Lee Elkin
@lelkin.bsky.social
AI, Decision Theory, Epistemology, Ethics
lelkin.com
lelkin.com
I was recently asked what I thought about the rationalists. I immediately thought we were talking about Descartes and Leibniz.
November 3, 2025 at 7:31 AM
I was recently asked what I thought about the rationalists. I immediately thought we were talking about Descartes and Leibniz.
Spotted at H&M. Take it however you want.
October 25, 2025 at 9:47 AM
Spotted at H&M. Take it however you want.
Do you still feel bad about yourself after getting a journal rejection?
October 4, 2025 at 8:44 AM
Do you still feel bad about yourself after getting a journal rejection?
Sure, let me Dutch book myself.
August 30, 2025 at 5:59 PM
Sure, let me Dutch book myself.
Proceduralists concerned with fair representation should view belief aggregation/opinion pooling as a credal cake-cutting problem.
August 29, 2025 at 2:45 PM
Proceduralists concerned with fair representation should view belief aggregation/opinion pooling as a credal cake-cutting problem.
Random thought: alignment should entail coalitional envy-freeness. But envy-freeness is likely violated in many cases since states of alignment are often Pareto dominated by the status quo.
August 26, 2025 at 4:26 PM
Random thought: alignment should entail coalitional envy-freeness. But envy-freeness is likely violated in many cases since states of alignment are often Pareto dominated by the status quo.
Maybe my most boomerish moment: back in my day, we were taught argument from analogy is bad mmmk.
August 11, 2025 at 10:52 AM
Maybe my most boomerish moment: back in my day, we were taught argument from analogy is bad mmmk.
Hot,humid summer day in Hong Kong. But what a view.
July 27, 2025 at 8:56 AM
Hot,humid summer day in Hong Kong. But what a view.
I’ll be joining the University of Hong Kong 🇭🇰 this month to work broadly on AI Welfare. Big topic, but bigger life event.
July 1, 2025 at 2:37 PM
I’ll be joining the University of Hong Kong 🇭🇰 this month to work broadly on AI Welfare. Big topic, but bigger life event.
Lots of AI debates would disappear if we brought eliminative materialism back in style.
June 1, 2025 at 11:44 PM
Lots of AI debates would disappear if we brought eliminative materialism back in style.
I asked my five-year-old nephew if he is a human. He says, “Of course, I am a HUMEAN!”
May 28, 2025 at 12:29 AM
I asked my five-year-old nephew if he is a human. He says, “Of course, I am a HUMEAN!”
Controversial take: AI MADE METAPHYSICS GREAT AGAIN
May 12, 2025 at 4:20 PM
Controversial take: AI MADE METAPHYSICS GREAT AGAIN
Has the following claim been explicitly made by P(doom) > 0 folk:
P(doom | mistreatment) >= P(doom)
'mistreatment' refers to some form of mistreatment of developed AI systems. #AISafety #AIAlignment
P(doom | mistreatment) >= P(doom)
'mistreatment' refers to some form of mistreatment of developed AI systems. #AISafety #AIAlignment
May 10, 2025 at 3:45 PM
Has the following claim been explicitly made by P(doom) > 0 folk:
P(doom | mistreatment) >= P(doom)
'mistreatment' refers to some form of mistreatment of developed AI systems. #AISafety #AIAlignment
P(doom | mistreatment) >= P(doom)
'mistreatment' refers to some form of mistreatment of developed AI systems. #AISafety #AIAlignment
New paper on algorithmic fairness in ensemble learning. philpapers.org/rec/STEDIF
We show:
(1) the linear pool of models preserves some fairness properties if the models individually have such properties;
(2) ensembles mitigate the accuracy-fairness tradeoff.
We show:
(1) the linear pool of models preserves some fairness properties if the models individually have such properties;
(2) ensembles mitigate the accuracy-fairness tradeoff.
Rush T. Stewart & Lee Elkin, Diving into Fair Pools: Algorithmic Fairness, Ensemble Forecasting, and the Wisdom of Crowds - PhilPapers
Is the pool of fair predictive algorithms fair? It depends, naturally, on both the criteria of fairness and on how we pool. We catalog the relevant facts for some of the ...
philpapers.org
March 17, 2025 at 2:13 PM
New paper on algorithmic fairness in ensemble learning. philpapers.org/rec/STEDIF
We show:
(1) the linear pool of models preserves some fairness properties if the models individually have such properties;
(2) ensembles mitigate the accuracy-fairness tradeoff.
We show:
(1) the linear pool of models preserves some fairness properties if the models individually have such properties;
(2) ensembles mitigate the accuracy-fairness tradeoff.
Bayes Business School’s slogan is “Always Learning.” Wonder if that was intentional.
February 21, 2025 at 12:01 AM
Bayes Business School’s slogan is “Always Learning.” Wonder if that was intentional.
Reposted by Lee Elkin
Reposted by Lee Elkin
New Element on opinion pooling out now and open access!
It was great fun writing this with @lelkin.bsky.social. Hopefully a useful intro to different ways we can combine, aggregate, pool probabilistic opinions, as well as a survey of the connections to questions about disagreement, testimony, etc.
It was great fun writing this with @lelkin.bsky.social. Hopefully a useful intro to different ways we can combine, aggregate, pool probabilistic opinions, as well as a survey of the connections to questions about disagreement, testimony, etc.
Opinion Pooling
Cambridge Core - Twentieth-Century Philosophy - Opinion Pooling
cup.org
February 5, 2025 at 10:24 AM
New Element on opinion pooling out now and open access!
It was great fun writing this with @lelkin.bsky.social. Hopefully a useful intro to different ways we can combine, aggregate, pool probabilistic opinions, as well as a survey of the connections to questions about disagreement, testimony, etc.
It was great fun writing this with @lelkin.bsky.social. Hopefully a useful intro to different ways we can combine, aggregate, pool probabilistic opinions, as well as a survey of the connections to questions about disagreement, testimony, etc.
It’s finally here! www.cambridge.org/core/element...
Opinion Pooling
Cambridge Core - Philosophy: General Interest - Opinion Pooling
www.cambridge.org
January 31, 2025 at 5:08 PM
It’s finally here! www.cambridge.org/core/element...
Okay, where do stand now on the accuracy of imprecise credences? Still living with the impossibility w.r.t strictly proper scoring rules? @wiglet1981.bsky.social @kevinzollman.com @jeffhelzner.bsky.social @jessicahullman.bsky.social
January 29, 2025 at 4:35 PM
Okay, where do stand now on the accuracy of imprecise credences? Still living with the impossibility w.r.t strictly proper scoring rules? @wiglet1981.bsky.social @kevinzollman.com @jeffhelzner.bsky.social @jessicahullman.bsky.social
Looking for papers arguing against deferring to experts.
January 27, 2025 at 9:33 AM
Looking for papers arguing against deferring to experts.
I'm surprised by the surprise toward DeepSeek-R1. Anyone familiar with work on collective intelligence in the social sciences and division of cognitive labor in philosophy of science can tell you that Mixture of Experts is going to do very well for you.
January 23, 2025 at 4:12 PM
I'm surprised by the surprise toward DeepSeek-R1. Anyone familiar with work on collective intelligence in the social sciences and division of cognitive labor in philosophy of science can tell you that Mixture of Experts is going to do very well for you.
Experience is starting to tell me that imprecise probability/credence is the QAnon of epistemology and decision theory.
December 11, 2024 at 9:54 AM
Experience is starting to tell me that imprecise probability/credence is the QAnon of epistemology and decision theory.
Making the job market season coincide with the holiday season seems like a cruel joke.
December 10, 2024 at 3:41 PM
Making the job market season coincide with the holiday season seems like a cruel joke.