text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# Optimizing for Stories (vs Optimizing Reality)
_Epistemic status: highly confident in the basic distinct, though it's not at all profound. Further details, models, and advice are somewhat speculative despite being drawn from varying amounts of observation. Essay a little rushed since otherwise I’d be unlikely to pu... | https://www.lesswrong.com/posts/fBgzL2gbavQaswsJc/optimizing-for-stories-vs-optimizing-reality |
# Sequence introduction: non-agent and multiagent models of mind
A typical paradigm by which people tend to think of themselves and others is as _consequentialist agents_: entities who can be usefully modeled as having beliefs and goals, who are then acting according to their beliefs to achieve their goals.
This is o... | https://www.lesswrong.com/posts/M4w2rdYgCKctbADMn/sequence-introduction-non-agent-and-multiagent-models-of |
# Reframing Superintelligence: Comprehensive AI Services as General Intelligence
Since the [CAIS technical report](https://www.fhi.ox.ac.uk/wp-content/uploads/Reframing_Superintelligence_FHI-TR-2019-1.1-1.pdf) is a gargantuan 210 page document, I figured I'd write a post to summarize it. I have focused on the earlier ... | https://www.lesswrong.com/posts/x3fNwSe5aWZb5yXEG/reframing-superintelligence-comprehensive-ai-services-as |
# What emotions would AIs need to feel?
[*this post is completely fluffy and superficial*]
When talking about AIs, the issue of emotions often comes up. I normally finesse the question by saying that we don’t really know if AIs would have things we would recognise and emotions, and would you stop pestering me, I have... | https://www.lesswrong.com/posts/iA6YqiJkZddanNk9x/what-emotions-would-ais-need-to-feel |
# Alignment Newsletter #40
Find all Alignment Newsletter resources [here](http://rohinshah.com/alignment-newsletter/). In particular, you can [sign up](http://eepurl.com/dqMSZj), or look through this [spreadsheet](https://docs.google.com/spreadsheets/d/1PwWbWZ6FPqAgZWOoOcXM8N_tUCuxpEyMbN1NYYC02aM/edit?usp=sharing) of ... | https://www.lesswrong.com/posts/kuL7YmsuQJ9v6xNhK/alignment-newsletter-40 |
# Predictors as Agents
\[UPDATE: I have concluded that the argument in this post is wrong. In particular, consider a generative model. Say the generative model has a choice between two possible 'fixed point' predictions A and B, and currently assigns A 70% probability and B 30% probability. Then the target distributio... | https://www.lesswrong.com/posts/G2EupNcdtigdyNhL2/predictors-as-agents |
# AlphaGo Zero and capability amplification
[AlphaGo Zero](https://deepmind.com/blog/alphago-zero-learning-scratch/) is an impressive demonstration of AI capabilities. It also happens to be a nice proof-of-concept of a [promising alignment strategy](https://ai-alignment.com/benign-model-free-rl-4aae8c97e385).
How Alp... | https://www.lesswrong.com/posts/HA3oArypzNANvXC38/alphago-zero-and-capability-amplification |
# mindlevelup 3 Year Review
_\[An overview of why I didn’t write as much in 2018. Aversions snowballed, and the blog format led to having expectations about content I found sometimes hard to satisfy. A look back at some of my favorite posts of last year, as well as what I plan to do in 2019.\]_
In 2018, not counting ... | https://www.lesswrong.com/posts/SYzRnvKPQegX4fAs9/mindlevelup-3-year-review |
# Book Review: The Structure Of Scientific Revolutions
When I hear scientists talk about Thomas Kuhn, he sounds very reasonable. Scientists have theories that guide their work. Sometimes they run into things their theories can’t explain. Then some genius develops a new theory, and scientists are guided by that one. So... | https://www.lesswrong.com/posts/HcjL8ydHxPezj6wrt/book-review-the-structure-of-scientific-revolutions |
# What are questions?
BONUS CONTENT
_(Meta: I have been paralyzed with indecisive perfectionism on this question for way too long the past day. This dynamic is no small part of why I have never posted an article or barely any comments on this site in eight years of reading it and being part of its culture. Every addi... | https://www.lesswrong.com/posts/cRQASMBhnHqZX5MWq/what-are-questions |
# The 3 Books Technique for Learning a New Skilll
When I'm learning a new skill, there's a technique I often use to quickly gain the basics of the new skill without getting drowned in the plethora of resources that exist. I've found that just 3 resources that cover the skill from 3 separate viewpoints(along with eithe... | https://www.lesswrong.com/posts/oPEWyxJjRo4oKHzMu/the-3-books-technique-for-learning-a-new-skilll |
# No surjection onto function space for manifold X
[*Note: highly technical. Skip if topology is not your thing*]
In his post on [formal open problems in decision theory](https://www.lesswrong.com/posts/5bd75cc58225bf06703753a9/formal-open-problem-in-decision-theory), Scott asked whether there could exist a topologic... | https://www.lesswrong.com/posts/eqi83c2nNSX7TFSfW/no-surjection-onto-function-space-for-manifold-x |
# Open Thread January 2019
If it’s worth saying, but not worth its own post, you can put it here.
Also, if you are new to LessWrong and want to introduce yourself, this is the place to do it. Personal stories, anecdotes, or just general comments on how you found us and what you hope to get from the site and community... | https://www.lesswrong.com/posts/BtN6My9bSvYrNw48h/open-thread-january-2019 |
#
LW Update 2019-1-09 – Question Updates, UserProfile Sorting
First update of 2019!
Streamlined Mobile Above-the-Fold
---------------------------------
On mobile, logged out users now see cleaner, slightly sleeker collections of Rationality: A-Z, Codex and HPMOR. Logged in users see only one recommended sequence in... | https://www.lesswrong.com/posts/kKsm224r6zcnShLej/lw-update-2019-1-09-question-updates-userprofile-sorting |
# What is narrow value learning?
Ambitious value learning aims to achieve superhuman performance by figuring out the underlying latent "values" that humans have, and evaluating new situations according to these values. In other words, it is trying to infer the criteria by which we judge situations to be good. This is ... | https://www.lesswrong.com/posts/vX7KirQwHsBaSEdfK/what-is-narrow-value-learning |
# Book Recommendations: An Everyone Culture and Moral Mazes
_Epistemic Status: Casual_
I highly recommend [An Everyone Culture,](https://www.amazon.com/Everyone-Culture-Deliberately-Developmental-Organization-ebook/dp/B01BO2ITX2) by Robert Kegan, and Moral Mazes, by [Robert Jackall](https://www.amazon.com/Moral-Mazes... | https://www.lesswrong.com/posts/YEr7gv4ZMEpqWW4SG/book-recommendations-an-everyone-culture-and-moral-mazes |
# Combat vs Nurture & Meta-Contrarianism
My initial reaction to [Combat vs Nurture](https://www.lesswrong.com/posts/ExssKjAaXEEYcnzPd/conversational-cultures-combat-vs-nurture) was to think "I already [wrote](http://lo-tho.blogspot.com/2014/12/epistemic-trust.html) [about](http://lo-tho.blogspot.com/2015/06/epistemic-... | https://www.lesswrong.com/posts/WB49uKgMkQRbKaHme/combat-vs-nurture-and-meta-contrarianism |
# Non-Consequentialist Cooperation?
_This is a very rough intuition pump for possible alternatives to value learning._
In broad strokes, the goal of [(ambitious) value learning](https://www.alignmentforum.org/s/4dHMdK5TLN6xcqtyc/p/5eX8ko7GCxwR5N9mN) is to define and implement a notion of cooperation (or helpfulness) ... | https://www.lesswrong.com/posts/F9vcbEMKW48j4Z6h9/non-consequentialist-cooperation |
# What shape has mindspace?
Epistemological status: Babbling.
Let a map each mind in mindspace to how aligned it is. We are trying to optimize a. To that end, lemmata are helpful which talk about the [shape](https://ncatlab.org/nlab/show/stuff,+structure,+property) of mindspace. That's why we try to call it a space ... | https://www.lesswrong.com/posts/XN6whdbSJENRnfjux/what-shape-has-mindspace |
# Megaproject management
Megaproject management is a new-ish subfield of project management. Originally considered to be the special case of project management where the budgets were enormous (billions of dollars), it is developing into a separate specialization because of the high complexity and tradition of failure ... | https://www.lesswrong.com/posts/XBQXDcGs3NhqRscxH/megaproject-management |
# Hierarchical system preferences and subagent preferences
During NeurIPS, I had a discussion with [Dylan Hadfield-Menell](https://people.eecs.berkeley.edu/~dhm/) about how to fomalise my [methods for extracting human values](https://www.lesswrong.com/posts/rcXaY3FgoobMkH2jc/figuring-out-what-alice-wants-part-ii).
He... | https://www.lesswrong.com/posts/iutXWSDd56ieAiyTi/hierarchical-system-preferences-and-subagent-preferences |
# One Website To Rule Them All?
_Epistemic status: I'm not the first person to think of this, and I'm not sure if this is possible. But I want to flesh out some options for consideration and get feedback. Sorry if it's long._
WANTED:
**A crowd-sourced site that reliably presents important pros and cons on any/all to... | https://www.lesswrong.com/posts/65MFhshvi2q26STGN/one-website-to-rule-them-all |
# A general framework for evaluating aging research. Part 1: reasoning with Longevity Escape Velocity
Summary
-------
To this day there is a lack of systematic research to evaluate a cause area with immense potential: aging research. This is the first of a series of posts in which I'll try to begin research to addres... | https://www.lesswrong.com/posts/iuybRR8xqbF92J2cK/a-general-framework-for-evaluating-aging-research-part-1-1 |
# Why is so much discussion happening in private Google Docs?
I've noticed that when I've been invited to view and comment on AI-safety related draft articles (in Google Docs), they tend to quickly attract a lot of extensive discussion, including from people who almost never participate on public forums like LessWrong... | https://www.lesswrong.com/posts/hnvPCZ4Cx35miHkw3/why-is-so-much-discussion-happening-in-private-google-docs |
# Does freeze-dried mussel powder have good stuff that vegan diets don't?
Personal background for this question:
- I'm a vegan for ethical reasons, meaning that I don't want to contribute to the farming of beings that can suffer.
- I think that oysters and mussels probably aren't sentient.
- Oysters and mussels seem p... | https://www.lesswrong.com/posts/vrF99dji7FoqvAawQ/does-freeze-dried-mussel-powder-have-good-stuff-that-vegan |
# Ambitious vs. narrow value learning
_(Re)Posted as part of the AI Alignment Forum sequence on_ _[Value Learning](https://www.alignmentforum.org/s/4dHMdK5TLN6xcqtyc)._
> _**Rohin's note:** The definition of narrow value learning in the previous post focused on the fact that the resulting behavior is limited to some ... | https://www.lesswrong.com/posts/SvuLhtREMy8wRBzpC/ambitious-vs-narrow-value-learning |
# this Website uses Bayesian networks to evaluate claims - RootClaim
I found out this website, [RootClaim](https://www.rootclaim.com), about a year and a half ago (a bit after it was founded), much before finding LW. that's how i found out about Bayesian probability, and i remember getting crazy at the potential of it... | https://www.lesswrong.com/posts/bK5Md78PgtHsASjzB/this-website-uses-bayesian-networks-to-evaluate-claims |
# Comments on CAIS
Over the last few months I’ve talked with Eric Drexler a number of times about his Comprehensive AI Services (CAIS) model of AI development, and read most of [his technical report](https://www.fhi.ox.ac.uk/wp-content/uploads/Reframing_Superintelligence_FHI-TR-2019-1.1-1.pdf) on the topic. I think th... | https://www.lesswrong.com/posts/HvNAmkXPTSoA4dvzv/comments-on-cais |
# Modernization and arms control don’t have to be enemies.
On [a recent panel on nuclear modernization and arms control at Brookings](https://www.brookings.edu/events/falling-apart-the-politics-of-new-start-and-strategic-modernization/), there were a few points made about the potential synergies between arms control a... | https://www.lesswrong.com/posts/2Hc8Bb84hhadHJLF9/modernization-and-arms-control-don-t-have-to-be-enemies |
# [Speech] Worlds That Never Were
_This is the speech I gave at the 2018 Bay Area Secular Solstice. It was written to fit the bill of 'concrete visions of the future.' There is one small change here from the version I gave : the Solstice organizers had me change the second-to-last sentence of the speech, because the e... | https://www.lesswrong.com/posts/mv9x67gqfZ5ceBnKm/speech-worlds-that-never-were |
# Time Biases
Book review: Time Biases: A Theory of Rational Planning and Personal
Persistence, by Meghan Sullivan.
I was very unsure about whether this book would be worth reading, as it
could easily have been focused on complaints about behavior that experts
have long known are mistaken.
I was pleasantly su... | https://www.lesswrong.com/posts/fzEZyZA9n6kFGKLFg/time-biases |
# Dutch-Booking CDT
_\[This post is now superseded by [a much better version of the argument](https://www.lesswrong.com/posts/X7k23zk9aBjjpgLd3/dutch-booking-cdt-revised-argument).\]_
In [a previous post](https://www.alignmentforum.org/posts/tpWfDLZy2tk97MJ3F/a-rationality-condition-for-cdt-is-that-it-equal-edt-part-... | https://www.lesswrong.com/posts/wkNQdYj47HX33noKv/dutch-booking-cdt |
# What are the open problems in Human Rationality?
LessWrong has been around for 10+ years, CFAR's been at work for around 6, and I think there have been at least a few other groups or individuals working on what I think of as the "Human Rationality Project."
I'm interested, especially from people who have invested s... | https://www.lesswrong.com/posts/gX8fcAwk3HGkFyJgk/what-are-the-open-problems-in-human-rationality |
# Directions and desiderata for AI alignment
_Note: This is the first post from_ _**part four: what needs doing**_ _of_ _[the sequence on iterated amplification](https://www.alignmentforum.org/s/EmDuGeRw749sD3GKd). The fourth part of the sequence describes some of the black boxes in iterated amplification and discusse... | https://www.lesswrong.com/posts/kphJvksj5TndGapuh/directions-and-desiderata-for-ai-alignment |
# When is CDT Dutch-Bookable?
I gave a [rough sketch of a Dutch Book against CDT,](https://www.alignmentforum.org/s/fgHSwxFitysGKHH56/p/wkNQdYj47HX33noKv) with an example. I expect it can be turned into a fairly general theorem. What setting should we use to formalize this? What are the necessary assumptions? What's t... | https://www.lesswrong.com/posts/TJT2oBMGaZTE7f2z2/when-is-cdt-dutch-bookable |
# Towards formalizing universality
(_[Cross-posted at ai-alignment.com](https://ai-alignment.com/towards-formalizing-universality-409ab893a456)_)
The scalability of [iterated amplification](https://arxiv.org/pdf/1810.08575.pdf) or [debate](https://arxiv.org/abs/1805.00899) seems to depend on whether large enough team... | https://www.lesswrong.com/posts/M8WdeNWacMrmorNdd/towards-formalizing-universality |
# CDT=EDT=UDT
_Epistemic status: I no longer endorse the particular direction this post advocates, though I'd be excited if someone figured out something that seems to work. I still endorse most of the specific observations._
So... what's the deal with counterfactuals?
Over the past couple of years, I've been writin... | https://www.lesswrong.com/posts/WkPf6XCzfJLCm2pbK/cdt-edt-udt |
# The Tether Theory and the Concrete, Subtle and Causal tiers
_Epistemic status: A bunch of meditation and meditation theory. A previously undocumented theory of this corner of consciousness._
Original post on my blog [bearlamp](http://bearlamp.com.au/the-tether-theory-and-the-concrete-subtle-and-causal-tiers/)
... | https://www.lesswrong.com/posts/yBhNgJwc7p5Bks3QR/the-tether-theory-and-the-concrete-subtle-and-causal-tiers |
# Anthropics: Full Non-indexical Conditioning (FNC) is inconsistent
I'll be using [Full non-indexical conditioning](https://arxiv.org/abs/math/0608592) (FNC) as an example of anthropic reasoning. Given that, it's useful to know that FNC is (time) inconsistent: the expected future probability is not the same as the cur... | https://www.lesswrong.com/posts/jH3NfxoNgKTh9r5KW/anthropics-full-non-indexical-conditioning-fnc-is |
# Anthropic probabilities: answering different questions
What is the answer to the question of anthropic probabilities? I've claimed before that there was no answer - [anthropic decision theory](https://arxiv.org/abs/1110.6437) (ADT) was the only way to go.
But actually, there are answers - the problem is simply that... | https://www.lesswrong.com/posts/nxRjC93AmsFkfDYQj/anthropic-probabilities-answering-different-questions |
# Human-AI Interaction
The importance of feedback
--------------------------
Consider trying to program a self-driving car to drive from San Francisco to Los Angeles -- _with no sensors_ that allow it to gather information as it is driving. This is possible in principle. If you can predict the exact weather condition... | https://www.lesswrong.com/posts/4783ufKpx8xvLMPc6/human-ai-interaction |
# What are the components of intellectual honesty?
I have a pretty strong intuition that "intellectual honesty" points to a specific thing, with a characteristic set of behaviors (both outward (in a conversation/writing) and inward (patterns of thinking)). My concept of intellectual honesty also seems to largely coinc... | https://www.lesswrong.com/posts/YKecvqYkAmBEm8Ngp/what-are-the-components-of-intellectual-honesty |
# Is there a.. more exact.. way of scoring a predictor's calibration?
On platforms like PredictionBook, a user's credibility will be measured by looking at a histogram of all of their predictions (edit: After being told about "brier score", I have become able to notice that PredictionBook totally shows a brier score r... | https://www.lesswrong.com/posts/sg6Xw5wcKL5kxXnsR/is-there-a-more-exact-way-of-scoring-a-predictor-s |
# What AI Safety Researchers Have Written About the Nature of Human Values
Overview
--------
It is clear to most AI safety researchers that the idea of “human values” is underdefined, and this concept should be additionally formalized before it can be used in (mostly mathematical) models of AI alignment. In other wor... | https://www.lesswrong.com/posts/GermiEmcS6xuZ2gBh/what-ai-safety-researchers-have-written-about-the-nature-of |
# Book Summary: Consciousness and the Brain
One of the fundamental building blocks of much of consciousness research, is that of [Global Workspace Theory (GWT)](https://en.wikipedia.org/wiki/Global_workspace_theory). One elaboration of GWT, which focuses on how it might be implemented in the brain, is the Global Neuro... | https://www.lesswrong.com/posts/x4n4jcoDP7xh5LWLq/book-summary-consciousness-and-the-brain |
# A Framework for Internal Debugging
When I'm debugging myself or others for unproductive behavioral, emotional, or mental patterns (hereby referred to as patterns) , I use this quasi-linear framework to make sure that I don't miss any critical steps and the issue stays solved. I've used this framework successfully wi... | https://www.lesswrong.com/posts/mFvuQTzHQiBCDEKw6/a-framework-for-internal-debugging |
# Buy shares in a megaproject
I think we should create a separate legal construct for megaprojects.
I recently wrote about megaproject management [here](https://www.lesswrong.com/posts/XBQXDcGs3NhqRscxH/megaproject-management). In the outline for Oxford's [Major Programme Management](https://www.sbs.ox.ac.uk/programm... | https://www.lesswrong.com/posts/exw3RJJbsPEmXvMFC/buy-shares-in-a-megaproject |
# The reward engineering problem
Today we usually train reinforcement learning agents to perform narrow tasks with simple goals. We may eventually want to train RL agents to behave “well” in open-ended environments where there is no simple goal.
Suppose that we are trying to train an RL agent **A**. In each episode,... | https://www.lesswrong.com/posts/4nZRzoGTqg8xy5rr8/the-reward-engineering-problem |
# And My Axiom! Insights from 'Computability and Logic'
Foreword
========
Max Tegmark's _[Our Mathematical Universe](https://www.amazon.com/Our-Mathematical-Universe-Ultimate-Reality/dp/0307599809)_ briefly touches on a captivating, beautiful mystery:
> The... | https://www.lesswrong.com/posts/n8neHaCubKG3HDWhT/and-my-axiom-insights-from-computability-and-logic |
# Some Thoughts on My Psychiatry Practice
I’ve noticed a marked change in my clientele after going into private practice.\[1\] Of course I expected class differences-- I charge full fee and don’t take insurance. But there are differences that are not as predictable as ‘has more money’. During residency I worked at a h... | https://www.lesswrong.com/posts/phgigGpzxHLTb9jrQ/some-thoughts-on-my-psychiatry-practice |
# Clothing For Men
Nothing fundamentally rational about this post, but it _is_ instrumental. It will be useful to those that want to dress better. It's already been discussed that dressing well improved one's social life; I wanted to put it in one place _how_ to reach that.
TL,DR
* The point of dressing well is to... | https://www.lesswrong.com/posts/675ovpkyyCNxdJcH8/clothing-for-men |
# Alignment Newsletter #41
_Building AI systems that require informed consent_
Find all Alignment Newsletter resources [here](http://rohinshah.com/alignment-newsletter/). In particular, you can [sign up](http://eepurl.com/dqMSZj), or look through this [spreadsheet](https://docs.google.com/spreadsheets/d/1PwWbWZ6FPqAg... | https://www.lesswrong.com/posts/YnBFravZQ5qm6Nmyh/alignment-newsletter-41 |
# In SIA, reference classes (almost) don't matter
This is [another](https://www.lesswrong.com/posts/jH3NfxoNgKTh9r5KW/anthropics-full-non-indexical-conditioning-fnc-is) write-up of a fact that is generally known, but that I haven't seen proven explicitly: the fact that [SIA](https://en.wikipedia.org/wiki/Self-indicati... | https://www.lesswrong.com/posts/MdvwkgKnbxNdbNRao/in-sia-reference-classes-almost-don-t-matter |
# The questions and classes of SSA
Important summary: **For SSA, the reference class to be used is defined in the question asked.**
I [said](https://www.lesswrong.com/posts/nxRjC93AmsFkfDYQj/anthropic-probabilities-answering-different-questions) that [SSA](https://en.wikipedia.org/wiki/Self-sampling_assumption) deri... | https://www.lesswrong.com/posts/wdEGA3RM7ESqHXKMX/the-questions-and-classes-of-ssa |
# Solving the Doomsday argument
The [Doomsday argument](https://en.wikipedia.org/wiki/Doomsday_argument) gives an anthropic argument for why we might expect doom to come reasonably soon. It's known that the Doomsday argument works under [SSA](https://en.wikipedia.org/wiki/Self-sampling_assumption), but not under [SIA]... | https://www.lesswrong.com/posts/igZ2irpE45wJaezq2/solving-the-doomsday-argument |
# Anthropics is pretty normal
In this post, I'll defend these claims:
1. The common understanding of anthropic reasoning is wrong.
2. There are interesting reasons for that error.
3. Anthropic updating is the same as normal updating. For example, our survival is evidence that the world is safer than we thought.
4. F... | https://www.lesswrong.com/posts/uAqs5Q3aGEen3nKeX/anthropics-is-pretty-normal |
# Finance Followups
[Cross-posted from Putanumonit.com](https://putanumonit.com/2019/01/01/finance-followups/).
* * *
[In Defense of Finance](https://putanumonit.com/2018/12/14/defense-of-finance/) generated over 110 comments across WordPress, [LessWrong](https://www.lesswrong.com/posts/kPi7MNgPrR3P3Lzav/in-defense-... | https://www.lesswrong.com/posts/hgR93JeqyZkNetgXi/finance-followups |
# EA Funds: Long-Term Future fund is open to applications until Feb. 7th
(Crossposted [to the EA Forum](https://forum.effectivealtruism.org/posts/oFeGLaJ5bZBBRbjC9/ea-funds-long-term-future-fund-is-open-to-applications-until) feel free to comment on either post with any questions)
The Long-Term Future Fund doing anot... | https://www.lesswrong.com/posts/dvGE8JSeFHtmHC6Gb/ea-funds-long-term-future-fund-is-open-to-applications-until |
# The Very Repugnant Conclusion
I haven't seen the [very repugnant conclusion](https://pdfs.semanticscholar.org/110e/2189a930e4de714b81db2c137b2548b9bcca.pdf) mentioned much here, so I thought I'd add it, as I need it as an example in a [subsequent post](https://www.lesswrong.com/posts/msJexF6sxojRHRLbi/synthesising-d... | https://www.lesswrong.com/posts/6vYDsoxwGQraeCJs6/the-very-repugnant-conclusion |
# Synthesising divergent preferences: an example in population ethics
This post is an example of how one could go about sythesising contradictory human preferences and meta-preferences, using a somewhat simplified version of the method [of this post](https://www.lesswrong.com/posts/Y2LhX3925RodndwpC/resolving-human-va... | https://www.lesswrong.com/posts/msJexF6sxojRHRLbi/synthesising-divergent-preferences-an-example-in-population |
# How could shares in a megaproject return value to shareholders?
In my [previous post](https://www.lesswrong.com/posts/exw3RJJbsPEmXvMFC/buy-shares-in-a-megaproject) on buying shares in a [megaproject](https://www.lesswrong.com/posts/XBQXDcGs3NhqRscxH/megaproject-management), several people asked the most reasonable ... | https://www.lesswrong.com/posts/949YSz6Qtt9ctj7XZ/how-could-shares-in-a-megaproject-return-value-to |
# Bundle your Experiments
The point of this post feels almost too obvious to be worth saying,
yet I doubt that it's widely followed.
People often avoid doing projects that have a low probability of
success, even when the expected value is high. To counter this bias, I
recommend that you mentally combine many suc... | https://www.lesswrong.com/posts/4LKEdcGfLDHaazP4j/bundle-your-experiments |
# Doing Despite Disliking: Self‐regulatory Strategies in Everyday Aversive Activities
A large survey of self-regulatory strategies and how reportedly effective they are.
Much more in the typical "self-regulatory" paradigm for self-help psychology, where people are assumed to have a lot more control over the strategie... | https://www.lesswrong.com/posts/qNZHjQqvRGSMcR836/doing-despite-disliking-self-regulatory-strategies-in |
# Reward uncertainty
In my [last post](https://www.alignmentforum.org/posts/4783ufKpx8xvLMPc6/human-ai-interaction), I argued that interaction between the human and the AI system was necessary in order for the AI system to “stay on track” as we encounter new and unforeseen changes to the environment. The most obvious ... | https://www.lesswrong.com/posts/ZiLLxaLB5CCofrzPp/reward-uncertainty |
# What math do i need for data analysis?
I always had fairly good mathematical thinking (I think) and loved learning about beautiful concepts in math - but i didn't learn much at all in school (cause i had the choice). You can say i was "utilitarian" regarding learning math, i didn't do it if i didn't see how it can e... | https://www.lesswrong.com/posts/RiYqxz6kb3vHjDbCb/what-math-do-i-need-for-data-analysis |
# Why not tool AI?
An extremely basic question that, after months of engaging with AI safety literature, I'm surprised to realize I don't fully understand: why not tool AI?
AI Safety scenarios seem to conceive of AI as an autonomous agent. Is that because of the current machine learning paradigm, where we're setting ... | https://www.lesswrong.com/posts/jkxkMTGfZDzBEaaY8/why-not-tool-ai |
# Less Competition, More Meritocracy?
Analysis of the paper: [Less Competition, More Meritocracy](https://poseidon01.ssrn.com/delivery.php?ID=467087112103027084086072030010100081113017041087020051010016066024098094122101107089004033055004033104044080098012007118093070062033006049092011069120082025093007093086016046020... | https://www.lesswrong.com/posts/idFsH3kWbkKFkawAy/less-competition-more-meritocracy |
# Capability amplification
(_Note: In the past I have referred to this process as ‘bootstrapping’ or ‘policy amplification,’ but those terms are too broad — there are other dimensions along which policies can be amplified, and ‘bootstrapping’ is used all over the place._)
Defining the “intended behavior” of a powerfu... | https://www.lesswrong.com/posts/t3AJW5jP3sk36aGoC/capability-amplification-1 |
# Announcement: AI alignment prize round 4 winners
We (Zvi Mowshowitz and Vladimir Slepnev) are happy to announce the results of the fourth round of the [AI Alignment Prize](https://www.lesswrong.com/posts/juBRTuE3TLti5yB35/announcement-ai-alignment-prize-round-3-winners-and-next), funded by Paul Christiano. From July... | https://www.lesswrong.com/posts/nDHbgjdddG5EN6ocg/announcement-ai-alignment-prize-round-4-winners |
# Following human norms
So far we have been talking about how to learn “values” or “instrumental goals”. This would be necessary if we want to figure out how to build an AI system that does exactly what we want it to do. However, we’re probably fine if we can keep learning and building better AI systems. This suggests... | https://www.lesswrong.com/posts/eBd6WvzhuqduCkYv3/following-human-norms |
# Do the best ideas float to the top?
### It may depend on what we mean by “best”.
_Epistemic status: I understand very little of anything_.
_Speculation about potential applications: regulating a logical prediction market, e.g. [logical induction](https://intelligence.org/files/LogicalInductionAbridged.pdf); cons... | https://www.lesswrong.com/posts/ciSyaM9prQDS72Cbh/do-the-best-ideas-float-to-the-top |
# Disentangling arguments for the importance of AI safety
_Note: my views have shifted significantly since writing this post. I now consider items 1, 2, 3, and 6.2 to be different facets of one core argument, which I call the "second species" argument, and which I explore in depth in [this report](https://www.alignmen... | https://www.lesswrong.com/posts/JbcWQCxKWn3y49bNB/disentangling-arguments-for-the-importance-of-ai-safety |
# Game Analysis Index
This post links to this blog’s posts discussing game design, balance, economics and related topics, as well as any strategy posts. It does not contain new content.
Much of the blog is relevant to gaming, but these are the explicitly on-topic posts.
#### Eternal Sequence
[Eternal, and Hearthsto... | https://www.lesswrong.com/posts/BrD84JT2N7oAyR4dW/game-analysis-index |
# Alignment Newsletter #42
Cooperative IRL as a definition of human-AI group rationality, and an empirical evaluation of theory of mind vs. model learning in HRI
Find all Alignment Newsletter resources [here](http://rohinshah.com/alignment-newsletter/). In particular, you can [sign up](http://eepurl.com/dqMSZj), or l... | https://www.lesswrong.com/posts/2dt8miopNAvhKZPNf/alignment-newsletter-42 |
# Should questions be called "questions" or "confusions" (or "other")?
_Or, what is the best way to think about "what is a question?" on LW?_
The LW Team just had a retreat where we thought through a lot of high level strategy. We have a lot of ideas building off of the "questions" feature.
One thing that struck us... | https://www.lesswrong.com/posts/h7u9prJqbDk7dQ5WM/should-questions-be-called-questions-or-confusions-or-other |
# Stale air / high CO2 may decrease your cognitive function
If you're in a closed space, you may want to open a window.
before the industrial revolution, the atmosphere had 300 parts-per-million (PPM) of CO2. today, this number is already above 400 on average, and 500 in urban areas.
but CO2 doesn't just effect the ... | https://www.lesswrong.com/posts/DtQzYXBTwaWXbhCoS/stale-air-high-co2-may-decrease-your-cognitive-function |
# Too Smart for My Own Good
Originally posted at [sandymaguire.me](http://sandymaguire.me/blog/too-smart/)
I want to share a piece of [ridiculously obvious advice][obvious] today.
[obvious]: http://mindingourway.com/obvious-advice/
I've got a bad habit, which is being too smart for my own good. Which is to say,
wh... | https://www.lesswrong.com/posts/66BsjQtX7wAGhP4tB/too-smart-for-my-own-good |
# The Relationship Between Hierarchy and Wealth
_Epistemic Status: Tentative_
I’m fairly anti-hierarchical, as things go, but the big challenge to all anti-hierarchical ideologies is “how feasible is this in real life? We don’t see many examples around us of this working well.”
Backing up, for a second, what do we m... | https://www.lesswrong.com/posts/kddyK3rrHxpJFyJjo/the-relationship-between-hierarchy-and-wealth |
# Learning with catastrophes
A _catastrophe_ is an event so bad that we are not willing to let it happen even a single time. For example, we would be unhappy if our self-driving car _ever_ accelerates to 65 mph in a residential area and hits a pedestrian.
Catastrophes present a theoretical challenge for traditional m... | https://www.lesswrong.com/posts/qALeGJ9nPcs9eC9Af/learning-with-catastrophes |
# Link: That Time a Guy Tried to Build a Utopia for Mice and it all Went to Hell
Video: [https://www.youtube.com/watch?v=5m7X-1V9nOs](https://www.youtube.com/watch?v=5m7X-1V9nOs)
Text version: http://www.todayifoundout.com/index.php/2018/12/that-time-a-guy-tried-to-build-a-utopia-for-mice-and-it-all-went-to-hell/
"I... | https://www.lesswrong.com/posts/unQKzkDn6wEHKtujG/link-that-time-a-guy-tried-to-build-a-utopia-for-mice-and-it |
# Shanghai LessWrong Meetup
（[中文翻译](http://hpmor.lofter.com/post/382652_12d80c733)）
I'm inviting to start a new Shanghai LW meetup, with focus on a broader discussion about less wrong in the AI safety and cryptography in context of the long-term outcomes for life and humanity. I'd suggest the common theme of the meet... | https://www.lesswrong.com/events/zR4atrRmiaqGLvjYj/shanghai-lesswrong-meetup |
# One-step hypothetical preferences
Human preferences are time-inconsistent, and also contradictory.
That, by itself, is not a huge problem, but it's also the case that few human preferences are present at any given moment. At the moment, I'm focused on finding the best explanation to get my ideas through to you, the... | https://www.lesswrong.com/posts/i6hWWcKyxBPj7ELT6/one-step-hypothetical-preferences |
# S-Curves for Trend Forecasting
**Epistemic Status**: Innovation research and business research is notoriously low quality, and so all the ideas here should be viewed through that lense. What's impressive about the S-curve and evolution trends literature is how remarkably self-consistent it is using a wide variety of... | https://www.lesswrong.com/posts/oaqKjHbgsoqEXBMZ2/s-curves-for-trend-forecasting |
# The human side of interaction
The last few posts have motivated an analysis of the human-AI system rather than an AI system in isolation. So far we’ve looked at the notion that the AI system should [get feedback from the user](https://www.alignmentforum.org/s/4dHMdK5TLN6xcqtyc/p/4783ufKpx8xvLMPc6) and that it could ... | https://www.lesswrong.com/posts/eD9T4kiwB6MHpySGE/the-human-side-of-interaction |
# Is Agent Simulates Predictor a "fair" problem?
It's a simple question, but I think it might help if I add in context. In the paper introducing [Functional Decision Theory](https://arxiv.org/pdf/1710.05060.pdf), it is noted that it is impossible to design an algorithm that can perform well on all decision problems si... | https://www.lesswrong.com/posts/HvLcGmr2APwqauFzW/is-agent-simulates-predictor-a-fair-problem |
# From Personal to Prison Gangs: Enforcing Prosocial Behavior
_This post originally appeared [here](http://seekingquestions.blogspot.com/2017/10/from-personal-to-prison-gangs-enforcing.html); I've updated it slightly and posted it here as a follow-up to [this post](https://www.lesswrong.com/posts/kddyK3rrHxpJFyJjo/the... | https://www.lesswrong.com/posts/sYt3ZCrBq2QAf3rak/from-personal-to-prison-gangs-enforcing-prosocial-behavior |
# Thoughts on reward engineering
_Note: This is the first post from **part five: possible approaches** of [the sequence on iterated amplification](https://www.alignmentforum.org/s/EmDuGeRw749sD3GKd). The fifth section of the sequence breaks down some of these problems further and describes some possible approaches._
... | https://www.lesswrong.com/posts/NtX7LKhCXMW2vjWx6/thoughts-on-reward-engineering |
# For what do we need Superintelligent AI?
Most of human problems could be solved by humans or slightly above human AIs. Which practically useful tasks – which we could define now – require very high level of superintelligence? I see two main such tasks:
1) Prevention of the creation of other potentially dangerous su... | https://www.lesswrong.com/posts/udcoqsJc8stfhinMf/for-what-do-we-need-superintelligent-ai |
# Future directions for narrow value learning
Narrow value learning is a huge field that people are already working on (though not by that name) and I can’t possibly do it justice. This post is primarily a list of things that I think are important and interesting, rather than an exhaustive list of directions to pursue... | https://www.lesswrong.com/posts/MxadmSXHnoCupsWqx/future-directions-for-narrow-value-learning |
# Building up to an Internal Family Systems model
Introduction
============
[Internal Family Systems (IFS)](https://selfleadership.org/about-internal-family-systems.html) is a psychotherapy school/technique/model which lends itself particularly well for being used alone or with a peer. For years, I had noticed that m... | https://www.lesswrong.com/posts/5gfqG3Xcopscta3st/building-up-to-an-internal-family-systems-model |
# Río Grande: judgment calls
In a particularly bad recent bout of anxiety, I learned what seemed to me to be a new and exciting mental move, which I named ‘making a judgment call’. If you had asked me previously whether I made judgment calls, I would have said ‘yes’ however, so describing what I am talking about here ... | https://www.lesswrong.com/posts/mb3QttP9RG8i8ueht/rio-grande-judgment-calls |
# Confessions of an Abstraction Hater
I've written about the [cost of abstraction](http://250bpm.com/blog:86) before. Once you are in the IT industry for couple of decades and once you've read couple of millions lines on legacy code you become healthily suspicious of any kind of abstraction. Not that we can do without... | https://www.lesswrong.com/posts/ivzkPAREG7dAiDJjn/confessions-of-an-abstraction-hater |
# Freely Complying With the Ideal: A Theory of Happiness
_Epistemic status_: somewhat empirical, having gone from badly depressed to astoundingly happy within a year through _a lot_ of different experiences (and no drugs). I don’t know whether what I do that makes me happy is idiosyncratic to me and a subgroup of peo... | https://www.lesswrong.com/posts/NYvsn97vmemuAM9zd/freely-complying-with-the-ideal-a-theory-of-happiness |
# Prediction Contest 2018: Scores and Retrospective
Way back in April 2018, I [announced a Prediction Contest](https://www.lesswrong.com/posts/sYW9oMXhLHCvaCoLX/prediction-contest-2018), in which the person who made the best predictions on a bunch of questions on PredictionBook ahead of a 1st July deadline would win a... | https://www.lesswrong.com/posts/8e8iWCCdEoprSxRc6/prediction-contest-2018-scores-and-retrospective |
# "The Unbiased Map"
“Long have we suffered under the tyranny of maps.
Biased maps which show topography, but not population.
Wretched maps which speak of religion, but not languages.
Divisive maps which paint with the color of Party, but not the color of economic conditions.
Dirty maps which show crop yield acros... | https://www.lesswrong.com/posts/be46uj5oPRsAv6eCT/the-unbiased-map |
# A small example of one-step hypotheticals
Just a small example of what [one-step hypotheticals](https://www.lesswrong.com/posts/i6hWWcKyxBPj7ELT6/one-step-hypothetical-preferences) might mean in theory and in practice.
This involves a human H pricing some small object:
:
* But I t... | https://www.lesswrong.com/posts/Q7WiHdSSShkNsgDpa/how-much-can-value-learning-be-disentangled |
# Can there be an indescribable hellworld?
Can there be an indescribable hellworld? What about an un-summarisable one?
By hellworld, I mean a world of very low value according to our value scales - maybe one where large number of simulations are being tortured (aka [mind crimes](https://philosophicaldisquisitions.blo... | https://www.lesswrong.com/posts/rArsypGqq49bk4iRr/can-there-be-an-indescribable-hellworld |
# Towards equilibria-breaking methods
All of this is vague intuition.
[Much](http://slatestarcodex.com/2014/07/30/meditations-on-moloch/) has been [said](https://www.lesswrong.com/posts/x5ASTMPKPowLKpLpZ/moloch-s-toolbox-1-2) about the problems of Nash equilibria. I've been wondering about how to quantify these sorts... | https://www.lesswrong.com/posts/CDY74XtLoPGd9urNj/towards-equilibria-breaking-methods |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.