’m getting better than that benchmark. Not because I’m a luddite. And definitely not because I romanticize the past. But because I’m catching myself being too crutchful every day, especially with the launch of Anthropic’s Fable 5 model.\n\nAt ProductMind, we are believers in the rise of the HyperCreator, or what folks are calling the Product Builder; although we think it’s so much more than that. We think AI will help us reach peak performance, if it’s affordable (@tedyang even wrote a book about it). But we also believe that the creative spark inside the best AI-enhanced projects comes from unassisted and highly skilled high-performers who can meld their drive, expertise, and creativity to new tools. Who can exercise top-tier judgment.\n\n*You should not shrink* as you gain your new powers. The thing to understand about AI and automation is that **even when you understand how it works, you can lose the ability to do things yourself without occasional intentional practice.** And that loss changes the quality of what you can create.\n\nI am an engineer, so I feel like I can engage in some professional self-disparagement. I admire mathematicians way more than our tribe. The idealized image of a professional mathematician working out a problem on a blackboard always fascinates me because they often need to work out the proof in long form. Engineers however, we love automation; anything that makes our lives easier and more abstracted is fair game. The abacus comes out, yep, adopted. The slide rule comes out - adopted. Log tables came out, we adopted the heck out of them. Graphing calculators, computers… holy crap, software that programs graphic UI on computers, sign me up! We jump into everything that means less effort, and we tend to lose our mastery. Many mathematicians still do it by hand when it comes to the hard stuff.\n\n## The programmer’s paradox\n\nConsider the programmer who truly understands computer architecture—someone who knows x86 assembly, who could write machine code if needed, who grasps every layer from silicon to screen. Now give them C++, then JavaScript, then modern frameworks. Watch them code exclusively at high levels of abstraction for a year.\n\nThey haven’t forgotten *how* computers work. They can still explain instruction sets and memory management. But something has changed. The intimate connection between thought and execution has weakened. The instincts that come from working close to the metal have dulled.\n\n**Knowledge persisted. Capability atrophied.**\n\nNow imagine the developer who never learned the fundamentals—who started with high-level frameworks and never looked beneath. When things break in unexpected ways, when performance matters, when novel problems emerge, they’re lost in ways they can’t even articulate. The loss is “irretrievable, untraceable, unknowable” because they never possessed what they lost.\n\n## Automation can be a cognitive trap\n\nWe’re living through this same dynamic with AI across every knowledge domain. You use AI to write. To analyze. To strategize. And you should—because you can’t run around being Clark Kent when everyone else is Superman. The competitive landscape has shifted. Manual operation in an era of AI-achievable speeds isn’t noble—it’s obsolete.\n\nBut there is a personal performance trap: **automation doesn’t just save you time. It rewires how you think. **There is a lot of research to back this up, even though the research field is still early. The net of it is that if you frame it in terms of **dual-process cognition**: **System 1** is fast, intuitive, pattern-based cognition; **System 2** is slower, effortful, analytical cognition. The current generation of AI UX is designed to interrupt your System 2 thinking, causing less cognition, less skill sharpening, and more dependence. (You can check it out at the bottom of this article.)\n\nWhen AI handles the scaffolding, you stop building mental frameworks for structure. When it generates variations, you stop developing the pattern recognition that comes from iteration. When it fills gaps in your argument, you stop noticing what’s missing in your own thinking.\n\nEven worse—especially if you understand *how* to write, how to think, how to analyze—you start to trust the shortcut. Not because you’re lazy, but because the output is good enough. Sometimes better than good enough. And slowly, imperceptibly, the muscle weakens.\n\n## Defeat the cycle\n\nHere’s my rule, and I recommend it to every client, every leader, every creative professional:\n\n**Use automation every day. Practice manually every week.**\n\nLet AI enhance your drafts, refine your prose, and extend your arguments. Do the creative core, get upscaled, output, then edit by hand.\n\nBut once a week—\n\n*every*week— do regular work completely from scratch.No AI assist. No enhancement. No polish.\n\nJust you and the blank page.\n\nThis isn’t about proving something. It’s about **maintaining the capability that makes you valuable in an automated world.**\n\nBecause here’s what manual practice preserves:\n\nThe struggle that generates genuine insight\n\nThe gap-awareness that identifies what’s missing\n\nThe revision instincts that come from seeing your own inadequacy\n\nThe voice that emerges only through unmediated expression\n\n## You’re preserving something important\n\nWhen you create manually, you’re not just maintaining skills. You’re preserving **the substrate of originality.**\n\nAI can enhance what you create. It can extend patterns, fill structures, and polish expression. But it can’t generate the novel combinations that come from your specific constellation of experiences wrestling with a specific problem at a specific moment.\n\nThat wrestling—that frustration of finding the right word, that recognition of conceptual gaps, that grinding iteration toward clarity—*is* the creative process. Automate it away entirely, and you’ve automated away the very thing that makes you irreplaceable.\n\nAI used to be trained on everyone’s originality, which means it’s the precious ingredient. So it’s important to preserve yours.\n\n## Finding balance\n\nThe future isn’t a choice between human capability and AI amplification. It’s maintaining both in productive tension.\n\nUse AI. Get better at prompting it, directing it, and extracting value from it. But never let it replace the foundational practice that keeps your capabilities sharp.\n\n**Once a week, turn off the machines and remember what you can do.**\n\nYour future self—the one who can still think originally when the tools change, who can still solve novel problems when the automation fails, who still has something unique to offer when everyone has access to the same AI—will thank you.\n\n*What’s your weekly manual practice? Reply and let me know what skills you’re intentionally preserving in an automated world.*", "url": "https://wpnews.pro/news/do-it-by-hand-intentional-practice-in-an-ai-world", "canonical_source": "https://substack.productmind.co/p/do-it-by-hand-intentional-practice", "published_at": "2026-07-16 17:00:42+00:00", "updated_at": "2026-07-16 17:05:01.095935+00:00", "lang": "en", "topics": ["artificial-intelligence", "ai-ethics", "ai-tools", "ai-products"], "entities": ["Anthropic", "ProductMind", "Claude", "Ted Yang"], "alternates": {"html": "https://wpnews.pro/news/do-it-by-hand-intentional-practice-in-an-ai-world", "markdown": "https://wpnews.pro/news/do-it-by-hand-intentional-practice-in-an-ai-world.md", "text": "https://wpnews.pro/news/do-it-by-hand-intentional-practice-in-an-ai-world.txt", "jsonld": "https://wpnews.pro/news/do-it-by-hand-intentional-practice-in-an-ai-world.jsonld"}}