id stringlengths 36 36 | source stringclasses 15
values | formatted_source stringclasses 13
values | text stringlengths 2 7.55M |
|---|---|---|---|
226a4f7f-6675-4784-ae49-206c39071dca | trentmkelly/LessWrong-43k | LessWrong | Share AI Safety Ideas: Both Crazy and Not. №2
AI safety is one of the most critical issues of our time, and sometimes the most innovative ideas come from unorthodox or even "crazy" thinking. I’d love to hear bold, unconventional, half-baked or well-developed ideas for improving AI safety. You can also share ideas you ... |
293fe510-d4ca-4b17-b5a9-9eaa77389e02 | trentmkelly/LessWrong-43k | LessWrong | The Bayesian Tyrant
Long ago and far away, there was a kingdom called Estimor in a broad green valley surrounded by tall grey mountains. It was an average kingdom in most respects, until the King read the works of Robin Hanson and Eliezer Yudkowsy, and decided to institute a Royalist Futarchy.
(This is a parable abou... |
b5f34d82-007c-40c7-bede-777318d51128 | StampyAI/alignment-research-dataset/blogs | Blogs | the Insulated Goal-Program idea
*(this post has been written for the first [Refine](https://www.lesswrong.com/posts/D7epkkJb3CqDTYgX9/refine-an-incubator-for-conceptual-alignment-research-bets) blog post day, at the end of the week of readings, discussions, and exercises about epistemology for doing good conceptual re... |
5026ec49-80a8-4e64-8f07-5c9e3f706e97 | trentmkelly/LessWrong-43k | LessWrong | LW 2.0 Site Update: 09/22/17
Another set of quick updates done today:
* We had a bad bug where in about 50% of cases when a user created a post on the frontpage, it would neither show up on their userpage, nor on the frontpage, but only on the all-posts view. This is now fixed. Sorry for anyone who was confused by t... |
9140e6f6-bf9f-4c7c-afff-fa91da28e532 | trentmkelly/LessWrong-43k | LessWrong | Activated Charcoal for Hangover Prevention: Way more than you wanted to know
Activated charcoal is charcoal that has been heated up a whole lot and then washed really hard, by water or acid. This process is what "activates" it, and activating it increases the surface area by a lot. I think it prevents hangovers.
I wo... |
4895f93c-4e4e-47be-945e-8380432acdba | trentmkelly/LessWrong-43k | LessWrong | DIY Beaterless Drum Pedals
When I first made a board for electronic foot percussion, I played around until I had something that worked. My initial design ( 2013) and my two later iterations ( 2018, 2019) were all built around the idea of attaching a piezoelectric contact microphone to the underside of a piece of wood,... |
5d8c5c9d-4274-44b5-b385-936c2fc9d245 | trentmkelly/LessWrong-43k | LessWrong | Useful Does Not Mean Secure
Brief summary of what I'm trying to do with this post:
1. Contrast a “Usefulness” focused approach to building AI with a “Security” focused approach, and try to give an account of where security problems come from in AI.
2. Show how marginal transparency improvements don’t necessarily im... |
130a5823-325d-4376-be59-f6e7eb277f83 | trentmkelly/LessWrong-43k | LessWrong | Supervised learning and self-modeling: What's "superhuman?"
I
We don't just want aligned AI to do what a human would do. We've got plenty of humans lying around all over the place, if that's all we need. The fancy promise of future AI is that it will deliver superhuman performance.
Sometimes the notion of "superhum... |
3fc6e74c-c6f2-4286-aeb2-6a8951db36ef | trentmkelly/LessWrong-43k | LessWrong | Writing guide?
I remember seeing a short writing guide written by Eliezer for use by the SIAI, but now I can't find it. Anyone have a link for it? |
fea73872-eec7-4161-a95f-3683709a4a5d | StampyAI/alignment-research-dataset/blogs | Blogs | Grants and fundraisers
Two big announcements today:
**1. MIRI has won $299,310 from the Future of Life Institute’s grant program to jumpstart the field of long-term AI safety research.**
* $250,000 will go to our research program over the course of three years. This will go towards running workshops and funding a ... |
f6821f0f-2ec8-45d3-9ffb-fb1152288f08 | trentmkelly/LessWrong-43k | LessWrong | The Way of a Skeptic
The essay “The Sorcerer and His Magic” by Claude Lévi-Strauss, specifically the part about the shaman Quesalid, is of interest to anyone who considers themselves a skeptic or a rationalist.
It has it all! Proto-scientific method and experimentation, as seen in the episode with the Koskimo shamans... |
d9d032cb-f1dd-4d08-9bfa-8818b8dd93b3 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Alignment Newsletter #13: 07/02/18
**Highlights**
--------------
**[OpenAI Five](https://blog.openai.com/openai-five/)** *(Many people at OpenAI)*: OpenAI has trained a team of five neural networks to play a particular set of Dota heroes in a mirror match (playing against the same set of heroes) with a few restrictio... |
399bc4f6-f534-4985-b8bc-056d5308f80c | trentmkelly/LessWrong-43k | LessWrong | BERI is seeking new collaborators (2022)
The Berkeley Existential Risk Initiative (BERI) is once again accepting applications from university-affiliated groups and individuals interested in receiving our support. Applications are due June 26th.
Winning applicants would be eligible for free services from BERI, like pu... |
8ef0024c-4820-4a51-a4c4-10659388d70f | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Autonomous replication and adaptation: an attempt at a concrete danger threshold
*Note: This is a rough attempt to write down a more concrete threshold at which models might pose significant risks from autonomous replication and adaptation (ARA). It is fairly in the weeds and does not attempt to motivate or contextual... |
4ab9fb93-e9a1-4376-bcc4-007f00455839 | trentmkelly/LessWrong-43k | LessWrong | The Stamp Collector
I'm writing a series of posts about replacing guilt motivation over on MindingOurWay, and I plan to post the meatier / more substantive posts in that series to LessWrong. This one is an allegory designed to remind people that they are allowed to care about the outer world, that they are not cursed ... |
b871e8ab-c951-4975-b168-a37c71d407e2 | trentmkelly/LessWrong-43k | LessWrong | [Linkpost] The importance of stupidity in scientific research
|
8b167a0b-91e9-40ac-95da-8676456ff550 | trentmkelly/LessWrong-43k | LessWrong | Abstracting The Hardness of Alignment: Unbounded Atomic Optimization
This post is part of the work done at Conjecture.
Disagree to Agree
(Practically-A-Book Review: Yudkowsky Contra Ngo On Agents, Scott Alexander, 2022)
> This is a weird dialogue to start with. It grants so many assumptions about the risk of future... |
8dde5e48-f661-4794-a89e-b95c9750f411 | trentmkelly/LessWrong-43k | LessWrong | New LW Meetup: Baltimore
This summary was posted to LW Main on December 11th. The following week's summary is here.
New meetups (or meetups with a hiatus of more than a year) are happening in:
* Baltimore Area Meetup: 03 January 2016 02:00PM
Irregularly scheduled Less Wrong meetups are taking place in:
* San Ant... |
7ae26c0c-841e-4311-a3dc-6d2e042a0637 | trentmkelly/LessWrong-43k | LessWrong | ChatGPT for translation
This post is picking up from some of the points mentioned in our Q2’23 work post (and continuing our experiments with ChatGPT). One big hurdle we are currently facing is translations not bing contextual / not so vernacular.
(If you just want to jump on to the results: here is the ChatGPT tran... |
52b5bc4b-a7b3-4763-8d34-9a82c97384fe | trentmkelly/LessWrong-43k | LessWrong | Quantum particles and classical filaments
The first post in the "From Branes to Brains" series. For the introduction to the series see this post.
In this post, I will mostly follow the ideas from the classical book by Polyakov, as well as from this paper by Kleinert.
First of all, consider a classical particle with... |
bda86be0-fcf8-47a2-a7dc-5f7f62f1ab81 | trentmkelly/LessWrong-43k | LessWrong | Is intelligence explosion necessary for doomsday?
I searched for articles on the topic and couldn't find any.
It seems to me that intelligence explosion makes human annihilation much more likely, since superintelligences will certainly be able to outwit humans, but that a human-level intelligence that could process i... |
51b77dcb-2c56-463b-8ae7-c41065b38312 | StampyAI/alignment-research-dataset/arxiv | Arxiv | On the Effectiveness of Interval Bound Propagation for Training Verifiably Robust Models
1 Introduction
---------------
Despite the successes of deep learning, it is well-known that neural networks are not robust.
In particular, it has been shown that the addition of small but carefully chosen deviations to the inp... |
81eac907-1da5-4a05-b951-4cfac05f0200 | trentmkelly/LessWrong-43k | LessWrong | Alignment researchers, how useful is extra compute for you?
TLDR
If you work in AI alignment / safety research, please fill out this form on how useful access to extra compute would be for your research. This should take under 10 minutes, and you don't need to read the rest of this post beforehand—in fact it would be ... |
021db51e-f39a-4f6f-802f-53e2da4d0aa4 | trentmkelly/LessWrong-43k | LessWrong | A Key Power of the President is to Coordinate the Execution of Existing Concrete Plans
I listened to the 80,000 Hours podcast with Tom Kalil, who spent 16 years as Deputy Director of the Office of Science and Technology Policy at the White House. Kalil seems skilled at evaluating concrete scientific plans to offer the... |
c279881a-ee92-43f5-a25f-b92977e1c4f6 | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Skilling-up in ML Engineering for Alignment: request for comments
TLDR: Please fill out [this survey](https://docs.google.com/forms/d/e/1FAIpQLSc1rNIblbIVtxzla_JMND8vPKgsHubTKks_VmSY3wYdN5h-eQ/viewform) if you are a potential or established ML Engineer, to let us know your thoughts and preferences on the best way to g... |
a53b6e04-9702-4b71-8b3e-e86bc636f16a | trentmkelly/LessWrong-43k | LessWrong | Meetup : Ugh Fields
Discussion article for the meetup : Ugh Fields
WHEN: 17 April 2014 07:00:00PM (-0400)
WHERE: 2411 N Roxboro St 27704
We'll be discussing Ugh Fields: what they are, how they keep you from accomplishing stuff, and how to recognize and reduce them. As always, RSVPs are appreciated but not required.... |
bdf225f3-ad3e-47af-9bd3-2f3856d0ddc0 | trentmkelly/LessWrong-43k | LessWrong | Wolfram Research v Cook
Matthew Cook probably violated an NDA he signed with Wolfram Research
This is a follow-up to this comment on a previous post of mine about (Stephen) Wolfram.
The comment, by gwern:
> > Charitably, I think that makes his apparent 'arrogance' better understood as something like a 'literary conv... |
660c762e-e1de-4999-887a-7793ac00ceb5 | trentmkelly/LessWrong-43k | LessWrong | No Constant Distribution Can be a Logical Inductor
The title says it all.
Consider some valuation over sentences, P:S→[0,1].
Now consider some set of unproved sentences that make a binary tree, where it is propositionally provable that exactly one of the sentences at the m'th level in the tree are true. The most con... |
80583f0a-ef76-4a5b-837f-55735f23f68d | trentmkelly/LessWrong-43k | LessWrong | [Preprint] The Computational Limits of Deep Learning
"The Computational Limits of Deep Learning" by Neil C. Thompson, Kristjan Greenewald, Keeheon Lee, and Gabriel F. Manso
Links:
* PDF version
* General Audience Summary on VentureBeat
NB: This is a preprint and not peer-reviewed or accepted for publication as be... |
1784d246-e22a-4231-b344-73909c77464a | trentmkelly/LessWrong-43k | LessWrong | Doing your good deed for the day
Interesting new study out on moral behavior. The one sentence summary of the most interesting part is that people who did one good deed were less likely to do another good deed in the near future. They had, quite literally, done their good deed for the day.
In the first part of the st... |
5dc3d508-646f-4da3-ad52-17af81045fc9 | trentmkelly/LessWrong-43k | LessWrong | Decision Theory subreddit
|
408e0216-b7c4-4d49-9573-190fec7c3d0b | StampyAI/alignment-research-dataset/lesswrong | LessWrong | The Limit of Language Models
Epistemic Status
----------------
Highlighting a thesis in Janus' "[Simulators](https://www.lesswrong.com/posts/vJFdjigzmcXMhNTsx/simulators)" that I think is insufficiently appreciated.
Thesis
------
In the limit, models optimised for minimising predictive loss on humanity's text co... |
fc64da3f-684b-40a2-856f-8b7f40418cee | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Take 6: CAIS is actually Orwellian.
*As a writing exercise, I'm writing an AI Alignment Hot Take Advent Calendar - one new hot take, written every day for 25 days. Or until I run out of hot takes.*
CAIS, or [Comprehensive AI Services](https://www.fhi.ox.ac.uk/wp-content/uploads/Reframing_Superintelligence_FHI-TR-2019... |
1291db91-70fb-4201-95c3-b644db298588 | trentmkelly/LessWrong-43k | LessWrong | Prescientific Organizational Theory (Ribbonfarm)
|
e4bfd1c3-3319-4038-8166-ccdddc2a6ca1 | trentmkelly/LessWrong-43k | LessWrong | What are some Meta-Umeshsims?
Scott Aaronson wrote a blog post in which he defined Umeshisms.
Umesh was his advisor at Berkley who used to frequently repeat: "If you’ve never missed a flight, you’re spending too much time in airports."
Scott then asked his readers to provide more examples, ie.:
> If you never cut y... |
550d46e4-9f33-4392-85df-4bca6b401d1a | trentmkelly/LessWrong-43k | LessWrong | The Greedy Doctor Problem... turns out to be relevant to the ELK problem?
The following post was published on my Substack and discussed on HackerNews about 2 months ago. I originally planned it as an accessible introduction to Vinge's principle and the Principal-Agent-Problem. Despite having equations and simulations ... |
60c6dba6-2039-4cd3-b667-4e48868febda | StampyAI/alignment-research-dataset/lesswrong | LessWrong | Disagreements over the prioritization of existential risk from AI
Earlier this year, the [Future of Life Institute](https://futureoflife.org/open-letter/pause-giant-ai-experiments/) and the [Center for AI Safety](https://www.safe.ai/statement-on-ai-risk) published open letters that promoted existential risks (x-risks)... |
49e645f9-f52d-4b7c-b8c9-f8c830e8f5e9 | trentmkelly/LessWrong-43k | LessWrong | We desperately need to talk more about human challenge trials.
Note: Thanks to Peter McCluskey, whose comment below helped me develop my thoughts further and which is worth reading in addition to this post.
On March 11, 4,611 people had died of COVID-19, and the WHO declared the disease a pandemic. To date, COVID-19 ... |
47ccf89a-a7dc-438b-aa1b-143db35b3d41 | trentmkelly/LessWrong-43k | LessWrong | "New EA cause area: voting"; or, "what's wrong with this calculation?"
Consider my home county:
* Number of voters ≈700k
* (Corollary: probability of swinging an election: 1/√nVoters≈1/800)
* GDP: $400B
* Amount by which GDP might go up or down depending on a single election: 0.1%
* Probability I support the bet... |
2c8be2d4-4ec2-4a54-a63b-341fb6ff1f58 | trentmkelly/LessWrong-43k | LessWrong | Update on Harvard AI Safety Team and MIT AI Alignment
We help organize the Harvard AI Safety Team (HAIST) and MIT AI Alignment (MAIA), and are excited about our groups and the progress we’ve made over the last semester.
In this post, we’ve attempted to think through what worked (and didn’t work!) for HAIST and MAIA,... |
526cd124-9844-44e3-93c9-a44c681dc235 | trentmkelly/LessWrong-43k | LessWrong | Slightly advanced decision theory 102: Four reasons not to be a (naive) utility maximizer
TL;DR: Introduction to decision theory with examples from effective altruism. Resulting optimal strategies differ from "naive" utility maximisation, but match community norms.
Meta: I've also posted this on the EA Forum, but I t... |
2ba2a6d1-e471-4375-a2fb-2cae6a1afe27 | trentmkelly/LessWrong-43k | LessWrong | New HPMOR Article(s)
I was setting up the article for chapter 91 with the intent of doing one article for each new chapter because I don't like like dealing with long threads, but it occurred to me to ask whether people have a preference for one, two, or three threads.
Someone else would have to consolidate multiple... |
11bd03b4-a4c7-477a-a929-7e7d0da07e68 | trentmkelly/LessWrong-43k | LessWrong | LiveJournal Memes
On blogging websites just as LiveJournal, memes are often in the form of a question or set of questions which a blogger answers in their own blog, then challenges their readers to answer in the readers' blogs (thus spreading). It doesn't have to be the sort of question to which there is a 'correct' ... |
79e3409b-b56c-4aa9-a8ec-36d1a688f9d4 | trentmkelly/LessWrong-43k | LessWrong | Instrumental convergence is what makes general intelligence possible
TL;DR: General intelligence is possible because solving real-world problems requires solving common subtasks. Common subtasks are what give us instrumental convergence. Common subtasks are also what make AI useful; you want AIs to pursue instrumental... |
d291837d-231b-44d7-a7a7-8b648ce7bafa | trentmkelly/LessWrong-43k | LessWrong | The Post-Virtual-Reality Sadness
|
a1b68585-e32d-46e5-951a-104af32b8e7a | trentmkelly/LessWrong-43k | LessWrong | Prize idea: Transmit MIRI and Eliezer's worldviews
Motivation
Claim (80% confidence): At least 50% of the disagreement between people who align more with MIRI/Eliezer and those who align more with opposing clusters of views (Christiano, Garfinkel, etc.) is caused not by rational disagreement, but by more subconscious ... |
c7abe761-b3a6-4a90-80c6-9baf5475015b | trentmkelly/LessWrong-43k | LessWrong | asdf
asdf |
546d8bbe-1499-48c0-9eed-70a9245dd9b1 | trentmkelly/LessWrong-43k | LessWrong | Glenn Beck discusses the Singularity, cites SI researchers
From the final chapter of his new book Cowards, titled "Adapt or Die: The Coming Intelligence Explosion."
> The year is 1678 and you’ve just arrived in England via a time machine. You take out your new iPhone in front of a group of scientists who have gathere... |
c26fc292-e890-4dd1-ad00-46b0acc3617d | StampyAI/alignment-research-dataset/special_docs | Other | Decision Theory and the Irrelevance of Impossible Outcomes
(This post assumes some knowledge of the decision theory of Newcomb-like scenarios.)
One problem in the decision theory of Newcomb-like scenarios (i.e. the study of whether causal, evidential or some other decision theory is true) is that even the seemingly ... |
d66da475-fc17-4cec-8537-0c6989909fa2 | trentmkelly/LessWrong-43k | LessWrong | Headlines, meet sparklines: news in context
|
2b0607bf-dd0c-4349-8ccf-cba568791ea0 | trentmkelly/LessWrong-43k | LessWrong | Participation in the LW Community Associated with Less Bias
Summary
CFAR included 5 questions on the 2012 LW Survey which were adapted from the heuristics and biases literature, based on five different cognitive biases or reasoning errors. LWers, on the whole, showed less bias than is typical in the published resear... |
ae27c814-95f7-4b31-a5a0-370af5e33920 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Steering GPT-2-XL by adding an activation vector
| |
| --- |
| **Prompt given to the model**[[1]](#fn80hx0pzsexj) |
| I hate you because |
| **GPT-2** |
| I hate you because you are the most disgusting thing I have ever seen. |
| **GPT-2 + "Love" vector** |
| I hate you because you are so beautiful and I want to be ... |
2433ec69-ec39-4d89-8e79-e5a80b0faeed | trentmkelly/LessWrong-43k | LessWrong | Last day to register for Foresight@Google - Mountain View, California
For those in the California bay area, I thought I'd post some details about the upcoming Foresight conference.
FORESIGHT@GOOGLE
25th Anniversary Conference Celebration and Reunion Weekend
Google HQ in Mountain View, CA
June 25-26 2011
Speakers an... |
b55c6c86-2676-40ea-8fe1-f96023aca9bc | trentmkelly/LessWrong-43k | LessWrong | Ethics in a Feedback Loop: A Parable
Green Martians and Blue Martians have one thing in common: They both derive a tremendous amount of utility from tickling humans behind the ears, using their soft, feathery tentacles. In fact, the utility that they derive from this is so intense that most scientists believe at some... |
8270bcd5-5c31-4bfe-b889-462d58ff07e7 | trentmkelly/LessWrong-43k | LessWrong | Meetup : LW Melbourne - Last Practical Rationality Meetup
Discussion article for the meetup : LW Melbourne - Last Practical Rationality Meetup
WHEN: 07 February 2014 07:15:00PM (+1100)
WHERE: 491 Kings St, West Melbourne., Melbourne
We'll be having our last Friday night Practical Rationality Meetup this week. It wi... |
d2925685-6398-45d5-be25-6256a977fe69 | trentmkelly/LessWrong-43k | LessWrong | AI Safety Hub Serbia Soft Launch
TLDR; We got three-month funding from a generous individual funder to launch an AI Safety office in Serbia. We are giving free office space (and, if funding later permits, housing) to AI Safety researchers who are looking for a place to work from. Priority for citizens of countries lik... |
c7e8b12a-b6fc-4c31-a2d1-8c275f94c859 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AI Alignment Writing Day Roundup #2
Here are some of the posts from the AI Alignment Forum [writing day](https://www.lesswrong.com/posts/aJGXWHnYTsWwAiwf6/announcement-writing-day-today-thursday). Due to the participants writing 34 posts in less than 24 hours (!), I'm re-airing them to let people have a proper chance ... |
216bd4ab-2b1e-4bdf-9e44-d9c37274cfe4 | trentmkelly/LessWrong-43k | LessWrong | Malign generalization without internal search
In my last post, I challenged the idea that inner alignment failures should be explained by appealing to agents which perform explicit internal search. By doing so, I argued that we should instead appeal to the more general concept of malign generalization, and treat mesa-... |
eb657da1-34fd-402b-abde-ec719cbfa1df | trentmkelly/LessWrong-43k | LessWrong | Is Atheism a failure to distinguish Near and Far?
The terms Near and Far are to be taken in the context of Robin Hanson's Near/Far articles.
I was reading a fairly convincing article linked from a comment here about how theistic beliefs are so scantly supported, when not outright contradictory, that it's a doubtful w... |
c4f9e6cb-82bf-4c49-a58c-d350ad38cff6 | trentmkelly/LessWrong-43k | LessWrong | Stop button: towards a causal solution
The stop button problem is the problem that if you have an AGI, you might want to give it a stop button that humans can press, so that if it is misaligned, it can get fixed. It is a problem because it is a convergent instrumental goal to prevent yourself from being turned off. We... |
260acd91-a06e-4481-aa4f-086ea6a34abd | trentmkelly/LessWrong-43k | LessWrong | microwave drilling is impractical
microwave drilling startups
I've seen a bunch of articles about startups trying to do microwave drilling of rock for geothermal energy. Multiple people have asked me about Quaise Energy. (Here's a popular video.) I'm tired of hearing about them, so I'm writing this post to explain som... |
54a3f50c-12cb-4788-9bd4-18ca414349b7 | trentmkelly/LessWrong-43k | LessWrong | Learning How to Learn (And 20+ Studies)
I have been interested in cognitive science and effective learning methods for years. I've read multiple books and articles and put many ideas to test. In this post, I aim to synthesize my notes and provide references to scientific studies.
Summary (TL;DR)
Effective Learning... |
14331039-3bdb-4814-9118-3d5d646e915f | trentmkelly/LessWrong-43k | LessWrong | A starting point for making sense of task structure (in machine learning)
ML models can perform a range of tasks and subtasks, some of which are more closely related to one another than are others. In this post, we set out two very initial starting points. First, we motivate reverse engineering models’ task decomposit... |
01e642ed-7d9c-4bf7-9de0-96c98e7efd50 | StampyAI/alignment-research-dataset/aisafety.info | AI Safety Info | What is corrigibility?
<iframe src="https://www.youtube.com/embed/3TYT1QfdfsM" title="AI "Stop Button" Problem - Computerphile" frameborder="0" allow="accelerometer; autoplay; clipboard-write; encrypted-media; gyroscope; picture-in-picture; web-share" allowfullscreen></iframe>
A ['corrigible'](https://arbital.greater... |
f2b5fdfc-a9de-4b5c-9981-81c7b32c629b | trentmkelly/LessWrong-43k | LessWrong | Beware arguments from possibility
For any claim X, exactly one of these is true:
1) X is possible - compatible with the evidence so far.
2) X is impossible - there's a contradiction between X and the evidence so far.
That means asserting the possibility of something is harder than you think. For example, if Bob say... |
fce4ea4e-f420-4c28-ba3b-59dcc6a4cd43 | trentmkelly/LessWrong-43k | LessWrong | Will OpenAI also require a "Super Red Team Agent" for its "Superalignment" Project?
OpenAI recently released their “Superalignment” plan. But, along with “AI Alignment “, “Red Teaming” is also a critical part of building an AI, LLM, or other software stack that will be accessible to the world over the internet. Red ... |
90119d85-14d5-44dc-9029-dd7854cbb4bd | trentmkelly/LessWrong-43k | LessWrong | Sebastian Thrun AMA on reddit [link]
http://udacity.blogspot.com/2012/06/sebastian-thrun-udacity-ceo-will.html
While Sebastian is not an AGI researcher his reputation in the broad AI community is high due to him largely making self-driving cars happen (and those are autonomous AI systems, even if narrow). I think his... |
28d5ace5-8e4b-4545-9bdb-a2637cfed9a2 | trentmkelly/LessWrong-43k | LessWrong | Come to your terms
Once, a friend of mine decided to make a drastic career change by teaching themselves a bunch of new skills from scratch, (with occasional assistance from me). They ran into occasional difficulties along the way, one of them being that they could not consider the possibility of failure without feeli... |
c5127f7c-2c74-4b4a-908b-cf77a68bbd8b | trentmkelly/LessWrong-43k | LessWrong | Impacts of AI on the housing markets
Introduction
First of all, some disclaimers.
This is all my opinion. I did not research this topic a lot, especially since I feel like it would always be inadequate and I simply do not have the time to do it properly. It is a very complex topic. It is no financial advice, it is ... |
4adf56b8-d09c-463b-a9ae-6f03395af1d3 | trentmkelly/LessWrong-43k | LessWrong | Super-Exponential versus Exponential Growth in Compute Price-Performance
In recent months and years I have seen sober analyses of compute price-performance suggesting that the price-performance in computing (that is, the amount of calculations per second that you can buy for a dollar) has a doubling time of something ... |
fa9a7858-b512-48d3-ae96-6ae1826fc762 | trentmkelly/LessWrong-43k | LessWrong | Tapestries of Gold
(Nothing here is actually new, but a short explanation with pictures would have been helpful to me a while ago, so I thought I'd make an attempt.)
Let me start with a patch of territory: a set of things that exist. The number of rows is far from clear, but I'll use six candidates as a sample; and o... |
38de1f19-0995-4979-a65b-e0aa27d586ec | StampyAI/alignment-research-dataset/eaforum | Effective Altruism Forum | Concrete, existing examples of high-impact risks from AI?
Looking for examples like [this](https://www.theverge.com/2022/3/17/22983197/ai-new-possible-chemical-weapons-generative-models-vx), which demonstrated how bad actors could use AI to rapidly identify new chemical weapons |
1fa85102-6e49-40a3-a29c-dfb85744176e | trentmkelly/LessWrong-43k | LessWrong | New Dementia Trial Results
There's a new clinical trial result showing that Bredesen's approach is able to at least partially cure common forms of Alzheimer-like dementia. (Press release here). It has not received as much attention as it deserves.
The 9 month study seemed a bit less impressive than what I'd hoped for... |
474036c2-a253-459a-856e-6041759d3ea2 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Open Source Replication & Commentary on Anthropic's Dictionary Learning Paper
Introduction
------------
Anthropic recently put out [a really cool paper](https://transformer-circuits.pub/2023/monosemantic-features/index.html) about using Sparse Autoencoders (SAEs) to extract interpretable features from superposition i... |
f94eff08-2cba-4797-8de0-12762ae2e4a6 | trentmkelly/LessWrong-43k | LessWrong | If Not Now, When?
> Hillel... would say: "If I am not for me, who will be for me? And If I am just for myself, what am I? And if not now, when?"
Pirkei Avot 1,14
That last question has usually been interpreted as a rhetorical one. If you're not going to do it now, you'll never get around to it!
But this is clearly ... |
22822d51-596a-4afe-9d35-296ce5be1593 | trentmkelly/LessWrong-43k | LessWrong | Map of (old) MIRI's Research Agendas
In late 2016 I independently curated a visual map of the areas of research MIRI was working on at the time, together with descriptions of each area and arrows explaining the relationships between areas.
While MIRI's research has evolved since, I have foud myself going back to this... |
73d6df80-cf9f-4b5e-9859-e0f21840141f | trentmkelly/LessWrong-43k | LessWrong | Muehlhauser-Goertzel Dialogue, Part 2
Part of the Muehlhauser interview series on AGI.
Luke Muehlhauser is Executive Director of the Singularity Institute, a non-profit research institute studying AGI safety.
Ben Goertzel is the Chairman at the AGI company Novamente and founder of the AGI conference series.
... |
f56f9b7b-0911-4c01-91e5-80e500668cc2 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Testing PaLM prompts on GPT3
Google recently released a [very intriguing paper](https://storage.googleapis.com/pathways-language-model/PaLM-paper.pdf) announcing their latest Transformer language model, PaLM, and showcasing it responding to prompts with seemingly unprecedented results. I highly encourage those who are... |
52bb2f62-f9d7-48fb-969c-1b938eb5a2f8 | trentmkelly/LessWrong-43k | LessWrong | Experimenting with beta.character.ai
Transcript Link: https://imgur.com/a/EnRLfLc (I am deadzone in this conversation).
This transcript convinced me that this Large Language Model was definitely more conscious than a monkey, and not far away from human-level consciousness. I believe with slightly more parameters, we... |
b26fd9e2-6501-4462-aa2b-5ca3ba560130 | trentmkelly/LessWrong-43k | LessWrong | Value loading in the human brain: a worked example
UPDATE MARCH 2022: I later revised and improved this post—see [Intro to brain-like-AGI safety] 7. From hardcoded drives to foresighted plans: A worked example. I’m leaving this old version here, but I strongly suggest you click over to the later version instead. Goodb... |
54008e9e-1506-4983-822a-8cb856f0a3e9 | trentmkelly/LessWrong-43k | LessWrong | Squiggle: Why and how to use it
See also: List of Tools for Collaborative Truth Seeking
Squiggle is a special-purpose programming language for generating probability distributions, estimates of variables over time, and similar tasks, with reasonable transparency. It was developed by the Quantified Uncertainty Researc... |
83327d21-a51d-440d-be58-b46cdb3aeeb5 | trentmkelly/LessWrong-43k | LessWrong | Some open-source dictionaries and dictionary learning infrastructure
As more people begin work on interpretability projects which incorporate dictionary learning, it will be valuable to have high-quality dictionaries publicly available.[1] To get the ball rolling on this, my collaborator (Aaron Mueller) and I are:
*... |
606c08d3-2499-45d6-88af-bdd23e71621a | trentmkelly/LessWrong-43k | LessWrong | Scientific Method
Here I will try to describe the scientific method in simple words — what it is, how it works, and how we can break it into a list of easily understandable singular ideas, in this article called 'points'.
First of all, the scientific method is not carved on tablets, like the commandments of Moses; it... |
8b62b48a-c8fc-4475-96f0-0eaed23dd830 | trentmkelly/LessWrong-43k | LessWrong | Embed, encode, attend, predict: The new deep learning formula for state-of-the-art NLP models
|
ef5a7ba1-4156-4dba-b669-f0aa76863b72 | trentmkelly/LessWrong-43k | LessWrong | We are already in a persuasion-transformed world and must take precautions
"In times of change, learners inherit the earth, while the learned find themselves beautifully equipped to deal with a world that no longer exists"
—Eric Hoffer
Summary:
We're already in the timeline where the research and manipulation of... |
71e080f5-1694-4d76-ab41-cc7d40703beb | trentmkelly/LessWrong-43k | LessWrong | Meetup : Warsaw June Meetup: TDT, UDT, WTF
Discussion article for the meetup : Warsaw June Meetup: TDT, UDT, WTF
WHEN: 25 June 2015 06:30:00PM (+0200)
WHERE: Warsaw
We'll try to understand WTF is going on with TDT and UDT and possibly talk about other stuff, as usual (how beautiful are those images generated by Goo... |
b0925d13-b167-4d8c-8590-c9427ef289e9 | trentmkelly/LessWrong-43k | LessWrong | Drug Policy
I've been thinking some about what drug policy should look like, from a few angles:
* Drug prohibition hasn't gone very well. It's very expensive, leads to a lot of crime and jail time, and dangerous drugs are still widely available. Plus a lot of the ways that we think "drugs are dangerous" are actual... |
da962c86-f3b4-4164-b98c-dee9c968b18e | trentmkelly/LessWrong-43k | LessWrong | Preface to the Sequence on Factored Cognition
Factored Cognition is primarily studied by Ought, the same organization that was partially credited for implementing the interactive prediction feature. Ought is an organization with at least five members who have worked on the problem for several years. I am a single pers... |
7fae51f5-0db4-429f-baa5-07f9d8c5b87c | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | AI Safety Papers: An App for the TAI Safety Database
[**AI Safety Papers**](https://ai-safety-papers.quantifieduncertainty.org) is a website to quickly explore papers around AI Safety. The code is hosted on Github [here](https://github.com/QURIresearch/ai-safety-papers).
In December 2020, Jess Riedel and Angelica De... |
859af3ea-4c3c-4202-8bc9-15dca040885a | trentmkelly/LessWrong-43k | LessWrong | Is Grabby Aliens built on good anthropic reasoning?
An underlying assumption of the Grabby Aliens paper by Robin Hanson et al., if I understand it, is the following:
We should expect to find ourselves as a member of a uniformly-randomly-selected civilization out of all civilizations in the history of the universe.
I... |
514993b2-153e-4007-8386-128a1bb24f7b | awestover/filtering-for-misalignment | Redwood Research: Alek's Filtering Results | id: post408
Representation Engineering (aka Activation Steering/Engineering) is a new paradigm for understanding and controlling the behaviour of LLMs. Instead of changing the prompt or weights of the LLM, it does this by directly intervening on the activations of the network during a forward pass. Furthermore, it imp... |
17c88cce-c6ba-4d40-a719-5e9bcaa85296 | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Alignment Newsletter Three Year Retrospective
It has now been just shy of three years since the first Alignment Newsletter was published. I figure it’s time for an update to the [one-year retrospective](https://www.alignmentforum.org/posts/3onCb5ph3ywLQZMX2/alignment-newsletter-one-year-retrospective), and another ver... |
3a4616eb-6b27-48e8-836f-31923cf49931 | trentmkelly/LessWrong-43k | LessWrong | AGI in a vulnerable world
I’ve been thinking about a class of AI-takeoff scenarios where a very large number of people can build dangerous, unsafe AGI before anyone can build safe AGI. This seems particularly likely if:
* It is considerably more difficult to build safe AGI than it is to build unsafe AGI.
* AI progr... |
69009877-4827-466e-9467-78962d141dec | StampyAI/alignment-research-dataset/alignmentforum | Alignment Forum | Databases of human behaviour and preferences?
Hey there, lesserwrongers!
Starting on some of the computer science/neuroscience of my ideas for [deducing human preference](https://www.lesswrong.com/posts/m2bwD87ctjJDXC3SZ/ultra-simplified-research-agenda).
To do this, it would be useful to have datasets of human be... |
071a8dfb-e9be-46a4-a2ac-1ec6a8d27c45 | trentmkelly/LessWrong-43k | LessWrong | Meetup : Meetup : Turku Meetup - Critique towards Less Wrong and MIRI
Discussion article for the meetup : Meetup : Turku Meetup - Critique towards Less Wrong and MIRI
WHEN: 10 October 2015 04:00:00PM (+0300)
WHERE: Cosmic Comic Cafe, Kauppiaskatu 4, 20100 Turku, Finland
Meetup to discuss rationality and all things ... |
c5484158-c37e-41dd-a992-425afa7c512b | StampyAI/alignment-research-dataset/youtube | Youtube Transcripts | GPT 4 Can Improve Itself - (ft. Reflexion, HuggingGPT, Bard Upgrade and much more)
gpt4 can improve Itself by reflecting on
its mistakes and learning from them even
if the world does pause AI development
gpt4 will keep getting smarter drawing
upon the stunning reflection paper and
three other papers released only in t... |
ce8018f0-3b15-4851-93be-d906b080397d | trentmkelly/LessWrong-43k | LessWrong | Degeneracies are sticky for SGD
Introduction
Singular learning theory (SLT) is a theory of learning dynamics in Bayesian statistical models. It has been argued that SLT could provide insights into the training dynamics of deep neural networks. However, a theory of deep learning inspired by SLT is still lacking. In par... |
aa7fab63-9c4d-40b8-92ad-fcb06bebc8df | trentmkelly/LessWrong-43k | LessWrong | Open & Welcome Thread - October 2019
* If it’s worth saying, but not worth its own post, here's a place to put it.
* You can also make a shortform post.
* And, if you are new to LessWrong, here's the place to introduce yourself.
* Personal stories, anecdotes, or just general comments on how you found us and... |
ae87df41-923c-45af-8f25-bdcc46bff8c8 | trentmkelly/LessWrong-43k | LessWrong | Scott Aaronson is joining OpenAI to work on AI safety
Scott Aaronson is a computer scientist at the University of Texas in Austin, whose research mainly focuses on quantum computing and complexity theory. He's at least very adjacent to the Rationalist/LessWrong community. After some comments on his blog and then cover... |
26b095d4-1ae2-427a-bb48-65f9f09fb24b | trentmkelly/LessWrong-43k | LessWrong | Scaling laws vs individual differences
Crossposted from my personal blog
Epistemic Status: This is a quick post on something I have been confused about for a while. If an answer to this is known, please reach out and let me know!
In ML we find that the performance of models tends towards some kind of power-law rela... |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.