text stringlengths 300 320k | source stringlengths 52 154 |
|---|---|
# LessWrong discussed in New Ideas in Psychology article
Hi LessWrong,
I wanted to let you know about a recent article in New Idea in Psychology written by myself and Dr. Dario Krpan (assistant professor of psychology at London School of Economics) that discusses LessWrong and Scott Alexander (Siskind) as examples ... | https://www.lesswrong.com/posts/8g4zvCHZ4B4iQAfhK/lesswrong-discussed-in-new-ideas-in-psychology-article |
# Omicron Post #5
A few key questions determine our fate. We are rapidly getting closer to the answers. That doesn’t mean we have the _magnitudes_ of the answers. The magnitudes _matter,_ and they matter _a lot._ It’s still a stronger understanding than even a few days ago. Next time will be stronger still.
My guess ... | https://www.lesswrong.com/posts/2CjAYW7TuM8cu3WWR/omicron-post-5 |
# Conversation on technology forecasting and gradualism
This post is a transcript of a multi-day discussion between Paul Christiano, Richard Ngo, Eliezer Yudkowsky, Rob Bensinger, Holden Karnofsky, Rohin Shah, Carl Shulman, Nate Soares, and Jaan Tallinn, following up on the Yudkowsky/Christiano debate in [1](https://w... | https://www.lesswrong.com/posts/nPauymrHwpoNr6ipx/conversation-on-technology-forecasting-and-gradualism |
# Covid 12/9: Counting Down the Days
Omicron is coming. We are not preparing for it. That is the news that matters.
_In addition to that,_ Delta is a growing problem as we face a winter wave. Future problems don’t protect us from current problems, no matter how inevitable and predictable the current problems might be... | https://www.lesswrong.com/posts/4zrg2xYJabiXQfbW3/covid-12-9-counting-down-the-days |
# Combining Forecasts
Often there are multiple ways to forecast the same thing, and we'd like a way of combining the forecasts together. For instance, consider the following question:
> Will _Will_ be among the top 3 most sold nonfiction books on Amazon for the week of November 21?
_(This was posed to the course sta... | https://www.lesswrong.com/posts/3Dzr7X9hCFaxEwYPq/combining-forecasts |
# Are big brains for processing sensory input?
Elephants have bigger brains than human beings. But human beings are smarter. What's all the cranial matter doing?
I used to believe that a larger organism's brain needs to generate larger electrical impulses to move its muscles. But that's not how electrical engineering... | https://www.lesswrong.com/posts/MxbXroGxXg7yEDFX7/are-big-brains-for-processing-sensory-input |
# The Promise and Peril of Finite Sets
Has this ever happened to you?
> But how can we decide whether this {species/moral theory/consensus node/quantum world} is actually 2 different things, or just 1 thing? Or does it even count at all?
The underlying cause of such dilemmas is that somebody, somewhere in the setup ... | https://www.lesswrong.com/posts/qLaShfcnXGnYeKFJW/the-promise-and-peril-of-finite-sets |
# There is essentially one best-validated theory of cognition.
There are many theories of cognition. But if you want to work within a framework with the following properties:
* Explains the major cognitive phenomena we know about.
* Fits experimental data well, *down to human reaction times,* in a wide variety of... | https://www.lesswrong.com/posts/NB9QrBa335GDijuyn/there-is-essentially-one-best-validated-theory-of-cognition |
# Understanding Gradient Hacking
*This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the* [*Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program*](https://www.lesswrong.com/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-program-under-evan-hubinger)*.*
*Ad... | https://www.lesswrong.com/posts/bdayaswyewjxxrQmB/understanding-gradient-hacking |
# [Linkpost] Chinese government's guidelines on AI
China (PRC) published a new AI policy: "*Ethical Norms for the New Generation Artificial Intelligence*" (2021-09-26).
A surprisingly large part of the document's content is dedicated to the topics related to AI X-risk (although the existential risk is never addresse... | https://www.lesswrong.com/posts/ZGrAbfKQbNEazQpkM/linkpost-chinese-government-s-guidelines-on-ai |
# The Plan
This is a high-level overview of the reasoning behind my research priorities, written as a Q&A.
What’s your plan for AI alignment?
----------------------------------
Step 1: sort out our fundamental confusions about agency
Step 2: ambitious value learning (i.e. build an AI which correctly learns human va... | https://www.lesswrong.com/posts/3L46WGauGpr7nYubu/the-plan |
# What role should evolutionary analogies play in understanding AI takeoff speeds?
*Cross-posted to the EA forum* [*https://forum.effectivealtruism.org/posts/aSDnzAm85a3Pi87rm/what-role-should-evolutionary-analogies-play-in*](https://forum.effectivealtruism.org/posts/aSDnzAm85a3Pi87rm/what-role-should-evolutionary-ana... | https://www.lesswrong.com/posts/teD4xjwoeWc4LyRAD/what-role-should-evolutionary-analogies-play-in |
# Moore's Law, AI, and the pace of progress
It seems to be a minority view nowadays to believe in Moore's Law, the routine doubling of transistor density roughly every couple of years, or even the much gentler claim, that [There's Plenty \[more\] Room at the Bottom](https://en.wikipedia.org/wiki/There%27s_Plenty_of_Ro... | https://www.lesswrong.com/posts/aNAFrGbzXddQBMDqh/moore-s-law-ai-and-the-pace-of-progress |
# Extrinsic and Intrinsic Moral Frameworks
The Western intellectual tradition comes from Christianity. Christianity comes from Judaism. Judaism comes from Zoroastrianism. Zoroastrianism has many ideas recognizable within Christianity.
- Zoroastrianism has a dualistic cosmology of good and evil.
- Zoroastrianism has an... | https://www.lesswrong.com/posts/zxvxCzW4y4HuLCGrN/extrinsic-and-intrinsic-moral-frameworks |
# Anxiety and computer architecture
Computer architecture seems like a great analogy for some of the anxiety that I feel.
## Call stack fatigue
From [Call stacks in everyday life](https://adamzerner.bearblog.dev/call-stacks-in-everyday-life/)
> For example, before writing this article I was getting my dog ready for... | https://www.lesswrong.com/posts/ErmHntpsMaZL9S65R/anxiety-and-computer-architecture |
# D&D.Sci GURPS Dec 2021: Hunters of Monsters
Intro (skippable)
=================
"Are you really the smartest member of the Hunters' Guild?"
"I'm the smartest at fighting! What's the difference?"
"Well, you're just about smart enough to write, at the very least"
"And you're just about short enough, and just about... | https://www.lesswrong.com/posts/3TDeQJLQBPmosgG6t/d-and-d-sci-gurps-dec-2021-hunters-of-monsters |
# What on Earth is a Series I savings bond?
_crossposted from: [https://blog.rossry.net/series-i/](https://blog.rossry.net/series-i/)_
_(Not investment advice, of course.)_
**Summary:** The [Series I savings bond](https://www.treasurydirect.gov/indiv/products/prod_ibonds_glance.htm) is a US government bond offered t... | https://www.lesswrong.com/posts/AC8YvKyGJrTG3JsyF/what-on-earth-is-a-series-i-savings-bond |
# Transforming myopic optimization to ordinary optimization - Do we want to seek convergence for myopic optimization problems?
While reading some papers, I found a way to take a [myopic](https://www.lesswrong.com/tag/myopia) optimization problem $\mathcal{L}$, and construct a non-myopic optimization problem $\mathcal{... | https://www.lesswrong.com/posts/Ayt24gxcjfY3tDmwK/transforming-myopic-optimization-to-ordinary-optimization-do |
# EA Dinner Covid Logistics
Yesterday we hosted an [EA](https://en.wikipedia.org/wiki/Effective_altruism) dinner, the first largeish (~22 people) gathering inside our house since before the pandemic. It went well, and it was really good to see people. I wanted to write a bit about what sort of precautions we took for ... | https://www.lesswrong.com/posts/bhSpqbNjDjKsjZDLS/ea-dinner-covid-logistics |
# Magna Alta Doctrina
Continuing topics from an earlier 2015 blog post, [The Brain as a Universal Learning Machine](https://www.lesswrong.com/posts/9Yc7Pp7szcjPgPsjf/the-brain-as-a-universal-learning-machine), this new series will begin with updated distillations of deep learning's best learnings, then interweave brai... | https://www.lesswrong.com/posts/iNaLHBaqh3mL45aH8/magna-alta-doctrina |
# Some abstract, non-technical reasons to be non-maximally-pessimistic about AI alignment
I basically agree with Eliezer’s picture of things in the [AGI interventions post](https://www.lesswrong.com/posts/CpvyhFy9WvCNsifkY/discussion-with-eliezer-yudkowsky-on-agi-interventions).
But I’ve seen some readers rounding of... | https://www.lesswrong.com/posts/vT4tsttHgYJBoKi4n/some-abstract-non-technical-reasons-to-be-non-maximally |
# Nuclear war anthropics
In the most recent episode of Rationally Speaking ("Humanity on the precipice", 10 Dec. 2021), Julia Galef and Toby Ord discuss the risk of a nuclear exchange during the Cold War:
> J.G. \[55:28\]: "If you had asked me to put a probability on each case, like taking nuclear near misses—for exa... | https://www.lesswrong.com/posts/zYTyhAo8eQiaCuEAB/nuclear-war-anthropics |
# For and Against Lotteries in Elite University Admissions
People (whose names often rhyme with Palcolm Sadwell) [sometimes](https://behavioralscientist.org/do-college-admissions-by-lottery/) [suggest](https://www.pushkin.fm/episode/the-powerball-revolution/) replacing the admissions systems of highly selective univer... | https://www.lesswrong.com/posts/Bzz2SHcBQpPiwnFuu/for-and-against-lotteries-in-elite-university-admissions |
# Redwood's Technique-Focused Epistemic Strategy
Imagine you’re part of a team of ML engineers and research scientists, and you want to help with alignment. Everyone is ready to jump in the fray; there’s only one problem — how are you supposed to do applied research when you don’t really know how AGI will be built, wh... | https://www.lesswrong.com/posts/2xrBxhRhde7Xddt38/redwood-s-technique-focused-epistemic-strategy |
# Teaser: Hard-coding Transformer Models
Transformer models are incredibly powerful for natural language tasks (and they are starting to find uses in many other fields of machine learning). Unfortunately, it is nigh-impossible to interpret what goes on inside them. OR IS IT???
In this post, I am trying to gauge poten... | https://www.lesswrong.com/posts/Lq6jo5j9ty4sezT7r/teaser-hard-coding-transformer-models |
# Baby Nicknames
Lily and Anna rarely call Nora by her name. Instead, it's always nicknames, with lots of variety. For a week starting on Thanksgiving I counted how often they used different terms:
| 21 | Norkit (Anna favorite) |
| --- | --- |
| 20 | Nwetzel (Lily favorite) |
| 16 | Norky |
| 9 | Cuddleezel |
| ... | https://www.lesswrong.com/posts/YwP7cPpHS69i5pSXu/baby-nicknames |
# Hard-Coding Neural Computation
Previously: [Teaser: Hard-coding Transformer Models](https://www.lesswrong.com/posts/Lq6jo5j9ty4sezT7r/teaser-hard-coding-transformer-models)
Introduction
============
Transformer models are incredibly powerful for natural language tasks (and they are starting to find uses in many ot... | https://www.lesswrong.com/posts/HkghiK6Rt35nbgwKA/hard-coding-neural-computation |
# Summary of the Acausal Attack Issue for AIXI
**Attention conservation notice:** To a large extent, this is redundant with [Paul's previous post about](https://ordinaryideas.wordpress.com/2016/11/30/what-does-the-universal-prior-actually-look-like/) [this](this{]}(https://ordinaryideas.wordpress.com/2016/11/30/what-d... | https://www.lesswrong.com/posts/YbahERfcjTu7LZNQ6/summary-of-the-acausal-attack-issue-for-aixi |
# What’s the backward-forward FLOP ratio for Neural Networks?
**Summary:**
------------
1. *Classic settings*, i.e. deep networks with convolutional layers and large batch sizes, ***almost always have backward-forward FLOP ratios close to 2:1***.
2. Depending on the following criteria we can encounter **ratios betw... | https://www.lesswrong.com/posts/fnjKpBoWJXcSDwhZk/what-s-the-backward-forward-flop-ratio-for-neural-networks |
# Understanding and controlling auto-induced distributional shift
*This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the* [*Stanford Existential Risks Initiative ML Alignment Theory Scholars (MATS) program*](https://www.lesswrong.com/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-pr... | https://www.lesswrong.com/posts/rTYGMbmEsFkxyyXuR/understanding-and-controlling-auto-induced-distributional |
# Solving Interpretability Week
For original motivation, see [solving corrigibility week](https://www.lesswrong.com/posts/Lv3emECEjkCSHG7L7/solve-corrigibility-week). I'll state what I learned from corrigibility week, why I didn't post one last week, and the updated format for interpretability.
What I Noticed
-----... | https://www.lesswrong.com/posts/jbdDxmhxBygDqbQMD/solving-interpretability-week |
# Analysis of Bird Box (2018)
[Caveat lector: hastily written, speculative. Also banal maybe.]
[SPOILER ALERT for the film Bird Box (2018) on Netflix. IMO it's not amazing but is fairly good, possibly worth watching spoiler-free. This post will spoil most of the events of the movie. Be warned that it's a horror movi... | https://www.lesswrong.com/posts/GMa9Jzh45wPobKjc7/analysis-of-bird-box-2018 |
# Omicron Post #6
Previous posts: [#1](https://thezvi.wordpress.com/2021/11/26/omicron-variant-post-1-were-fed-its-never-over/), [#2](https://thezvi.wordpress.com/2021/11/29/omicron-variant-post-2/), [#3](https://thezvi.wordpress.com/2021/12/02/omicron-post-3/), [#4](https://thezvi.wordpress.com/2021/12/06/omicron-pos... | https://www.lesswrong.com/posts/Rxz8i4AB7EZESoDFK/omicron-post-6 |
# Language Model Alignment Research Internships
I'm [Ethan Perez](https://ethanperez.net/), a final year PhD student at NYU, working on aligning language models with human preferences. I'm looking to hire research interns to work on projects in this space, starting early 2022. I expect candidates to have strong softwa... | https://www.lesswrong.com/posts/vHcGGrnzcshybrCJD/language-model-alignment-research-internships |
# Enabling More Feedback for AI Safety Researchers
*Cross-posted from the* [*EA forum*](https://forum.effectivealtruism.org/posts/6qaaRnu6oN4pdAnWF/enabling-more-feedback)*.*
### **Acknowledgements**
*Thank you to JJ Hepburn for creating this initiative & providing the ideas described in this post, and thank you to ... | https://www.lesswrong.com/posts/AaKodoAETap2A98Kq/enabling-more-feedback-for-ai-safety-researchers |
# Mystery Hunt 2022
[Teacher's password](https://www.lesswrong.com/posts/StbTTmFZD6jZrCCH8/teacher-s-password-the-lesswrong-mystery-hunt-team) will be participating in the [2022 MIT Mystery Hunt](https://www.mitmh2022.com)!
**What is the MIT Mystery Hunt?**
The MIT Mystery Hunt is a puzzle-solving competition, in wh... | https://www.lesswrong.com/events/iBZsonppCgagfNAzm/mystery-hunt-2022 |
# Consequentialism & corrigibility
Background 1: Preferences-over-future-states (a.k.a. consequentialism) vs ~~Preferences-over-trajectories~~ other kinds of preferences
======================================================================================================================================
*(Note: The o... | https://www.lesswrong.com/posts/KDMLJEXTWtkZWheXt/consequentialism-and-corrigibility |
# Zvi’s Thoughts on the Survival and Flourishing Fund (SFF)
Epistemic Status: **The views expressed are mine alone. They are various degrees of strong and weak, and are various degrees of weakly or strongly held.** Due to time constraints, I’m not confident I was careful to mark every claim here with the proper ‘I thi... | https://www.lesswrong.com/posts/kuDKtwwbsksAW4BG2/zvi-s-thoughts-on-the-survival-and-flourishing-fund-sff |
# Interlude: Agents as Automobiles
I ended up writing a long rant about agency in my review of Joe Carlsmith’s report on x-risk from power-seeking AI. I’ve polished the rant a bit and posted it into this sequence. The central analogy between APS-AI and self-propelled machines (“Auto-mobiles”) is a fun one, and I suspe... | https://www.lesswrong.com/posts/cxkwQmys6mCB6bjDA/interlude-agents-as-automobiles |
# ARC is hiring!
The [Alignment Research Center](https://alignmentresearchcenter.org/) is hiring researchers; if you are interested, please apply!
(**Update: We have wrapped up our hiring "round" for early 2022, but are still accepting researcher applications on a rolling basis** [**here**](https://docs.google.com/fo... | https://www.lesswrong.com/posts/dLoK6KGcHAoudtwdo/arc-is-hiring |
# ARC's first technical report: Eliciting Latent Knowledge
ARC has published a report on [Eliciting Latent Knowledge](https://docs.google.com/document/d/1WwsnJQstPq91_Yh-Ch2XRL8H_EpsnjrC1dwZXR37PC8/edit?usp=sharing), an open problem which we believe is central to alignment. We think reading this report is the clearest... | https://www.lesswrong.com/posts/qHCDysDnvhteW7kRd/arc-s-first-technical-report-eliciting-latent-knowledge |
# Should we rely on the speed prior for safety?
*This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the* [*Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program*](https://www.lesswrong.com/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-program-under-evan-hu... | https://www.lesswrong.com/posts/iALu99gYbodt4mLqg/should-we-rely-on-the-speed-prior-for-safety |
# Ngo's view on alignment difficulty
This post features a write-up by Richard Ngo on his views, with inline comments.
Color key:
| Chat | Google Doc content | Inline comments |
| --- | --- | --- |
13\. Follow-ups to the Ngo/Yudkowsky conversation
=================================================
13.1. Alignment di... | https://www.lesswrong.com/posts/gf9hhmSvpZfyfS34B/ngo-s-view-on-alignment-difficulty |
# Robin Hanson's "Humans are Early"
This post was nominated for the 2021 Review by Algon. The post is short, so I'm including the full text:
> Imagine that advanced life like us is terribly rare in the universe. So damn rare that if we had not shown up, then our region of the universe would almost surely have forever... | https://www.lesswrong.com/posts/qmLw98jJvQEnaGpbF/robin-hanson-s-humans-are-early |
# The Natural Abstraction Hypothesis: Implications and Evidence
_This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the [Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program](https://www.lesswrong.com/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-program-... | https://www.lesswrong.com/posts/Fut8dtFsBYRz8atFF/the-natural-abstraction-hypothesis-implications-and-evidence |
# 2020 Review: The Discussion Phase
*LessWrong is currently doing a* [*major review of 2020*](https://www.lesswrong.com/posts/M9kDqF2fn3WH44nrv/the-2020-review-preliminary-voting) *— looking back at old posts and considering which of them have stood the test of time.*
*There are three phases:*
* ***Preliminary Vot... | https://www.lesswrong.com/posts/vQfyDLkrKvufaZTwt/2020-review-the-discussion-phase |
# We'll Always Have Crazy
### *Content warning: preparadigmatic meander; half-baked models; thinking-out-loud. Unlike the median Duncan essay.*
* * *
Logan Strohl's response to the first draft of the below:
> i really wish this were a more-effective-for-me invitation to help you figure something out. i think that's... | https://www.lesswrong.com/posts/g5WziBKDE8M2Xbu4H/we-ll-always-have-crazy |
# Leverage
If you want to change the world then you need to act on scale. A single human being cannot act on scale directly. You need leverage. **Leverage** is mechanism by which you can cause an impact several orders of magnitude bigger than yourself.
Sam Altman [gives five examples of leverage](https://blog.samaltm... | https://www.lesswrong.com/posts/od4zNmfujwEcDtWRq/leverage |
# Framing approaches to alignment and the hard problem of AI cognition
_(Thanks to Adam Shimi and Steven Byrnes for feedback.)_
_(Edit: unsurprisingly, my views have updated some after writing this post. Specifically, I now put much higher credence on solving inner alignment via verification or relaxed adversarial tr... | https://www.lesswrong.com/posts/Cj6PBGSjqkPfXbwCF/framing-approaches-to-alignment-and-the-hard-problem-of-ai |
# My Overview of the AI Alignment Landscape: A Bird's Eye View
_**Disclaimer:** I recently started as an interpretability researcher at Anthropic, but I wrote this doc before starting, and it entirely represents my personal views not those of my employer_
_Intended audience: People who understand_ why _you might thin... | https://www.lesswrong.com/posts/SQ9cZtfrzDJmw9A2m/my-overview-of-the-ai-alignment-landscape-a-bird-s-eye-view |
# More power to you
When people say things like “we were promised [flying cars](https://rootsofprogress.org/where-is-my-flying-car),” I sometimes wonder, “who *promised* this?”
I guess this is what they mean. From a 1959 ad that ran in the *LA Times:*
*.*
**TL... | https://www.lesswrong.com/posts/R67qpBj5doGcjQzmc/universality-and-the-filter |
# Motivations, Natural Selection, and Curriculum Engineering
_Epistemic status: I am professionally familiar with contemporary machine learning theory and practice, and have a long-standing amateur interest in natural history. But (despite being a long-term lurker on LW and AF) I feel very uncertain about the nature a... | https://www.lesswrong.com/posts/RwYh4grJs4pbJdTh3/motivations-natural-selection-and-curriculum-engineering |
# Five Missing Moods
Read [this](https://www.econlib.org/archives/2016/01/the_invisible_t.html) Bryan Caplan post, it's short and direct. His thesis is:
> You can learn a lot by comparing the mood reasonable proponents _would_ hold to the mood actual proponents _do_ hold.
I could quibble, but at the very least this ... | https://www.lesswrong.com/posts/Eew7wvSjo3SarHjCR/five-missing-moods |
# An Open Letter to the Monastic Academy and community members
I am sharing a link here so that this account is locatable to members of this community from which the Monastic Academy actively recruits. This account describes my personal experiences, observations of organizational practices and misconduct, and knowledg... | https://www.lesswrong.com/posts/z8eNwLzMAmS9zijBN/an-open-letter-to-the-monastic-academy-and-community-members |
# Elicitation for Modeling Transformative AI Risks
*This post is part 8 in our sequence on Modeling Transformative AI Risk. We are building a model to understand debates around existential risks from advanced AI. The model is made with Analytica software, and consists of nodes (representing key hypotheses and cruxes) ... | https://www.lesswrong.com/posts/Kz9NHBMeJxzSwb7R9/elicitation-for-modeling-transformative-ai-risks |
# Where can one learn deep intuitions about information theory?
I'm currently going through [Brilliant](https://brilliant.org/)'s course on "Knowledge and Uncertainty". I just got through the part where it explains what Shannon entropy is. I'm now watching a wave of realizations cascade in my mind. For instance, I now... | https://www.lesswrong.com/posts/jRKHuNuy7EAxxvFub/where-can-one-learn-deep-intuitions-about-information-theory |
# Omicron Post #7
Here are some better worlds we might have lived in, but don’t.
1. In a better world, I could focus on this full time and also maybe even hire a research assistant, and be better able to scour for information.
2. In a better world than that, there would be a department at a newspaper, or some other... | https://www.lesswrong.com/posts/XrzPey4cwhPeHL6QF/omicron-post-7 |
# Subway Slides
Here is the current layout of [Davis Square station](https://en.wikipedia.org/wiki/Davis_station_(MBTA)):
[](https://www.jefftk.com/davis-layout-big.png)
If you're coming in from the primary headhouse, on the right, your path is really indirect:
[![](https... | https://www.lesswrong.com/posts/gpQtmBEsRuAxXv5fE/subway-slides |
# What Caplan’s "Missing Mood" Heuristic Is Really For
Bryan Caplan [wrote and quoted](https://www.econlib.org/archives/2015/12/the_meaning_of_2.html) the idea of a "proper reaction," the "specific package of moods" that a "reasonable person" holding an intellectual position would have. If this mood is missing, it is ... | https://www.lesswrong.com/posts/xwH3Amd82qGcAa7CX/what-caplan-s-missing-mood-heuristic-is-really-for |
# The "Other" Option
_For this post, I commissioned Misha Yagudin, a top-ranked forecaster, to provide feedback and commentary. I include selected comments as quotes throughout._
For open-ended questions, it's easy to underestimate the number of plausible outcomes. For instance, while practicing for the forecasting c... | https://www.lesswrong.com/posts/tvjTDZrgP9FMy5Wqz/the-other-option |
# Reviews of “Is power-seeking AI an existential risk?”
(Edited 10/14/22 to add Lifland review. Edited 7/10/23 to add Levinstein review. Edited 10/18/23 to add superforecaster reviews.)
Open Philanthropy solicited reviews of my draft report [“Is power-seeking AI an existential risk?”](https://docs.google.com/document... | https://www.lesswrong.com/posts/qRSgHLb8yLXzDg4nf/reviews-of-is-power-seeking-ai-an-existential-risk |
# Covid 12/16: On Your Marks
The main action continues to be at [the Omicron post](https://thezvi.substack.com/p/omicron-post-7). There’s still some background developments and near term developments worth noticing, but for now nothing that seems as important or that seems terribly urgent, so read that first.
Executi... | https://www.lesswrong.com/posts/pCbSJBtWTp5tLhc8e/covid-12-16-on-your-marks |
# Evidence Sets: Towards Inductive-Biases based Analysis of Prosaic AGI
***Epistemic Status**:* [*Exploratory*](https://www.alignmentforum.org/posts/Hrm59GdN2yDPWbtrd/feature-idea-epistemic-status)*. My current but-changing outlook with limited exploration & understanding for ~60-80hrs.*
***Acknowledgements**: This p... | https://www.lesswrong.com/posts/aW5CPtqtvs2EYKMMK/evidence-sets-towards-inductive-biases-based-analysis-of |
# The Case for Radical Optimism about Interpretability
Summary:
1. I argue we can probably build ML systems that are both high-performance and much more interpretable than current systems.
2. I discuss some reasons why current ML interpretability lags behind what should be possible.
3. I highlight some interestin... | https://www.lesswrong.com/posts/LHCSZbhbtoLpr7B7u/the-case-for-radical-optimism-about-interpretability |
# Blog Respectably
The best way to learn how to write is to take classes. The best way to take classes is via a university degree. I earned my master's in blogging at the University of Pens.
Besides teaching you how to write, a university provides necessary credentials. You wouldn't be reading this post if I weren't ... | https://www.lesswrong.com/posts/7comfTbgr8kEzC2ZM/blog-respectably |
# Venture Granters, The VCs of public goods, incentivizing good dreams
*Epistemic Status: Early concept*
The Gist of it: To support sane risktaking, an institution should generally grant career capital in proportion to a quantitative record of the harms *and* benefits of a decisionmaking entity's past funding decisio... | https://www.lesswrong.com/posts/NY9nfKQwejaghEExh/venture-granters-the-vcs-of-public-goods-incentivizing-good |
# DL towards the unaligned Recursive Self-Optimization attractor
Consider this abridged history of recent ML progress:
>A decade or two ago, computer vision was a field that employed dedicated researchers who designed specific increasingly complex feature recognizers (SIFT, SURF, HoG, etc.) These were usurped by deep... | https://www.lesswrong.com/posts/S54P4qzkMSb6iwqjx/dl-towards-the-unaligned-recursive-self-optimization |
# Important ML systems from before 2012?
I am the main coordinator of a group studying trends in Machine Learning.
Together, we are building a [public dataset](https://docs.google.com/spreadsheets/d/1AAIebjNsnJj_uKALHbXNfn3_YsT6sHXtCU0q7OIPuc4/edit#gid=0) annotating milestone ML systems between 1950 and today. You ca... | https://www.lesswrong.com/posts/eYGsJC8hwECeGrjii/important-ml-systems-from-before-2012 |
# Introducing the Principles of Intelligent Behaviour in Biological and Social Systems (PIBBSS) Fellowship
*Cross-posted to* [*EA Forum*](https://forum.effectivealtruism.org/posts/Ckont9EtqkenegLYv/introducing-the-principles-of-intelligent-behaviour-in)
Introduction
============
How would you go about scientifically... | https://www.lesswrong.com/posts/4Tjz4EJ8DozE9z5nQ/introducing-the-principles-of-intelligent-behaviour-in |
# Universal counterargument against “badness of death” is wrong
All the usually offered counterarguments to the idea that death is evil boil down to one universal counterargument:
*If there is no death, then there will be a bad phenomenon X (overpopulation, cessation of progress, eternal dictatorship, boredom), there... | https://www.lesswrong.com/posts/ukBrtPGnELCRJk5SP/universal-counterargument-against-badness-of-death-is-wrong |
# How I became a person who wakes up early
This is an extended anecdotal description of a life change that honestly still surprises me; I went from the classic internet nerd who stays up until 3 and sleeps until noon, to a person who reliably goes to bed before 10pm and wakes up at 6:30am every morning. This didn't ha... | https://www.lesswrong.com/posts/D5EnCvPiXEqGa6xrL/how-i-became-a-person-who-wakes-up-early |
# Disentangling Perspectives On Strategy-Stealing in AI Safety
*This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program.*
*Additional thanks to Ameya Prabhu and Callum McDougall for their thoughts and... | https://www.lesswrong.com/posts/CtGwGgxfoefiwfcor/disentangling-perspectives-on-strategy-stealing-in-ai-safety |
# Occupational Infohazards
[content warning: discussion of severe mental health problems and terrifying thought experiments]
This is a follow-up to [my recent post](https://www.lesswrong.com/posts/MnFqyPLqbiKL8nSR7/my-experience-at-and-around-miri-and-cfar-inspired-by-zoe) discussing my experience at and around MIRI ... | https://www.lesswrong.com/posts/pQGFeKvjydztpgnsY/occupational-infohazards |
# Elevated Air Purifier Cubes
I've been [a](https://www.jefftk.com/p/comparing-options-for-safer-events) [big](https://www.jefftk.com/p/covid-christmas) [proponent](https://www.jefftk.com/p/ea-dinner-covid-logistics) of [box fan air purifier cubes](https://cleanarair.com/), but they are bulky and fragile. If your ceil... | https://www.lesswrong.com/posts/AYSznMsqtHfNo4bjR/elevated-air-purifier-cubes |
# Should I delay having children to take advantage of polygenic screening?
I would like to know whether I should delay having children to take advantage of polygenic screening technology. I imagine this could be valuable for other aspiring parents to know as well, and would (probably) have positive externality for asp... | https://www.lesswrong.com/posts/tg2fzDSjzyEgo2QFr/should-i-delay-having-children-to-take-advantage-of |
# Don't Influence the Influencers!
*This post was written under Evan Hubinger’s direct guidance and mentorship, as a part of the* [*Stanford Existential Risks Institute ML Alignment Theory Scholars (MATS) program*](https://www.lesswrong.com/posts/FpokmCnbP3CEZ5h4t/ml-alignment-theory-program-under-evan-hubinger)*.*
*... | https://www.lesswrong.com/posts/Eg9FE2iYp3ngySsMD/don-t-influence-the-influencers |
# Beware using words off the probability distribution that generated them.
Suppose Alice is thinking about some feature of the real world. To make the graphs easier to draw, this example is a continuous 1d variable. Alice has a probability distribution that looks like this.
![](https://39669.cdn.cke-cs.com/rQvD3VnunX... | https://www.lesswrong.com/posts/9wvJ4MXFvmA5bJpoS/beware-using-words-off-the-probability-distribution-that |
# SGD Understood through Probability Current
*My previous post about SGD was an intro to this model. That post concerned a model of a loss landscape on two "datapoints". In this post I attempt to build a new model of SGD and validate it, with mixed success, but it is sort of interesting.*
Gradient Variance
----------... | https://www.lesswrong.com/posts/n85tApvZndCJx4xFb/sgd-understood-through-probability-current |
# Business Writing Example #1
One of my friends is a saleswoman in Shenzhen who sells lead-acid batteries. She sends out lots of emails and gets few responses. When she does get a response, she often hands it off to someone else in her company who is better at closing sales.
Here is a letter my friend sent to a custo... | https://www.lesswrong.com/posts/BFvCKqmnGDxdztaK2/business-writing-example-1 |
# Business Writing Example #2
After [Business Writing Example #1](https://www.lesswrong.com/posts/BFvCKqmnGDxdztaK2/business-writing-example-1), the customer sent an inquiry to my friend.
> Please check possibilities to my inquiry form attachment. Please confirm prices and lead time too. Please also pay attention on ... | https://www.lesswrong.com/posts/dCsqiyZtsvBLTJojg/business-writing-example-2 |
# Buckets & Bayes
*Author's note: this essay was originally published pseudonymously in 2017. It's now being permanently rehosted at this link. I'll be rehosting a small number of other upper-quintile essays from that era over the coming weeks.*
* * *
I once had a conversation with a couple of the cofounders of the ... | https://www.lesswrong.com/posts/Kbrm4KKbEWWSeBiQE/buckets-and-bayes-1 |
# Law of No Evidence
This is a reference post for the Law of No Evidence.
Scott Alexander did us all a public service this week with his post [The Phrase “No Evidence” Is a Red Flag for Bad Science Communication.](https://astralcodexten.substack.com/p/the-phrase-no-evidence-is-a-red-flag) If you have not yet read it ... | https://www.lesswrong.com/posts/j9HoG56Y6KuopSzdn/law-of-no-evidence |
# Parenting: "Try harder next time" is bad advice for kids too
A post from last year that really stuck with me is Neel Nanda's ["Stop pressing the Try Harder button"](https://www.lesswrong.com/posts/FY6bZAxHTcMi5nhyb/stop-pressing-the-try-harder-button). Key excerpt:
> And every time I thought about the task, I resol... | https://www.lesswrong.com/posts/avRxtmxberoKzcFJJ/parenting-try-harder-next-time-is-bad-advice-for-kids-too |
# A Short Note on My Value Notion
Value of something
------------------
When I have $100 to spend, I can buy ~50 cups of coffee, ~10 regular meals, or a 1-night stay at a hotel, etc. Accordingly it feels like the value of $100 *should be* ~50 cups of coffee, or ~10 meals, or a one-night stay at a hotel. If I can’t pu... | https://www.lesswrong.com/posts/aMAu6nJP7vJedBnST/a-short-note-on-my-value-notion |
# Omicron Paths
While there is still a lot we don't know about Omicron, things are starting to come together. We know it is able to infect people who've previously had covid or been vaccinated, and that it spreads extremely quickly. Combine this with the US having very limited tracking of variants, and what should we ... | https://www.lesswrong.com/posts/dFPk23n8gXC7Ati89/omicron-paths |
# Prizes for last year's 2019 Review
About a year ago, LessWrong held its second annual review, where we looked over the best posts from 2019. The LessWrong team offered $2000 in prizes for the top post authors, and (up to) $2000 in prizes for the best reviews of those posts.
Once [again](https://www.lesswrong.com/po... | https://www.lesswrong.com/posts/i864SgyydDxgBuiJG/prizes-for-last-year-s-2019-review |
# Omicron Post #8
I have fallen mildly ill, as have my wife and son. So far we don’t have a positive Covid-19 test, and everyone is maximally vaccinated, but given the timing the obvious conclusions do seem likely. Wish me well, and hopefully Christmas will not be cancelled and I can keep the posts mostly going. I do ... | https://www.lesswrong.com/posts/ykidxwif4T6w5Akmg/omicron-post-8 |
# Researcher incentives cause smoother progress on benchmarks
_(Epistemic status: likely. That said, this post isn't thorough; I wanted to write quickly.)_
Let's look at [the state of the art in ImageNet][sotaimage].[^meaninglessgarbage]
The curve looks pretty smooth especially for the last 7 years.
However, there d... | https://www.lesswrong.com/posts/3b79GzkPXLfHyxRhv/researcher-incentives-cause-smoother-progress-on-benchmarks |
# Book Launch: The Engines of Cognition
*The Engines of Cognition* is a newly published collection of essays by the LessWrong Community. They're currently available with Prime at [**Amazon US**](https://www.amazon.com/dp/1736128515?), where they'll arrive in time for Christmas, and you can read on to learn more about ... | https://www.lesswrong.com/posts/mvPfao35Moah8py46/book-launch-the-engines-of-cognition |
# Quick Poll: Booster Reactions
Lots of people are getting covid boosters now. To help myself and others plan I did an extremely informal poll on Twitter and Facebook about how people’s booster side effects compared to their second dose. Take home message: **boosters are typically easier than second shots, but they’re... | https://www.lesswrong.com/posts/QHBBeSYqdFdnNcFut/quick-poll-booster-reactions |
# On (Not) Reading Papers
**The emperor's new paper.**
============================
In a recent [lunch conversation](https://universalprior.substack.com/p/applied-mathematical-logic-for-the), a colleague of mine explained her system for reading scientific papers. She calls the system tongue-in-cheek "*not* reading th... | https://www.lesswrong.com/posts/72QA8qk9g6wZNDWeS/on-not-reading-papers |
# Perpetual Dickensian Poverty?
Several of my friends have been sharing screenshots of this Twitter post:
> Time for your annual reminder that, according to *A Christmas Carol*, Bob Cratchit makes 15 shillings a week. Adjusted for inflation, that's $530.27/wk, $27,574/yr, or $13.50/hr.
>
> Most Americans on minimum ... | https://www.lesswrong.com/posts/tJMnF7bgTCCeTEAmy/perpetual-dickensian-poverty |
# What questions do you have about doing work on AI safety?
As part of the work [Nonlinear](http://nonlinear.org/) is doing to try and improve the state of AI safety research, we’re going to be talking with a range of researchers. One of our aims is to help people who are unsure about doing AI safety research figure o... | https://www.lesswrong.com/posts/64GQH24XhzbQfwMHb/what-questions-do-you-have-about-doing-work-on-ai-safety |
# Experiences raising children in shared housing
[Crosspost from The Whole Sky](https://thewholesky.wordpress.com/2021/10/17/experiences-in-raising-children-in-shared-housing/).
Sometimes I see posts about people’s hope to raise children in a group housing situation, and it often seems overly optimistic to me. In pa... | https://www.lesswrong.com/posts/dyJfGeWo5GX2u6NGi/experiences-raising-children-in-shared-housing |
# Confusion about Sequences and Review Sequences
I just made a series of attempted actions on the site that ended up with me being confused. Am I missing ways to use the site, or are these ways missing?
I was reading the page for the [2019 review book launch](https://www.lesswrong.com/posts/mvPfao35Moah8py46/the-engi... | https://www.lesswrong.com/posts/jjCnLWekDvxTQCvAu/confusion-about-sequences-and-review-sequences |
# Two (very different) kinds of donors
This post describes a very simple and very important distinction between two kinds of donors/two kinds of donations.
I apologize if the content of this (short) post is obvious to you. Repeated experience has led me to believe that it is *not* obvious to many people, and can som... | https://www.lesswrong.com/posts/cBaWaJoYSkFQhNj5L/two-very-different-kinds-of-donors |
# Worldbuilding exercise: The Highwayverse.
This is an exercise where I take a world which is otherwise the same as ours except for one physical law being changed, and try to work out what that world would look like.
This is meant to exercise my ability to work out how things are connected to other things - how does ... | https://www.lesswrong.com/posts/fEw4KmqjgAKDWqwYM/worldbuilding-exercise-the-highwayverse |
# Six Specializations Makes You World-Class
> If you want an average, successful life, it doesn't take much planning. Just stay out of trouble, go to school, and apply for jobs you might like. But if you want something extraordinary, you have two rather: 1) Become the best at one specific thing. 2) become very good (t... | https://www.lesswrong.com/posts/7xJyAqPAa93DJbSQz/six-specializations-makes-you-world-class |
# Morality and constrained maximization, part 1
(Cross-posted from [Hands and Cities](https://handsandcities.com/2021/12/21/morality-and-constrained-maximization-part-1/))
Instrumental rationality is about achieving your goals. But morality, famously, sometimes demands that you don’t.
Suppose, for example, that you ... | https://www.lesswrong.com/posts/KoENChyLzfsaojE3Q/morality-and-constrained-maximization-part-1 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.