TE
科技回声
首页24小时热榜最新最佳问答展示工作
GitHubTwitter
首页

科技回声

基于 Next.js 构建的科技新闻平台,提供全球科技新闻和讨论内容。

GitHubTwitter

首页

首页最新最佳问答展示工作

资源链接

HackerNews API原版 HackerNewsNext.js

© 2025 科技回声. 版权所有。

The climate cost of the AI revolution

2 点作者 harryvederci大约 2 年前

1 comment

rvz大约 2 年前
Yet there exists no efficient alternative to training, fine-tuning and even just with basic inference on so-called &#x27;AI&#x27; models and opaque black-boxes for deep learning which just contributes more wanton garbage outputs and generative BS that is easy to break, trick and confuse. All for it to burn up the planet further by once again retraining, rebuilding or fine-tuning the whole model over again wasting large amounts of CO2 for what? More broken AI models.<p>This is why companies have to keep announcing things like <i>&#x27;reducing our carbon footprint&#x27;</i> nonsense and at the same time build more data centers continuing to burn up the planet.<p>Such greenwashing virtue signalling responses to appease the gullible. It is no better than Bitcoin mining. At least with cryptocurrencies like Ethereum, PoS has reduced there energy efficiency by 99.99%, meaning there is a efficient alternative to PoW.<p>NO such energy efficient alternative to deep learning training, fine-tuning exists after decades of AI hype.
评论 #35174706 未加载