Difference between revisions of "Recursive self-improvement"

From Lesswrongwiki
Jump to: navigation, search
Line 1: Line 1:
'''Recursive self-improvement''' is an approach to [[Artificial General Intelligence]] that allows a system to make adjustments to its own functionality resulting in improved performance. The system could then feedback on itself with each cycle reaching ever higher levels of intelligence resulting in either a hard or soft [[AI takeoff]].
+
'''Recursive self-improvement''' is an approach to [[Artificial General Intelligence]] that allows a system to make adjustments to its own functionality resulting in improved performance. The system could then feedback on itself with each cycle reaching ever higher levels of intelligence resulting in either a hard or soft [[AI takeoff]]. This is distinct from simple "self-improvement". The word "recursive" means here that the modification ''improves its ability to improve''.
  
Such a system would be bound by the limitations of its hardware. The software would be searching for cognitive algorithms that were the most optimal.
+
Recursively self-improving AI is considered to be the push behind the [[intelligence explosion]]. While any sufficiently intelligent AI will be able to improve its own intelligence, [[Seed AI]] is the category for which recursive self-improvement is built in from the start as the primary method of gaining intelligence. Alternative architectures which would have a harder time self-improving may include neural networks, or large "hand-coded" projects like [[Cyc]].
  
==Blog Posts==
+
==Self-improvement in humans==
  
 +
The human species has made an enormous amount of progress since evolving around fifty thousand years ago. This is because we can pass on knowledge and infrastructure from previous generations. This is a type of self-improvement, but it is not ''recursive''. If we never learned to modify our own brains, then we would eventually reach the point where making new discoveries required more knowledge than could be gained in a human lifetime. All human progress to date has been limited by the hardware we are born with, which is the same hardware Homo sapiens were born with fifty thousand years ago. ''True'' recursive self-improvement will come when we discover how to modify or augment our own brains in order to be more intelligent. This would lead us to more quickly being able to discover how to become even more intelligent. Assuming the rate is fast enough, this initiates a positive feedback loop.
 +
 +
==Blog posts==
 
*[http://lesswrong.com/lw/we/recursive_selfimprovement/ Recursive Self Improvement] by Eliezer Yudkowsky
 
*[http://lesswrong.com/lw/we/recursive_selfimprovement/ Recursive Self Improvement] by Eliezer Yudkowsky
 
*[http://lesswrong.com/lw/w5/cascades_cycles_insight/ Cascades, Cycles, Insight...] by Eliezer Yudkowsky
 
*[http://lesswrong.com/lw/w5/cascades_cycles_insight/ Cascades, Cycles, Insight...] by Eliezer Yudkowsky
 
*[http://lesswrong.com/lw/w6/recursion_magic/ ...Recursion, Magic] by Eliezer Yudkowsky
 
*[http://lesswrong.com/lw/w6/recursion_magic/ ...Recursion, Magic] by Eliezer Yudkowsky
  
==External Links==
+
==See also==
 +
*[[Intelligence explosion]]
 +
*[[Singularity]]
 +
*[[Seed AI]]
 +
*[[Gödel machine]]
 +
*[[AI takeoff]]
  
 +
==External links==
 
*[http://intelligence.org/upload/LOGI/seedAI.html Seed AI] description from Singularity Institute.
 
*[http://intelligence.org/upload/LOGI/seedAI.html Seed AI] description from Singularity Institute.
 
 
*[http://intelligence.org/upload/artificial-intelligence-risk.pd Risks from Artificial Intelligence] by Eliezer Yudkowsky.
 
*[http://intelligence.org/upload/artificial-intelligence-risk.pd Risks from Artificial Intelligence] by Eliezer Yudkowsky.
 
 
*[http://www.xuenay.net/Papers/DigitalAdvantages.pdf Advantages of Artificial Intelligence] by Kaj Sotala
 
*[http://www.xuenay.net/Papers/DigitalAdvantages.pdf Advantages of Artificial Intelligence] by Kaj Sotala
 
 
*[http://commonsenseatheism.com/wp-content/uploads/2011/02/Good-Speculations-Concerning-the-First-Ultraintelligent-Machine.pdf Speculations Concerning the First Ultraintelligent Machine] by I.J. Good
 
*[http://commonsenseatheism.com/wp-content/uploads/2011/02/Good-Speculations-Concerning-the-First-Ultraintelligent-Machine.pdf Speculations Concerning the First Ultraintelligent Machine] by I.J. Good
 
==See Also==
 
 
*[[Evolutionary algorithm]]
 
*[[Intelligence explosion]]
 
*[[Singularity]]
 
*[[Seed AI]]
 
*[[Gödel machine]]
 
*[[AI takeoff]]
 

Revision as of 12:54, 30 June 2012

Recursive self-improvement is an approach to Artificial General Intelligence that allows a system to make adjustments to its own functionality resulting in improved performance. The system could then feedback on itself with each cycle reaching ever higher levels of intelligence resulting in either a hard or soft AI takeoff. This is distinct from simple "self-improvement". The word "recursive" means here that the modification improves its ability to improve.

Recursively self-improving AI is considered to be the push behind the intelligence explosion. While any sufficiently intelligent AI will be able to improve its own intelligence, Seed AI is the category for which recursive self-improvement is built in from the start as the primary method of gaining intelligence. Alternative architectures which would have a harder time self-improving may include neural networks, or large "hand-coded" projects like Cyc.

Self-improvement in humans

The human species has made an enormous amount of progress since evolving around fifty thousand years ago. This is because we can pass on knowledge and infrastructure from previous generations. This is a type of self-improvement, but it is not recursive. If we never learned to modify our own brains, then we would eventually reach the point where making new discoveries required more knowledge than could be gained in a human lifetime. All human progress to date has been limited by the hardware we are born with, which is the same hardware Homo sapiens were born with fifty thousand years ago. True recursive self-improvement will come when we discover how to modify or augment our own brains in order to be more intelligent. This would lead us to more quickly being able to discover how to become even more intelligent. Assuming the rate is fast enough, this initiates a positive feedback loop.

Blog posts

See also

External links