id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
337fc0f6-7acd-4ac9-953b-7d46a508d817 | trentmkelly/LessWrong-43k | LessWrong | When discussing AI risks, talk about capabilities, not intelligence
Public discussions about catastrophic risks from general AI systems are often derailed by using the word “intelligence”. People often have different definitions of intelligence, or associate it with concepts like consciousness that are not relevant to... |
415d192f-62db-4e52-a4ef-1c86387d1281 | trentmkelly/LessWrong-43k | LessWrong | There Should Be More Alignment-Driven Startups
Many thanks to Brandon Goldman, David Langer, Samuel Härgestam, Eric Ho, Diogo de Lucena, and Marc Carauleanu, for their support and feedback throughout.
Most alignment researchers we sampled in our recent survey think we are currently not on track to succeed with alignm... |
f50302ae-b117-4cd7-8318-fb65a7ef7c1a | trentmkelly/LessWrong-43k | LessWrong | MetaAI: less is less for alignment.
Summary
In May 2023, MetaAI submitted a paper to arxiv called LIMA: Less Is More for Alignment. It's a pretty bad paper and (in my opinion) straightforwardly misleading. Let's get into it.
The Superficial Alignment Hypothesis
The authors present an interesting hypothesis about LLM... |
ba573990-f689-4a24-90f4-c5e4d2393c28 | trentmkelly/LessWrong-43k | LessWrong | Corporate ecology
Direct competition is resource intensive. Just to compete, species and companies have to invest heaps of energy in long trunks and roots, extra hunting and massive advertising campaigns for instance, instead of expanding or improving production. To avoid these costs they move into niches. Where there... |
62d2858c-b108-4a3c-bd46-ef3977105352 | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] The Fear of Common Knowledge
Today's post, The Fear of Common Knowledge was originally published on 09 July 2008. A summary (taken from the LW wiki):
> A discussion of an interesting kind of lie, in which someone tells a lie that the person they're speaking to knows is a lie, but doesn't know that the per... |
cc2d4401-0845-4d1f-a291-a5cf518e0fb6 | trentmkelly/LessWrong-43k | LessWrong | Relevance of the STEPS project to (F)AI teams
The View Point Reasearch Institute http://www.vpri.org/ (which by the way brought us the squeek etoys) has a project called STEPS Toward The Reinvention of Programming.
The goal is basically to create a complete operating system, tool chain and program suite in 20000 line... |
d423e242-70d0-43fa-9645-80b756508a77 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Best resources to learn philosophy of mind and AI?
I would like to learn some of the philosophy of mind and AI. I wonder if anyone could point to some of the best resources on this topic.
I am able to teach myself on the technical and programming side of LLMs, but I am kind of lost on the philosophy side. |
bd13afe6-6241-4a9a-8d6c-8e0d4af46529 | trentmkelly/LessWrong-43k | LessWrong | Contra shard theory, in the context of the diamond maximizer problem
A bunch of my response to shard theory is a generalization of how niceness is unnatural. In a similar fashion, the other “shards” that the shard theory folk want to learn are unnatural too.
That said, I'll spend a few extra words responding to the a... |
e98f26f4-7158-43b4-b83d-8851623f2274 | trentmkelly/LessWrong-43k | LessWrong | How could the universe be infinitely large?
I ask this question in the context of thinking about infinite ethics. It seems to me that I would need to give up at least one of three beliefs about the universe in order to coherently think that it is infinitely large. They are:
(i) at some past point in time t the univer... |
74321ea2-dae3-41c2-885e-65c96ae8fed9 | trentmkelly/LessWrong-43k | LessWrong | The Relationship Between the Village and the Mission
Epistemic Status: Braindump, not as well thought out as I’d like.
Previously:
* Project Hufflepuff
* The Archipelago Model of Community Standards
* The Craft is not the Community
* What is Rationalist Berkeley's Community Culture?
* Dragon Army Retrospective
... |
5a19748a-5eed-4224-b5d3-69653762544a | StampyAI/alignment-research-dataset/blogs | Blogs | some simulation hypotheses
*(thanks to [Alexander](https://www.lesswrong.com/users/self-embedded-agent) for conversations that led to this post)*
some simulation hypotheses
--------------------------
what a strange time to live in, right on the verge of building an AI which will dictate the fate of the cosmos for ... |
6353a2ad-3a8d-40f4-a51f-a26f959e1dac | trentmkelly/LessWrong-43k | LessWrong | Meetup : Dallas Meetup
Discussion article for the meetup : Dallas Meetup
WHEN: 09 July 2016 01:30:00PM (-0500)
WHERE: 3725 Belt Line Rd, Addison, TX 75001
It's been too long since we last had a Dallas meetup. We will be meeting at Dunn Brothers Coffee. Look for the sign that says "LessWrong" on it.
Feel free to co... |
d09faa03-2e1c-410d-b16c-5d2fee71bfae | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Beliefs at different timescales
.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: none;... |
a81b64ad-6df7-4c27-959e-5266fcc2fd36 | trentmkelly/LessWrong-43k | LessWrong | An Introduction to Evidential Decision Theory
Introduction
This blog post serves as an introduction to Evidential Decision Theory (EDT), designed for readers with no prior background. While much has been written about EDT’s application to well-known problems like the Smoking Lesion problem, and Newcomb-like problems, ... |
cca8ce10-477e-4016-a282-5ef571fa99ce | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | How to hedge investment portfolio against AI risk?
Has anyone read anything on this? Would love an article recommendation.
By AI risk I mean the risk that AI greatly increases productivity, shifts the value of assets. It seems like the conventional strategy of investing in the S&P would be flawed if new AI firms sud... |
d9e15d61-b445-4d42-a119-441722780675 | trentmkelly/LessWrong-43k | LessWrong | Double Crux In A Box
Summary: Teaching a particular way of debating, where you work together to find out what about the world you actually disagree on. It involves practicing by working through existing disagreement.
Tags: Small, Repeatable, Investment
Purpose: Double Cruxing is more productive method of investigati... |
8e7d3960-2fb3-4c78-a805-270c3e5f1cb4 | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post3878
An initial draft looking at three ways of getting information out of Oracles, information that's useful and safe - in theory. One thing I may need to do, is find slightly better names for them ^_^ Good and safe uses of AI Oracles Abstract: An Oracle is a design for potentially high power artificial intell... |
0bf053bd-3a36-4fa5-b677-cad110a2b6a0 | trentmkelly/LessWrong-43k | LessWrong | $300 Fermi Model Competition
Summary
Task: Make an interesting and informative Fermi estimate
Prize: $300 for the top entry
Deadline: February 16th, 2025
Results Announcement: By March 1st, 2025
Judges: Claude 3.5 Sonnet, the QURI team
Motivation
LLMs have recently made it significantly easier to make Fermi estimate... |
91e421e0-266f-44bf-ad41-92f78a37fd02 | trentmkelly/LessWrong-43k | LessWrong | How can one be less wrong, if their conversation partner loses the interest on discussing the topic with them?
I must admit that I'm a fan of this video:
I understand that there are topics that you are interested, and there are not. I, for example, am currently interested with this topic and ignore the rest of other... |
35b6ceeb-f939-4197-be0c-f37e20438a69 | trentmkelly/LessWrong-43k | LessWrong | Individual Rationality Needn't Generalize to Rational Consensus
tl;dr
Organizations that enforce rationality at the collective level can get very different voting outcomes than organizations that enforce rationality at the individual level, per known results in social choice theory. This has implications for real-worl... |
dfe397c7-e99b-49ca-b831-e2f101762d7c | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post457
People talk about model weight security being really hard and crucial around the advent of AGI. (E.g. RAND report , Leopold ; see here for some distinctions in these threat models that I think are important.) But I think that the thinking on this has not been sufficiently attentive to the fact that during ... |
5f55fc09-40fd-4042-a8c8-1fde26757ade | trentmkelly/LessWrong-43k | LessWrong | Interpersonal Morality
Followup to: The Bedrock of Fairness
Every time I wonder if I really need to do so much prep work to explain an idea, I manage to forget some minor thing and a dozen people promptly post objections.
In this case, I seem to have forgotten to cover the topic of how morality applies to more than... |
33c596ae-297d-472e-a31f-57de3e5458af | trentmkelly/LessWrong-43k | LessWrong | Phil Trammell on Economic Growth Under Transformative AI
This is a transcript with slides for the latest episode (audio, youtube) of "The Inside View", a podcast I host about the future of AI progress. I interview Phil Trammell, an Oxford PhD student in economics and research associate at the Global Priorities Institu... |
56122950-0af8-4c8f-8547-acc8abb7330f | trentmkelly/LessWrong-43k | LessWrong | The proper use of regret?
Bryan Caplan in Parenthood as the Trump of All Past Regret explains: whatever tiny alterations he makes to his life before he had his children, would result in him not having the precise children he has today whom he so dearly loves, and therefore he does not regret a thing.
This is seems to... |
f96f3145-37de-4571-9ff2-915c9a34a419 | LDJnr/LessWrong-Amplify-Instruct | LessWrong | "GreyThumb.blog offers an interesting analogue between research on animal breeding and the fall of Enron. Before 1995, the way animal breeding worked was that you would take the top individual performers in each generation and breed from them, or their parents. A cockerel doesn't lay eggs, so you have to observe daught... |
8c3c731b-43b8-4176-a954-517fa99a2193 | trentmkelly/LessWrong-43k | LessWrong | Slowing AI: Crunch time
It is often asserted that time will be more valuable and higher leverage in "crunch time," shortly before critical models are deployed.
This brief post provides a starting point for consideration of three questions:
1. What is more time (before dangerous AI) good for?
2. What is different n... |
c5e80310-e4b6-409f-baea-dc269b8cde64 | trentmkelly/LessWrong-43k | LessWrong | Meetup In a Box: Year In Review
Summary: Three rounds of questions, where the meetup thinks about their past year in a few frames of reference, then discusses their insights.
Tags: Small, Medium, One-Off, Experiment
Purpose: Year in Review is something more useful than New Year's Resolutions - more like a team retro... |
3aef04bf-5347-4562-8426-0bff26203558 | trentmkelly/LessWrong-43k | LessWrong | User-inclination-guessing algorithms: registering a goal
Reading post about Community Notes made me recall a recent idea and post it here.
It's commonly known that ranking algorithms based on estimating user's views and inclinations can be incredibly useful in recommendation systems, advertising, and content curation... |
7c02a984-aa4c-4f80-8748-03cea113d98a | trentmkelly/LessWrong-43k | LessWrong | Forecast AI 2027
AI 2027 is a forecasting scenario by Daniel Kokotajlo, Scott Alexander, Thomas Larsen, Eli Lifland, and Romeo Dean. It predicts AI agents will begin automating AI research and development in 2027, triggering an intelligence explosion and arms race between the US and China. Most cognitive work becomes ... |
7c95f1d4-cbe7-4f30-9a0c-77a35c893619 | trentmkelly/LessWrong-43k | LessWrong | Gödel, Escher, Bach in the age of LLMs
Look, nothing I can say about GEB will be news to anyone reading LessWrong, but I wrote all this up to organize my thoughts, so I might as well post it here for anyone who hasn't read the book or wants a wave of nostalgia about that time they got obsessed with it as a teenager. I... |
eddcb0b7-0d02-4c0d-afdf-7b7a61671aef | trentmkelly/LessWrong-43k | LessWrong | Procedural Executive Function, Part 2
The Off Road project has since been folded into Rethink Wellbeing, but I’ve continued working to better understand and treat Executive Dysfunction. You can read more about the project’s origins here.
If you haven't, I suggest reading the start of my overview and exploration of Ex... |
71ab7e28-a2c0-4904-83b9-5bf89ae8b04b | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | How do new models from OpenAI, DeepMind and Anthropic perform on TruthfulQA?
**Authors:** [**Owain Evans**](https://owainevans.github.io/)**,**[**Stephanie Lin**](https://www.linkedin.com/in/sylin07/) **(Oxford),**[**Jacob Hilton**](https://www.jacobh.co.uk/) **(OpenAI)**
**NB: Updated (May 14) with additional re... |
b38e8e1b-4d76-4a68-9360-f82adaf6bc9a | trentmkelly/LessWrong-43k | LessWrong | Too much feedback can be a bad thing
Didn't have the time to read the article itself, but based on the abstract, this certainly sounds relevant for LW:
> Recent advances in information technology make it possible for decision makers to track information in real-time and obtain frequent feedback on their decisions. Fr... |
0d77afc2-179d-404d-b50a-00bcfd2ef0db | trentmkelly/LessWrong-43k | LessWrong | SHRDLU, understanding, anthropomorphisation and hindsight bias
EDIT: Since I didn't make it sufficiently clear, the point of this post was to illustrate how the GOFAI people could have got so much wrong and yet still be confident in their beliefs, by looking at what the results of one experiment - SHRDLU - must have f... |
f24172dd-f3d0-4650-a709-b91fdc91fd77 | trentmkelly/LessWrong-43k | LessWrong | To Be Decided #1
(Preface: This is the first edition of a quarterly email newsletter I started earlier this year called To Be Decided. I'm posting this as an experiment; if response here is positive, I'll post the two issues that have gone out since then as well as future issues as they come out. Feedback welcome!)
W... |
57f41c8a-9ae8-4c91-a1d9-f3010e3536a0 | StampyAI/alignment-research-dataset/arxiv | Arxiv | Certified Adversarial Defenses Meet Out-of-Distribution Corruptions: Benchmarking Robustness and Simple Baselines
1 Introduction
---------------
Developing machine learning (ML) systems that are robust to adversarial variations in the test data is critical for applied domains that require ML safety [[20](#bib.bib20... |
aa9707ad-d88d-45c1-9cd4-8335cb6a2c6f | trentmkelly/LessWrong-43k | LessWrong | POWERplay: An open-source toolchain to study AI power-seeking
We're open-sourcing POWERplay, a research toolchain you can use to study power-seeking behavior in reinforcement learning agents. POWERplay was developed by Gladstone AI for internal research.
POWERplay's main use is to estimate the instrumental value that... |
900043b4-3899-44bc-95b6-99d888248786 | StampyAI/alignment-research-dataset/arxiv | Arxiv | The AI Index 2022 Annual Report
Artificial Intelligence
Index Report 2022
2
Artificial Intelligence
Index Report 2022INTRODUCTION TO THE
AI INDEX REPORT 2022
Welcome to the fifth edition of the AI Index Report! The latest edition includes data from a broad set of academic,
private, and nonprofit organizations as we... |
92ceee7f-1a89-44f5-b401-932ffd4ef3c1 | trentmkelly/LessWrong-43k | LessWrong | Open Thread, March 2011
Has been a lot of open threads in the past, but not recently. It is for the discussion of Less Wrong topics that have not appeared in recent posts. If a discussion gets good enough, open a new top-level post. Besides, I have an issue to discuss it here. Enjoy!
|
319c840e-ffa9-462d-b155-d1c508ac2554 | trentmkelly/LessWrong-43k | LessWrong | My new AI control ideas are being posted
I'm posting my new AI control ideas at Less Wrong here.
I will crosspost some of the most interesting ones, but most of them will be only on Less Wrong.
Let me know what you think. |
29ab46ac-5624-499c-a151-bc7ba1ef1743 | trentmkelly/LessWrong-43k | LessWrong | Data on FED and macroeconomy
Turned out, it's devastating to search for time series of standard macro data as: money supply, savings, data on consumer basket across different classes is also missing.
I am surprised (this is really contradictory with common sense) of shortage of high-quality data in the US on the subj... |
0d83b415-bdb9-4f50-b962-a9fb84960f2f | trentmkelly/LessWrong-43k | LessWrong | The Twins
Someone is pregnant with twins. At birth, the twins are taken blindfolded to separate rooms. Both rooms are identical on the interior; outside, there is a ‘1’ printed on one of the doors and a ‘2’ printed on the other. Once inside, the blindfolds are removed, and the twins are left alone in their separate ro... |
60d97ffb-1d86-4c62-b90c-f92b51fc7c7d | trentmkelly/LessWrong-43k | LessWrong | Writing and enthusiasm
Steve Pavlina explains that the method he'd been taught in school-- a highly structured writing process of organizing what to say before it's written-- tends to produce dull writing, but starting from enthusiasm results in articles which are a pleasure to write and are apt to be more fun and mem... |
d7d6c9e3-4e51-4d59-af0d-afade0a0dd92 | trentmkelly/LessWrong-43k | LessWrong | Double Podcast Drop on AI Safety
TL;DR
Two days ago, I released two podcast episode:
* "One Big Bad Bill" on muckrAIkers ➡️ Igor Krawczuk and I discuss the substantial bill that Trump is attempting to push through the legislature, and I get a little animated... (IMO it's justified)
Spotify, Apple Podcasts, YouTu... |
ad78ce2f-563a-45ec-a86b-e7fbee096675 | trentmkelly/LessWrong-43k | LessWrong | Case Study II
Introduction and motivation
My last excruciatingly-thorough analysis was well-received. Since I posted it, I’ve thought of – and had pointed out to me, h/t johnswentworth – some more tricks I could have tried. And as luck would have it, the person who provided the last dataset has released a similar one... |
330bc218-2047-4fac-aede-19cdce0182a2 | trentmkelly/LessWrong-43k | LessWrong | Gauging interest for a Rio de Janeiro meetup group.
Who would be interested? Please reply to this thread or msg in private.
Thanks!
Edit: It's Rio de Janeiro, RJ, in Brazil.
It seems it is going to happen, discussion also on fb now:
http://www.facebook.com/Ierfh/posts/108940625936257?comment_id=28607&ref=notif¬if... |
289865b1-fbdf-4059-b2b4-625f4ad52f8e | StampyAI/alignment-research-dataset/blogs | Blogs | Cortés, Pizarro, and Afonso as precedents for takeover
*Daniel Kokotajlo, 29 February 2020*
*Epistemic status: I am not a historian, nor have I investigated these case studies in detail. I admit I am still uncertain about how the conquistadors were able to colonize so much of the world so quickly. I think my ignoran... |
22e4e3be-dee4-4f6e-81d1-81a9551f9ab9 | trentmkelly/LessWrong-43k | LessWrong | [Linkpost] Value extraction via language model abduction
This is my first post on lesswrong. I'll merely be linkposting content on epistemics and alignment here while getting more familiar with the culture.
tl;dr:
> We attempt to automatically infer one’s beliefs from their writing in three different ways. Initial r... |
7751e659-5763-47b1-9dc5-6941f7143dec | trentmkelly/LessWrong-43k | LessWrong | Meta Decision Theory and Newcomb's Problem
Hi all,
As part of my PhD I've written a paper developing a new approach to decision theory that I call Meta Decision Theory. The idea is that decision theory should take into account decision-theoretic uncertainty as well as empirical uncertainty, and that, once we acknowle... |
cd7424ce-6c25-4457-804b-efdbd810cd5e | StampyAI/alignment-research-dataset/arxiv | Arxiv | Designing a Safe Autonomous Artificial Intelligence Agent based on Human Self-Regulation
Designing a Safe Autonomous Artificial Intelligence Agent based on Human Self -Regulation
Mark Muraven
University at Albany
January 5, 2017
Author contact: muraven@albany.edu
Human Self -Regulation for AI 2
Abs... |
db4443b0-1657-4f55-bb06-30c41a5dc068 | trentmkelly/LessWrong-43k | LessWrong | Success Buys Freedom
This is Part 2 of my series shamelessly ripping off the excellent writings of alkjash. Part 1 is here
----------------------------------------
I. Parables
A. Albert Einstein
Albert Einstein worked in a patent office. In 1905 he published 4 groundbreaking papers on the photoelectric effect, Br... |
981c70d4-e4bf-471d-bcc4-2eea2e27493c | trentmkelly/LessWrong-43k | LessWrong | Omicron Post #10
Reporting in the wake of Christmas remains spotty, but this fog of war was anticipated, and we must do what we can with what we do know. Mostly, what we’ve learned reinforces the picture outlined in Post #9, with reduced severity. Also with reduced generation time which we confirmed a bit via a study ... |
e18e7eef-a6e0-4994-8ae3-3c41f0efe9da | trentmkelly/LessWrong-43k | LessWrong | What are the limits of superintelligence?
This post is mostly concerned with a superintelligent AI performing recursive self-improvement, this analysis is done to help make sense of the take off speed of such an operation.
Plausibility and Limits
Before considering upper limits, it may be worth considering whether g... |
391d7335-f708-4fcb-953a-4bcca5e369f2 | trentmkelly/LessWrong-43k | LessWrong | Choosing to Choose?
,,,,,,,,,,,,,,,,,,,,,,,,,,,,,,
This is related to wireheading, utility functions, taste, and rationality. It is a series of puzzles meant to draw attention to certain tensions in notions of rationality and utility functions for embedded agents. I often skip the particulars of embedded situations. I... |
bbcc4d12-5bb6-4d39-9dbc-6d709205bff6 | trentmkelly/LessWrong-43k | LessWrong | Non-directed conceptual founding
[Metadata: crossposted from https://tsvibt.blogspot.com/2022/08/non-directed-conceptual-founding.html. Written 13 June 2022. I'm fairly likely to not respond to comments promptly. If you're especially interested in chatting, my gmail address is: tsvibtcontact ]
In trying to understan... |
08e668f5-ad33-461d-9198-f50cf49a6a94 | trentmkelly/LessWrong-43k | LessWrong | Making a Secular Solstice Songbook
After this year's secular solstice several people were saying they'd be interested in getting together to sing some of these songs casually. This is a big part of what we sang at the post-EAG music party, but one issue was logistical: how do you get everyone on the same words and cho... |
c4d8fd0d-d501-4ab4-95a2-ff095576cd80 | trentmkelly/LessWrong-43k | LessWrong | AI Forecasting Resolution Council (Forecasting infrastructure, part 2)
UPDATE October 18, 2020: The AI Resolution Forecasting Council has been retired due to lack of demand for its services. I'm keeping this post up for historical value.
---------------
This post introduces the AI Forecasting Resolution Council, a g... |
af4bf10e-30b0-4341-a2eb-21541e644dbf | trentmkelly/LessWrong-43k | LessWrong | On working 80%
A year ago, I decided to reduce my employment level from 100% to 80% and to take Fridays off.
My main motivation was to have some time for myself: Relax, reduce my stress level from work, have more time for side projects, do more sports, and maybe spend more time with my wife without the kids. This was... |
7d36385b-a63a-40d7-97c8-1a6de84f0102 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | AI governance student hackathon on Saturday, April 23: register now!
[*Cross-posted from the EA Forum*](https://forum.effectivealtruism.org/posts/2s3gEnHiduSmYZX9R/ai-governance-student-hackathon-on-saturday-april-23)
[Columbia Effective Altruism](https://www.columbia-ea.org/) and [Roosevelt at Columbia](http://roose... |
72bff036-a727-4439-b20e-7557ebfed89e | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Towards Better Milestones for Monitoring AI Capabilities
.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-sp... |
6fa40e5e-7d02-4a27-879d-2b298e37f072 | trentmkelly/LessWrong-43k | LessWrong | Intuitions on Universal Behavior of Information at a Distance
What’s special about information at a distance? Some examples of what I mean:
* What information about a little patch of pixels in a photo is relevant to another little patch of pixels far away in the same photo?
* Relationship between exact voltages in ... |
09add7ba-7f35-45f7-abaf-7eff3842ed54 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Fear mitigated the nuclear threat, can it do the same to AGI risks?
It seems that only a small minority of the general public realizes that misaligned and powerful AIs might be a serious threat to the world as we know it, and the problems might happen not that far into the future.
Nobody wants humanity to go extinct... |
887e8f69-249f-4d82-b8fd-d2507328356a | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Specification gaming: the flip side of AI ingenuity
*(Originally posted to the* [*Deepmind Blog*](https://deepmind.com/blog/article/Specification-gaming-the-flip-side-of-AI-ingenuity)*)*
**Specification gaming** is a behaviour that satisfies the literal specification of an objective without achieving the intended out... |
17894598-b62f-4a42-adab-63d83959de46 | trentmkelly/LessWrong-43k | LessWrong | Competence/Confidence
This "essay" is a bit of an experiment, format-wise. I'd like to draw the reader's attention to a certain way of thinking about and evaluating progress that's been useful to me in a wide variety of domains. It's an exercise I run through when I'm feeling stuck, which almost always provides some... |
5ae4b30b-496a-45bf-8a35-d007dc879ef6 | trentmkelly/LessWrong-43k | LessWrong | What Do AI Safety Pitches Not Get About Your Field?
When I was first introduced to AI Safety, coming from a background studying psychology, I kept getting frustrated about the way people defined the and used the word "intelligence". They weren't able to address my questions about cultural intelligence, social evolutio... |
c6db4c3e-7f00-41ee-a941-102a79063851 | trentmkelly/LessWrong-43k | LessWrong | Ediburgh LW Meetup Sunday 15th May, 2pm
Location: Delhi Cafe, 67 Nicolson Street
Map: http://maps.google.co.uk/maps?ie=UTF8&t=h&cid=1874860554950886070
I will be there with the Oxford Book of Aphorisms (for no particular reason)
Also check out: http://www.facebook.com/event.php?eid=161542077233135. Since we touch... |
ca756199-7530-4ef7-8abe-081316745bb3 | trentmkelly/LessWrong-43k | LessWrong | Is ambition rational?
I don't understand the people around me who are working so very hard to succeed. It strikes me as irrational. Why do you do it?
A long time ago I reasoned that it is more efficient to strive for a 90% rather than 100% on a test because both yield the same "A". This morphed into a way of life. I ... |
552c5c81-d439-44ad-9f54-11646e065461 | trentmkelly/LessWrong-43k | LessWrong | Conversation on technology forecasting and gradualism
This post is a transcript of a multi-day discussion between Paul Christiano, Richard Ngo, Eliezer Yudkowsky, Rob Bensinger, Holden Karnofsky, Rohin Shah, Carl Shulman, Nate Soares, and Jaan Tallinn, following up on the Yudkowsky/Christiano debate in 1, 2, 3, and 4.... |
e9c69606-9901-4266-ad61-522f060a2a89 | trentmkelly/LessWrong-43k | LessWrong | Saving Lives Reduces Over-Population—A Counter-Intuitive Non-Zero-Sum Game
AN ARGUMENT AGAINST SAVING LIVES
When making the argument for charitable giving, a response I have often received is that saving lives in areas that are already over-populated will perpetuate over-population-solving nothing. It’s a concern I sh... |
c7d3134d-8896-4cc2-a8b4-b04f8ca476dc | trentmkelly/LessWrong-43k | LessWrong | What are the most exciting developments from non-Europe and/or non-Northern-Hemisphere?
My vague sense of "macrohistory of the world" is fairly Euro-centric, and Northern Hemisphere Centric. And I'm not sure how much of that is to with "well, for good or for ill Europe kinda took over the world and everything is entan... |
c0edbcd4-28bc-4e92-b24d-24c354d5cc82 | trentmkelly/LessWrong-43k | LessWrong | Counterfactual Planning in AGI Systems
Counterfactual planning is a design approach for creating a range of safety mechanisms that can be applied in hypothetical future AI systems which have Artificial General Intelligence.
My new paper Counterfactual Planning in AGI Systems introduces this design approach in full. I... |
599eb6db-b42d-44e3-a3d9-da80a6da2f40 | trentmkelly/LessWrong-43k | LessWrong | A couple productivity tips for overthinkers
1. If you find that you’re reluctant to permanently give up on to-do list items, “deprioritize” them instead
I hate the idea of deciding that something on my to-do list isn’t that important, and then deleting it off my to-do list without actually doing it. Because once it’s ... |
1c8199ce-d919-4afa-b815-5dad5b3a028d | trentmkelly/LessWrong-43k | LessWrong | Reward Hacking from a Causal Perspective
Post 4 of Towards Causal Foundations of Safe AGI, preceded by Post 1: Introduction, Post 2: Causality, Post 3: Agency, and Post 4: Incentives.
By Francis Rhys Ward, Tom Everitt, Sebastian Benthall, James Fox, Matt MacDermott, Milad Kazemi, Ryan Carey representing the Causal In... |
57af5208-91f8-4d7b-9b87-5b57c9e7fd0b | trentmkelly/LessWrong-43k | LessWrong | Dealing with Awkwardness
This post introduces my model of awkwardness and the methods I’ve used to overcome it.[1]
Awkwardness used to be a big struggle in my life - limiting my agency through excessive risk aversion. I can vividly recall a conversation where I was visibly shaking and stammering throughout - I had le... |
d23c3128-63e6-4862-b00c-b8ae4c96b583 | StampyAI/alignment-research-dataset/blogs | Blogs | Costs of extinction risk mitigation
We very roughly estimate that the annual cost of reducing the probability of human extinction by 0.01% is within the range of $1.1 billion to $3.5 trillion.
Introduction
------------
This article is intended to be usable in a Cost-Benefit Analysis (CBA) analysis of extinction r... |
c59e88e0-8adb-482d-8f83-2054469bdda7 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | [Appendix] Natural Abstractions: Key Claims, Theorems, and Critiques
This is the appendix to [Natural Abstractions: Key Claims, Theorems, and Critiques](https://www.lesswrong.com/posts/gvzW46Z3BsaZsLc25/natural-abstractions-key-claims-theorems-and-critiques-1). It contains additional details that we expect are only re... |
e60f9fd4-256c-44d4-9793-fc858e8ba9c8 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Edinburgh Weekday LW Meetup
Discussion article for the meetup : Edinburgh Weekday LW Meetup
WHEN: 21 September 2011 01:00:00PM (+0100)
WHERE: Delhi Cafe, 67 Nicolson Street, Edinburgh, Scotland, United Kingdom.
The Edinburgh LW meetups resume. The topic shall be: Value Complexity & Instrumental Rationality... |
cc8f4fbf-9b32-4896-9f74-4bf80439c099 | trentmkelly/LessWrong-43k | LessWrong | Kids NCurses Messenger
Yesterday I wrote a program that lets me and Lily (7y) send messages back and forth:
The laptop is an eeepc 700. It was the first mass produced netbook and I liked it a lot. By 2010 I wasn't using it anymore, and it ended up in the basement. Recently I found it, I noticed that while it is s... |
c944bf59-7241-4663-92b2-a80b3b9ee6d7 | trentmkelly/LessWrong-43k | LessWrong | Misery Pits
This is an attempt at a description, more intensional than extensional, of a phenomenon that can adversely affect communities. I'm trying to avoid having anyone specific (especially specific-and-local) in mind while I write it; I just mean to define the phenomenon as a reference point and wave tentatively ... |
cd35ceed-03e6-4861-b64e-9d136afb58a7 | StampyAI/alignment-research-dataset/blogs | Blogs | KOLSITAN, a tiny video game
KOLSITAN, a tiny video game
---------------------------
[**KOLSITAN**](kolsitan) is a tiny video game I made in about ten days, with music by [Chiaro](https://chiaro.bandcamp.com/).
You can access the source code and the various resources used [here](kolsitan/kolsitan.zip); there's also... |
ef76868c-201e-43a5-b1f7-f87238285139 | trentmkelly/LessWrong-43k | LessWrong | Updates from Comments on "AI 2027 is a Bet Against Amdahl's Law"
AI 2027 is a Bet Against Amdahl's Law was my attempt to summarize and analyze "the key load-bearing arguments AI 2027 presents for short timelines". There were a lot of great comments – every time I post on LW is a learning experience. In this post, I'm ... |
d59d63ee-ce86-4dc2-928b-643a9480c499 | trentmkelly/LessWrong-43k | LessWrong | Thoughts on civilization collapse
Epistemic status: an idea I believe moderately strongly, based on extensive reading but not rigorous analysis.
We may have a dramatically wrong idea of civilization collapse, mainly inspired by movies that obsess over dramatic tales of individual heroism.
Traditional view:
In a ... |
cd9db891-6859-4962-b88d-c8dc1b3731ef | trentmkelly/LessWrong-43k | LessWrong | Multivariate estimation & the Squiggly language
(Talk given at an event on Sunday 16th of August. Ozzie Gooen is responsible for the talk, Justis Mills edited the transcript.
If you're a curated author and interested in giving a 5-min talk, which will then be transcribed and edited, sign up here.)
Ozzie: This imag... |
fcf56d0b-8b54-4a8a-9cb7-2df3625309b4 | trentmkelly/LessWrong-43k | LessWrong | The VNM independence axiom ignores the value of information
Followup to : Is risk aversion really irrational?
After reading the decision theory FAQ and re-reading The Allais Paradox I realized I still don't accept the VNM axioms, especially the independence one, and I started thinking about what my true rejection cou... |
cc63b027-f5ce-4ce8-8d49-7c333c6d91ab | trentmkelly/LessWrong-43k | LessWrong | LW UI features you might not have tried
Every once in a while I see a comment describing a problem with lesswrong that is fixable with existing features, if only the user knew about them. This post is an attempt to make these highly useful features more widely known.
I am technically a mod on lesswrong, which is why... |
790071a0-6b2c-4419-b80f-4b49f95735a7 | trentmkelly/LessWrong-43k | LessWrong | Practical Limits to Giving Now and The Haste Consideration
Quite a few people have talked about whether or not it's better to donate money or donate time now rather than save and donate more money later. While there's lots of nuance on these positions, there does seem to be roughly these two camps -- weighing in (mos... |
621ace52-5c77-402f-9846-115e8c66fba0 | trentmkelly/LessWrong-43k | LessWrong | AI companies should be safety-testing the most capable versions of their models
AI companies should be safety-testing the most capable versions of their models
Only OpenAI has committed to the strongest approach: testing task-specific versions of its models. But evidence of their follow-through is limited.
You might... |
8ae5b6a9-997d-4a0c-96b8-d8e2c992fb15 | trentmkelly/LessWrong-43k | LessWrong | Strawman Yourself
One good way to ensure that your plans are robust is to strawman yourself. Look at your plan in the most critical, contemptuous light possible and come up with the obvious uncharitable insulting argument for why you will fail.
In many cases, the obvious uncharitable insulting argument will still be ... |
46011129-53b7-40dc-91e0-0c8b6e48cb15 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | [linkpost] Sharing powerful AI models: the emerging paradigm of structured access
I have written a new GovAI blog post - [link here](https://www.governance.ai/post/sharing-powerful-ai-models).
How should labs should share large AI models? I argue for a "structured access" approach, where outsiders interact with the m... |
87a28314-2628-4207-91a6-e66887e51f56 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Question 2: Predicted bad outcomes of AGI learning architecture
**Architecture-independent risks: human and AGI misalignment**
--------------------------------------------------------------
Let’s say as AGI safety researchers, we’ve done our research and soul-searching, and we’ve placed our personal bets about the mo... |
5bc538a2-3290-44b4-88ac-d167b6a4b8d5 | trentmkelly/LessWrong-43k | LessWrong | [SEQ RERUN] Eliezer's Meta-Level Determinism
Today's post, Eliezer’s Meta-Level Determinism was originally published on June 23, 2008. A summary:
> Eliezer has suggested that the most dramatic changes in the history of life are the result of events that changed the optimization process at work. This is experimenta... |
458fb0e3-619f-4e7c-83bc-59a2c6cf6374 | trentmkelly/LessWrong-43k | LessWrong | Humans are Insecure Password Generators
And They're Currently Being Cracked
(A crosspost of Humans are Insecure Password Generators)
Any password longer than 12 characters or so is invulnerable to brute force attacks. So why do so many longer ones keep getting breached?
Because humans don't pick passwords random... |
381cec37-3e84-4d77-86d2-c5e8807a50fd | StampyAI/alignment-research-dataset/lesswrong | LessWrong | The Evil Genie Puzzle
Foolish you. You decided to rub a lamp in the hope that a genie would appear and you found one, only it's an Evil Genie. He tells you only have two options: either wishing for your perfect life or being pelted with rotten eggs. You start thinking about how wonderful your new life will be, but of ... |
62ffa3b4-e9c3-4ae0-9f0d-44d2878baca5 | trentmkelly/LessWrong-43k | LessWrong | Updating Utility Functions
This post will be about AIs that “refine” their utility function over time, and how it might be possible to construct such systems without giving them undesirable properties. The discussion relates to corrigibility, value learning, and (to a lesser extent) wireheading.
We (Joar Skalse and J... |
6ddece16-672a-4bee-b20b-a021c7e24f89 | trentmkelly/LessWrong-43k | LessWrong | Proof Length and Logical Counterfactuals Revisited
Update: This version of the Trolljecture fails too; see the counterexample due to Sam.
In An Informal Conjecture on Proof Length and Logical Counterfactuals, Scott discussed a "trolljecture" from a MIRI workshop, which attempted to justify (some) logical counterfactu... |
05abf766-647c-4f4a-ad34-272e2c1e5a54 | StampyAI/alignment-research-dataset/aisafety.info | AI Safety Info | What is Ought's research strategy?
[Ought](https://ought.org/) is a research lab that "[develops mechanisms for delegating high-quality reasoning to advanced machine learning systems](https://ought.org/)." Its research focuses on methods for breaking up complex, hard-to-verify tasks into simpler, easier-to-verify task... |
adf7b038-8b59-4b85-9eb8-6970d2823585 | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Alignment is hard. Communicating that, might be harder
*Crossposted from the EA Forum:* [*https://forum.effectivealtruism.org/posts/PWKWEFJMpHzFC6Qvu/alignment-is-hard-communicating-that-might-be-harder*](https://forum.effectivealtruism.org/posts/PWKWEFJMpHzFC6Qvu/alignment-is-hard-communicating-that-might-be-harder)... |
87476687-8e6a-431a-9e99-e5193c5d9af4 | trentmkelly/LessWrong-43k | LessWrong | Simple Improvement to College Football Overtime Rules
For a long time I have thought that a simple tweak to the NCAAF overtime rules would fix several problems with no downsides. The Alabama-LSU game reinforced my impression of this by showing what we could have every time.
PROPOSED: In overtime, teams cannot kick an... |
b8a275ec-2c12-4d51-a898-8b0b044e6413 | trentmkelly/LessWrong-43k | LessWrong | A Self-Respect Feedback Loop
This is a followup to Affordance Widths.
Epistemic Status: It’s only a model
Okay! This is something I’ve been trying to explain for awhile, but I think I have a handy chart for it now.
Here’s how it works:
A person can actually regulate how much self-respect they feel, and show. O... |
d31a5139-9591-4789-a8e6-e9d62cf3fd54 | trentmkelly/LessWrong-43k | LessWrong | Power dynamics as a blind spot or blurry spot in our collective world-modeling, especially around AI
Where I’m coming from
***Epistemic status: personal experience***
In a number of prior posts, and in ARCHES, I’ve argued that more existential safety consideration is needed on the topic of multi-principal/multi-agen... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.