AI 工具的效果取决于你的判断力—

AI 工具的效果取决于你的判断力——这正是关键所在
AI tools are only as good as your judgment

原始链接: https://theaileverageweekly.com/posts/your-ai-tools-are-only-as-good-as-your-judgment-and-that-s-the-point.html

工程界对 AI 依赖的焦虑常被误诊为懒惰，而真正的问题在于“推卸责任”。当工程师不加审视地接受 AI 生成的代码时，就会积累必然导致生产故障的“技术债”。解决之道并非减少 AI 的使用，而是以“对抗性”方式使用它。不要将 AI 视为先知，而应将其视为一名需要严格审查其工作的过度自信的初级开发者。通过主动引导 AI 反驳其自身的解决方案——质疑边界条件、安全漏洞和潜在假设，工程师便能从被动的消费者转变为主动的架构师。归根结底，未来工程师最宝贵的技能并非精巧的提示词（prompting），而是进行深度、审慎判断的能力。通过采用“生成、审问、修订”的工作流，你不仅不会让专业技能退化，反而能使其磨练得更加锋利。削弱工程判断力的并非 AI 本身，而是被动地使用它。如何在两者之间做出选择，完全掌握在你手中。

这篇 Hacker News 的讨论聚焦于如何有效利用人工智能工具，主要探讨了使用 AI 来提升个人技能与使用 AI 来规避努力之间的矛盾。一位用户提倡“副驾驶”模式：作者不应让 AI 起草文本，而应将其用作编辑，由 AI 提供批判性反馈并指出薄弱环节。通过反复根据这些反馈进行修改，用户可以磨练自己的技艺，将 AI 视为个人成长的工具。在对自己的作品感到满意后，他们建议再使用模型进行最终润色，以捕捉可能遗漏的改进之处。然而，评论区对此持怀疑态度。多位用户指出，这篇文章本身似乎就是由大语言模型撰写的，这削弱了其论点本身的可信度。另一些人则反驳了“提升判断力”这一前提，指出在职场环境中，效率要求往往迫使用户为了速度而非质量而依赖 AI。归根结底，这场辩论对比了将 AI 视为良师益友的理想化愿景，与企业要求优先考虑产出而非匠心精神的现实情况。

原文

There's a quiet anxiety spreading through engineering teams right now: Am I becoming dependent on AI? Is my judgment atrophying?

My take: that's the wrong question. The right one is whether you're using AI in a way that sharpens your judgment or replaces it. Those are genuinely different modes of use, and most engineers drift into the second one without noticing.

The Dependency Trap Is Real — But Misdiagnosed

The common critique is that AI tools make engineers lazy. I don't think that's it. The problem isn't laziness — it's abdication. When you accept a generated solution without interrogating it, you're not saving time. You're deferring a debt that compounds interest.

The engineer who copy-pastes an AI-generated auth middleware without reading it isn't moving faster. They're moving faster now and slower — much slower — when that middleware silently fails in a production edge case at 2am.

But here's where I'll stake the actual opinion: the solution isn't to use AI less. It's to use it adversarially.

Adversarial Use, Concretely

What does adversarial use look like? You treat the AI output as a first draft from a smart-but-overconfident junior engineer. You don't reject it reflexively and you don't accept it wholesale. You interrogate it.

Here's a prompt pattern I've baked into my actual workflow:

Here's the solution you proposed: [paste output]

Now argue against it. What are the edge cases this doesn't handle? 
What assumptions did you make that might not hold in a production system? 
What would you change if you knew this code would be read by a senior 
engineer in a security audit?

Run that after any non-trivial AI-generated solution. What comes back is almost always useful — missed error states, implicit assumptions about input shape, security surface area that got glossed over. And critically: you are now thinking alongside the tool, not just consuming its output.

That loop — generate, interrogate, revise — is where judgment lives. It's where you stay sharp.

The Real Skill Isn't Prompting

The engineers who will be dangerous with AI five years from now aren't the ones who have memorized the best prompt templates. They're the ones who can look at any generated output — code, architecture diagram, spec, test suite — and immediately ask the right skeptical questions.

That skill is built by practice. Adversarial prompting is one way to practice it deliberately rather than accidentally.

AI doesn't erode engineering judgment. Passive AI use does. The distinction matters, and it's entirely within your control.

I break down one concrete AI workflow like this every week in The AI Leverage Weekly — practical, no fluff, free. Subscribe: https://theaileverageweekly.beehiiv.com/subscribe?utm_source=devto&utm_medium=article&utm_campaign=medium_w3