id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
ddb96906-331a-4098-aa82-aefb92919d9b
trentmkelly/LessWrong-43k
LessWrong
Technical Predictions Related to AI Safety If you want to create a policy of nuclear safety then you need to understand: * Manufacturing nuclear weapons requires access to uranium. * Uranium enrichment is a big, slow activity that's difficult to hide from enemy nation-states. * Nuclear weapons are so expensive onl...
7ef7f822-6df8-4d3b-937c-556edc3faa4e
trentmkelly/LessWrong-43k
LessWrong
Searching for post on Community Takeover I am searching for an old LessWrong post about something like "Community takeover": The idea that a successful community (Rationalist/AI Risk) could attract people from outside, who would attempt to co-opt/hijack the agenda to focus on something else. This is inspired by the s...
d0131424-8f5e-4e99-b276-a9a42e0dd6f0
trentmkelly/LessWrong-43k
LessWrong
Prokaryote Multiverse. An argument that potential simulators do not have significantly more complex physics than ours Definitions "Universe" can no longer be said to mean "everything", such a definition wouldn't be able to explain the existence of the word "multiverse". I define universe as a region of existence that,...
aa71c08d-532c-49c6-bce2-8f6e700ee5ed
StampyAI/alignment-research-dataset/lesswrong
LessWrong
What an actually pessimistic containment strategy looks like Israel as a nation state has an ongoing national security issue involving Iran.  For the last twenty years or so, Iran has been covertly developing nuclear weapons. Iran is a country with a very low opinion of Israel and is generally diplomatically opposed ...
c244a51c-b409-4a62-81c6-f05baa19e08c
trentmkelly/LessWrong-43k
LessWrong
[Fun][Link] Alignment SMBC Comic Alt text: "We just have to create new bigotries faster than they can learn to live in harmony." Bonus panel: The author seems to allow sharing (there are multiple sharing buttons at the bottom). Here is the full link to the comic again: https://www.smbc-comics.com/comic/align  Here ...
674d49de-4c15-416b-b5b0-aed71b1c0388
trentmkelly/LessWrong-43k
LessWrong
Covid 1/13/22: Endgame Important note, fleshed out a bit more in Omicron Post #14: From various sources, I have become convinced that rapid tests taken from nose swabs are likely to often be several days slower at detecting infections than rapid tests that use throat swabs. So if you are testing with the goal of actu...
fd7b7d48-5c9c-4d56-8caa-47ca60aa38ea
trentmkelly/LessWrong-43k
LessWrong
Is a near-term, self-sustaining Mars colony impossible? Some argue that Elon Musk's plan to build a self-sustaining colony on Mars is effectively impossible for the foreseeable future without discontinuous tech progress. (See e.g. Jess Riedel's claim here, which Robin Hanson agrees with.) Is there a simple explanatio...
77c987a0-39fe-4f08-ba9f-7f7e5738f612
trentmkelly/LessWrong-43k
LessWrong
Reminder: London meetup, Sunday 2pm, near Holborn Reminder: London meetup on Sunday May 1 at 14:00 at the Shakespeares Head (official page) on Kingsway near Holborn Tube station. Note that there's more than one pub in London with that name, so make sure you get the right one.  As always, we'll have a big picture of a ...
7989abbf-1acc-41db-8b75-8d0490300484
trentmkelly/LessWrong-43k
LessWrong
Research agenda: Supervising AIs improving AIs [This post summarizes some of the work done by Owen Dudney, Roman Engeler and myself (Quintin Pope) as part of the SERI MATS shard theory stream.] TL;DR Future prosaic AIs will likely shape their own development or that of successor AIs. We're trying to make sure they d...
dee1d702-2153-4616-bc97-85e53bc56e15
StampyAI/alignment-research-dataset/arxiv
Arxiv
Discriminator Rejection Sampling 1 Introduction --------------- Generative Adversarial Networks (GANs) (Goodfellow et al., [2014](#bib.bib6)) are a powerful tool for image synthesis. They have also been applied successfully to semi-supervised and unsupervised learning (Springenberg, [2015](#bib.bib25); Odena, [2016...
6dabd756-01a9-44dd-a2f5-51d873161411
trentmkelly/LessWrong-43k
LessWrong
Another Argument Against Eliezer's Meta-Ethics I think I've found a better argument that Eliezer's meta-ethics is wrong. The advantage of this argument is that it doesn't depend on the specifics of Eliezer's notions of extrapolation or coherence. Eliezer says that when he uses words like "moral", "right", and "should...
9c285751-6986-48ba-a159-11df76c44ea9
trentmkelly/LessWrong-43k
LessWrong
First Strike and Second Strike First-strike weapons destroy your enemy's ability to retaliate. Second-strike weapons preserve your ability to retaliate after a first strike. Missile defense systems constitute first-strike technology. Suppose you build an effective[1] missile defense system. You could launch a pre-emp...
efea6a7e-7804-4d36-a40c-510060857056
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Swimming Upstream: A Case Study in Instrumental Rationality > One data point for careful planning, the unapologetic pursuit of fulfillment, and success. Of particular interest to up-and-coming AI safety researchers, this post chronicles how I made a change in my PhD program to work more directly on AI safety, overcomi...
8fe8b0b0-e32b-4959-bf47-b1d3fc789eed
trentmkelly/LessWrong-43k
LessWrong
A Developmental Framework for Rationality [A way of looking at rationality as a set of transitioning worldviews that operate under different assumptions. The short summary of progression looks like techniques → habits → introspection → feelings → multiple ontologies. Skills which appear to be at odds with one another ...
1ed97984-ae80-47a2-bf8f-48b8c5aa246d
trentmkelly/LessWrong-43k
LessWrong
Meetup : San Francisco Meetup: Stories Discussion article for the meetup : San Francisco Meetup: Stories WHEN: 24 October 2016 06:15:00PM (-0700) WHERE: 1769 15th St, San Francisco, CA We'll be meeting to tell each other stories! Ideally, come with a story from your life (doesn't have to be about anything in partic...
926c9c96-16be-41df-93b7-7a5bc9070a77
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
The Inter-Agent Facet of AI Alignment *We argue that* [*intent alignment and capability robustness*](https://www.alignmentforum.org/posts/SzecSPYxqRa5GCaSF/clarifying-inner-alignment-terminology) *do not capture the full “chain of command” (i.e., “alignment” without any qualifier) from the programmer's intentions to a...
43c6edf8-da7d-418f-b845-fb7bd9b0b96a
trentmkelly/LessWrong-43k
LessWrong
Hyderabad Meetup : July 8, 2012 There has been no Lesswrong meetup in India in 2012. I suggest having a meetup in the afternoon on Sunday, July 8 at Cafe Coffee Day, NEXT TO Airtel Showroom, Madhapur Main Road, Madhapur, Hyd (Landmark: OPPOSITE Krishe Sapphire) Me and couple of my friends commit ourselves to be th...
f889f989-e45e-49fc-bda5-9a277fb9f19a
trentmkelly/LessWrong-43k
LessWrong
Four mindset disagreements behind existential risk disagreements in ML I've previously written that good ideas and conversations about AGI seem to have propagated through ML weirdly slowly. A different weird phenomenon I observe is that the field's relative inaction about AGI seems less based on a confident set of be...
a183ab5f-e5e2-46f0-b893-905814bedcba
trentmkelly/LessWrong-43k
LessWrong
To Predict What Happens, Ask What Happens When predicting conditional probability of catastrophe from loss of human control over AGI, there are many distinct cruxes. This essay does not attempt a complete case, or the most generally convincing case, or addressing the most common cruxes. Instead these are my best gues...
a2ba874f-ed7d-44b9-97a5-4b3fc8ef4a80
trentmkelly/LessWrong-43k
LessWrong
"Thank you for updating" Would it be supercilious to thank someone for updating?  I know I would feel uncomfortable doing it, but I often feel the urge to do so anyway. There seems to be something vicious about thanking them.  My own estimation of my own belief has not changed in these situations.  I feel fairly sati...
4e3d6f60-0f9b-4808-8a88-78a15a1d1c87
trentmkelly/LessWrong-43k
LessWrong
bumping into financial unknown unknowns subtitle: "I think there are a few black boxes in my professional life as well." Current problem: my finances are a little tight these days. How I imagine personal finances is kind of like a stream. I need as much water coming in as going out. I'd like to create more value fo...
049f54bc-6e25-4edc-8c75-cabf6f81a139
trentmkelly/LessWrong-43k
LessWrong
We Should Prepare for a Larger Representation of Academia in AI Safety Epistemic Status: I had the idea for the post a few days ago and quickly wrote it down while on a train. I'm very curious about other perspectives. TL;DR: The recent increased public interest in AI Safety will likely lead to more funding for and m...
6c14f88a-58ec-46b9-b6bb-4d3d71ab04b5
trentmkelly/LessWrong-43k
LessWrong
AI #122: Paying The Market Price If you are Meta, and you want to attract top AI talent, you have a problem, because no one wants to work for you or on your products. So it is going to cost you. Mark Zuckerberg has decided he will pay what it takes to get at least some very good talent. If you are the rest of us, esp...
87994be4-9cd2-4846-8a0c-674deb50c605
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Escaping the Löbian Obstacle Earlier this year, when looking for an inroad to AI safety, I learned about the Löbian Obstacle, which is a problem encountered by 'purely logical' agents when trying to reason about and trust one another. In the original paper of Yudkowsky and Herreshoff [1], they show that a consequence ...
838c2daf-997b-4945-b1d7-5439d04f81bc
trentmkelly/LessWrong-43k
LessWrong
Are funds (such as the Long-Term Future Fund) willing to give extra money to AI safety researchers to balance for the opportunity cost of taking an "industry" job? I do not feel like I am ready to apply for funding yet, but I was wondering are funds (such as Long-Term Future Fund) willing to give extra money to AI saf...
36164c2c-6476-43e8-962c-dd1335521bed
trentmkelly/LessWrong-43k
LessWrong
How to Price a Futures Contract All investments are risky. The techniques explained in this series are exceptionally risky. I am not a registered investor, attorney, advisor, broker, banker, lawyer, dealer or anything of the sort. All opinions expressed here are for my personal exploration. I present it to you for ent...
432b4840-f3ff-4aec-a6bd-4028b9916556
trentmkelly/LessWrong-43k
LessWrong
The Roots of Progress events in Austin, November 4–6 I’ll be in Austin, TX in November for a few events: Talk on nuclear power “The Energy of Tomorrow: The Promise, Failure, and Possible Rebirth of Nuclear Power”: > In the 1950s, nuclear power was seen as the energy of the future. Today, it is stagnating on the sid...
20118e1d-0d80-4bfb-a9da-bafe802d4573
StampyAI/alignment-research-dataset/arbital
Arbital
William Frankena's list of terminal values "Life, consciousness, and activity; health and strength; pleasures and satisfactions of all or certain kinds; happiness, beatitude, contentment, etc.; truth; knowledge and true opinions of various kinds, understanding, wisdom; beauty, harmony, proportion in objects contemplat...
f1bbb07a-0b64-4f4d-8019-29883f9a9331
trentmkelly/LessWrong-43k
LessWrong
(Yet Another) Map for AI Risk Discussion Full Size I made this map for myself, but it quickly turned into a tool which I use to help find productive discussion - for example, in steering towards a common understanding of key divergences. I also find that it helps me introduce the topic to friends who are not aware o...
01b0ec11-debf-498d-a0bc-12748b4150ce
trentmkelly/LessWrong-43k
LessWrong
Schelling Shifts During AI Self-Modification Introduction In this essay, I name and describe a mechanism that might break alignment in any system that features an originally aligned AI improving its capabilities through self-modification. I doubt that this idea is new, but I haven't yet seen it named and described in ...
a7504bc6-e21a-41c6-b433-8b62b1a5c2c8
trentmkelly/LessWrong-43k
LessWrong
Is my perception of reality bounded by my intelligence? Hello all, Here I am living in the Netherlands, 2019 AD. In my perception there is no two-way explicit communication with non-human intelligence, no space travel. Artifical intelligence is in it's infancy (at least, as seen by the general public). All my evidenc...
531b5022-b716-4398-b931-64ba48585908
trentmkelly/LessWrong-43k
LessWrong
Investing for the Long Slump I have no crystal ball with which to predict the Future, a confession that comes as a surprise to some journalists who interview me.  Still less do I think I have the ability to out-predict markets.  On every occasion when I've considered betting against a prediction market - most recently...
efd61224-faa3-47ab-a555-4467f4d4882b
trentmkelly/LessWrong-43k
LessWrong
Demands for Particular Proof: Appendices Appendices to:  You're Entitled to Arguments, But Not (That Particular) Proof (The main article was getting long, so I decided to move the appendices to a separate article which wouldn't be promoted, thus minimizing the size of the article landing in a promoted-article-only-re...
b6b66a19-cea4-4e6f-9614-f3b1b82e14c1
trentmkelly/LessWrong-43k
LessWrong
[Link] On the Height of a Field Mark Eichenlaub posted a great little case-study about the difficulty of updating beliefs, even over trivial matters like the slope of a baseball field. The basic story of Bayes-updating assumes the likelihood of evidence in different states is obvious, but feedback between observations...
63d2835f-5aa9-479f-bc62-190c025b6e7f
trentmkelly/LessWrong-43k
LessWrong
Autopoietic systems and difficulty of AGI alignment I have recently come to the opinion that AGI alignment is probably extremely hard. But it's not clear exactly what AGI or AGI alignment are. And there are some forms of aligment of "AI" systems that are easy. Here I operationalize "AGI" and "AGI alignment" in some di...
9397fde7-ac57-4edc-a8d8-a9b3410a13c8
StampyAI/alignment-research-dataset/blogs
Blogs
Methodology for discontinuous progress investigation AI Impacts’ [discontinuous progress investigation](https://aiimpacts.org/discontinuous-progress-investigation/) was conducted according to methodology outlined on this page. Details ------- Contributions to the [discontinuous progress investigation](https://aiim...
4fe4f715-12e7-4266-9a3c-51828aa69069
trentmkelly/LessWrong-43k
LessWrong
The Most Important Century: The Animation This is a linkpost for the Rational Animations' video based on The Most Important Century sequence of posts by Holden Karnofsky. Below, the whole script of the video. Matthew Barnett has written most of it. Several paragraphs were written by Ajeya Cotra, during the feedback ...
8ab18984-4e11-495a-bde2-3a6b02acc26a
trentmkelly/LessWrong-43k
LessWrong
Is Molecular Nanotechnology "Scientific"? Prerequisite / Read this first:  Scientific Evidence, Legal Evidence, Rational Evidence Consider the statement "It is physically possible to construct diamondoid nanomachines which repair biological cells."  Some people will tell you that molecular nanotechnology is "pseudosc...
18290745-9929-4405-8f65-f6ea72c677bc
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Reflections on the PIBBSS Fellowship 2022 Last summer, we ran the first iteration of the PIBBSS Summer Research Fellowship. In this post, we share some reflections on how the program went. Note that this post deals mostly with high-level reflections and isn’t maximally comprehensive. It primarily focusses on informat...
16370f16-4409-4841-a2ab-ae59227fb6d0
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
It’s not obvious that getting dangerous AI later is better Intro ===== This post was originally intended as a follow-up post to Josh’s [Are short timelines actually bad?](https://www.lesswrong.com/posts/YC6ZjCQPPuKQb49TQ/are-short-timelines-actually-bad), but given the [AI pause debate](https://forum.effectivealtruis...
c607a43c-1186-4a78-9feb-663c75fc725a
trentmkelly/LessWrong-43k
LessWrong
MINE: Free tool for detecting novel associations in large data sets I was waiting for someone more knowledgeable to post something about this, but it's been a couple days and thought I'd bring it to LW's attention. > The maximal information coefficient (MIC) is a measure of two-variable dependence developed with the ...
1ec93b56-3b7e-4f35-9be3-9a5b49e8e3fd
trentmkelly/LessWrong-43k
LessWrong
NYU Code Debates Update/Postmortem TL;DR We designed an ambitious scalable oversight experimental setup, where we had people with no coding/programming experience try to answer coding questions (“Which of these two outputs is the correct output of the given function on this input?”), using LLMs that debate or are argu...
fc35f233-b4b0-4caa-94b7-1d788a23e03e
trentmkelly/LessWrong-43k
LessWrong
Losing Your Religion: Analytic Thinking Can Undermine Belief It appears that Scientific American have found out about raising the sanity waterline. Plus one successful prediction by Mr. Yudkowsky, I presume. Since I couldn't bring myself to quote any particular bit of the article, deeming all of them to be of substanc...
4d9b94ea-72b1-45b5-8af0-446fbe33c6df
trentmkelly/LessWrong-43k
LessWrong
Experimental longtermism: theory needs data The series explains my part in the response to COVID, my reasons for switching from AI alignment work to the COVID response for a full year, and some new ideas the experience gave me. While it is written from my (Jan Kulveit's) personal perspective, I co-wrote the text with ...
e183caa6-eabc-4bdb-ae31-671049d29b10
trentmkelly/LessWrong-43k
LessWrong
Peter Thiel's AGI discussion in his startups class @ Stanford [link] http://blakemasters.tumblr.com/post/24464587112/peter-thiels-cs183-startup-class-17-deep-thought Some perspectives on AI risk that might be interesting. Peter is (the primary?) donor for SI, and an investor in AGI startup Vicarious Systems.  
e1629764-333b-4081-8321-4b0a37e525e5
trentmkelly/LessWrong-43k
LessWrong
How to get useful work out of theologians [LINK] I've offered theologians a new argument they can develop against physicalism. Should they pursue it, we may actually get some useful work out of them. See here. (The intelligence explosion animation was constructed from Anna Salamon's slides for her Minicamp x-risk pre...
c4ea2cf1-b25e-4f47-a7b8-21e11a5b1e9d
trentmkelly/LessWrong-43k
LessWrong
"Why Try Hard" Essay targeted at non rationalists Hello everyone, This is a follow-up to my last post about optimizing, which I intended to spread to nonrationalist friends of mine.  Initial feedback let me know that a lot of the language was a put-off, as well as the dry style and lack of opposing counters to argume...
3cec6874-c678-48cb-ad67-4abf21e16625
trentmkelly/LessWrong-43k
LessWrong
How To: A Workshop (or anything) Author's note: while the whole piece is a 55min read, there's a ~7min standalone essay at the top, and then a break and a shift into a different kind of content.  You really actually for real can just read the 7 minutes and then stop, if that satisfies the need you had going in. Or any...
4b3642e6-78f7-4c6f-b920-132e95ab7511
trentmkelly/LessWrong-43k
LessWrong
Outer alignment and imitative amplification Understanding the outer alignment problem What really is outer alignment? In “Risks from Learned Optimization,” we defined outer alignment in the context of machine learning as “aligning the specified loss function with the intended goal.” But that's not a perfectly well-def...
a290eea2-83d5-4c71-8f2c-f5760cca084d
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
[Intro to brain-like-AGI safety] 13. Symbol grounding & human social instincts *Part of the* [*“Intro to brain-like-AGI safety” post series*](https://www.alignmentforum.org/s/HzcM2dkCq7fwXBej8)*.* 13.1 Post summary / Table of contents ===================================== In [the previous post](https://www.alignment...
fea07528-db6c-4296-948a-e360dd74ff55
trentmkelly/LessWrong-43k
LessWrong
Hugo Awards - HP:MoR This year I'm a supporting member of WorldCon for the first time, and I noticed that the Hugo's have a category for Best Fan Writer ("Any person whose writing has appeared in semiprozines or fanzines or in generally available electronic media during 2010."). I do believe Harry Potter and the Metho...
6decb651-bb51-4a77-881c-642f53bec725
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Open Philanthropy's AI governance grantmaking (so far) *Written by* [*Luke Muehlhauser*](https://forum.effectivealtruism.org/users/lukeprog)*. To save time on formatting, I've left footnotes linked to the original source.* --- When the Soviet Union began to fracture in 1991, the world was forced to reckon with the...
5ab33db9-a009-4cb2-98f9-7d714a0e864f
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Scoring forecasts from the 2016 “Expert Survey on Progress in AI” Summary ------- This document looks at the predictions made by AI experts in [The 2016 Expert Survey on Progress in AI](https://aiimpacts.org/2016-expert-survey-on-progress-in-ai/), analyses the predictions on ‘Narrow tasks’, and gives a Brier score to...
6e7da1bd-7eb7-4732-8b58-75746e6fce67
StampyAI/alignment-research-dataset/arxiv
Arxiv
Bayesian Policy Optimization for Model Uncertainty 1 Introduction --------------- At its core, real-world robotics is about operating under uncertainty. An autonomous car must drive alongside human drivers it has never met, under road conditions which are different from the day before. An assistive home robot must ...
b429415b-3ba7-4cc6-8668-2974a4e184f3
trentmkelly/LessWrong-43k
LessWrong
Weekly LW Meetups This summary was posted to Main on February 26th. The following week's summary is here. Irregularly scheduled Less Wrong meetups are taking place in: * Baltimore Area Meetup: Futurology / Open Discussion: 28 February 2016 03:00PM * Cologne meetup: 27 February 2016 05:00PM * European Community We...
be8d6645-bfce-420a-b5a7-5a2519ec29a0
trentmkelly/LessWrong-43k
LessWrong
Intrinsic Drives and Extrinsic Misuse: Two Intertwined Risks of AI Given their advanced capabilities, future AI systems could pose significant risks to society. Some of this risk stems from humans using AI systems for bad ends (misuse), while some stems from the difficulty of controlling AI systems “even if we wanted ...
dd43a661-e5f3-40f9-a89c-11edbc1c7e45
trentmkelly/LessWrong-43k
LessWrong
Progress links and tweets, 2023-05-03 The Progress Forum * Reflections from seeing the Starship launch firsthand * When America received a message from the future: the 1939 World’s Fair Announcements * Speculative Technologies recruiting “field strategists.” “Come design a research program to unlock powerful mate...
e16f8809-9b0d-498c-92e2-91d8df587fa4
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Simplified bio-anchors for upper bounds on AI timelines *This is a replication and distillation of the extensive analysis made by Ajeya Cotra’s* [*Draft report on AI timelines*](https://www.lesswrong.com/posts/KrJfoZzpSDpnrv9va/draft-report-on-ai-timelines)*. In this work, I aim at making the argument as concise and c...
c8932b65-8d96-4704-87a0-5141ad45cb91
trentmkelly/LessWrong-43k
LessWrong
Harmony Intelligence is Hiring! Hey folks! Pleased to announce we have a new open position for a Founding Research Engineer at Harmony Intelligence. You’ll be responsible for measuring and identifying dangerous AI capabilities across various domains: cybersecurity, biosecurity, persuasion and manipulation, self-exfilt...
269f4382-7683-4230-b3c6-30462317fab8
trentmkelly/LessWrong-43k
LessWrong
Cultish Countercultishness In the modern world, joining a cult is probably one of the worse things that can happen to you. The best-case scenario is that you’ll end up in a group of sincere but deluded people, making an honest mistake but otherwise well-behaved, and you’ll spend a lot of time and money but end up with...
e8f93a4d-4926-4cfc-8a37-f7d216c05193
trentmkelly/LessWrong-43k
LessWrong
The Bayesian Agent Followup to: Reinforcement Learning: A Non-Standard Introduction, Reinforcement, Preference and Utility A reinforcement-learning agent interacts with its environment through the perception of observations and the performance of actions. A very abstract and non-standard description of such an agent ...
18cf70e6-6d6c-4225-80e3-cbfe0b87f422
trentmkelly/LessWrong-43k
LessWrong
What should an Einstein-like figure in Machine Learning do? Suppose you are an extremely talented but still unknown machine learning researcher (like Einstein of patent-clerk days), and you discover something very important, maybe a few tricks that allow you to train neural nets 4 orders of magnitude faster than anyon...
56e09640-39a3-4bae-b372-7a6d70c6811e
trentmkelly/LessWrong-43k
LessWrong
Testing "True" Language Understanding in LLMs: A Simple Proposal The Core Idea What if we could test whether language models truly understand meaning, rather than just matching patterns? Here's a simple thought experiment: 1. Create two artificial languages (A and B) that bijectively map to the same set of basic con...
b2940f97-4fde-460a-a7b1-aea19a6fe347
trentmkelly/LessWrong-43k
LessWrong
Notes from a mini-replication of the alignment faking paper Key takeaways * This post contains my notes from a 30-40 hour mini replication of Greenblatt et al.’s alignment faking paper. * This was a significant paper because it provided some evidence for potential catastrophic risk from AI misalignment. (more) * My...
13f32a9c-ed9d-4013-8478-5eb673d955fa
StampyAI/alignment-research-dataset/lesswrong
LessWrong
What if we solve AI Safety but no one cares Suppose that next year, AI Safety is solved, the solution is approved by Eliezer Yudkowsky, etc. How do we actually get people to follow this solution? It seems to me a lot of people or companies will ignore any AI Safety solutions for the same reasons they are currently i...
fea201ac-d42c-44d5-9153-73b2823c4798
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Mitigating existential risks associated with human nature and AI: Thoughts on serious measures. Howdy World, I'm a 65-year old evolutionary behavioral ecologist, pretty well-trained and still avidly teaching and learning. I specialize in the evolution of social and sexual behavior in all species, including humans. I'...
283d7913-db0d-42df-acd2-589ae0215fd5
trentmkelly/LessWrong-43k
LessWrong
Geoffrey Hinton - Full "not inconceivable" quote CBS interview with Geoffrey Hinton published March 25th, 2023: In the last two days I've seen several people share news articles about Hinton saying it is "not inconceivable" that AI will wipe out humanity in this interview. The articles often don't give the full con...
5323e153-63a6-4620-af53-81b3eb23d77d
trentmkelly/LessWrong-43k
LessWrong
Generative, Episodic Objectives for Safe AI Introduction Current discourse around AI safety focuses on creating some new paradigm for training AI models to make them safe, such as assistance games (Russell 2021) or corrigibility (Soares 2014, 2018). The prevailing view seems to be that the only thing keeping current A...
66d92384-cc04-4994-823c-d3e22c314c97
trentmkelly/LessWrong-43k
LessWrong
Timeless Identity Crisis Followup/summary/extension to this conversation with SilasBarta So, you're going along, cheerfully deciding things, doing counterfactual surgery on the output of decision algorithm A1 to calculate the results of your decisions, but it turns out that a dark secret is undermining your efforts.....
968a0def-5f07-4b43-a292-963c8f022220
trentmkelly/LessWrong-43k
LessWrong
The Singularity Wars (This is a introduction, for  those not immersed in the Singularity world, into the history of and relationships between SU, SIAI [SI, MIRI], SS, LW, CSER, FHI, and CFAR. It also has some opinions, which are strictly my own.) The good news is that there were no Singularity Wars.  The Bay Area ha...
b225033e-bdcf-4b87-907a-f1e15a9aa58d
trentmkelly/LessWrong-43k
LessWrong
More money with less risk: sell services instead of model access OpenAI is currently charging 100,000 times less per line of code than professional US devs.[1] An LLM's code output is of course less reliable than a professional's. And it is hard to use a text-completion API effectively in large projects. What should...
7500626a-4ae2-4a1f-9140-d94dbe0aa391
trentmkelly/LessWrong-43k
LessWrong
Research on how pattern-finding contributes to memorization? I notice that I have a much easier time memorizing things when I find a pattern. Often, I fail to think to even look for a pattern. As an example, I had a hard time memorizing the unit circle until I realized that the numerator is just counting up and down f...
45874669-947a-4cda-a700-5e1c83eba596
trentmkelly/LessWrong-43k
LessWrong
fimfiction.net LessWrong group There is now a LessWrong group on fimfiction, to let LWers on fimfiction find each other and collect stories that might be of interest to them.  (That?  Which?  Grammar Nazis, help!)
1cc705fc-5c53-4d51-9270-266d6837023f
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Finite Factored Sets: Polynomials and Probability In this post, given a [finite factored set](https://www.lesswrong.com/s/kxs3eeEti9ouwWFzr) F=(S,B).mjx-chtml {display: inline-block; line-height: 0; text-indent: 0; text-align: left; text-transform: none; font-style: normal; font-weight: normal; font-size: 100%; font-s...
9e868d89-3e65-4437-950f-44a79d2cf167
trentmkelly/LessWrong-43k
LessWrong
Forecasting Newsletter: April 2020 A forecasting digest with a focus on experimental forecasting. * You can sign up here. * You can also see this post on the Effective Altruism Forum here * And the post is archived here The newsletter itself is experimental, but there will be at least five more iterations. Feel f...
241ce48a-d42c-4433-9c6f-c28460984445
trentmkelly/LessWrong-43k
LessWrong
Not Even Evidence Introduction. I've been thinking about Anthropic Arguments recently, and noticed a disturbing lack of hard hitting (as in, can be shown to be true or false given our knowledge of physics and such) thought experiments on the topic. The self sampling assumption is as follows: > All other things equal...
2d094146-d10d-4d8d-b8dd-6c7dbef352cb
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Note: Experimental, trying to repost a Twitter thread as a Lesswrong post to see if people like it (so if you like it, or don't, tell me).Let's talk a little bit about oscillating motivation, shadow values, non-coercion, and the problem of revolving productivity systems.We've all had the feeling that new productivity ...
f4c0adc9-c282-4f33-8d0b-fcec7fda440b
trentmkelly/LessWrong-43k
LessWrong
[Link] Book Review: ‘The AI Does Not Hate You’ by Tom Chivers (Scott Aaronson) Scott Aaronson uploaded a review about The AI Does Not Hate You, a book by Tom Chivers. The book: > This is a book about AI and AI risk. But it's also more importantly about a community of people who are trying to think rationally about i...
d96f4e7e-2a89-42cb-b3e4-33a9e12ca99b
trentmkelly/LessWrong-43k
LessWrong
Colonization models: a programming tutorial (Part 1/2) Introduction Are we alone in the universe?  How likely is our species to survive the transition from a Type 0 to a Type II civilization?  The answers to these questions would be of immense interest to our race; however, we have few tools to reason about these que...
a5e693e5-46b7-4669-a095-3d1ef5a451d3
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Whole Brain Emulation: Looking At Progress On C. elgans Being able to treat the pattern of someone's brain as software to be run on a computer, perhaps in parallel or at a large speedup, would have a huge impact, both socially and economically.  Robin Hanson thinks it is the [most likely route to artificial intelligen...
0793b70d-76e4-44a6-adbb-d35e83e8ccd1
trentmkelly/LessWrong-43k
LessWrong
Avoiding coronavirus spread from family members, roommates, extended family and close friends Mostly based on this article https://www.erinbromage.com/post/the-risks-know-them-avoid-them * The key variables: how many people are in a space, how much airflow is around me, and how long will you be in the environment? ...
b0db70c0-f088-40b1-bcb7-129ecc330e37
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington D.C.: Meta Meetup Discussion article for the meetup : Washington D.C.: Meta Meetup WHEN: 03 January 2016 03:00:00PM (-0500) WHERE: Reynolds Center x-posted from list. Gathering in courtyard from 3:00pm, hard start 3:30pm - until closing (7:00 pm). We'll be meeting to talk about the Less Wrong D...
6962ad5c-15d7-4a47-804b-a35a1e11bd33
trentmkelly/LessWrong-43k
LessWrong
Eisenhower's Atoms for Peace Speech Along with other AI governance researchers, I have been interested in better understanding historical examples of widespread international cooperation. Perhaps the most well-studied analogy can be found in the case of nuclear energy & nuclear weapons. I believe we can learn a lot f...
eaa74684-f192-4480-84f8-8d451e28cee5
trentmkelly/LessWrong-43k
LessWrong
Analog rewritable tablet I am requesting suggestions for an analog-ish rewritable tablet. Details below. I have a paper and pencil task list for work that I repeatedly erase and write over as I complete various tasks. After a month or two, it gets hard to read and I have to copy everything onto a new list. I'm lookin...
1c4b9e3b-e54f-435c-8663-b1cd4e2725a4
trentmkelly/LessWrong-43k
LessWrong
Meetup : Atlanta - Filtering Ideas Discussion article for the meetup : Atlanta - Filtering Ideas WHEN: 09 August 2014 07:00:00PM (-0400) WHERE: 2388 Lawrenceville Hwy. Unit L. Decatur, GA 30033 This meetup, we'll be talking about recognizing confusion, doublethink, and how to filter good and true ideas from false a...
0d577df6-8b88-449c-b888-682e9eac27f2
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post0 Summary: We have previously argued that scalable oversight methods can be used to provide guarantees on low-stakes safety – settings where individual failures are non-catastrophic. However, if your reward function (e.g. honesty) is compatible with many possible solutions then you also need to avoid having fr...
24d7e1a8-a9c9-464d-8bf5-eb31357955c2
trentmkelly/LessWrong-43k
LessWrong
Slopworld 2035: The dangers of mediocre AI None of this article was written with AI assistance. Introduction There have been many, many, many attempts to lay out scenarios of AI taking over or destroying humanity. What they tend to have in common is an assumption that our doom will be sealed as a result of AI becomi...
482a805c-d770-480d-95ab-82116354afa3
trentmkelly/LessWrong-43k
LessWrong
A Guide For LLM-Assisted Web Research It's hard to imagine doing web research without using LLMs. Chatbots may be the first thing you turn to for questions like: What are the companies currently working on nuclear fusion and who invested in them? What is the performance gap between open and closed-weight models on the...
c120385c-b5bd-4b29-96d6-d6b82e3f1d24
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Eliezer Yudkowsky's original Sequences have been edited, reordered, and converted into an ebook! Rationality: From AI to Zombies is now available in PDF, EPUB, and MOBI versions on intelligence.org (link). You can choose your own price to pay for it (minimum $0.00), or buy it for $4.99 from Amazon (link). The contents...
4a5e50a4-efc3-471d-845d-359ba46d9034
trentmkelly/LessWrong-43k
LessWrong
Prescriptive vs. descriptive and objective vs. subjective definitions Imagine you're writing a Field Guide to Boats, and you want to know what you should include in your field guide. Barges? Rafts? These things? You want something like a dictionary definition of boat. A descriptive definition that includes anything...
b7ace1c4-9cc3-43db-bd7a-0e2b8d79ddb1
trentmkelly/LessWrong-43k
LessWrong
Meetup : SLC, UT: Free Will and Rationality Checklists Discussion article for the meetup : SLC, UT: Free Will and Rationality Checklists WHEN: 15 September 2012 03:00:00PM (-0600) WHERE: Calvin S Smith Library, 810 East 3300 South, Salt Lake City, UT We'll be in the back conference room 3:00 Greetings, Introductio...
ef28c47f-e4d7-4d1e-88d8-c887409d2966
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Decision Theory *(A longer text-based version of this post is also available on MIRI's blog* [*here*](https://intelligence.org/2018/10/31/embedded-decisions/)*, and the bibliography for the whole sequence can be found* [*here*](https://intelligence.org/embedded-agency)*.)* ![](https://res.cloudinary.com/lesswrong-2-...
5738e9c9-de1c-4358-a76b-c3bfe2464409
trentmkelly/LessWrong-43k
LessWrong
How would you use video gamey tech to help with AI safety? I've been working on AI safety for a while now. It's going better than expected, but I have finite hours. The more hours I spend on safety, the less I can spend on business-oriented things. Those business-oriented things are long term (4+ years) high risk atte...
d66beef8-0366-41fd-87f1-cce818f91b27
trentmkelly/LessWrong-43k
LessWrong
Universality Unwrapped Introduction Informally, a universal system is universal with respect to any computation; and it is a universal system with respect to a given computation if it understands every set of beliefs that can be ascribed to the computation. The intuition is that the system can reverse engineer most or...
bf801667-91d7-446b-bef8-bb46616212d7
trentmkelly/LessWrong-43k
LessWrong
Apply to MATS 7.0! Applications are open for the ML Alignment & Theory Scholars (MATS) Winter 2024-25 Program, running Jan 6-Mar 14, 2025. First-stage applications are due Oct 6! MATS is a twice-yearly, 10-week AI safety research fellowship program operating in Berkeley, California, with an optional 4-month extension...
7b9e27f8-97c3-4248-87eb-9a78ab9ad546
trentmkelly/LessWrong-43k
LessWrong
Romantic idealism: true love conquers almost all More romantic people tend to be vocally in favor of more romantic fidelity in my experience. If you think about it though, faith in romance is not a very romantic ideal. True love should overcome all things! The highest mountains, the furthest distances, social classes,...
dc73e349-4d31-41ff-af13-648203cd6c3b
trentmkelly/LessWrong-43k
LessWrong
Editing Advice for LessWrong Users Who am I? I'm an editor who does feedback requests for LessWrong. Many of these feedback requests are for proofreading/clarity of language, so I have lots of experience reading community draft posts and helping improve them. I want to share some high level takeaways - low effort self...
80780dbd-a86e-4966-a9d4-18716332e885
trentmkelly/LessWrong-43k
LessWrong
CAIS-inspired approach towards safer and more interpretable AGIs Epistemic status: a rough sketch of an idea Current LLMs are huge and opaque. Our interpretability techniques are not adequate. Current LLMs are not likely to run hidden dangerous optimization processes. But larger ones may. Let's cap the model size at...
2b367ab4-ce92-41f0-9dd8-91ac8b4d41a0
trentmkelly/LessWrong-43k
LessWrong
Why Glen Weyl is a technocrat In 2019 Glen Weyl wrote the article Why I Am Not A Technocrat where he asserted that while he isn't a technocrat the rationality community is technocratic. I think he suffers from a case of projection because he fails to understand the problem of technocracy. Weyl presents the challenge ...
ad2d622d-b498-4be9-acc5-011c24a7c916
trentmkelly/LessWrong-43k
LessWrong
Detached Lever Fallacy This fallacy gets its name from an ancient sci-fi TV show, which I never saw myself, but was reported to me by a reputable source (some guy at an SF convention).  Anyone knows the exact reference, do leave a comment. So the good guys are battling the evil aliens.  Occasionally, the good guys ha...