Based on 3 visible X reactions from 6 accounts; directional sample.
Ask a question below.
Published answers will appear here.
A stray meta thought: this was my first time writing a paper without any intention of publishing it in a journal. Econ academia rewards papers that are pretty, high tech, and surprising. Such papers are valuable! I love consuming them and also try to produce. But there’s a different kind of value in applying basic and careful econ reasoning to urgent questions. I wish more people felt like they could do so without worrying about whether it’ll be published in journal X. Maybe it won’t; so what? All the worse for academia. Many of us got into economics because it’s a powerful language for understanding the world… and understanding takes many forms
@AlthoffLukas We don't even recognize that AI is an ill-posed, inverse problem; we are getting way ahead of ourselves. https://x.com/BetaTomorrow/status/2066435380623385000?s=20
@AlthoffLukas Super exciting!
We use the graphical model to give crisp minimal formalizations of a number of RSI-related considerations, isolating the key elasticities & superelasticities we want to measure: (1) bottlenecks on data, inference, experiments, & training compute; (2) acceleration in narrow capabilities without broad capabilities; (3) growth spurts due to algorithm-specific acceleration; (4) economic feedback loops.
Very happy to share the first paper from @ElasticityInst: The Economics of Recursive Self-Improvement. Two parts: (1) a graphical representation of feedback loops, to formalize a variety of RSI-related arguments, where each arrow represents responsiveness (elasticity); (2) a survey of existing evidence with a loose calibration & a “wish list” of evidence that would help us calibrate better.
The Elasticity Institute paper maps feedback loops in AI research labor.
A stray meta thought: this was my first time writing a paper without any intention of publishing it in a journal. Econ academia rewards papers that are pretty, high tech, and surprising. Such papers are valuable! I love consuming them and also try to produce. But there’s a different kind of value in applying basic and careful econ reasoning to urgent questions. I wish more people felt like they could do so without worrying about whether it’ll be published in journal X. Maybe it won’t; so what? All the worse for academia. Many of us got into economics because it’s a powerful language for understanding the world… and understanding takes many forms
@AlthoffLukas We don't even recognize that AI is an ill-posed, inverse problem; we are getting way ahead of ourselves. https://x.com/BetaTomorrow/status/2066435380623385000?s=20
We use the graphical model to give crisp minimal formalizations of a number of RSI-related considerations, isolating the key elasticities & superelasticities we want to measure: (1) bottlenecks on data, inference, experiments, & training compute; (2) acceleration in narrow capabilities without broad capabilities; (3) growth spurts due to algorithm-specific acceleration; (4) economic feedback loops.
Very happy to share the first paper from @ElasticityInst: The Economics of Recursive Self-Improvement. Two parts: (1) a graphical representation of feedback loops, to formalize a variety of RSI-related arguments, where each arrow represents responsiveness (elasticity); (2) a survey of existing evidence with a loose calibration & a “wish list” of evidence that would help us calibrate better.
@METR_Evals @ConstellOrg Shout out specifically to foundational work on RSI by @TomDavidsonX, @Ph_Aghion, @ChadJonesEcon, @bfjo
Worth paying attention to https://twitter.com/testingham/status/2076723049609801995
👀 https://twitter.com/testingham/status/2076723049609801995
Based on 3 visible X reactions from 6 accounts; directional sample.
Ask a question below.
Published answers will appear here.
@METR_Evals @ConstellOrg Shout out specifically to foundational work on RSI by @TomDavidsonX, @Ph_Aghion, @ChadJonesEcon, @bfjo