Bounded Regret
  • Home

Appendix: More Is Different In Other Domains

a year ago 5 min read
In Future ML Systems Will Be Qualitatively Different, I argued that we should expect ML systems to exhibit emergent capabilities. My main support for this was four historical examples of emergence in ML.
Read Now Read Later
Jacob Steinhardt
By: Jacob Steinhardt

Empirical Findings Generalize Surprisingly Far

a year ago 7 min read
Previously, I argued that emergent phenomena in machine learning mean that we can't rely on current trends to predict what the future of ML will be like. In this post, I will argue
Read Now Read Later
Jacob Steinhardt
By: Jacob Steinhardt

ML Systems Will Have Weird Failure Modes

a year ago 8 min read
Previously, I've argued that future ML systems might exhibit unfamiliar, emergent capabilities, and that thought experiments provide one approach towards predicting these capabilities and their consequences. In this post I’ll describe a
Read Now Read Later
Jacob Steinhardt
By: Jacob Steinhardt

Anchor Weights for ML

a year ago 2 min read
In the previous post, I talked about several "anchors" that we could use to think about future ML systems, including current ML systems, humans, ideal optimizers, and complex systems. In fact,
Read Now Read Later
Jacob Steinhardt
By: Jacob Steinhardt

Thought Experiments Provide a Third Anchor

a year ago 5 min read
Previously, I argued that we should expect future ML systems to often exhibit "emergent" behavior, where they acquire new capabilities that were not explicitly designed or intended, simply as a result
Read Now Read Later
Jacob Steinhardt
By: Jacob Steinhardt
Newer Posts
Page 2 of 15
Older Posts
Powered by Ghost
Bounded Regret