Diabloto96 | v1.8.0Mar 19th 2023 | (-2) | ||
Ruben Bloom | v1.7.0Oct 1st 2020 | (+227) | ||
Ruben Bloom | v1.6.0Jul 26th 2020 | C-Class on the basis of having some explanation of concept plus some elaboration in conjuction with a really strong posts list. | ||
Ruben Bloom | v1.5.0May 4th 2020 | |||
Ruben Bloom | v1.4.0Apr 17th 2020 | (+10/-10) | ||
Ruben Bloom | v1.3.0Apr 17th 2020 | (+9/-9) | ||
Ruben Bloom | v1.2.0Apr 17th 2020 | (+806/-8) | ||
Ruben Bloom | v1.1.0Apr 17th 2020 | (+559/-528) | ||
Vladimir Nesov | v0.0.3May 27th 2010 | (+12) /* See also */ | ||
Vladimir Nesov | v0.0.2May 27th 2010 | (+14) /* See also */ |
In Goodhart Taxonomy, Scott Garrabrant identifiedidentifies four kinds of Goodharting:
In Goodhart Taxonomy,Taxonomy, Scott Garrabrant identified four kinds of Goodharting:
Goodhart's Law states that when a proxy for some value becomes the target of optimization pressure, the proxy will cease to be a good proxy. ConsiderOne form of Goodhart is demonstrated by the Soviet story of a factory graded on how many shoes they produced (a good proxy for productivity) – they soon began producing a higher number of tiny shoes. Useless, but the numbers look good.
In Goodhart Taxonomy, Scott Garrabrant identified four kinds of Goodharting:
Goodhart's lawLaw states that oncewhen a certain indicator of success is made aproxy for some value becomes the target of optimization pressure, the proxy will cease to be a social or economic policy, it will losegood proxy. Consider the information content that would qualify it to play suchSoviet story of a role.
Goodhart's Law is of particular relevance to AI Alignment by blogospheroid
Goodhart's Law is of particular relevance to AI Alignment. Suppose you have something which is generally a good proxy for "the stuff that humans care about", it would be dangerous to have a powerful AI optimize for the proxy, in accordance with Goodhart's law, the proxy will breakdown.