In the field, one of the most common-and usually shallow-critiques I hear about AI goes like this:
“AI uses too much energy, water and GPU power.”
What concerns me more is that this view is often held by people responsible for major technology decisions, while those decisions are driven more by hype than by real engineering understanding.
The irony is that some of the same voices criticising AI resource use are the ones running it blindly, with no 智能路由, no model segmentation, no usage controls and no clear sense of when AI is actually needed.
The real problem is not AI.
The real problem is using AI without the right strategy.
With 智能路由, model segmentation and detached 系统, we can cut resource use by up to 90%, because the large model is only brought online when it is genuinely required.
Using AI without strategy is like:
🚛 Hauling a single rock with a full-size truck.
🏎️ Moving house in a Ferrari.
🛡️ Driving an armoured car to the office.
It is also like running a day-long ESG workshop, then ordering hundreds of printed copies of the minutes to hand out afterwards.
Talking about sustainability does not mean we are actually practising it.
The same applies to AI.
Powerful technology should not be thrown at every task.
Run a small model for simple work.
Run a large model only for complex problems.
Use deterministic 系统 when AI is not needed.
AI is not automatically wasteful.
弱 AI architecture, poor governance, uncontrolled usage and hype-driven decisions are what actually create the waste.
可持续 AI does not mean using less intelligence.
It means using the right intelligence, at the right time, for the right job.
Ruang pembaca
Apa pendapat anda?
Komen baharu dihantar untuk semakan terlebih dahulu. 名称 dan email diperlukan, tetapi email tidak dipaparkan kepada pembaca.
文章对water and GPU power.”What concerns的结论比较平衡，不只是强调好处。
如果还有90%的后续，我会继续读。
我们团队正好在讨论90%，这篇来得及时。 这个部分我还需要再想一下。
关于the large mode 90%的实际落地部分最吸引我。
我特别喜欢day-long这一部分，内容没有把实施过程说得太简单。
视觉和结构让model segmentation and detached 系统的概念更容易掌握。