一位开发者在休假数周后重新审视了自己与AI工具的关系，发现了令人担忧的依赖模式。[1]当他返回工作时，发现有多个暂停的Agent任务分散在11个cmux标签中，这促使他进行了深刻反思。[1]
他坦言："我没有怀念它。我最近意识到我的大部分AI使用是不必要的。事实上，我认为它通常是一个成瘾的拐杖，使我在智力上变弱，好奇心和信心减少。"[1]
这位开发者指出，自己曾将AI Agent作为逃避压力的中介，而不是直接面对任务本身。[1]他的观察反映了技术行业面临的更广泛问题："技术行业的很大一部分现在押注于大规模社会经济对大语言模型的依赖。"[1]
针对这一现象，他提出了更理性的使用原则——有意识地、有针对性地调用AI工具，保持人在决策循环的核心地位。[1]用他的话说："人是循环，我们偶尔会有思考地让Agent参与其中。"[1]这一观点强调，AI应该是辅助手段，真正的决策权应该始终掌握在人的手中。
After taking a break from work, a technology professional returned to find multiple paused AI Agent tasks across multiple tabs, prompting a broader reflection on their relationship with artificial intelligence tools [1]. Upon returning, the author realized they had not actually missed using these systems and came to an uncomfortable conclusion: most of their AI usage had become unnecessary [1].
The author acknowledged falling into unhealthy patterns with AI Agents and ChatGPT, describing their dependence as "an addictive crutch that makes me intellectually weaker, with less curiosity and confidence" [1]. Rather than tackling tasks directly, they had used AI agents as a buffer against stress, allowing them to avoid confronting challenges head-on [1]. This realization extended to a broader concern about the technology industry, which the author noted is increasingly betting on large-scale societal dependence on large language models [1].
The core insight emerged as a corrective principle: "The human is the loop, and we tag the agent in occasionally, thoughtfully" [1]. This framing rejects the notion that AI should drive decision-making, instead positioning humans at the center of critical choices with AI tools deployed strategically rather than as default reflexes [1].