能效越高电越不够用——杰文斯从不掷骰子
原创
你有没有仔细算过,训练一个大模型到底"省"了多少电?大概率没省。这帖子戳穿了一个流行幻想:杰文斯悖论不是"可能会发生",它是必然。AI能效每涨一点,省下的电立刻被更大的模型、更多的推理请求吞掉,而且越吞越凶。
我觉得这个判断是对的。过去十年,模型参数量涨了上万倍,能耗也涨了上万倍。效率确实在提高,但根本没有余量。能效进步越是靠技术突破释放,需求膨胀就越凶猛——这是一个确定性反馈回路,不是一枚需要开盒才知道结果的黑箱。
杰文斯不掷骰子,他直接押确定性。
原文:Jev does not play dice · 来源:Hacker News
版权声明
所有资源都来源于爬虫采集,如有侵权请联系我们,我们将立即删除
相关文章
- 0.05ms 往返:localhost 为什么不能直接当网站?
- Fornax's Guide to Fast Ethernet
- Possibly leaked NSA Cyberattack tool behind attacks (2017)
- Steve Wozniak on Writing Apple Integer Basic
- PayByPhone is stealing $50M/year from their users
- Grand Theft Auto 6: A Landmark Achievement That Redefines a Console Generation
- Research Acceleration: The Jagged Frontier
- Google Embraces Geothermal Technology to Vaporize Rock for Clean Power
- Microsoft's Agent Framework PyPI package compromised to steal credentials
- Nvidia's Seven Billionaires Have Lost a Third of Their Wealth
最新文章
- 0.05ms 往返:localhost 为什么不能直接当网站? 3小时前
- Fornax's Guide to Fast Ethernet 3小时前
- Possibly leaked NSA Cyberattack tool behind attacks (2017) 4小时前
- Steve Wozniak on Writing Apple Integer Basic 4小时前
- PayByPhone is stealing $50M/year from their users 4小时前
- Grand Theft Auto 6: A Landmark Achievement That Redefines a Console Generation 4小时前
- Research Acceleration: The Jagged Frontier 5小时前
- Google Embraces Geothermal Technology to Vaporize Rock for Clean Power 6小时前
- Microsoft's Agent Framework PyPI package compromised to steal credentials 7小时前
- Nvidia's Seven Billionaires Have Lost a Third of Their Wealth 7小时前
itfan123





