id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
2d83b527-10be-4331-a917-a5aa333935a0
trentmkelly/LessWrong-43k
LessWrong
AI & wisdom 2: growth and amortised optimisation Written for the AI Impacts essay competition on the automation of wisdom and philosophy Growth modes and direct v amortised optimisation Amortised optimisation is more important when relevant histories are long and growth is slow. If growth suddenly picks up, the flex...
c6b6fb6f-3687-42cb-b017-f739d9190949
StampyAI/alignment-research-dataset/lesswrong
LessWrong
What's Your Cognitive Algorithm? *Epistemic Status: I'm neither a neuroscientist nor an ML researcher, but am trying to figure out "what kinds of human thought are actually possible to replicate on silicon right now?".* *Here's my best guess of how human cognition works. Please tear it apart!*   When I looked at GP...
1b09c240-1a2d-4c0e-8f22-facf7b89f7ef
trentmkelly/LessWrong-43k
LessWrong
IEEE released the first draft of their AI ethics guide
7d101631-30e6-48d8-baa2-c793c15ce67f
trentmkelly/LessWrong-43k
LessWrong
Anthropic AI made the right call I've seen a number of people criticize Anthropic for releasing Claude 3 Opus, with arguments along the lines of: > Anthropic said they weren't going to push the frontier, but this release is clearly better than GPT-4 in some ways! They're betraying their mission statement! I think th...
96c38b83-1292-4000-a5ef-1f3d54e38780
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
Where can I find mentorship and advice for becoming a researcher? There are multiple programmes you can apply to if you want to try becoming a researcher. If accepted to these programs, you will get funding and mentorship. [aisafety.training](https://aisafety.training/) provides an up-to-date list of which programs ha...
b6f4be7a-95ed-4096-9680-4b4cbc9e4e83
trentmkelly/LessWrong-43k
LessWrong
Meetup : Scrum: A Philosophy of Life? Discussion article for the meetup : Scrum: A Philosophy of Life? WHEN: 18 November 2015 07:00:00PM (-0800) WHERE: 10850 West Pico Blvd, Los Angeles, CA 90064, USA How to Find Us: Go into the Westside Pavillion. The Landmark Theater has an adjacent Wine Bar (all ages welcome), l...
3ff0b249-fa1d-4ab0-acd0-a2a8dd43b010
trentmkelly/LessWrong-43k
LessWrong
Open thread, 21-27 April 2014 If it's worth saying, but not worth its own post (even in Discussion), then it goes here. Thread started before the end of the last thread to ecourage Monday as first day.
da78c462-1974-4ff6-985e-75dd22cbf8f5
StampyAI/alignment-research-dataset/blogs
Blogs
New paper: “Incorrigibility in the CIRL Framework” [![Incorrigibility in the CIRL Framework](https://intelligence.org/files/IncorrigibilityCIRL.png)](https://arxiv.org/abs/1709.06275) MIRI assistant research fellow Ryan Carey has a new paper out discussing situations where good performance in [Cooperative Inverse Re...
e83a2f03-1e71-4e5b-bd69-94be8b4d40ad
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Vingean Reflection: Open Problems .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-size-adjust: none; letter-spacing: normal; word-wrap: normal; word-spacing: normal; white-space: nowrap; float: non...
3e7fbd63-1e4c-4009-90eb-79c909a081a7
trentmkelly/LessWrong-43k
LessWrong
How should potential AI alignment researchers gauge whether the field is right for them? None
dd3ecbf4-4515-4e35-a795-76b5c788a98b
StampyAI/alignment-research-dataset/blogs
Blogs
Six Dimensions of Operational Adequacy in AGI Projects **Editor’s note:**The following is a lightly edited copy of a document written by Eliezer Yudkowsky in November 2017. Since this is a snapshot of Eliezer’s thinking at a specific time, we’ve sprinkled reminders throughout that this is from 2017. A background note:...
2a8044de-36b9-4ab8-aeee-398a5aff7fcc
trentmkelly/LessWrong-43k
LessWrong
Map and Territory and "Paths" Hi everyone, Recently have started a discussion group amongst some of my friends and aquaintences in an attempt to study and improve our rationality. What I tend to do, is using the sequences as 'source' material, write up a lesson plan that attempts to be a bit more accesible to the la...
42fe3b19-9071-4ba8-9ad3-9243a850b2f9
trentmkelly/LessWrong-43k
LessWrong
The Open Agency Model Notes on AI for complex, consequential problems Eric Drexler Centre for the Governance of AI University of Oxford Introduction This document argues for “open agencies” — not opaque, unitary agents — as the appropriate model for applying future AI capabilities to consequential tasks that call f...
ee261a78-560a-46ef-9d66-a2261de8d062
trentmkelly/LessWrong-43k
LessWrong
[LINK] How to calibrate your confidence intervals In the book "How to Measure Anything" D. Hubbard presents a step-by-step method for calibrating your confidence intervals, which he has tested on hundreds of people, showing that it can make 90% of people almost perfect estimators within half a day of training. I've b...
1f896ec8-6753-4b70-a3d2-553b273c2d29
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Almost one year ago, in April 2007, Matthew C submitted the following suggestion for an Overcoming Bias topic: "How and why the current reigning philosophical hegemon (reductionistic materialism) is obviously correct [...], while the reigning philosophical viewpoints of all past societies and civilizations are obvious...
508c6d04-e87b-48d6-bf13-7da7f18cde27
trentmkelly/LessWrong-43k
LessWrong
Why anything that can be for-profit, should be In a recent post I surveyed different types of funding models, including nonprofit models such as universities or private foundations, and for-profit models such as startups. Although we need both models, I believe that for-profit models are underrated today. In what foll...
9f8a43da-9892-4007-8a14-9ac70b539af4
trentmkelly/LessWrong-43k
LessWrong
Q&A with Michael Littman on risks from AI [Click here to see a list of all interviews] > Michael L. Littman is a computer scientist. He works mainly in reinforcement learning, but has done work in machine learning, game theory, computer networking, Partially observable Markov decision process solving, computer solvin...
09f7eabb-fcc9-4a1c-805f-2000bc3634e5
trentmkelly/LessWrong-43k
LessWrong
A Ground-Level Perspective on Capacity Building in International Development I've been enjoying the blog/podcast Statecraft, which interviews powerful professionals in government and how they approach important problems, because its exposing me to many unfamiliar perspectives. In the spirit of Statecraft, but only bei...
f9290917-b2f8-4c47-b262-6bbfb99a039c
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Epistemic status: Speculative, just having fun. This piece isn't well-cited, but I can pull up sources as needed - nothing about mole-rats is my original research. A lot of this piece is based on Wikipedia.When I wrote about “weirdness” in the past, I called marine invertebrates, archaea viruses, and Florida Man stori...
b569e2df-4fc2-43ba-b2a5-8a8125b89db0
trentmkelly/LessWrong-43k
LessWrong
Deep Learning is cheap Solomonoff induction? Background  Lucius:  I recently held a small talk presenting an idea for how and why deep learning generalises. It tried to reduce concepts from Singular Learning theory back to basic algorithmic information theory to sketch a unified picture that starts with Solomonoff ind...
7ab45759-c9dc-4407-a217-dc1a2991e549
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Infra-Bayesianism Unwrapped **Introduction** ================ [Infra-Bayesianism](https://www.alignmentforum.org/s/CmrW8fCmSLK7E25sa) is a recent theoretical framework in AI Alignment, coming from Vanessa Kosoy and Diffractor (Alexander Appel). It provides the groundwork for a learning theory of RL in the non-realiza...
27600b21-29be-4e3f-8001-42a4750b6e37
trentmkelly/LessWrong-43k
LessWrong
Dont focus on updating P doom Motivation: Improving group epistemics. TL; DR (Changes to) P doom/alignment difficulty are a shibboleth dominating conversations, distorting epistemics. Instead, focus on updates to your gears level models. Focus on near and concrete details instead of far and vague abstractions.  Peo...
484ab110-89cc-4b3e-a881-b1c25a6e7c94
trentmkelly/LessWrong-43k
LessWrong
Updating my AI timelines Earlier this year I offered to bet people who had short AI timelines. While it wasn't my intention to be known as "a long AI timelines guy", I have begun feeling that was how people perceived me. Nonetheless, in the last few months, I've modified my views substantially. Thus, I offer this sho...
121b1847-e8c6-4e97-b0b5-9db11e1e7f5a
trentmkelly/LessWrong-43k
LessWrong
Rationality Quotes November 2013 Another month has passed and here is a new rationality quotes thread. The usual rules are: * Please post all quotes separately, so that they can be upvoted or downvoted separately. (If they are strongly related, reply to your own comments. If strongly ordered, then go ahead and post ...
64b26cf2-5e2c-4246-ab7a-a23596a22d7a
trentmkelly/LessWrong-43k
LessWrong
Is the confirmation bias really a bias? The confirmation bias is a widely discussed cognitive phenomenon that has drawn considerable attention in psychology and behavioural economics.  Traditionally, the search for confirmatory information has been portrayed as a bias, suggesting a systematic deviation from rational ...
eb981540-a2c5-48e0-aa78-7eb5e80e720b
StampyAI/alignment-research-dataset/lesswrong
LessWrong
AIs and Gatekeepers Unite! > "Bah, everyone wants to be the gatekeeper. What we NEED are AIs." >         -- Schizoguy > > Some of you have expressed the opinion that the [AI-Box Experiment](http://www.yudkowsky.net/essays/aibox.html) doesn't seem so impossible after all.  That's the spirit!  Some of you even thin...
900a2a01-b3ec-409f-9c75-d2a7a500f8df
trentmkelly/LessWrong-43k
LessWrong
Could markets be called optimization proccesses? Edit: This is old material. It may be out of date. Most posters here seem to agree1 that: * Intelligence at least human intelligence is an optimization process. * Evolution is an optimization process. * Other optimization processes may exist. Taking these as a g...
5e9a8fd3-4349-481a-9d58-0a01eff562d4
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
an Evangelion dialogue explaining the QACI alignment plan *this post was written by [Tamsin Leake](https://carado.moe) at [Orthogonal](https://orxl.org).* *thanks to [Julia Persson](https://www.lesswrong.com/users/juliahp) and [mesaoptimizer](https://mesaoptimizer.com/) for their help putting it together.* this p...
cafd60a3-bfbf-4996-b968-ead08a5dc935
trentmkelly/LessWrong-43k
LessWrong
Memetic downside risks: How ideas can evolve and cause harm This post was written for Convergence Analysis. Overview We introduce the concept of memetic downside risks (MDR): risks of unintended negative effects that arise from how ideas “evolve” over time (as a result of replication, mutation, and selection). We di...
ae559b51-0868-4824-81f2-6e135449f6e2
trentmkelly/LessWrong-43k
LessWrong
Cortés, AI Risk, and the Dynamics of Competing Conquerors Cortés's conquest of the Aztecs reflects current anxieties surrounding AI risks. His technological and tactical edge over the Aztecs parallels the potential superiority of smarter-than-human AIs attempting to take over earth. This historical parallel implies th...
1818d96b-d40b-4e05-9e23-a470ec9b1a36
trentmkelly/LessWrong-43k
LessWrong
Some informal ramblings abaut what a defenition of reality might look like. I were originally going to just chat about this on IRC but last few times I did that I ended up concludin it'd have been better to make an inchoherent discusion post on it and i just got some karma to burn, so here goes nothing. Sorry for the ...
5253bedc-0cfe-4c59-a2ca-c9824c42abdd
trentmkelly/LessWrong-43k
LessWrong
Logits, log-odds, and loss for parallel circuits Today I’m going to discuss how to think about logits like a statistician, and what this implies about circuits. This post doesn’t have any prerequisites other than perhaps a very basic statistical background that can be adequately recovered from the AI-generated “glossa...
7519677d-ee19-4a3d-a7d0-856b0288c692
trentmkelly/LessWrong-43k
LessWrong
Jason Crawford on the non-linear model of innovation: SSC Online Meetup Jason Crawford on the non-linear model of innovation. Please signup here by an hour before the event, and we will send you an invitation https://forms.gle/XXk3hPNkcEJJbHFC9 IMPORTANT: As Daylight Savings Time approaches an end, note the times: ...
31273222-9b41-4321-b432-04d6076b4e1f
trentmkelly/LessWrong-43k
LessWrong
Machine Learning Analogy for Meditation (illustrated) Here's an illustrated rendition of a semiformal explanation of certain effects of meditation. It was inspired by, but differs significantly from, Kaj's post on meditation. Some people appreciated gjm's transcription for readability. ...
a21b7d9b-6627-49da-bb81-2e7d360093e5
trentmkelly/LessWrong-43k
LessWrong
A Proof of Occam's Razor Related to: Occam's Razor If the Razor is defined as, “On average, a simpler hypothesis should be assigned a higher prior probability than a more complex hypothesis,” or stated in another way, "As the complexity of your hypotheses goes to infinity, their probability goes to zero," then it can...
badc8d68-1fb4-4d76-a051-c8876abd5b6e
trentmkelly/LessWrong-43k
LessWrong
Long-term Donation Bunching? Let's say you're a relatively well-off American with an income of $100k and have taken the Giving What We Can pledge to donate 10% of your money to the most effective charities you can find. The standard deduction in the US is currently $12k, which means if you were to donate $10k/year it ...
708761dc-e797-43d7-a922-9ff8be00bcac
trentmkelly/LessWrong-43k
LessWrong
International Scientific Report on the Safety of Advanced AI: Key Information I thought that the recently released International Scientific Report on the Safety of Advanced AI seemed like a pretty good summary of the state of the field on AI risks, in addition to being about as close to a statement of expert consensus...
1a1ae630-b685-4aab-b0f4-56b38928d2dd
trentmkelly/LessWrong-43k
LessWrong
Conservatism in neocortex-like AGIs (Related to Stuart Armstrong’s post last summer on Model Splintering.) Let’s say I have a human personal assistant. Call him Ahmed. Ahmed is generally trying to help me. Sometimes a situation comes up where Ahmed feels surprised and confused. Other times, a situation comes up where...
9cf6f585-bda8-4459-841f-65494870341d
trentmkelly/LessWrong-43k
LessWrong
Weekly LW Meetups: Melbourne, Canberra, Zurich, Mountain View There are upcoming irregularly scheduled Less Wrong meetups in: * Melbourne practical rationality meetup: 04 November 2011 07:00PM * Canberra Meetup Saturday November 5th: 05 November 2011 11:00AM * First Zurich LW Meetup: 12 November 2011 03:00PM Noti...
3a175e10-8c33-4955-8362-e1ed6990abbb
trentmkelly/LessWrong-43k
LessWrong
Mistral Large 2 (123B) seems to exhibit alignment faking UPDATE: Recent work with improved AF and compliance gap classifiers disagrees with our results. We recommend using the improved classifiers.  Summary We wanted to briefly share an early takeaway from our exploration into alignment faking: the phenomenon appear...
e9c3794d-75f0-434e-9c44-2df258c4e7a3
trentmkelly/LessWrong-43k
LessWrong
Answering questions honestly instead of predicting human answers: lots of problems and some solutions This post is the result of work I did with Paul Christiano on the ideas in his “Teaching ML to answer questions honestly instead of predicting human answers” post. In addition to expanding upon what is in that post in...
26547f28-3cf7-4e7d-aa03-71659d821f36
trentmkelly/LessWrong-43k
LessWrong
Quotes on Existential Risk Similar to the AGI Quotes thread and the monthly Rationality Quotes threads, this is a thread for memorable quotes about existential risk. * Please post all quotes separately, so that they can be voted up/down separately.  (If they are strongly related, reply to your own comments. If stron...
09d041d1-d79a-4617-acd7-689e1a0a3c8e
trentmkelly/LessWrong-43k
LessWrong
Another problem with AI confinement: ordinary CPUs can work as radio transmitters Excerpts: > Computer tech expert William Entriken has created a C library that can make a computer emit radio waves even if the device doesn't include any radio transmission hardware. > > His program, called System Bus Radio, is writte...
971cd8c3-e2a8-4c46-b01f-5d5748e77776
trentmkelly/LessWrong-43k
LessWrong
Pulse and Glide Cycling About 10 years ago I stopped bicycling very much, because my knees hurt. Since then I've mostly gotten around by walking, running, and public transport. With COVID, however, I've started bicycling more, and initially my nees were hurting again a bit. A couple weeks ago I bicycled about half an ...
a5c541ae-029b-426f-a967-d4f732962aa5
StampyAI/alignment-research-dataset/blogs
Blogs
“Biological anchors” is about bounding, not pinpointing, AI timelines I previously summarized Ajeya Cotra’s [“biological anchors” method for forecasting for transformative AI](https://www.cold-takes.com/forecasting-transformative-ai-the-biological-anchors-method-in-a-nutshell/), aka **“Bio Anchors.”** Here I want to t...
aa241961-5068-4b04-8ee4-b40558b9a540
StampyAI/alignment-research-dataset/lesswrong
LessWrong
The Shard Theory Alignment Scheme *Generated as part of SERI MATS, Team Shard, under John Wentworth.* *Thanks to Logan Smith, Garrett Baker, Michael Einhorn, Quintin Pope, and Thomas Kwa for chatting about related topics. All mistakes are my own.* The Alignment Scheme ==================== The shard theory alignment...
6c07f886-c058-4f38-aed9-1708e936583b
trentmkelly/LessWrong-43k
LessWrong
Meetup : Cincinnati/Columbus: Memorisation exercise (NB, time changed) Discussion article for the meetup : Cincinnati/Columbus: Memorisation exercise (NB, time changed) WHEN: 18 November 2012 02:00:00PM (-0500) WHERE: 4934 Juniper Way Beavercreek, OH 45440 Please notice changed time! We will meet at Choe's Asian G...
9c0e9e56-6501-4464-9529-794a5010ef6d
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Timeless Identity Today's post, Timeless Identity was originally published on 03 June 2008. A summary (taken from the LW wiki):   > How can you be the same person tomorrow as today, in the river that never flows, when not a drop of water is shared between one time and another? Having used physics to comp...
90083e03-9394-44b7-b4f6-2d71d5bd7adb
trentmkelly/LessWrong-43k
LessWrong
Book review: Artificial Intelligence Safety and Security Book review: Artificial Intelligence Safety and Security, by Roman V. Yampolskiy. This is a collection of papers, with highly varying topics, quality, and importance. Many of the papers focus on risks that are specific to superintelligence, some assuming that ...
c0f758f7-a644-4e53-b5cd-ea173f5835f2
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Urging an International AI Treaty: An Open Letter > We call on governments worldwide to actively respond to the potentially catastrophic risks posed by advanced artificial intelligence (AI) systems to humanity, encompassing threats from misuse, systemic risks, and loss of control. We advocate for the development and r...
f444f649-67f6-477f-8018-47ad91ca1be0
trentmkelly/LessWrong-43k
LessWrong
If we live in a simulation, what does that imply? If we live in a simulation, what does that imply about the world of our simulators and our relationship to them? [1] Here are some proposals, often mutually contradictory, none stated with anything near certainty. 1. The simulators are much like us, or at least are o...
0274c072-a58b-4bc6-82a4-70019d39311c
trentmkelly/LessWrong-43k
LessWrong
Separating the roles of theory and direct empirical evidence in belief formation: the examples of minimum wage and anthropogenic global warming I recently asked two questions on Quora with similar question structures, and the similarities and differences between the responses were interesting. Question #1: Anthropoge...
2accdf3d-d406-4966-9d84-2342139f77cf
trentmkelly/LessWrong-43k
LessWrong
Scheduling Algorithm for a PhD Student This post was inspired by Christian and Griffiths Algorithms to live by. Unlike other self-help books, it argues that the optimal scheduling strategy depends on your goals. If you want to minimize lateness of your single latest output, you should use Earliest Due Date. If you wan...
e4542052-376e-4818-846c-883c24c8b8cc
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Einstein's Speed Today's post, Einstein's Speed was originally published on 21 May 2008. A summary (taken from the LW wiki):   > Albert was unusually good at finding the right theory in the presence of only a small amount of experimental evidence. Even more unusually, he admitted it - he claimed to know ...
3e6a24bc-622f-4a8e-8df9-25da7d56b646
trentmkelly/LessWrong-43k
LessWrong
Ideological Inference Engines: Making Deontology Differentiable* * Rather, making deontology play well with differentiable systems trained end-to-end. This post is part of my hypothesis subspace sequence, a living collection of proposals I'm exploring at Refine. Preceded by oversight leagues, and followed by represen...
f4c65c6a-281d-4f64-a073-34d9ef5aa973
StampyAI/alignment-research-dataset/special_docs
Other
“Betting on the Past” by Arif Ahmed [This post assumes knowledge of decision theory, as discussed in Eliezer Yudkowsky’s [Timeless Decision Theory](http://intelligence.org/files/TDT.pdf) and in Arbital’s [Introduction to Logical Decision Theory](https://arbital.com/p/logical\_dt/).] I recently discovered an interest...
df0795c6-f165-42a2-b216-94e1e13f80f6
StampyAI/alignment-research-dataset/blogs
Blogs
CSRBAI talks on preference specification We’ve uploaded a third set of videos from our recent [Colloquium Series on Robust and Beneficial AI](https://intelligence.org/colloquium-series/) (CSRBAI), co-hosted with the Future of Humanity Institute. These talks were part of the week focused on preference specification in ...
af923309-b99c-4e74-9c62-2800e5919151
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Equilibrium and prior selection problems in multipolar deployment To [avoid catastrophic conflict in multipolar AI scenarios](https://www.alignmentforum.org/posts/DbuCdEbkh4wL5cjJ5/preface-to-eaf-s-research-agenda-on-cooperation-conflict-and), we would like to design AI systems such that AI-enabled actors will tend t...
b66926c9-5d31-4f31-a58a-9418b9cbe6ca
trentmkelly/LessWrong-43k
LessWrong
Recent updates to gwern.net (2016-2017) > Previously: 2011; 2012-2013; 2013-2014; 2014-2015; 2015-2016 “Every season hath its pleasures; / Spring may boast her flowery prime, / Yet the vineyard’s ruby treasures / Brighten Autumn’s sob’rer time.” Another year of my completed writings, sorted by topic: AI: * Why ...
ea719cff-87ad-4cec-92a6-4b75d9f78c29
trentmkelly/LessWrong-43k
LessWrong
Meta:
b00be031-6555-4762-b8b0-09a11fb5d51a
trentmkelly/LessWrong-43k
LessWrong
Looking for remote writing partners (for AI alignment research) I'm looking for other junior researchers to form a distributed writing group for mutual support. Please get in touch if you're interested. Who? One to three other people and I. I do AI alignment research independently. I'm working on the Farlamp projec...
43007641-7f9e-476b-ab38-7ff677949c09
trentmkelly/LessWrong-43k
LessWrong
[Linkpost] Scaling Laws for Generative Mixed-Modal Language Models In this paper authors explore the scaling properties of mixed-modal generative models, discovering new scaling laws that unify the contributions of individual modalities and the interactions between them. I find most interesting that they have found so...
9eacf1d9-f3de-4930-8570-17532e9ef91f
trentmkelly/LessWrong-43k
LessWrong
Risks of downloading alien AI via SETI search Alexei Turchin. Risks of downloading alien AI via SETI search Abstract: This article examines risks associated with the program of passive search for alien signals (SETI—the Search for Extra-Terrestrial Intelligence). In this paper we propose a scenario of possible vulner...
831f3dd5-f0e4-4a3d-9b13-574633801ab4
trentmkelly/LessWrong-43k
LessWrong
Meal Replacements in 2025? I'm considering meal replacements for 1-2 meals a day, and recall Soylent and Mealsquares were popular meal replacements ~5 years ago when I visited the Berkeley rationalist scene. I haven't found any recent posts discussing the newer options like Huel or about long-term effects.  Does anyo...
0e2100a9-bcb6-46dd-8f67-248fa6651e74
trentmkelly/LessWrong-43k
LessWrong
My recent posts Over at medium, I'm continuing to write about AI control; here's a roundup from the last month. Many of these seem like interesting things to discuss here; would it be better to post each of these as a link when I write it? #Strategy * Prosaic AI control argues that AI control research should first...
cf118d8e-88fd-4d2d-ada0-f675cdb0211b
trentmkelly/LessWrong-43k
LessWrong
Safety engineering, target selection, and alignment theory This post is the latest in a series introducing the basic ideas behind MIRI's research program. To contribute, or learn more about what we've been up to recently, see the MIRI fundraiser page. Our 2015 winter funding drive concludes tonight (31 Dec 15) at midn...
68524de3-736d-4cfb-998a-079e05715fb8
trentmkelly/LessWrong-43k
LessWrong
Forum Digest: Updateless Decision Theory Summary: This is a quick expository recap, with links, of the posts on this forum on the topic of updateless decision theory, through 3/19/15. Read this if you want to learn more about UDT, or if you're curious about what we've been working on lately! Introduction Updateless ...
0667c787-b650-46af-b8cb-f4eda148fa10
trentmkelly/LessWrong-43k
LessWrong
Idols of the Mind Pt. 2 (Novum Organum Book 1: 53-68) This is the fifth post in the Novum Organum sequence. For context, see the sequence introduction. We have used Francis Bacon's Novum Organum in the version presented at www.earlymoderntexts.com. Translated by and copyright to Jonathan Bennett. Prepared for LessWro...
e66d1dc3-dfcb-4cbd-9fe1-891a1e890f9d
trentmkelly/LessWrong-43k
LessWrong
'The Battle for Compassion': ethics in a world of accelerating change The Battle for Compassion: Ethics in an Apathetic Universe is a new book by first-time author Jonathan Leighton. He argues for 'negative utilitarianism plus', which means negative utilitarianism that nevertheless values existence. Chapter 13, 'Where...
637fbecb-9e57-4315-b702-36a21ef1e814
StampyAI/alignment-research-dataset/arxiv
Arxiv
Improving Social Welfare While Preserving Autonomy via a Pareto Mediator 1 Introduction --------------- We often choose to delegate our decisions to algorithms. What should a decision-support system do when multiple people choose to delegate their actions to the same algorithm? When someone uses an app to get direc...
50756681-c022-4a95-94f4-0c16b160cea4
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
President Biden Issues Executive Order on Safe, Secure, and Trustworthy Artificial Intelligence Released today (10/30/23) this is crazy, perhaps the most sweeping action taken by government on AI yet.  Below, I've segmented by x-risk and non-x-risk related proposals, excluding the proposals that are geared towards pr...
74d8715c-b0f6-4cfc-991d-afd29a228016
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
How can I work on AI policy? We’re still working on a more detailed answer, but in the meantime, here are some introductory resources to help you on your way: - 80,000 Hours has a [career review](https://80000hours.org/career-reviews/ai-policy-and-strategy/) making the case for AI policy as an impactful area to work ...
273c5227-b812-4df6-a486-dee038a5d4b3
trentmkelly/LessWrong-43k
LessWrong
Mental Blinders from Working Within Systems In make an extraordinary effort (and in other posts), Eliezer talks about how uncommon goals require uncommon means. He also notes that people seem incredibly unwilling to consider these unusual routes: > Even so, I think that we could do with more appreciation of the virtu...
7236a41c-c5e1-4060-a360-174e1cf113d4
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
The many counterfactuals of counterfactual mugging .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-size-adjust: none; letter-spacing: normal; word-wrap: normal; word-spacing: normal; white-space: n...
7b719553-6b15-4bb9-bdb3-0f2c0bc8230a
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Is AI Safety dropping the ball on privacy? ![](https://res.cloudinary.com/lesswrong-2-0/image/upload/f_auto,q_auto/v1/mirroredImages/WCevxhGtmnPhWH3ah/nngk0b8yejx3mjmaljpb)**TL;DR** ========= The lack of privacy-preserving technologies facilitates better predictive models of human behavior. This accelerates several e...
f27a3b7e-caaa-4cd2-842d-7cd21cc1c6c3
StampyAI/alignment-research-dataset/lesswrong
LessWrong
A Quick Guide to Confronting Doom Within the last two weeks, two sets of things happened: Eliezer Yudkowsky shared a post expressing extreme pessimism about humanity's likelihood of surviving AGI, and a number of AI research labs published new, highly impressive results. The combination of these two has resulted in a ...
73a384d5-f38e-4727-a85e-f4b6dafd0874
trentmkelly/LessWrong-43k
LessWrong
[Link] What do conservatives know that liberals don't (and vice versa)? I am a PhD student currently conducting research on political polarization and persuasion.  I am running an experiment that requires a database of trivia questions which conservatives are likely to get correct, and liberals are likely to get wron...
a0805d53-ce3f-4e64-960e-8e00da1d10b6
trentmkelly/LessWrong-43k
LessWrong
cheaper sodium electrolysis sodium electrolysis Aluminum metal is a widely-used material. It costs ~$2.5/kg. A significant fraction of its production cost is electricity. Currently, Na metal is produced by Downs cell electrolysis of NaCl. Making sodium metal with electrolysis requires much less energy per mass of sod...
f884b8e0-b183-4bdd-9935-4cd70a295a9b
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Notes on Caution This post examines the virtue of **caution**. As with my other posts in this sequence, I’m less interested in breaking new ground and more in gathering and synthesizing whatever wisdom I could find on the subject. I wrote this not as an expert on the topic, but as someone who wants to learn more about...
7c03aa34-2ad2-4429-bbf3-b7214e51e1b2
trentmkelly/LessWrong-43k
LessWrong
Integers as Compression I would like to propose a conception of what integers are. This is something of a misnomer, I don't actually want to claim that there is a one true conception of what integers are, but nonetheless this is a convenient way of describing what this post is about. I actually believe that there's mu...
bf8d1b08-f2bd-420f-83da-1edf88f28a29
trentmkelly/LessWrong-43k
LessWrong
Recursive Cognitive Refinement (RCR): A Self-Correcting Approach for LLM Hallucinations I’m an independent researcher who has arrived here, at AI safety through an unusual path, outside the standard academic or industry pipelines. Along this journey, I encountered the recurring problem of large language models exhibit...
a4590f53-ce19-4c1b-bce5-309579205717
trentmkelly/LessWrong-43k
LessWrong
Instrumental Rationality 4.2: Creating Habits [Instrumental Rationality Sequence 4.2/7] [Part two of a three-part series of habits.] [We go over three techniques for creating habits: TAPs, Systematic Planning, and Scaling Up.] Techniques: Creating Habits [The Techniques section has been broken up into sections: on...
4c4a0d95-4e2d-48bf-9ddc-005b4ccaa75e
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post3884 A putative new idea for AI control; index here . In a previous post on AIs in virtual worlds , I described the idea of a utility function that motivates the AI to do operate in a virtual world with certain goals in mind, but to shutdown immediately if it detects that the outside world -- us -- is having a...
925db498-e574-4c8d-8031-d02d0421bf3b
trentmkelly/LessWrong-43k
LessWrong
Evaluating “What 2026 Looks Like” So Far Summary In 2021, @Daniel Kokotajlo wrote What 2026 Looks Like, in which he sketched a possible version of each year from 2022 - 2026. In his words: > The goal is to write out a detailed future history (“trajectory”) that is as realistic (to [him]) as [he] can currently manage ...
8d28a11b-04a7-42b8-8c9f-f4bfca7d4d45
trentmkelly/LessWrong-43k
LessWrong
Survey advice Things I believe about making surveys, after making some surveys: 1. If you write a question that seems clear, there’s an unbelievably high chance that any given reader will misunderstand it. (Possibly this applies to things that aren’t survey questions also, but that’s a problem for another time.) 2....
f7259470-9270-4b16-ae09-2ef90dde8db2
trentmkelly/LessWrong-43k
LessWrong
Priorities for the UK Foundation Models Taskforce The UK government recently established the Foundation Models Taskforce, focused on AI safety and backed by £100M in funding. Founder, investor and AI expert Ian Hogarth leads the new organization. The establishment of the Taskforce shows the UK’s intention to be a lea...
6247f7a4-b42b-4ac2-96bc-203f618ee464
StampyAI/alignment-research-dataset/arxiv
Arxiv
Bridging In- and Out-of-distribution Samples for Their Better Discriminability 1 Introduction --------------- Detecting out-of-distribution (OOD) samples, i.e., samples from a distribution other than the distribution of the samples used for training (called ‘in-distribution (ID)’ samples), is a vital problem to cop...
ec38112d-552d-4908-9292-33d990f5d0c4
trentmkelly/LessWrong-43k
LessWrong
The AI Boom Mainly Benefits Big Firms, but long-term, markets will concentrate > 'Market values strongly suggest a need for review. Especially now that competition from China is curtailed.' Bengt Holmström, MIT > > 'I fear Google is becoming the new old Microsoft, before the antitrust case. They may not be doing subs...
a2b9d0ba-bdd8-4c94-b204-e6a1c21bf587
trentmkelly/LessWrong-43k
LessWrong
Regularly Scheduled: Day-of Reminders Two years ago I wrote Regularly Scheduled, a tool to make it easier for people to handle repeated events: I mostly use it for a recurring dinner we host with some friends from college. It works well, and has been running with basically no maintenance. After twice now forgetting ...
839a6bd2-2902-45fd-b2bc-f19d97587bb9
trentmkelly/LessWrong-43k
LessWrong
Effective altruism and criticism toward activism: Answer to a paradox Since a little while now, I have been exploring the notion of Effective Altruism - EA for short. My readings on the topic so far have been very interesting [1], and I would like to add my own idea that I deem important and have yet to read elsewhere...
7032029c-bac2-44c9-bd6d-a261c4b66dad
trentmkelly/LessWrong-43k
LessWrong
Specificity: Your Brain's Superpower Introduction What if there was a magic superpower that simultaneously enabled you to: * Demolish bad arguments * Judge startup ideas * Make scientific breakthroughs * Solve climate change * Understand “God” * Be emotionally mature * Teach concepts better * Draw better * B...
4c0d2e9b-8512-4136-9e54-133df53d6b3c
trentmkelly/LessWrong-43k
LessWrong
[LINK] Behind the Shock Machine: book reexamining Milgram obedience experiments There's a book called Behind the Shock Machine by psychologist Gina Perry, published just a week ago, which investigates the original Milgram obedience experiments. I haven't read it, but I've read a summary / editorial published in the Pa...
d3a1c6b9-2d34-4f58-93d1-80c5e57ac4f7
trentmkelly/LessWrong-43k
LessWrong
George Orwell's Prelude on Politics Is The Mind Killer I have found this most wonderful (if fairly lengthy) article, and thought you would enjoy having it brought to your attention.   It is so good, I think we should include it among the references in the "Politics is the Mind Killer" wiki page.  But, before that, I...
749a2750-7684-4df6-b0d3-439d7555e8c4
trentmkelly/LessWrong-43k
LessWrong
May 2013 Media Thread This is the monthly thread for posting media of various types that you've found that you enjoy. Post what you're reading, listening to, watching, and your opinion of it. Post recommendations to blogs. Post whatever media you feel like discussing! To see previous recommendations, check out the old...
2b0c67e0-3f91-45cf-9d41-9a19c55dda8c
trentmkelly/LessWrong-43k
LessWrong
Rationalist Sport This post is a bit of an experiment; Most of the time, Discussion post lay out an idea and this idea then get commented upon. This post, on the other hand, will be purely about discussion on a topic. If this works out well, I'll might post more of these in the future. On to the meat of this post:  ...
db57eb47-108a-4676-998b-0a27eb32f75e
trentmkelly/LessWrong-43k
LessWrong
You Can Get Fluvoxamine [TLDR: I paid $95 for a 10 minute video consultation with a doctor, told them I was depressed and wanted fluvoxamine, and got my prescription immediately.] I'm not a doctor, and this isn't medical advice. If you want information on the status of fluvoxamine as a Covid treatment, you can see th...
1d163493-0621-4b91-b8f3-b7577bb49964
trentmkelly/LessWrong-43k
LessWrong
[Team Update] Why we spent Q3 optimizing for karma In Q3 of 2019, the LessWrong team picked the growth of a single metric as our only goal. For the duration of this quarter, the overwhelming consideration in our decision-making what would most increase the target metric. Why target a simple, imperfect metric? The Le...
61981c5f-8f7d-486d-bf22-862df86ab491
awestover/filtering-for-misalignment
Redwood Research: Alek's Filtering Results
id: post1217 Thanks to Paul Christiano and Kate Woolverton for useful conversations and feedback. In " Conditioning Predictive Models ," we devote a lot of effort into trying to understand how likely predictive models are compared to other alternatives in realistic training regimes (and if we do get a predictive model...
ac6669b5-08f3-4ab3-89fa-2cbe96a409a2
trentmkelly/LessWrong-43k
LessWrong
Finance and AI Timelines This is in part a response to AI 2027, which I think is rather vague and somewhat unrealistic in projecting the financial aspects of the AI future.  As a jumping off point, AI 2027 projects that a year from now, OpenAI’s valuation will have roughly quadrupled to $1 trillion and data center inv...
2264b116-aab4-4c41-88af-80c592175d5f
trentmkelly/LessWrong-43k
LessWrong
Know thyself vs. know one another People often aspire to the ideal of honesty, implicitly including both honesty to themselves and honesty with others. Those who care about it a lot often aim to be as honest as they can bring themselves to be, across circumstances. If the aim is to get correct information to yourself ...