id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
95cac3db-938b-4339-96d8-b0e08ca16b17 | StampyAI/alignment-research-dataset/blogs | Blogs | Alignment of Language Agents
[](/?source=post_page-----9fbc7dd52c6c--------------------------------)Listen
Share
*By Zachary Kenton, Tom Everitt, Laura Weidinger, Iason Gabriel, Vladimir Mikulik and Geoffrey Irving*... |
92bc7901-9b5f-457d-a3a3-89783c1070fc | trentmkelly/LessWrong-43k | LessWrong | Plan B in AI Safety approach
Central claim: Ambitious alignment of superintelligent AI will not be solved in time, therefore different variants of plan B should be explored and implemented.
Below is a short overview and links to my works on AI safety. The central theme of all them is “plan B”. Thus it could be called... |
bb82b6f5-ae41-4657-9e8f-4dea1888db03 | StampyAI/alignment-research-dataset/special_docs | Other | G.K. Chesterton On AI Risk
\*[An SSC reader working at an Oxford library stumbled across a previously undiscovered manuscript of G.K. Chesterton’s, expressing his thoughts on AI, x-risk, and superintelligence. She was kind enough to send me a copy, which I have faithfully transcribed]\*
![](https://slatestarcodex.co... |
c163c7a0-9ca5-4e21-8593-5686005451e6 | trentmkelly/LessWrong-43k | LessWrong | [Linkpost] "Governance of superintelligence" by OpenAI
OpenAI has a new blog post out titled "Governance of superintelligence" (subtitle: "Now is a good time to start thinking about the governance of superintelligence—future AI systems dramatically more capable than even AGI"), by Sam Altman, Greg Brockman, and Ilya S... |
d1ac4465-349a-4d44-ac6b-30183e791b28 | trentmkelly/LessWrong-43k | LessWrong | Yes, avoiding extinction from AI *is* an urgent priority: a response to Seth Lazar, Jeremy Howard, and Arvind Narayanan.
The following is a post I published on my personal blog in response to criticisms published by Seth Lazar, Jeremy Howard, and Arvind Narayanan of Center for AI Safety's recent statement on AI existe... |
49b22288-2ac1-4191-bb9d-9bd4ec2a3b06 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Response to (Scott Alexander): Preschool: Much More Than You Wanted to Know
Previously (Here): The Case Against Education, The Case Against Education: Foundations, The Case Against Education: Splitting the Education Premium Pie and Considering IQ
I see Scott’s analysis of preschool as burying the lead.
I see his analy... |
370b2cd5-f56a-451e-9a45-389fd025f500 | trentmkelly/LessWrong-43k | LessWrong | Two Neglected Problems in Human-AI Safety
In this post I describe a couple of human-AI safety problems in more detail. These helped motivate my proposed hybrid approach, and I think need to be addressed by other AI safety approaches that currently do not take them into account.
1. How to prevent "aligned" AIs from un... |
1a5eec64-e9da-4edd-97ce-d36f26751d2c | trentmkelly/LessWrong-43k | LessWrong | Open & Welcome Thread - November 2019
* If it’s worth saying, but not worth its own post, here's a place to put it.
* You can also make a shortform post.
* And, if you are new to LessWrong, here's the place to introduce yourself.
* Personal stories, anecdotes, or just general comments on how you found us an... |
470e277f-1cb3-45b8-a1d3-b737d6936f69 | trentmkelly/LessWrong-43k | LessWrong | Fire Law Incentives
Pacific Gas & Electric is planning to spend $15-30b to bury power lines. I see why they're doing it: PG&E equipment sparked some of the worst fires in California history, including the 2018 Camp Fire which destroyed Paradise, but I'm not convinced that this is good for California overall.
Historic... |
ca9a53ec-181a-4e92-9dd4-84795d9e3abe | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Intelligence or Evolution?
Here are two kinds of super high-level explanation for how things turned out in the future:
1. Someone wanted things to turn out that way.
2. Selective forces favoured that outcome.
Which one is a better explanation?
This question is one I’m highly uncertain about, and often shows up as a... |
1b9ccb37-6de8-4a1c-8432-4a22006dff06 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Why does no one care about AI?
It's pretty much generally agreed upon in the EA community that the development of unaligned AGI is the most pressing problem, with some saying that we could have AGI within the next 30 years or so. In The Precipice, Toby Ord estimates the existential risk from unaligned AGI is 1 in 10 o... |
49e2a1a5-7c9d-4942-bfe1-f74dd9e790d1 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "This document describes precautions to take in a scenario like the Cuban Missile Crisis, where nuclear war seems plausibly imminent within the next days or weeks. This is not a guide for what to do if a missile is currently inbound and will strike within minutes or hours.OverviewIf tensions between nuclear powers are ... |
234c1fc4-4ed1-4347-8b85-37a1abbeba29 | trentmkelly/LessWrong-43k | LessWrong | [META] Introduction thread attitudes and discussions
So, we have three introduction threads, and recently have had several contentious ones, like AspiringKnitter's introduction as a theist or Bakkot's introduction as a proponent of infanticide. It seems to me like there's value in trying to come up with a unified poli... |
17575322-51dc-4987-9ff5-4d04f702fda3 | trentmkelly/LessWrong-43k | LessWrong | Why did we wait so long for the bicycle?
h/t alyssavance
> The bicycle, as we know it today, was not invented until the late 1800s. Yet it was a simple mechanical invention. It would seem to require no brilliant inventive insight, and certainly no scientific background.
> Why, then, wasn’t it invented much earlier?
... |
eb8c80ff-35e2-490d-8010-584b2c8451bf | trentmkelly/LessWrong-43k | LessWrong | Meetup : DC Meetup: Games and Conversation
Discussion article for the meetup : DC Meetup: Games and Conversation
WHEN: 21 October 2012 03:00:00PM (-0400)
WHERE: Kogod Courtyard
The next meetup will be at the courtyard by the Portrait Gallery at 3pm on Sunday 10/21, as per usual. The theme will be games and general ... |
9a8e5d24-4fde-4a0a-bd1b-41092f7c0675 | trentmkelly/LessWrong-43k | LessWrong | Meetup : San Antonio Meetup
Discussion article for the meetup : San Antonio Meetup
WHEN: 06 March 2016 02:00:00PM (-0600)
WHERE: 12651 Vance Jackson Rd #118, San Antonio, TX 78230
Bubble tea, frozen yogurt, and discussion at Yumi Berry! All are welcome!
New Meetup to discuss rationality and all things LessWrong a... |
dcb6f221-f763-4203-8608-3b5a3aaf2081 | trentmkelly/LessWrong-43k | LessWrong | Discovering Your Secretly Secret Sensory Experiences
In his recent excellent blog post, Yvain discusses a few "universal" (commonplace) human experiences that many people never notice they don't have, such as the ability to smell, see some colors, see mental pictures, and feel emotions. I was reminded of a longstandi... |
8ecdd332-c2fd-4028-96e1-a1988d6ced0a | trentmkelly/LessWrong-43k | LessWrong | Eight political demands that I hope we can agree on
Scott Alexander wrote in Book Review: History of the Fabian Society about how the Fabian society had eight political thesis that they agreed on.
I want to propose eight points that I think we can agree as a rationality movement about:
1. We want public policy that'... |
d0107e96-1e5d-4a81-b2a4-ea8c91c2cf31 | trentmkelly/LessWrong-43k | LessWrong | Changing the AI race payoff matrix
Suppose that AI capability research is done, but AI safety research is ongoing. Any of the major players can launch an AI at the press of a button to win the cosmos. The longer everyone waits, the lower the chance that the cosmos is paperclips. The default is that someone will press ... |
ad8b86d2-2ad7-40f1-8175-2f4ad748c3d1 | trentmkelly/LessWrong-43k | LessWrong | S-Expressions as a Design Language: A Tool for Deconfusion in Alignment
TL;DR: S-expressions are a minimal structure language for early-stage conceptual engineering (aka deconfusion). They are especially useful in alignment, where the hardest part is often figuring out what the problem even is. S-expressions do not e... |
c8aeb571-a703-4b43-a2a7-c0aba83ea5e1 | trentmkelly/LessWrong-43k | LessWrong | Turns Out Interruptions Are Bad, Who Knew?
I’ve been known to accuse people who say open offices are “fine with a few mitigations” of not paying attention to the cost of their mitigations. I believed they shrunk their thoughts down to the point that not much was lost from an interruption, at the cost of only being abl... |
01422589-b92a-47a2-966e-613995d3c8b5 | trentmkelly/LessWrong-43k | LessWrong | Expected Value- Millionaires Math
I credit reading this post years ago as the first step towards making me instinctively THINK in terms of expected value when considering opportunities. I consider it to be among the best "Why" resources for thinking in expected value, and wish it was more known among the LW crowd. |
9f6ce8ef-a2ba-40fa-9d67-9476e30436f4 | trentmkelly/LessWrong-43k | LessWrong | Faerie Rings
This post describes "Faerie rings", a copiable form of community that forms a building block for capturing the upsides: learning from each other, deeper relationships, and support to solve problems that are beyond one's abilities. I believe it counts as a nascent prototype for an Iterated Trust Kickstart... |
56d70835-bc4f-4049-9542-ace89d26a1e1 | trentmkelly/LessWrong-43k | LessWrong | When does training a model change its goals?
Here are two opposing pictures of how training interacts with deceptive alignment:
1. “goal-survival hypothesis”:[1] When you subject a model to training, it can maintain its original goals regardless of what the training objective is, so long as it follows through on dec... |
20962ffe-6911-4830-a8a3-29d99ffff7ba | trentmkelly/LessWrong-43k | LessWrong | Blink: The Power of Thinking Without Thinking (book review)
Summary
Blink is a book about people making unconscious emotional judgments, and how this isn't always a bad thing. In certain circumstances people subconsciously figure something out, and change their behavior based on this, far sooner than they gain conscio... |
1338bc56-3f1c-4ac1-910c-fca7c8069e7d | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Understanding Selection Theorems
*This post is a distillation of a corpus of ideas on Selection Theorems by* [*johnswentworth*](https://www.lesswrong.com/users/johnswentworth)*, mainly* [*this post*](https://www.lesswrong.com/posts/G2Lne2Fi7Qra5Lbuf/selection-theorems-a-program-for-understanding-agents)*. It was made ... |
9068949f-60cc-4798-b154-b29c995b99f6 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Review of "Fun with +12 OOMs of Compute"
Introduction
============
This review is part of a project with Joe Collman and Jérémy Perret to try to get as close as possible to peer review when giving feedback on the Alignment Forum. Our reasons behind this endeavor are detailed in [our original post](https://www.alignme... |
f55ecdb5-0032-4756-9be1-b4e1716a7c40 | trentmkelly/LessWrong-43k | LessWrong | Why is the impact penalty time-inconsistent?
I showed in a previous post that impact penalties were time-inconsistent. But why is this? There are two obvious possibilities:
1. The impact penalty is inconsistent because it includes an optimisation process over the possible polices of the agent (eg when defining the Q... |
fbe3cda1-eb73-40fd-a125-33c4475b349e | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Degamification
[Goodhart's Law](https://www.lesswrong.com/tag/simulacrum-levels) refers to the tendency that when someone sets a performance metric for a goal, the metric itself becomes a target of optimization, often at the expense of the goal it's supposed to measure. Some metrics are subject to imperfectly-aligned ... |
8f41556d-46e6-4585-a605-deb81e76c1fb | trentmkelly/LessWrong-43k | LessWrong | Branches of rationality
Related to: Go forth and create the art!, A sense that more is possible.
If you talk to any skilled practitioner of an art, they have a sense of the depths beyond their present skill level. This sense is important. To create an art, or to learn one, one must have a sense of the goal.
By co... |
05fb0ae9-f2a7-4d27-a4f4-18ab52b27959 | StampyAI/alignment-research-dataset/distill | Distill Scientific Journal | A Discussion of 'Adversarial Examples Are Not Bugs, They Are Features': Discussion and Author Responses
#rebuttal,
.comment-info {
background-color: hsl(54, 78%, 96%);
border-left: solid hsl(54, 33%, 67%) 1px;
padding: 1em;
color: hsla(0, 0%, 0%, 0.67);
}
#header-info {
margin-top: 0;
margin-bottom: 1.5rem;
... |
a0e975ea-da6e-441b-b182-1bbb0aeb179e | trentmkelly/LessWrong-43k | LessWrong | RFC: Meta-ethical uncertainty in AGI alignment
I'm working on writing a paper about an idea I previously outlined for addressing false positives in AI alignment research. This is the first completed draft of one of the subsections arguing for the adoption of a particular, necessary hinge proposition to reason about al... |
c7d8358d-6dc8-42b3-ae17-2c0a3a39c018 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | I'm interviewing Max Tegmark about AI safety and more. What shouId I ask him?
Next week for The 80,000 Hours Podcast I'm interviewing Max Tegmark — MIT physicist, founder of the Future of Life Institute, instigator of the 2015 Puerto Rico summit on AI safety, and author of 'Life 3.0: Being Human in the Age of Artifici... |
f34ef965-3f7e-4990-88ea-a3d702b4c503 | trentmkelly/LessWrong-43k | LessWrong | Enantiodromia
Oxford Languages defines enantiodromia as “the tendency of things to change into their opposites, especially as a supposed governing principle of natural cycles and of psychological development”. While Plato’s works are so influential that the project of Western philosophy is sometimes described as footn... |
8c623fc0-968a-45aa-9cb1-f8df63cdc4f6 | trentmkelly/LessWrong-43k | LessWrong | I was Wrong, Simulator Theory is Real
[Epistemic Status: Excitedly writing up my new thoughts. I literally just fixed one mistake, so its possible there are others. Not a finalized research product.]
Overview
Fixing a small bug in my recent study dramatically changes the data, and the new data provides significant e... |
bdec61a3-30bd-4580-89d8-e1259e45e95b | trentmkelly/LessWrong-43k | LessWrong | Mailing Binax kits
Is it safe with regard to test efficacy to mail Binax rapid covid test kits (i.e. to a friend) in the winter when the weather is below freezing? The official instructions say to store at above 2 degrees C, but they are on sale on Amazon. Going to visit a friend in another city, being able to mail th... |
7dcfb0be-36c7-491f-856a-6cc5b9ea6130 | StampyAI/alignment-research-dataset/special_docs | Other | How Will National Security Considerations Affect Antitrust Decisions in AI? An Examination of Historical Precedents
Introduction: The Confluence of AI, National Security, and Antitrust Artificial Intelligence (AI)-like past general purpose technologies such as railways, the internet, and electricity-is likely to have ... |
1b505a5b-b9dd-4d8e-907b-16fa2890e6d3 | trentmkelly/LessWrong-43k | LessWrong | Announcing the Alignment of Complex Systems Research Group
tl;dr: We’re a new alignment research group based at Charles University, Prague. If you’re interested in conceptual work on agency and the intersection of complex systems and AI alignment, we want to hear from you. Ideal for those who prefer an academic settin... |
0914e0f3-3f96-4bbb-adbd-ec6a12823142 | trentmkelly/LessWrong-43k | LessWrong | New Q&A by Nick Bostrom
Underground Q&A session with Nick Bostrom (http://www.nickbostrom.com) on existential risks and artificial intelligence with the Oxford Transhumanists (recorded 10 October 2011).
http://www.youtube.com/watch?v=KQeijCRJSog |
9db3a68a-7316-4d43-98bc-e22df2da969e | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | 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 psych... |
c3901ebf-e180-49ba-9579-3d1c5d1cd5f5 | trentmkelly/LessWrong-43k | LessWrong | Evan Hubinger on Inner Alignment, Outer Alignment, and Proposals for Building Safe Advanced AI
It’s well-established in the AI alignment literature what happens when an AI system learns or is given an objective that doesn’t fully capture what we want. Human preferences and values are inevitably left out and the AI, l... |
e8b25ba4-3d1a-47d3-97fc-c5684bbfce8d | trentmkelly/LessWrong-43k | LessWrong | Cost/Benefit analysis of School Closures in the US
TLDR: Covid 19 school closures may do more harm than good, because years of schooling affects overall life expectancy and quality of life for millions of children.
This is my very first post here, so if I apologize if I say something that has been said many times bef... |
6d661209-6a00-4465-ad36-cac21218470d | trentmkelly/LessWrong-43k | LessWrong | Ending British Columbia's anti-cryonics law
[Edit: I did not mean to post this, just save it as a draft (I only remember pressing the 'save and continue' button, not the 'submit' button. That shouldn't've posted it, right?).
Anyway, that's why it dissolves into slightly cryptic point form notes to myself at the end. ... |
70b15cd7-a953-4fbc-aeb2-d08c13e62680 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | A Tentative Timeline of The Near Future (2022-2025) for Self-Accountability
I don’t currently have the energy to set up an account on a prediction market (yes, I am remarkably lazy in some ways), but nonetheless want to have my personal predictions publicly available somewhere. This is primarily intuition-based, and n... |
c97a3d99-d7b0-4154-a9ac-e8925dccc533 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Data Taxation: A Proposal for Slowing Down AGI Progress
*Co-authored by Sammet, Joshua P. S. and Wale, William.*
Foreword
========
*This report was written for the **Policies for slowing down progress towards artificial general intelligence (AGI)** case of the AI governance hackathon organized by Apart Research.*
W... |
5f2b1d11-ddb3-40f1-85f0-6fe999708bc5 | trentmkelly/LessWrong-43k | LessWrong | Will the US have more than 100,000 new daily COVID-19 cases before January 1, 2022?
I am trying to work through a process where I can assess what the probability of this question being a yes is.
I am going to use this post as an information dump in order to be able to rationally come up with the best probability I ca... |
82dd5b54-1c17-470f-b2e9-5e256e72589e | StampyAI/alignment-research-dataset/arxiv | Arxiv | Using Causal Analysis to Learn Specifications from Task Demonstrations
1. Introduction
----------------
As we move from robots dedicated to specific pre-programmed tasks to more general purpose tasks, there is a need for easy re-programmability of these robots. A promising approach to such easy re-programming is Le... |
094dbfe4-8984-4fe9-81c0-d8e5c29d4b6e | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | 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 po... |
7d472962-7bc3-4c14-a28c-ad6f1b87ec19 | trentmkelly/LessWrong-43k | LessWrong | Outreach Thread
Based on an earlier suggestion, here's an outreach thread where you can leave comments about any recent outreach that you have done to convey rationality-style ideas broadly. The goal of having this thread is to organize information about outreach and provide community support and recognition for raisi... |
4a14d190-570a-4d13-8c09-e2a6e4306aa3 | trentmkelly/LessWrong-43k | LessWrong | Anthropic paradoxes transposed into Anthropic Decision Theory
Anthropic Decision Theory (ADT) replaces anthropic probabilities (SIA and SSA) with a decision theory that doesn't need anthropic probabilities to function. And, roughly speaking, ADT shows that total utilitarians will have a behaviour that looks as if it w... |
16eb1e76-9db0-4b3b-93fc-fac6a7869102 | trentmkelly/LessWrong-43k | LessWrong | Group Rationality Diary, June 16-30
This is the public group instrumental rationality diary for June 16-30.
> It's a place to record and chat about it if you have done, or are actively doing, things like:
>
> * Established a useful new habit
> * Obtained new evidence that made you change your mind about some beli... |
8a66e685-94e0-45f6-aa97-628c8a718608 | trentmkelly/LessWrong-43k | LessWrong | How could a friendly AI deal with humans trying to sabotage it? (like how present day internet trolls introduce such problems)
After close observation of behaviours online and in person, for a few years, it becomes ever clearer in my mind that a significant fraction (>1%) of the regular adult population would intentio... |
abb9c9e4-939c-483b-b43f-1f7e17043e75 | trentmkelly/LessWrong-43k | LessWrong | Stop pushing the bus
With the recent proposals about moratoriums and regulation, should we also start thinking about a strike by AI researchers and developers?
The reasoning I imagine as follows. AI capability is now growing really fast, and toward levels that will strongly affect the world. And AI safety lags behind... |
b1187c6c-6f2a-42df-ac06-0b5cc51182c8 | trentmkelly/LessWrong-43k | LessWrong | What ML gears do you like?
In John's recent post he mentions many people in ML not having good gears level models of what's going on.
To wit; what gears-level models do you know for ML? How much support is there for them? Are there "settled science" kind models that have tons of empirical support?
What gears-level m... |
ca4fe2e7-0610-4f99-a3fd-6e30ee0cc564 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | [Link] OpenAI: Learning to Play Minecraft with Video PreTraining (VPT)
<https://openai.com/blog/vpt/>
> We trained a neural network to play Minecraft by Video PreTraining (VPT) on a massive unlabeled video dataset of human Minecraft play, while using only a small amount of labeled contractor data. With fine-tuning, ... |
efd32a3c-c3b1-4324-a3a9-b7669ab048c9 | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] Formative Youth
Today's post, Formative Youth was originally published on 24 February 2009. A summary (taken from the LW wiki):
> People underestimate the extent to which their own beliefs and attitudes are influenced by their experiences as a child.
Discuss the post here (rather than in the comments... |
8563094c-8a08-478d-8ced-28890c11c0c4 | trentmkelly/LessWrong-43k | LessWrong | Unpopular ideas attract poor advocates: Be charitable
Unfamiliar or unpopular ideas will tend to reach you via proponents who:
* ...hold extreme interpretations of these ideas.
* ...have unpleasant social characteristics.
* ...generally come across as cranks.
The basic idea: It's unpleasant to promote ideas that... |
eb9da559-5f1e-4631-9ca5-dd686b376082 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "Originally posted at sandymaguire.me. I have to get you to drop modesty and say to yourself, "Yes, I would like to
do first-class work." Our society frowns on people who set out to do really
good work. You're not supposed to; luck is supposed to descend on you and you
do great things by chance. Well, that's a kind of ... |
4df0a577-40e4-4e56-9d39-12468dba8830 | trentmkelly/LessWrong-43k | LessWrong | A Different Kind of Ark: My failed attempt to build a bridge between universes
In the summer of 2020, I launched a website called The Vessel Project. It’s a weird, winding, conjecturing exploration into the nature of reality. Some would call it crazy. I don’t necessarily disagree. But the story behind its inception is... |
cad1ef8b-07a1-4ad7-b735-221a9edfc7d5 | trentmkelly/LessWrong-43k | LessWrong | What's the most easy, fast, efficient way to create and maintain a personal Blog?
I have seen a lot of members here have personal blogs and encourage creating anti-Ra content. Although I have some experience on creating python programs/android apps, I don't know much about web development. I started watching some html... |
b6890c39-7c1b-4330-81c6-03b630b09839 | trentmkelly/LessWrong-43k | LessWrong | Great rationality posts by LWers not posted to LW
Ever since Eliezer, Yvain, and myself stopped posting regularly, LW's front page has mostly been populated by meta posts. (The Discussion section is still abuzz with interesting content, though, including original research.)
Luckily, many LWers are posting potentially... |
2eb1a42b-79ad-4ffc-99b5-f6c733e2f5d8 | trentmkelly/LessWrong-43k | LessWrong | Being Likened to a Trolley Problem
(Epistemic status: Contains use of the trolley problem, eventually / the whole time, so what do you expect?)
(I'm aware that there are a lot of good, practical reasons to try and figure out whether someone likes you, and that many of these inform the weird extra-importance we place ... |
a7ed2045-2be6-47a7-b3aa-a9ca0c273a3d | trentmkelly/LessWrong-43k | LessWrong | Stupid questions thread, October 2015
This thread is for asking any questions that might seem obvious, tangential, silly or what-have-you. Don't be shy, everyone has holes in their knowledge, though the fewer and the smaller we can make them, the better.
Please be respectful of other people's admitting ignorance and ... |
9df95cf0-6982-4b05-9daa-87b42a775cfb | trentmkelly/LessWrong-43k | LessWrong | [LINK] Amanda Knox exonerated
Here are the New York Times, CNN, and NBC. Here is Wikipedia for background.
The case has made several appearances on LessWrong; examples include:
* You Be the Jury: Survey on a Current Event (December 2009)
* The Amanda Knox Test: How an Hour Beats a Year in the Courtroom (December... |
b9f0fc2d-fc41-4ede-b446-f7e179cb04eb | StampyAI/alignment-research-dataset/arxiv | Arxiv | RL with KL penalties is better viewed as Bayesian inference
1 Introduction
---------------
Large language models (LMs), such as GPT-3 Brown et al. ([2020](#bib.bib46 "Language models are few-shot learners")), tend to generate outputs that reflect undesirable features of their training data such as offensiveness (Ge... |
c6f0866d-4748-4b57-802c-0930dc91cef9 | trentmkelly/LessWrong-43k | LessWrong | Welcome to Less Wrong! (9th thread, May 2016)
Hi, do you read the LessWrong website, but haven't commented yet (or not very much)? Are you a bit scared of the harsh community, or do you feel that questions which are new and interesting for you could be old and boring for the older members?
This is the place for the ... |
89832bd3-d755-457f-8c76-882a9d44047a | trentmkelly/LessWrong-43k | LessWrong | Cambridge LW: Post-mortem
A brief post-mortem of this weeks LW meet-up.
Topic Selection
This week we had comparatively few people, so what worked this time may not generalise. At the beginning of the meet-up Douglas suggested we use one of his meetup formats. Of the three, skill focus was not usable without prepa... |
c8879173-1544-4cd3-9d71-1b524a059a11 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Games in Kocherga club: FallacyMania, Tower of Chaos, Training game
Discussion article for the meetup : Games in Kocherga club: FallacyMania, Tower of Chaos, Training game
WHEN: 23 March 2016 07:40:00PM (+0300)
WHERE: Moscow, B.Dorogomilovskaya, 5-2
Welcome to Moscow LW community makeshift games! In that g... |
cab4c8bd-5a5a-462f-8b1e-c0835f8d84d3 | StampyAI/alignment-research-dataset/blogs | Blogs | October 2021 Newsletter
Redwood Research is a new alignment research organization that just launched their [website](https://www.redwoodresearch.org/) and released an explainer about [what they're currently working on](https://www.alignmentforum.org/posts/k7oxdbNaGATZbtEg3/redwood-research-s-current-project). We're qu... |
390110b7-7c1a-4c84-8771-37084631a14e | trentmkelly/LessWrong-43k | LessWrong | Book summary: Unlocking the Emotional Brain
If the thesis in Unlocking the Emotional Brain (UtEB) is even half-right, it may be one of the most important books that I have read. Written by the psychotherapists Bruce Ecker, Robin Ticic and Laurel Hulley, it claims to offer a neuroscience-grounded, comprehensive model o... |
55142ed5-c558-4c58-9188-955f98a26c2e | trentmkelly/LessWrong-43k | LessWrong | Things That Make Me Enjoy Giving Career Advice
Introduction
I spend a bunch of my time having various kinds of mentoring conversations with people. I generally enjoy doing this, but some things make these conversations much more fun for me than others! This post is an attempt to compile some of these.
Obviously, writ... |
038823f1-5034-4af4-addf-7a87f887bec8 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | Governing Transformative Artificial Intelligence by Markus Anderljung
great great seen you all today so I'm
here to talk about sort of some slightly
more big-picture issues then you will
have heard about so far and so I'll be
talking about basically the issue of how
do we make sure that artificial
intelligence in the ... |
1ed9acb6-8eae-4f74-9be4-76b8ba185cf4 | trentmkelly/LessWrong-43k | LessWrong | What are the best ways of absorbing, and maintaining, knowledge?
Recently, I've collapsed (ascended?) down/up a meta-learning death spiral -- doing a lot less of reading actual informative content, than figuring out how to manage and acquire such content (as well as completely ignoring the antidote). In other words, I... |
5e50c550-64d3-4c65-a228-b774915e6c8e | StampyAI/alignment-research-dataset/arxiv | Arxiv | Hidden Incentives for Auto-Induced Distributional Shift
1 Introduction
---------------
Consider a content recommendation system whose performance is measured by accuracy of predicting what users will click. This system can achieve better performance by either 1) making better predictions, or 2) changing the distrib... |
07841103-6000-4d0b-aed2-905702c440bf | trentmkelly/LessWrong-43k | LessWrong | Modeling humanity's robustness to GCRs?
Cross-posted to the EA-Forum
Epistemic Status: I have spent less than 15 minutes thinking about / querying the Internet for answers on what I am going to ask here. As such, it might be the case that this idea has already been discarded due to implementation or theoretical diff... |
2c134269-f924-4237-bc0a-93741d1505f3 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | dalle2 comments
**(i.)**
========
On April 6, OpenAI announced “DALLE-2” to great fanfare.
There are two different ways that OpenAI talks to the public about its models: as research, and as marketable products.
These two ways of speaking are utterly distinct, and seemingly immiscible. The OpenAI blog contains vario... |
13b7ae26-3f23-43a8-8a07-53b25dee4754 | trentmkelly/LessWrong-43k | LessWrong | Should I Pursue a PhD?
Introduction
I'm currently flirting with the idea of trying for a math PhD 2 - 3 years down the line.
I'm currently on a Theoretical Computer Science Masters program at the University of [Redacted] in the United Kingdom.
(My program is 2 semesters of teaching (7 - 8 months) followed by a 9 - 1... |
432a9c83-8102-4aab-8908-ff108222b5ad | trentmkelly/LessWrong-43k | LessWrong | Chapter 77: SA, Aftermaths: Surface Appearances
Aftermath: Albus Dumbledore and -
The old wizard sat alone at his desk, in the unsilence of the Headmaster's office, amid the innumerable and unnoticed devices; his robes a gentle yellow, of soft fabric, not such clothing as he ordinarily wore before others. His wrinkle... |
a3891bb3-ecac-49ea-9326-cc43196a2f68 | trentmkelly/LessWrong-43k | LessWrong | [LDSL#5] Comparison and magnitude/diminishment
This post is also available on my Substack.
In the previous post on LDSL, I argued that an important part of causal inference is to look at the larger things. This implies a method of comparison to rank things by magnitude. [1]
Homogenous comparison
When the things to ... |
66ed3000-2dad-4f91-b868-9d6587bd2992 | trentmkelly/LessWrong-43k | LessWrong | A mostly critical review of infra-Bayesianism
Introduction
I wrote this post towards the end of my three and a half months long SERI MATS fellowship. I didn't get even close to the point where I could say that I understand infra-Bayesianism on a really detailed level (according to Vanessa there are only three people ... |
a1ee2134-727b-493d-b741-15f3b4011843 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Strongest real-world examples supporting AI risk claims?
[Manually cross-posted to LessWrong [here](https://www.lesswrong.com/posts/h66WSLKdShqzNsZL9/strongest-real-world-examples-supporting-ai-risk-claims).]
There are some great collections of examples of things like [specification gaming](https://docs.google.com/sp... |
805b7b40-ca0a-4f2b-bc88-0aa5eeb2c0d1 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Can't Unbirth a Child
**Followup to**: [Nonsentient Optimizers](/lw/x5/nonsentient_optimizers/)
Why would you want to *avoid* creating a sentient AI? "*Several reasons,*" [I said](/lw/x5/nonsentient_optimizers/). "Picking the simplest to explain first—I'm not ready to be a father."
So here is the *strongest* re... |
dc5a82ef-c4e9-4d24-bba1-ec251715a090 | trentmkelly/LessWrong-43k | LessWrong | Deontology and Tool AI
Please forgive me for being awkward on my first post, I just wanted to get these ideas out there to facilitate discussion on the topic. I also am sorry if I use any newly-discovered terminology incorrectly, but I am doing my best.
Introduction: Utilitarianism is Unaligned
Intuitively, I think ... |
aab3e5eb-2b8e-4632-a8ce-9ab1c22a1110 | trentmkelly/LessWrong-43k | LessWrong | «Boundaries» Sequence (Index Post)
In an upcoming short sequence of posts, I'm going to try to describe a causal pathway from
1. a key missing idea in the utility-theoretic foundations of game theory, leading to
2. some problems I think I see in effective altruism discourse, leading further to
3. gaps in some app... |
b24b9f7e-3e7d-4c46-9dc5-1feee4876153 | trentmkelly/LessWrong-43k | LessWrong | Inner Goodness
Followup to: Which Parts Are "Me"?, Effortless Technique
A recent conversation with Michael Vassar touched on - or to be more accurate, he patiently explained to me - the psychology of at least three (3) different types of people known to him, who are evil and think of themselves as "evil". In ascend... |
31033a16-5f0e-47b0-85b4-145a6f33934d | trentmkelly/LessWrong-43k | LessWrong | How would a person go about starting a geoengineering startup?
None |
dc7eb9a7-d802-4785-97ba-b91926edabd2 | trentmkelly/LessWrong-43k | LessWrong | Can I learn language faster? Or, perhaps, can I memorize the foreign words and recall them faster?
I've been learning Korean for a while now. I am finding a rather asymmetric aspect to my vocabulary learning process and results. I use Anki as the learning tool for vocabulary. What I find is that fairly quickly I get t... |
f6caef56-5666-4e17-bd10-bc53c7d89f87 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | What did AI Safety’s specific funding of AGI R&D labs lead to?
Funding I found after some googling:
* Tallinn's (and Musk’s) seed investments in DeepMind¹
* OpenPhil's $30M grant to OpenAI²
* FTX's $500M³, Tallinn's, Moskovitz’(and Schmidt’s)⁴ investments in Anthropic
I’m curious how you consider the consequences ... |
58a09529-840f-400e-b47f-eea3a7cb6b58 | trentmkelly/LessWrong-43k | LessWrong | Quantifying Risk
Epistemic Status: Providing a jumping-off point for further discussion, not intended to be a final authority by any means.
With COVID, a lot of people in the rationality community have had to fundamentally reassess what kind of risks they're willing to handle. We've used microCOVID extensively, tried... |
ad2f5b3f-b040-40e9-9115-2e256a5cc479 | StampyAI/alignment-research-dataset/arbital | Arbital | Generalized principle of cognitive alignment
A generalization of the [https://arbital.com/p/7g0](https://arbital.com/p/7g0) is that whenever we are asking how we want an AI algorithm to execute with respect to some alignment or safety issue, we might ask how we ourselves are thinking about that problem, and whether we... |
f9728180-2c38-4e32-8131-58b450db9561 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Ideas for studies on AGI risk
[This post from a few days ago](https://www.lesswrong.com/posts/8vGFoqcQdkk7n3vDz/where-s-the-foom) brought up the lack of preeminent AI papers on the topic of AI risk. The lack of such peer-reviewed studies is often brought up by critics as a way to dismiss the worries, so I'd like to co... |
96216a44-acbb-4ff5-8464-8870f5e80c6a | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | Roger Grosse | How can deep learning research inform long-term AI safety?
hi everyone
so great to see people again if i'm out
in scotland or maybe not in scotland
benson good to see you great to see
everyone
it's uh it's really great to have
everybody here today uh this is our
first seminar of the
uh of the new year a... |
a48af432-96bc-46c8-a8b4-620cf045fd94 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Contrast Pairs Drive the Empirical Performance of Contrast Consistent Search (CCS)
tl;dr
=====
[Contrast consistent search (CCS)](https://arxiv.org/abs/2212.03827)[[1]](#fnwnplf5vji3o) is a method by Burns *et al.* that consists of two parts:
1. Generate contrast pairs by adding pseudolabels to an unlabelled dataset... |
5b366bd5-b337-4664-ab07-9d78e9f7aa7b | trentmkelly/LessWrong-43k | LessWrong | My guide to lifelogging
I've defended the practice of lifelogging as a means of life extension here. In this post I'll provide a fairly comprehensive guide on how to lifelog. Since lifelogging exists on a spectrum from "taking a picture every so often" to "recording every single detail of your life, in uncompressed HD... |
45526f34-af3b-46e4-a577-b47a1f692908 | trentmkelly/LessWrong-43k | LessWrong | Against easy superintelligence: the unforeseen friction argument
In 1932, Stanley Baldwin, prime minister of the largest empire the world had ever seen, proclaimed that "The bomber will always get through". Backed up by most of the professional military opinion of the time, by the experience of the first world war, an... |
140a8718-950e-4ae1-900f-f998648e3865 | trentmkelly/LessWrong-43k | LessWrong | Lucid dreaming technique and study
|
bdd52d2d-2b2f-4891-b110-47956ebd3c33 | StampyAI/alignment-research-dataset/blogs | Blogs | unoptimal superintelligence doesn't lose
unoptimal superintelligence doesn't lose
----------------------------------------
i [previously](unoptimal-superint-loses.html) wrote a post about how a superintelligence with an unoptimal decision system likely loses to alien superintelligences that are more optimal, at the ... |
ec646e76-2c5b-42d9-af66-0d237814c7ab | StampyAI/alignment-research-dataset/arxiv | Arxiv | Truthful AI: Developing and governing AI that does not lie
Executive Summary & Overview
----------------------------
#### The threat of automated, scalable, personalised lying
Today, lying is a human problem. AI-produced text or speech is relatively rare, and is not trusted to reliably convey crucial information... |
bec9253e-3705-44e2-b1b6-d9c11af80463 | trentmkelly/LessWrong-43k | LessWrong | AXRP Episode 44 - Peter Salib on AI Rights for Human Safety
YouTube link
In this episode, I talk with Peter Salib about his paper “AI Rights for Human Safety”, arguing that giving AIs the right to contract, hold property, and sue people will reduce the risk of their trying to attack humanity and take over. He also te... |
4443ab58-c65e-4742-8487-20996d2d0ad9 | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | Attention - General - Theory, Info-Weighted Patterns, Attribution Patterns [rough early thoughts]
in our last couple of videos
we've studied and tried to reverse
engineer some really simplified versions
of a transformer
and the methods we used um really relied
on some of those simplifications
especially the fact that ... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.