Difference between revisions of "Changing selection pressures argument"
(Created page with "The '''changing selection pressures argument''' is an argument for continuous AI takeoff. It states that during most of hominid evolution, evolution was not optimizing for...") |
|||
| (3 intermediate revisions by the same user not shown) | |||
| Line 1: | Line 1: | ||
| − | The '''changing selection pressures argument''' is an argument for [[continuous AI takeoff]]. It states that during most of hominid evolution, evolution was not optimizing for | + | The '''changing selection pressures argument''' is an argument for [[continuous AI takeoff]]. It states that during most of hominid evolution, evolution was not optimizing for scientific ability or cultural accumulation (or some other factor that explains human dominance on Earth), but that at some point evolution did start optimizing for these things (i.e. there was a change in selection pressure towards scientific ability or cultural accumulation), and that this explains why humans are so much better than chimps at doing science despite only splitting off from chimps several million years ago. This becomes an argument for continuous AI takeoff because (the argument goes), during AI development there will not be a similar change in selection pressures, i.e. AI developers will always be optimizing for something like the scientific ability of AI systems. |
==History== | ==History== | ||
Latest revision as of 04:17, 30 August 2021
The changing selection pressures argument is an argument for continuous AI takeoff. It states that during most of hominid evolution, evolution was not optimizing for scientific ability or cultural accumulation (or some other factor that explains human dominance on Earth), but that at some point evolution did start optimizing for these things (i.e. there was a change in selection pressure towards scientific ability or cultural accumulation), and that this explains why humans are so much better than chimps at doing science despite only splitting off from chimps several million years ago. This becomes an argument for continuous AI takeoff because (the argument goes), during AI development there will not be a similar change in selection pressures, i.e. AI developers will always be optimizing for something like the scientific ability of AI systems.
History
The argument was first made by Paul Christiano in a 2018 blog post.[1]
The argument was named by Richard Ngo in 2020.[2]
