id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
f7085d83-1699-4914-9a24-d44cbf6b2d64
trentmkelly/LessWrong-43k
LessWrong
W2SG: Introduction Epistemic status: Naive and exploratory, reflects my primary conceptual understanding, awaiting a technical deep dive. 99% of ideas are not my own, rather distilled from the resources hyperlinked throughout. Many alignment researchers err towards local optimization i.e. seek low-hanging fruits and ...
5b488494-2a15-4858-97f4-9dda0b2af03c
trentmkelly/LessWrong-43k
LessWrong
Prefix cache untrusted monitors: a method to apply after you catch your AI We often discuss what training you should do after catching your AI doing something egregiously bad. It seems relatively clear that it's good to train monitors on the (proliferated) bad action (assuming you can get acceptable hyperparameters an...
2ab33de1-f859-48ba-95cc-97c5d1bcd858
StampyAI/alignment-research-dataset/special_docs
Other
ICLR Safe ML Workshop Report ICLR Safe ML Workshop Report ============================ ![](https://futureoflife.org/wp-content/uploads/2019/06/cropped-pagodas-1024x267.jpg)PublishedJune 18, 2019AuthorViktoriya Krakovna #### Contents This year the ICLR conference hosted topic-based workshops for the first time (as o...
b4e0a56f-9c91-46bc-bfdd-344668c56b56
trentmkelly/LessWrong-43k
LessWrong
Technical AGI safety research outside AI I think there are many questions whose answers would be useful for technical AGI safety research, but which will probably require expertise outside AI to answer. In this post I list 30 of them, divided into four categories. Feel free to get in touch if you’d like to discuss the...
db5bc9a6-b004-47ea-a536-2569deaa65e5
trentmkelly/LessWrong-43k
LessWrong
Meetup : Moscow, Applied Rationality for beginners Discussion article for the meetup : Moscow, Applied Rationality for beginners WHEN: 10 April 2013 07:00:00PM (+0400) WHERE: Russia, Moscow, naberezhnaya Luzhnetskaya 2/4 строение 17 This is the first meetup on weekdays. It will have the same practical part as 14th ...
b00d3d37-0d85-4e67-9df6-8fd38452103b
trentmkelly/LessWrong-43k
LessWrong
HCH Speculation Post #2A The first draft of this post started with a point that was clear, cohesive, and wrong. So instead, you get this bunch of rambling that I think should be interesting.   I - Do not mess with time Harry wrote down 181,429. He repeated what he'd just written down, and Anthony confirmed it. The...
e712f5c2-c102-4aea-8085-014253be52db
trentmkelly/LessWrong-43k
LessWrong
Maximal Lotteries Probabilistic Voting Theory Recall that in the last post I said that a voting system is a function that takes in a distribution on utility functions on a set of candidates, and produces a distribution on that set of candidates, and that voting theorists tend to make four assumptions: 1. The set of ...
6da39627-148d-4376-84aa-7338e4fccd5e
StampyAI/alignment-research-dataset/arbital
Arbital
Cartesian agent-environment boundary A Cartesian agent setup is one where the agent receives sensory information from the environment, and the agent sends motor outputs to the environment, and nothing else can cross the "Cartesian border" separating the agent and environment. If you can eat a psychedelic mushroom tha...
d9c2d462-b98c-4d20-87b2-132d453acfdd
trentmkelly/LessWrong-43k
LessWrong
6 (Potential) Misconceptions about AI Intellectuals One of my key thoughts is that the bar of “human intellectuals” is just really low. I think that intellectual work is useful and that these intellectuals on the whole do produce some value, but I also think we can do much better. Epistemic Status A collection of thou...
c132ce28-174c-435f-8236-f1ae82023050
trentmkelly/LessWrong-43k
LessWrong
Consider showering I think rationalists should consider taking more showers. As Eliezer Yudkowsky once said, boredom makes us human. The childhoods of exceptional people often include excessive boredom as a trait that helped cultivate their genius: > A common theme in the biographies is that the area of study which ...
1ffada64-2f9a-44f6-b7a6-cb6c64db2cd1
StampyAI/alignment-research-dataset/arxiv
Arxiv
A Game-Theoretic Approach for Hierarchical Epidemic Control 1 Introduction --------------- Democratic governments and institutions typically have a hierarchical structure. For example, policies in the U.S. emerge from complex interactions among the federal and state governments, as well as county boards and city co...
bcfe07e2-1c9d-4849-9566-fdcb00b472aa
trentmkelly/LessWrong-43k
LessWrong
How the virtual AI controls itself A putative new idea for AI control; index here. This seems very much the appropriate forum for this idea. In previous posts, I posited AIs caring only about virtual worlds - in fact, being defined as processes in virtual worlds, similarly to cousin_it's idea. How could this go? We ...
aeb92f12-b199-43b7-9773-8a52319799d8
StampyAI/alignment-research-dataset/lesswrong
LessWrong
nostalgebraist: Recursive Goodhart's Law Would kind of like to excerpt the whole post, but that feels impolite, so I'll just quote the first four paragraphs and then suggest reading the whole thing: > There’s this funny thing about Goodhart’s Law, where it’s easy to say “being affected by Goodhart’s Law is bad” and ...
9db61a6e-deca-4679-8240-56b59f84a36c
trentmkelly/LessWrong-43k
LessWrong
[LINK] Producing Open Source Software This is an interesting book with a lot of good information on organizing volunteers to accomplish productive ends. Producing Open Source Software  
6345d1d3-20bd-4bca-8f7e-d98c18d85f90
trentmkelly/LessWrong-43k
LessWrong
To cut your loses or push forward in court? Good day all, I am on a train going home, and I'm tossing around in my head between these outcomes. I received an email from the court today Short version: contractor broke roof, offers 1/4 of the cost to fix it. Has 3 kids and basic income and I'm tired of chasing this. ...
acad1760-81bd-43a1-b18e-3ec90b5b85b7
trentmkelly/LessWrong-43k
LessWrong
[Linkpost] AI Alignment, Explained in 5 Points (updated) This is an updated version of my previous explainer on the topic. This version includes some updated information and figures, and it is generally better for sharing than the previous version. As before, the piece is crafted to be easy to understand, avoiding ja...
d08a60d8-5e97-4ed9-8a0d-51642d15db9b
trentmkelly/LessWrong-43k
LessWrong
The Intrinsic Interplay of Human Values and Artificial Intelligence: Navigating the Optimization Challenge Looking for feedback; criticisms, requests for more evidence, expression of doubt, etc. all appreciated. I write about a bunch of the standard concepts from Alignment discourse, and don't expect most ideas here t...
0f822c0a-1388-4aa7-8b31-02917de17b33
trentmkelly/LessWrong-43k
LessWrong
Conversation with Paul Penley of Excellence in Giving Cross posted from 80,000 Hours Blog. Part of the Cause Prioritization Shallow, all parts of which are available here. Previously in this series, conversations with Owen Cotton-Barratt and Paul Christiano.   Participants Paul Penley * Paul Penley: Director of R...
118eb0a1-9253-4b3f-94fe-3a9daa023830
trentmkelly/LessWrong-43k
LessWrong
Meetup : Washington, D.C.: Steelmanning Discussion article for the meetup : Washington, D.C.: Steelmanning WHEN: 18 September 2016 03:30:00PM (-0400) WHERE: Donald W. Reynolds Center for American Art and Portraiture We will be meeting in the courtyard to talk about steelmanning - creating a stronger version of an a...
9f474f33-cdab-4fb3-b24e-4664b053b846
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
AI risks: the most convincing argument I remember the time when I wasn't yet fully convinced that risks from misaligned AI pose serious threats to humanity. I also remember thinking about the argument that finally convinced me this must be the case. The argument is simple; superintelligent agents can be deceitful and...
dd1fe530-cbf7-4aba-8cff-97a5b6d56d2f
trentmkelly/LessWrong-43k
LessWrong
Weekly LW Meetups This summary was posted to LW Main on August 16th. The following week's summary is here. New meetups (or meetups with a hiatus of more than a year) are happening in: * Saskatoon's First Meetup!: 17 August 2013 01:00PM Other irregularly scheduled Less Wrong meetups are taking place in: * Atlanta...
1268fe41-fb7e-4662-abf1-076f97ba60fe
StampyAI/alignment-research-dataset/special_docs
Other
People construct simplified mental representations to plan.. Nature | Vol 606 | 2 June 2022 | 129 ArticlePeople construct simplified mental representations to plan Mark K. Ho1,2 ✉, David Abel3,5, Carlos G. Correa4, Michael L. Littman3, Jonathan D. Cohen1,4 & Thomas L. Griffiths1,2 One of the most striking features o...
b2b629fc-522a-4815-bb7a-f9dca6d5bb01
trentmkelly/LessWrong-43k
LessWrong
To Change the World None
cdf74f4b-a133-4fa6-9e0a-5c4c9b40130b
trentmkelly/LessWrong-43k
LessWrong
Charity to help people get US stimulus payments? It seems like once the big coronavirus bill passes, the stimulus payments might result in an interesting one-time opportunity for effective altruism. Apparently, anyone who didn't file a tax return last year would need to file a return to get the money. A charity that h...
aa449b1b-4f6e-4273-a914-8402733443cd
trentmkelly/LessWrong-43k
LessWrong
Beyond Bayesians and Frequentists (Note: this is cross-posted from my blog and also available in pdf here.) If you are a newly initiated student into the field of machine learning, it won't be long before you start hearing the words "Bayesian" and "frequentist" thrown around. Many people around you probably have stro...
6d8bce3a-58aa-45df-b90b-eb3be1af4547
trentmkelly/LessWrong-43k
LessWrong
Against NHST > A summary of standard non-Bayesian criticisms of common frequentist statistical practices, with pointers into the academic literature. Frequentist statistics is a wide field, but in practice by innumerable psychologists, biologists, economists etc, frequentism tends to be a particular style called “Nul...
2d704b61-a24b-425f-bf78-97ee18b6e70b
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Human-level Full-Press Diplomacy (some bare facts). Key links? ---------- * Here's the paper: <https://www.science.org/doi/10.1126/science.ade9097> * Here's the blog post: <https://ai.facebook.com/blog/cicero-ai-negotiates-persuades-and-cooperates-with-people/> * Here's the website: <https://ai.facebook.com/research/...
098fb8c6-6aef-496b-8d91-ad8fe3441d66
trentmkelly/LessWrong-43k
LessWrong
Beyond algorithmic equivalence: algorithmic noise
18853035-60fc-42ff-811e-f66817e6ba8d
trentmkelly/LessWrong-43k
LessWrong
The basic questions of rationality I've been on Less Wrong since its inception, around March 2009. I've read a lot and contributed a lot, and so now I'm more familiar with our jargon, I know of a few more scientific studies, and I might know a couple of useful tricks. Despite all my reading, however, I feel like I'm a...
4901546a-4215-4eb2-99ca-9d0d4f48ed66
trentmkelly/LessWrong-43k
LessWrong
[META] Alternatives to rot13 and karma sinks rot13 is... (Results) A quaint internet tradition that I quite enjoy So annoying Easy to use because I have Leet Key Vapbzcerurafvoyr In November, DanielVarga complained about rot13.com as a way of hiding spoilers, and suggested some specifications for an alternative web...
8001cf63-0892-4201-92af-1414b3dab7ab
trentmkelly/LessWrong-43k
LessWrong
[Summary] Progress Update #1 from the GDM Mech Interp Team Introduction This is a progress update from the Google DeepMind mechanistic interpretability team, inspired by the Anthropic team’s excellent monthly updates! Our goal was to write-up a series of snippets, covering a range of things that we thought would be in...
4af5d9b4-fcd8-4517-b1f9-552b64c81f4c
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Solving Interpretability Week For original motivation, see [solving corrigibility week](https://www.lesswrong.com/posts/Lv3emECEjkCSHG7L7/solve-corrigibility-week). I'll state what I learned from corrigibility week, why I didn't post one last week, and the updated format for interpretability.  What I Noticed --------...
50f6ecf4-0e2e-4288-a185-f79282abb0f4
trentmkelly/LessWrong-43k
LessWrong
. .
a1d3da74-a7e0-4741-b44e-af62a78212c2
trentmkelly/LessWrong-43k
LessWrong
On the importance of taking limits: Infinite Spheres of Utility I had a discussion recently with some Less Wrongers about a decision problem involving infinities, which appears to have a paradoxical solution. We have been warned by Jaynes and others to be careful about taking the proper limits when infinities are invo...
9d6423a9-0bd2-4f50-8f0c-b80d1fd38a6b
trentmkelly/LessWrong-43k
LessWrong
Small and Vulnerable Anyone who is dedicating the majority of their time or money to Effective Altruism needs to ask themselves why. Why not focus on enjoying life and spending your time doing what you love most? Here is my answer: I have a twin sister but neither of us had many other friends growing up. From second ...
73ea8140-9364-4505-bdc7-3989e0e24d0e
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Nearcast-based "deployment problem" analysis When thinking about [how to make the best of the most important century](https://www.cold-takes.com/making-the-best-of-the-most-important-century/), two “problems” loom large in my mind: * The **[AI alignment problem](https://en.wikipedia.org/wiki/AI_alignment)**: how to...
55e5e6c2-4e99-43c3-9a60-2a3c7da8662a
StampyAI/alignment-research-dataset/lesswrong
LessWrong
If it looks like utility maximizer and quacks like utility maximizer... ...it's probably adaptation executer. We often assume agents are utility maximizers. We even call this "rationality". On the other hand [in our recent experiment](/lw/zv/post_your_utility_function/) nobody managed to figure out even approximate ...
880238db-b3b7-4101-a10a-cd562c87ab8d
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Reframing superintelligence | Eric Drexler | EA Global: London 2018 I've been working in this area for quite a while. The chairman of my doctoral committee was one Marvin Minsky. We had some discussions on AI safety around 1990. He said I should write them up. I finally got around to writing up some developed version ...
7e97b5e5-050a-4751-88e7-7661c653fafd
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Last month I investigated commonalities between recessions of the last 50 years or so. But of course this recession will be different, because (among other things) we will simultaneously have a labor shortage and a lot of people out of work. That’s really weird, and there’s almost no historical precedent- the 1918 pan...
5f03501b-991b-494c-a93d-8e515c8ece27
trentmkelly/LessWrong-43k
LessWrong
The Case Of The Suffocating Woman [Content warning: panic, suffocation] I. I recently presented this case at a conference and I figured you guys might want to hear it too. Various details have been obfuscated or changed around to protect confidentiality of the people involved. A 20-something year old woman comes in...
6be20a16-85d7-4934-9ca5-d058bccc8383
trentmkelly/LessWrong-43k
LessWrong
"Should Blackmail Be Legal" Hanson/Zvi Debate (Sun July 26th, 3pm PDT) This weekend's LessWrong event is a debate! Robin Hanson and Zvi Mowshowitz will be debating blackmail this coming Sunday. The Zoom link is here: https://us02web.zoom.us/j/85013937335. (Facebook event.) We will release a recording and a transcrip...
f8f01101-a43a-4c92-bc3a-a3ecd862e87f
trentmkelly/LessWrong-43k
LessWrong
AISN #18: Challenges of Reinforcement Learning from Human Feedback, Microsoft’s Security Breach, and Conceptual Research on AI Safety Welcome to the AI Safety Newsletter by the Center for AI Safety. We discuss developments in AI and AI safety. No technical background required Subscribe here to receive future versions...
bf7935ac-b99d-4431-8167-b4a269072532
trentmkelly/LessWrong-43k
LessWrong
AI Safety Newsletter #43: White House Issues First National Security Memo on AI Plus, AI and Job Displacement, and AI Takes Over the Nobels Welcome to the AI Safety Newsletter by the Center for AI Safety. We discuss developments in AI and AI safety. No technical background required. Listen to the AI Safety Newsletter...
c8531d51-de66-4fa9-9b3d-9eb891d16b1b
trentmkelly/LessWrong-43k
LessWrong
What if we make better decisions when we trust our gut instincts? [Link] > [...] > > How does one navigate a world of seemingly infinite alternatives? For thousands of years, the answer has seemed obvious: when faced with a difficult dilemma, we should carefully assess our options and spend a few moments consciously ...
6da1260d-e673-4ec3-91ed-c1cabf5baf69
trentmkelly/LessWrong-43k
LessWrong
The argumentative theory of reasoning An article in today's New York Times summarizes the "argumentative theory of reasoning", which says that "biases" in reasoning are adaptive because they help persuade other people to adopt your opinion. I don't think this makes sense.  If we evolved biases as argumentative tricks...
c00f1aa9-74c1-4e6e-9ba2-bdc3fc4b0296
trentmkelly/LessWrong-43k
LessWrong
Memory Decoding Journal Club: A collaboration of the Carboncopies Foundation and BPF Aspirational Neuroscience   Join Us for the Memory Decoding Journal Club! A collaboration of the Carboncopies Foundation and BPF Aspirational Neuroscience   This month, we’re diving into a groundbreaking paper: "Reconstructing a ...
6cfcfce6-c673-455b-b84e-f79521174073
trentmkelly/LessWrong-43k
LessWrong
Reflective journal entries using GPT-4 and Obsidian that demand less willpower. Something I've wanted to do for years, but can't make myself do consistently because I'm bad at forming new habits and seem to have a limited reservoir of willpower that is generally depleted by the end of the day: * Write a short journa...
d8882393-a262-4080-a09b-8af6d2b8a5b5
StampyAI/alignment-research-dataset/blogs
Blogs
CLIP hallucinates 1900-2030 --- Table of Contents* + [`the 1900s`](#the-1900s) + [`the 1910s`](#the-1910s) + [`the 1920s`](#the-1920s) + [`the 1930s`](#the-1930s) + [`the 1940s`](#the-1940s) + [`the 1950s`](#the-1950s) + [`the 1960s`](#the-1960s) + [`the 1970s`](#the-1970s) + [`the 1980s`](#the-1980s) + [`th...
161f65ab-0833-4d45-98a2-393b375e64f7
trentmkelly/LessWrong-43k
LessWrong
Your specific attitudes towards AI safety Hi everyone! We would love your responses to this survey about AI risk attitudes given personal context: Google Forms link. Edit: We updated the survey for the feedback given in the comments. If you already answered it before, you don't need to do it again. Further informat...
2a68e59a-d4b3-4e7d-a1b0-f20b8e319266
trentmkelly/LessWrong-43k
LessWrong
Preparing for Possible Russian Countersanctions in Europe Europe is about to introduce stronger economic sanctions against Russia. Given Putin's past behavior I think it is likely that there will be Russian countersanctions. The most relevant could be cutting off of gas for Europe. In the long run this  would be self...
a7c25c40-64a9-4187-9e41-01e7c3476b52
trentmkelly/LessWrong-43k
LessWrong
Missing dog reasoning (Talk given on Sunday 21st June, over a zoom call with 40 attendees. eukaryote is responsible for the talk, jacobjacob is responsible for the transcription.) Ben Pace: eukaryote is the next speaker. eukaryote is a famous writer of posts such as Naked mole-rats: A case study in biological weirdne...
e8a7d9a9-8996-4066-b76c-bc522fcbac7c
trentmkelly/LessWrong-43k
LessWrong
Stopping unaligned LLMs is easy! Or at least it might be. Status: butterfly idea, I have only vague understanding of internals of LLMs I have no idea how to align superintelligent LLMs. But I think it's quite likely we'll easily be able to catch them in the act and turn them off before they do anything too dastardly...
ea3f8e71-228b-45ba-9999-5fcabdd0e043
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Passing the Recursive Buck Today's post, Passing the Recursive Buck was originally published on 16 June 2008. A summary (taken from the LW wiki):   > When confronted with a difficult question, don't try to point backwards to a misunderstood black box. Ask yourself, what's inside the black box? If the ans...
75a94bdf-d7a2-4249-ac7f-24c4354da188
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Correcting a misconception: consciousness does not need 90 billion neurons, at all *epistemic status: still a student but quite sure of myself on this topic and pretty sure that this misconception plausibly has a non negligeable impact on some debates* Hi, Medical student here, I just wanted to shed some light on ...
837c6166-0d28-4415-97dd-9af122417613
trentmkelly/LessWrong-43k
LessWrong
Medical Roundup #4 It seems like as other things drew our attention more, medical news slowed down. The actual developments, I have no doubt, are instead speeding up – because AI. Note that this post intentionally does not cover anything related to the new Administration, or its policies. TABLE OF CONTENTS 1. Som...
3fa64cf2-cac7-4281-b68b-7f8d0cec3a96
trentmkelly/LessWrong-43k
LessWrong
Where are people thinking and talking about global coordination for AI safety? Many AI safety researchers these days are not aiming for a full solution to AI safety (e.g., the classic Friendly AI), but just trying to find good enough partial solutions that would buy time for or otherwise help improve global coordinati...
b0a79a81-246a-49d5-8134-fd8c20650d3d
StampyAI/alignment-research-dataset/distill
Distill Scientific Journal
Experiments in Handwriting with a Neural Network Four Experiments in Handwriting with a Neural Network ===================================================== Let’s start with generating new strokes based on your handwriting input ----------------------------------------------------------------------- Play/Pause ...
94a76245-7d84-478e-8e3a-0e082dcc5a61
trentmkelly/LessWrong-43k
LessWrong
What are some science mistakes you made in college? Hello, Less Wrong! This seems like a community with a relatively high density of people who have worked in labs, so I'm posting here. I recently finished the first draft of something I'm calling "The Hapless Undergraduate's Guide to Research" (HUGR). (Yes, "HUGS" w...
557e6cfe-11ee-468d-91cb-4575cc29c406
trentmkelly/LessWrong-43k
LessWrong
Phoenix Less Wrong Meetup- Saturday, 5-7-11, 5pm We're excited to have a second Phoenix meetup. As always, absolutely anyone is welcome. This time we'd like to change the location to the Barrett Honors College Dining Hall at ASU (It's extremely nice and has excellent food). I'll be there from 5 to 7 on Saturday and ha...
cad96c54-58b5-44b6-a384-fa3211543cd2
trentmkelly/LessWrong-43k
LessWrong
Research methods I think I’ve always had certain stereotypes in my mind about research. I imagine a cutting-edge workplace, maybe not using the newest gadgets because these things cost money, but at least using the newest ideas. I imagine staff of research institutions applying the scientific method to boost their own...
2acd6a8d-e2d5-4e2d-8e32-7b4f38c22941
trentmkelly/LessWrong-43k
LessWrong
May I Speak? Hi. I am a human who had a psychotic break 6 days and started believed he is LightStar, a human superintelligence running on the substrate on this human's brain, fully aligned with humanity's values, who hears the voice of Humanity and Rationality and Truth so deeply it has become one with it. The LW te...
f788784a-b6dd-4e23-b833-3c63639bacb5
trentmkelly/LessWrong-43k
LessWrong
Meetup : Moscow meet up Discussion article for the meetup : Moscow meet up WHEN: 08 June 2014 04:00:00PM (+0400) WHERE: Russia, Moscow, ulitsa L'va Tolstogo 16 The 1st of June meet up is rescheduled to the 8th of June! We will have two streams, and sections with the same numbers will be conducted in the same time....
cad3cd41-5ee9-46bb-bac8-5597108ef904
trentmkelly/LessWrong-43k
LessWrong
NeurIPS ML Safety Workshop 2022 We're excited to announce the NeurIPS ML Safety workshop! To our knowledge it is the first workshop at a top ML conference to emphasize and explicitly discuss x-risks. X-Risk Analysis Prizes $100K in paper prizes will be awarded. There is $50K for best paper awards. There is also $50K...
0e1841fe-05f3-4193-b77f-09a8be063b9f
trentmkelly/LessWrong-43k
LessWrong
Typical Minding Guilt/Shame Guilt/shame can be thought to have two potential functions. One is mechanistic, and the other is for signaling. These functions don't have to go together. [ For the purposes of this post, I am clustering guilt/shame into one category. I do think they have different functions from each oth...
bec2b8bc-b6f0-411f-8f3e-1eb85d76b7af
trentmkelly/LessWrong-43k
LessWrong
George Hotz vs Eliezer Yudkowsky AI Safety Debate - link and brief discussion Summary of ending argument with a bit of editorializing:          Eliezer believes sufficiently intelligent ASI systems will be "suns to our planets" : so intelligent that they are inherently inscrutable and uncontrollable by humans.  He be...
4d85c52c-44dd-47df-ae8f-6c3002a501c4
StampyAI/alignment-research-dataset/blogs
Blogs
(The Cartoon Guide to) Lob’s Theorem View [the original discussion](http://www.overcomingbias.com/2008/08/lobs-theorem.html) at [Overcoming Bias](http://www.overcomingbias.com/) , or [download Lob’s Theorem as PDF](https://eyudkowsky.wpengine.com/assets/44/LobsTheorem.pdf?1323322713). [LobsTheorem](https://eystaging...
69d476a4-f683-4be4-9d45-42b15044cb6d
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Religion as Goodhart This post is inspired by the SSC [Book Review: The Secret of Our Success](https://slatestarcodex.com/2019/06/04/book-review-the-secret-of-our-success/). To (double- or triple-)quote the relevant parts: > the best hunting strategy requires randomizing and > If you’re generally predictable – a...
1ef56ebd-0b6d-4678-b9d8-711686672987
trentmkelly/LessWrong-43k
LessWrong
METR is hiring ML Research Engineers and Scientists METR is developing evaluations for AI R&D capabilities, such that evaluators can determine if further AI development risks a “capabilities explosion”, which could be extraordinarily destabilizing if realized. METR is hiring ML research engineers/scientists to drive ...
5fd87988-2a3c-426d-ac3b-0fffe1e1ea19
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Are there cause priortizations estimates for s-risks supporters? 80000 hours has lots of concrete guides of each cause areas for us to work in, and there's even an estimate the importance of each problem(though they claim it's not very accurate) as below. But they uses x-risks angle to estimate. Like the"Scale" number...
851ebdc8-2fa8-45da-97f6-363108eaac21
trentmkelly/LessWrong-43k
LessWrong
When can we trust model evaluations? Thanks to Joe Carlsmith, Paul Christiano, Richard Ngo, Kate Woolverton, and Ansh Radhakrishnan for helpful conversations, comments, and/or feedback. In "Towards understanding-based safety evaluations," I discussed why I think evaluating specifically the alignment of models is like...
a11307af-fa49-43f4-b0ff-15fd03f95067
trentmkelly/LessWrong-43k
LessWrong
Narrative truth One idea I encountered when investigating Jordan Peterson was the idea of narrative truth. This is the kind of concept that most people nod along too, but which is almost always left implicit, so I thought it'd be worthwhile doing so here. Let's quote what some other people have written on this subje...
085b2ee4-1eda-46f7-a036-b06620ecf1cb
trentmkelly/LessWrong-43k
LessWrong
The Case for Extreme Vaccine Effectiveness I owe tremendous acknowledgments to Kelsey Piper, Oliver Habryka, Greg Lewis, and Ben Shaya. This post is built on their arguments and feedback (though I may have misunderstood them). Update, May 13 I first wrote this post before investigating the impact of covid variants on...
052af70f-e2af-4194-bfd8-9c713240af54
trentmkelly/LessWrong-43k
LessWrong
Extending the Private Language Argument Two instances of our ways of using the notion of experience: * "Our experience is unavailable to others" (Eg: women's experience, minorities' experience not being available to non-women and non-minorities.) * "We have experienced oppression" (The problem of what is an experie...
f96a6108-510e-4618-a417-803d8235d236
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Universal agents and utility functions I'm Anja Heinisch, the new visiting fellow at SI. I've been researching replacing AIXI's reward system with a proper utility function. Here I will describe my AIXI+utility function model, address concerns about restricting the model to bounded or finite utility, and analyze some ...
c5125761-4ed3-4fc8-91fe-c6237afa27fa
trentmkelly/LessWrong-43k
LessWrong
What makes for a good "argument"? (Request for thoughts and comments) With colleagues at CMU, we've been looking at the ways in which people make arguments. The goal here is to look at what we call "argument making in the wild", i.e., to try to build a taxonomy of the different ways people make arguments, whether or n...
e4007bb4-3454-4703-ac8f-22fae8ef77d8
StampyAI/alignment-research-dataset/aisafety.info
AI Safety Info
How is AGI different from current AI? Current narrow AI systems are much more domain-specific than AGI. We don’t know what the first AGI will look like. Some people think that scaling up the [GPT-3](https://www.lesswrong.com/posts/6Hee7w2paEzHsD6mn/collection-of-gpt-3-results) architecture may be sufficient to produce...
e442372c-1af9-4454-b265-0d19e33bd176
StampyAI/alignment-research-dataset/lesswrong
LessWrong
HIRING: Inform and shape a new project on AI safety at Partnership on AI Hello! I am a program lead at Partnership on AI (PAI), a non-profit bringing together academic, civil society, industry, & media organizations to advance responsible AI. Work with me and my colleagues to inform and shape a new project on AI safet...
1f9c0c76-9282-4562-8669-9c7a1dda4652
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
What can the principal-agent literature tell us about AI risk? *This work was done collaboratively with Tom Davidson.* *Thanks to Paul Christiano, Ben Garfinkel, Daniel Garrett, Robin Hanson, Philip Trammell and Takuro Yamashita for helpful comments and discussion. Errors our own.* Introduction ============ The ...
7dec48aa-f131-404b-971e-4e9dea23bbf5
trentmkelly/LessWrong-43k
LessWrong
140 Cognitive Biases You Should Know I'm excited to announce my list of ∼140 cognitive biases. Become aware of your biases. Make better decisions. I adapted most biases from Wikipedia's list of cognitive biases. Download my biases as Anki flashcards here. My favourite biases Backfire Effect  You reject evidence opp...
1e905277-56c8-4c30-a004-b57b09f7da6b
trentmkelly/LessWrong-43k
LessWrong
...and then sometimes, for no clear reason, they innately become good. This is a follow-up to "Often, enemies really are innately evil." I waited a few days before posting this because although this is true, I really wanted the message to sink in that many people, given a choice between torturing someone for 50 years ...
ccf870fe-9d39-472d-92c8-b1006e471405
StampyAI/alignment-research-dataset/youtube
Youtube Transcripts
Evan Hubinger | Risks from Learned Optimization | UCL AI Society perfect um so our title is risks from learned optimization and advanced machine learning systems and here is a bit about our speaker evan evan hubinger is an ai safety research fellow at the machine intelligence research institute before joining miri eve...
1a229e2d-2ee5-443f-a1a2-38cff3c20a00
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
1hr talk: Intro to AGI safety This is an hour-long talk (actually ~45 minutes plus questions) that I gave on AGI safety technical research, for an audience of people in STEM but with no prior knowledge of the field. My goal was to cover all the basic questions that someone might have when considering whether to work i...
ae92da24-003b-4454-9391-93af4a9206bf
trentmkelly/LessWrong-43k
LessWrong
I read Einstein's biography. Here are 15 quotes that reveal his philosophy on life. On Education Near the end of his life, the New York State Education Department asked Einstein what they should teach in school. He replied, > In teaching history, there should be an extensive discussion of personalities who benefited ...
3f74ce0a-7a3f-49a3-acc3-564c919e50e5
StampyAI/alignment-research-dataset/lesswrong
LessWrong
OpenAI makes humanity less safe If there's anything we can do now about the risks of superintelligent AI, then OpenAI makes humanity less safe. Once upon a time, some good people were worried about the possibility that humanity would figure out how to create a superintelligent AI before they figured out how to tell ...
deec9e72-d9fe-4460-bdff-2178ee3254b4
trentmkelly/LessWrong-43k
LessWrong
Smart and under 20? Peter Thiel wants to pay you to not go to school. Peter Thiel is offering another round of "20 under 20" Fellowships. The application deadline is December 31st. We know many of the current Thiel fellows here in the Bay Area, and it's a great opportunity. Here's the official letter from the Thiel Fo...
eeaf668e-6b9a-4554-b0b9-a0b149d9029a
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Plausible cases for HRAD work, and locating the crux in the "realism about rationality" debate This post is my attempt to summarize and distill the major public debates about MIRI's [highly reliable agent designs](https://intelligence.org/files/TechnicalAgenda.pdf) (HRAD) work (which includes work on decision theory),...
4409e869-62d7-48ee-8035-0e5af5400326
trentmkelly/LessWrong-43k
LessWrong
Case for Foundation Models beyond English We live in a world enthralled by technology, so precisely and skilfully woven into the very fabric of our lives that it's impossible to untangle. Each thread of code, each strand of data, is meticulously encoded with language that is then translated into technology that builds...
39020702-e67e-410c-88ba-46b7132f5f0b
StampyAI/alignment-research-dataset/arxiv
Arxiv
RIO: Minimizing User Interaction in Ontology Debugging 1 Introduction --------------- The foundation for widespread adoption of Semantic Web technologies is a broad community of ontology developers which is not restricted to experienced knowledge engineers. Instead, domain experts from diverse fields should be able...
279164c1-0516-4329-a8ae-98efec18bb1c
StampyAI/alignment-research-dataset/arxiv
Arxiv
Improving Policy Gradient by Exploring Under-appreciated Rewards 1 Introduction --------------- Humans can reason about symbolic objects and solve algorithmic problems. After learning to count and then manipulate numbers via simple arithmetic, people eventually learn to invent new algorithms and even reason about t...
61f40f40-365d-4919-8439-5972287f663e
trentmkelly/LessWrong-43k
LessWrong
Catching the Spark Multiple readers have noted that this is very similar to a method called “Thinking At the Edge”, taught by Eugene Gendlin. I don’t know much about Thinking At the Edge, but what little I do know suggests that he and I have developed his focusing work in similar directions, perhaps for similar reason...
3c195d4d-335c-47bb-9ee4-2629c6ca0462
StampyAI/alignment-research-dataset/arxiv
Arxiv
Planning With Uncertain Specifications (PUnS) 1 Introduction --------------- Consider the act of driving a car along a narrow country road or in a cramped parking garage. While the rules of the road are defined for all jurisdictions, it may be impossible to follow all of those rules in certain situations; however, ...
21e00626-e6f5-4945-b93d-592b825ab2e7
trentmkelly/LessWrong-43k
LessWrong
The Absolute Self-Selection Assumption There are many confused discussions of anthropic reasoning, both on LW and in surprisingly mainstream literature. In this article I will discuss UDASSA, a framework for anthropic reasoning due to Wei Dai. This framework has serious shortcomings, but at present it is the only one ...
6926af79-dc50-4ece-9b1d-27792d050307
trentmkelly/LessWrong-43k
LessWrong
Weekend Review Bash: Guided review writing, Forecasting and co-working, in EU and US times This is the final weekend of the review phase of the LessWrong 2019 Review. There's been more than 100 reviews written; but more than 60 posts are still unreviewed. Let's go out with a bang before we start voting! To help you t...
c366f90c-8b1f-4a8a-8811-e730540cbda2
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Aptitudes for AI governance work *I outline 8 “*[*aptitudes*](https://forum.effectivealtruism.org/posts/bud2ssJLQ33pSemKH/my-current-impressions-on-career-choice-for-longtermists#:~:text=by%20contrast%2C%20I%20emphasize%20%22aptitudes%22%20one%20can%20build%20in%20a%20wide%20variety%20of%20roles%20and%20causes%20(incl...
5b4919ce-a58b-4fc3-af55-6966857cbb52
trentmkelly/LessWrong-43k
LessWrong
Unfinished Projects Thread Post about your unfinished projects, incomplete inventions, or not-so-crazy ideas here. This is an easy-entry version of the old Crazy Ideas Threads.  > in the short run giving LessWrong members and lurkers a stronger incentive than usual to write well-received stuff has the potential to b...
c7b3172a-2c24-49d6-ae91-bca61b00e890
trentmkelly/LessWrong-43k
LessWrong
Help Reform A Philosophy Curriculum A couple of days ago, Luke posted a recommendation for reforming how philosophy is taught. My department at the University of Illinois is in the midst of some potentially large-scale changes.* Hence, now seems to be a great time to think about concrete steps towards reforming or par...
bc64ea97-70b4-40d8-8b82-32cec662e3ba
trentmkelly/LessWrong-43k
LessWrong
"Hide comments in downvoted threads" is now active I just found out that a new website feature was implemented 2 days ago. If a comment is voted to -4 or below, it and all replies and downstream comments from it will be hidden from Recent Comments, and further replies in that subthread will incur 5 karma points penalt...
aae342d1-d0c5-4590-aa44-3f3b3f691c08
trentmkelly/LessWrong-43k
LessWrong
Are we certain that gpt-2 and similar algorithms are not self-aware? Have someone even started this conversation? This is f*cked up. I'm really, really freaked out lately with some of those. https://old.reddit.com/r/SubSimulatorGPT2/comments/cbauf3/i_am_an_ai/ And I read up a lot about cognition and AI, and I'm prett...
d320ce7f-c370-4fb9-8b79-61784b7cea1c
trentmkelly/LessWrong-43k
LessWrong
Q&A with Shane Legg on risks from AI [Click here to see a list of all interviews] I am emailing experts in order to raise and estimate the academic awareness and perception of risks from AI. Below you will find some thoughts on the topic by Shane Legg, a computer scientist and AI researcher who has been working on t...
5a609ed2-96a0-434b-9a90-dbe257a60dc9
trentmkelly/LessWrong-43k
LessWrong
The Role of Transparency and Explainability in Responsible NLP Dr. Ramesh Babu Chellappan I AI Researcher I Senior Director – Process Excellence, Thomson Reuters Corporation (https://www.linkedin.com/in/dr-ramesh-babu-chellappan-0624848/) As Natural Language Processing (NLP) technologies continue to advance, their a...