肖莱:AI扩展只是乘数,更巧妙的方法扩展得更好
Briefing by News4U · First published
今天,弗朗索瓦·肖莱在X上发文,并引用图灵奖得主理查德·萨顿关于“蛮力计算不足以产生智能”的观点称,数据和算力扩展是放大某种方法泛化能力的“multipliers”(乘数)。他写道,数万亿美元投入可能让“very inefficient, low-generalization approaches look adequate”(非常低效、低泛化的方法看起来够用),但这并不意味着它们就是“the ultimate end-all-be-all of AI”(人工智能的终极答案)。肖莱表示,研究者应开发能够扩展的方法,否则“simply can't compete”(根本无法竞争),同时记住“scaling is only a multiplier: a cleverer approach will scale better”(扩展只是乘数:更巧妙的方法会扩展得更好)。 [1] [2]
来源
- @fchollet (X): Data scaling and compute scaling are *multipliers* for the generalization power of an approach.
@fchollet (X) · · Excerpt
- @richardssutton (X): Brute force computation is not enough for intelligence. Clever new methods are needed. The science of AI is far from finished. IMHO
@richardssutton (X) · · Excerpt