Chip Huyen: Building when it feels like there's nothing left to build - The Pragmatic Summit
Chip Huyen launched a side project that garnered 300,000 views in a week — and within a day, someone recreated it using AI. If whatever you build can be replicated instantly, what is the incentive to keep building? Huyen oscillates between excitement and despair, confronting a world where AI eliminates the moat of execution and even threatens to remove the need for imagination. She poses a raw, urgent question to the audience: why should we continue building software when AI makes duplication trivial and the competitive advantage of effort dissolves? The answer she explores is both personal and provocative.
Ключевые выводы
AI commoditizes execution: if you can describe a product, AI can replicate it, eroding traditional moats like data and removing the need for imagination.
The long tail of problems — culturally specific, human-preference driven, and involving irreversible actions in the real world — will remain fertile ground for human builders.
Human-to-human collaboration workflows (like code review) are outdated in an AI era; feedback should target instruction quality, not generated code.
The digital environment itself must evolve to be «agent-ready»: rate limits, web search, and interfaces are still designed for human patterns, not AI efficiency.
Building for joy and as personal expression (e.g., custom apps as gifts) offers a meaningful path forward, even when economic incentives blur.
Вкратце
Build not because your work is defensible or irreplaceable, but because problem-solving and creation are intrinsically rewarding — and the long tail of human-specific, culturally nuanced, irreversible problems will never be fully automated.
The «Ghibli Moment» of Software
If AI can generate any described image, it can generate any described software.
Huyen recounts launching a small side project that attracted 300,000 views in a week — only to receive an email the next day from someone who had recreated it entirely using AI. The incident crystallized her unease: whatever exists can now be replicated at near-zero marginal cost. She calls this the «Ghibli moment of software,» referencing how AI can generate images in any artistic style once that style is seen and described. The more products exist, the less imagination is needed; you simply point and say «do that.»
This dynamic collapses traditional competitive moats. Data, once considered defensible, is now merely expensive to acquire — and OpenAI competitors have demonstrated that with capital, datasets can be assembled quickly. Execution speed no longer matters when cloning is trivial. Huyen finds herself toggling between exhilaration (she can build anything) and despair (so can everyone else). The incentive structure for creation feels fundamentally broken when differentiation evaporates overnight.
«It feels great»
Audience member reveals the punchline: we build because it feels great.
“It feels great.”
The Long Tail of Problems AI Won't Solve
Outdated Workflows: The Death of Line-by-Line Code Review
Senior engineers review AI-generated code line by line; junior engineers ignore feedback they can't action.
Outdated Workflows: The Death of Line-by-Line Code Review
Huyen describes a team where a senior engineer reviews every pull request written by AI line by line, ostensibly for mentorship. The junior engineers don't read the feedback because they didn't write the code and don't know how to translate critiques into better AI prompts. The traditional code-review loop — designed for human authorship — has become theater. The real skill is now prompt engineering and AI instruction, not code syntax.
Making the World Agent-Ready
Digital infrastructure is built for humans; AI needs formats optimized for speed and machine interaction.
Why Keep Building? Three Answers
Learning, purpose, and joy remain valid reasons even when economic moats disappear.
Learning Through Problem-Solving Building sharpens your ability to identify and solve problems. The process itself — not the artifact — is the durable skill.
Sense of Purpose Creating something that solves a real problem, even if replicable, gives meaning. Contribution doesn't require exclusivity.
Intrinsic Joy Huyen now builds apps as birthday gifts for friends (e.g., a tea-tracking app). She advocates normalizing building for fun, not just economic return or competitive advantage.
The Artisan Software Future (Maybe)
Mass production made clothes cheap; wealth brought demand for custom tailoring back.
Huyen draws a parallel to the fashion industry: industrial looms democratized clothing, then disposable income revived demand for bespoke, hand-tailored garments. She's uncertain whether software will follow the same arc — will users prize «artisan apps» built by hand over AI-generated commodities? The question remains open, but she finds personal satisfaction in crafting custom tools for friends, treating software as gift and craft rather than product.
For now, AI makes her life happier by accelerating the parts of building she enjoys and eliminating drudgery. The exception: moments of existential dread when the economic logic of creation collapses. But those moments, she suggests, can be weathered by anchoring motivation in process, not defensibility.
Люди
Глоссарий
Отказ от ответственности: Это ИИ-сводка видео с YouTube, подготовленная в образовательных и справочных целях. Она не является инвестиционной, финансовой или юридической консультацией. Всегда проверяйте информацию по первоисточникам перед принятием решений. TubeReads не связан с автором контента.