Can you give an AI a goal of “minimally impact the world”?

From Stampy's Wiki

Alternate Phrasings

  • What about minimising the system's effects on the world?
  • What if we include a term for "don't have too big an impact"?

Canonical Answer

This is actually an active area of AI alignment research, called "Impact Measures"! It's not trivial to formalize in a way which won't predictably go wrong (entropy minimization likely leads to an AI which tries really hard to put out all the stars ASAP since they produce so much entropy, for example), but progress is being made. You can read about it on the Alignment Forum tag, or watch Rob's videos Avoiding Negative Side Effects and Avoiding Positive Side Effects

Stamps: plex


Canonical Question Info
(edits welcome)
Asked by: robertskmiles
OriginWhere was this question originally asked
Twitter (comment link)
Date: 2021/06/29


Discussion