id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
de0beab4-eb38-4633-a41f-b8967c3e7d21
trentmkelly/LessWrong-43k
LessWrong
Defending the non-central fallacy Aaron Bergman recently defended logical fallacies against the charge that they're bad arguments. His thesis wasn't new. Gwern highlighted "Bayesian Informal Logic and Fallacy" and "Fallacies as weak Bayesian evidence".  Defending logical fallacies sounds pretty fun, so I decided to d...
891ebc58-0ec8-497e-a18d-c892c4ce4cee
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Surprised By Brains Today's post, Surprised by Brains was originally published on 23 November 2008. A summary (taken from the LW wiki):   > If you hadn't ever seen brains before, but had only seen evolution, you might start making astounding predictions about their ability. You might, for instance, think...
6e71c6bc-1cbf-4156-8e06-2f46dcbccae1
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Cultish Countercultishness Today's post, Cultish Countercultishness was originally published on 30 December 2007. A summary (taken from the LW wiki):   > People often nervously ask, "This isn't a cult, is it?" when encountering a group that thinks something weird. There are many reasons why this question...
6a81ef4c-555e-4bd5-8a79-5964a707b18b
trentmkelly/LessWrong-43k
LessWrong
Beyond Reasonable Doubt? - Richard Dawkins [link] A new article looking at the jury system rationally and scientifically. Excerpt: > Courtroom dramas accurately portray the suspense that hangs in the air when the jury returns and delivers its verdict. All, including the lawyers on both sides and the judge, are on te...
79224219-f144-401b-b7e6-0a4402faa49c
trentmkelly/LessWrong-43k
LessWrong
How to eradicate the desire to check time-wasting sites Linkpost: https://gingerjumble.wordpress.com/2020/12/05/how-to-stop-compulsively-checking-messaging-sites/   I used to be constantly checking messaging apps. Discord, Telegram, WhatsApp (and Email). They are the most skinner-boxy things I use since I don’t use ...
8245cfb2-b22f-451d-a463-1a835739edcc
trentmkelly/LessWrong-43k
LessWrong
THE ARCHIVE Musk’s DOGE and the Data Rush: The Race to Secure the Ultimate Asset By: Jason Reid TL;DR The article speculates that Musk is strategically positioning himself in the AI race by leveraging political influence and initiatives like DOGE to secure exclusive access to vast surveillance data. With traditional...
74c91b8c-86bb-4374-a655-60627db5bb88
trentmkelly/LessWrong-43k
LessWrong
You can't understand human agency without understanding amoeba agency This is supposed to be a trivially true statement, and yet it sounds controversial somehow, doesn't it?  The difference between the two is quantitative, not qualitative, unless you explicitly or implicitly subscribe to the "humans are the only real...
580828ec-5ee0-427d-a43f-38f6c8aac259
trentmkelly/LessWrong-43k
LessWrong
Meetup : San Francisco Meetup: Rationality Diary Discussion article for the meetup : San Francisco Meetup: Rationality Diary WHEN: 07 December 2015 06:15:00PM (-0800) WHERE: 1597 Howard St. San Francisco, CA We'll be meeting to tell stories about when we tried to solve a problem in our lives, and how it went. This ...
0a76f500-c9ac-49cc-a8f3-318eeb901db0
trentmkelly/LessWrong-43k
LessWrong
A central AI alignment problem: capabilities generalization, and the sharp left turn (This post was factored out of a larger post that I (Nate Soares) wrote, with help from Rob Bensinger, who also rearranged some pieces and added some text to smooth things out. I'm not terribly happy with it, but am posting it anyway ...
8f98ac6a-c1bd-484a-850f-71014a15b36f
trentmkelly/LessWrong-43k
LessWrong
Ranking the "competition" based on optimization power Most long term users on Less Wrong understand the concept of optimization power and how a system can be called intelligent if it can restrict the future in significant ways. Now I believe that in this world, only institutions are close to superintelligence in any s...
e71153d3-2da5-4181-b726-7b79ccd9c671
trentmkelly/LessWrong-43k
LessWrong
Defining capability and alignment in gradient descent This is the first post in a series where I'll explore AI alignment in a simplified setting: a neural network that's being trained by gradient descent. I'm choosing this setting because it involves a well-defined optimization process that has enough complexity to be...
eabc4fe2-0896-40b6-81f0-b15ff1bf1284
trentmkelly/LessWrong-43k
LessWrong
Control Fraud   A recent post by Bruce Schneier on control fraud. Control fraud is a process of optimizing an organization for fraud, utilizing a position of power to suborn controls. From the abstract of the paper Bruce references: > Individual “control frauds” cause greater losses than all other forms of property ...
22f0b7d8-0498-4c96-bbb0-6dc4954bf1e5
trentmkelly/LessWrong-43k
LessWrong
Meetup : Melbourne, practical rationality Discussion article for the meetup : Melbourne, practical rationality WHEN: 01 June 2012 07:00:00PM (+1000) WHERE: 55 walsh st, west melbourne 3003 australia Practical rationality, as distinct from the social and rationality outreach meetups. Look for a social meetup on the ...
d8aea774-e9b2-49e2-87c6-721c57098336
trentmkelly/LessWrong-43k
LessWrong
AI Safety Unconference NeurIPS 2022 The AI Safety Unconference brings together persons interested in aspects of AI safety, from technical AI safety problems to issues of governance of AI. As an unconference, it aims to foster valuable social interactions between participants, through moderated discussion groups, one-o...
0248b72c-df7d-41db-82e7-3e0967dadf50
trentmkelly/LessWrong-43k
LessWrong
Progress links digest, 2023-12-15: Vitalik on d/acc, $100M+ in prizes, and more Job openings * Economic Innovation Group is looking for a spring research intern (via ROP fellow @cojobrien) * Future House (AI for bio) hiring for their Assessment Team. “We want Future House to be the world leader for evaluating the sc...
d59e5f57-d1db-4d0c-b9b0-fda4d90a7f4f
trentmkelly/LessWrong-43k
LessWrong
Mech Interp Challenge: October - Deciphering the Sorted List Model I'm writing this post to discuss solutions to the September challenge, and present the challenge for this October.  If you've not read the first post in this sequence, I'd recommend starting there - it outlines the purpose behind these challenges, and...
f49f3411-5544-4c78-87bd-cd5fdc0da4f9
trentmkelly/LessWrong-43k
LessWrong
Chapter 65: Contagious Lies Hermione Granger had read somewhere once, that one of the keys to staying thin was to pay attention to the food you ate, to notice yourself eating it, so that you were satisfied with the meal. This morning she'd made herself toast, and put butter on the toast, and cinnamon on the butter, an...
716c14e1-95bb-4e9f-99ad-d1654042caf4
trentmkelly/LessWrong-43k
LessWrong
The Correct Contrarian Cluster Followup to: Contrarian Status Catch-22 Suppose you know someone believes that the World Trade Center was rigged with explosives on 9/11.  What else can you infer about them?  Are they more or less likely than average to believe in homeopathy? I couldn't cite an experiment to verify it...
c9ea041f-92ea-4ad3-9384-46196daf1851
trentmkelly/LessWrong-43k
LessWrong
Types of recursion As a freshman in college I took an intro linguistics class where we spent a lot of time discussing universal grammar and recursive phrase structures. One of the examples we looked at I still don't fully understand - it illustrated two distinct forms of nested phrases that he mind handles very differ...
102f8ce7-a942-4e52-ab54-144fcd945f7a
trentmkelly/LessWrong-43k
LessWrong
Comments on Power Law Distribution of Individual Impact I had a discussion online yesterday, stemming from whether you should expect to be able to identify individuals who will most shape the long term future of humanity. It was on a discussion of whether CEA should have staff work on doing this full time, and I was e...
0b66ed16-b041-441b-8c9e-c03195c0741f
StampyAI/alignment-research-dataset/blogs
Blogs
2013 in Review: Fundraising **Update 04/16/2014:** At a donor’s [request](http://intelligence.org/2014/04/02/2013-in-review-fundraising/#comment-1318521618) I have replaced the Total Donations per Year chart with one that shows which proportion of the donations were from new and returning donors. Some tweaks were also...
eccc1fff-58fa-4f4a-9e3d-0fb550f85226
trentmkelly/LessWrong-43k
LessWrong
Pigeons outperform humans at the Monty Hall Dilemma [LINK] Humans overthink the problem, which lets their biases in. [Pop Article] [Journal Article]
ad065078-7852-4725-a85e-239d04f00da2
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Neural networks generalize because of this one weird trick > *Produced as part of the*[*SERI ML Alignment Theory Scholars Program*](https://serimats.org/)*- Winter 2022 Cohort* > > > *A big thank you to all of the people who gave me feedback on this post: Edmund Lao, Dan Murfet, Alexander Gietelink Oldenziel, Luci...
bf1c8148-b155-49cb-896b-4dff64f9aaaf
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Maximal lotteries for value learning Suppose Ann asks robot Bob to buy her favorite fruit. Bob doesn't know what Ann's favorite fruit is, but he knows that Ann likes either apples, bananas or oranges, each with probability 1/3, and doesn't like the other two fruits at all. At the groceries, for his money Bob can buy 1...
8d2af1af-1d79-4288-9118-5e8f63154c62
trentmkelly/LessWrong-43k
LessWrong
Low-effort review of "AI For Humanity" Stumbled across a book in the new section of the library: "AI For Humanity," by Andeed Ma, James Ong (founder of the think tank AIII, which is also the sound I make when thinking about AI risk), and Siok Siok Tan. It's a mass-market-ish book about, well, AI for humanity, by a sam...
94457226-f64b-40c5-a75a-5907178e453f
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Competitive safety via gradated curricula *Epistemic status: brainstorming some speculative research directions. Not trying to thoroughly justify the claims I’m making.* One way to think about the AI safety problem: there’s a spectrum of methods which each represent a different tradeoff between safety and ease of tra...
7af071c6-6f6f-4aa3-b793-276a9614db5d
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Debate on Instrumental Convergence between LeCun, Russell, Bengio, Zador, and More *An* *[actual debate](https://facebook.com/story.php?story_fbid=10156248637927143&id=722677142)* *about instrumental convergence, in a public space! Major respect to all involved, especially Yoshua Bengio for great facilitation.* *For ...
c650ae47-93d6-4511-a71e-dc4ad8d2e235
trentmkelly/LessWrong-43k
LessWrong
Survey for alignment researchers! UPDATE 3/9: Thanks to broad participation from the community, this and associated surveys have raised approximately $10,000 for high-impact alignment organizations. Given the reasonable sample size we now have, we are now going to pause donations for any subsequent responses. (However...
8b7e4587-52b0-4c20-a962-0434a5a3bc0c
trentmkelly/LessWrong-43k
LessWrong
Dreams of Friendliness Continuation of:  Qualitative Strategies of Friendliness Yesterday I described three classes of deep problem with qualitative-physics-like strategies for building nice AIs - e.g., the AI is reinforced by smiles, and happy people smile, therefore the AI will tend to act to produce happiness.  In...
c3e22c14-df0a-4b67-b5be-d9445f5ff9bc
trentmkelly/LessWrong-43k
LessWrong
The Sequences Highlights on YouTube This post consists of the Sequences Highlights compiled to a zoomer-readable format, i.e., video. If you'd prefer to watch these videos on YouTube proper rather than on this post, use this YouTube playlist.  These videos may or may not suit your mind well; as far as I can tell, the...
bfb854f5-cae6-4c68-a36e-ed9bc4b7b2fe
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Are we there yet? > *How good of an idea do the people on this website have in regards to predicting an AGI apocalypse?* > >   **I** get the impression that a lot of people seem to be very certain that it will happen relatively soon and I think I've figured out a way to test it. The spreadsheet for the data would ...
eddef481-05a6-4e9a-ace6-9a0fb5610f7b
trentmkelly/LessWrong-43k
LessWrong
‘GiveWell for AI Safety’: Lessons learned in a week On prioritizing orgs by theory of change, identifying effective giving opportunities, and how Manifund can help. Epistemic status: I spent ~20h thinking about this. If I were to spend 100+ h thinking about this, I expect I’d write quite different things. I was surpr...
df504b2c-5268-4ae0-a9ce-2ec49a36c198
trentmkelly/LessWrong-43k
LessWrong
Resources I send to AI researchers about AI safety If you're interested in seeing my up-to-date recommendations, please see the Arkose Resource Center! For the purpose of reducing the number of pages where I keep updated recommendations, I'm now retiring this post. However, you're welcome to look at the 2023 web arch...
a5029056-981d-4a92-a756-d77822324606
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Status Quo Engines - AI essay This essay for the Open Philanthropy AI Worldviews Contest is targeted at Question 1. AIs have recently accrued some impressive successes, raising both hopes and concerns that artificial general intelligence (AGI) may be just decades away. I will argue that the probability of an AGI, as ...
62e94758-64ba-4802-8180-cca7c916639a
trentmkelly/LessWrong-43k
LessWrong
Wrapping and Centering In the early days of the web screens weren't very large, and people normally wanted text to run from edge to edge. If you code up HTML by hand, with no styling, this is still what you'll get. As screens got larger, though, this could lead to lines that were too wide for easy reading. Sites have ...
6c753d16-4569-441d-ad41-f360f4edc9f7
trentmkelly/LessWrong-43k
LessWrong
Detach the grim-o-meter I'm betting that the last three posts have given many readers an incorrect impression about my demeanor. It's easy to read those posts and conclude that I must be a grim, brooding character who goes around with his jaw set all day long. Which is understandable, but silly. You don't need to car...
64679499-0b47-48e2-80e2-0d9d1a6e3233
trentmkelly/LessWrong-43k
LessWrong
On predictability, chaos and AIs that don't game our goals I want to thank @Ryan Kidd, @eggsyntax and Jeremy Dolan for useful discussions and for pointing me to several of the relevant resources (mentioned in this post) that I have used for linking my own ideas with those of others. Executive summary Designing an AI...
cf577b41-729f-4094-b7d5-13cc64229ad5
trentmkelly/LessWrong-43k
LessWrong
Meetup : Austin, TX Discussion article for the meetup : Austin, TX WHEN: 10 December 2011 01:30:00PM (-0600) WHERE: 2222B Guadalupe St Austin, Texas 78705 Once again, the Austin LW meetup is meeting at Caffe Medici at 1:30 PM. We sit on the second floor to the left, near (or often on) the stage. It's finals time, ...
a08c18ed-2a7d-4bf3-a3fc-e36381d6b522
trentmkelly/LessWrong-43k
LessWrong
What should AI safety be trying to achieve? This is the second of three posts summarizing what I learned when I interviewed 17 AI safety experts about their "big picture" of the existential AI risk landscape: how will artificial general intelligence (AGI) play out, how things might go wrong, and what the AI safety com...
2349cf70-e62f-4daf-b7e9-e9a4ce554d4c
trentmkelly/LessWrong-43k
LessWrong
Mosquito Net Fishing I recently saw a study claiming: > Distributed mosquito nets are intended to be used for malaria protection, yet increasing evidence suggests that fishing is a primary use for these nets, providing fresh concerns for already stressed coastal ecosystems. >   —The perverse fisheries consequences ...
dccdd1a2-c346-4426-b6c4-1c3d59110f15
StampyAI/alignment-research-dataset/lesswrong
LessWrong
An OpenAI board seat is surprisingly expensive The Open Philanthropy Project [recently bought](http://www.openphilanthropy.org/focus/global-catastrophic-risks/potential-risks-advanced-artificial-intelligence/openai-general-support) a seat on the board of the billion-dollar nonprofit AI research organization OpenAI for...
f4d07f6a-d644-4157-aa19-459673cc8e6d
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Continuous Improvement Today's post, Continuous Improvement was originally published on 11 January 2009. A summary (taken from the LW wiki):   > Humans seem to be on a hedonic treadmill; over time, we adjust to any improvements in our environment - after a month, the new sports car no longer seems quite ...
cf7151e3-5503-4c2a-b03d-2bcb832088e8
trentmkelly/LessWrong-43k
LessWrong
Forecasting Newsletter: December 2021 Highlights * Polymarket’s future is uncertain after it settled with the CFTC for $1.4M * Astral Codex Ten gives out $40k to forecasting projects * Many people, including Mathew Yglesias, write predictions for 2022. * Eli Lifland writes the reference piece on bottlenecks to imp...
b834a9e5-c5bd-4ce2-aa78-3d3a8a51c619
trentmkelly/LessWrong-43k
LessWrong
There is no sharp boundary between deontology and consequentialism The purpose of this post is to serve as storage for particular thought so I can link it in dialogues. I've seen logic along these lines several times: 1. Consequentialist reasoning is more complex than following deontological rules; 2. Neural netwo...
3f8f518e-d0bc-440a-a402-148096edb127
trentmkelly/LessWrong-43k
LessWrong
Viral Guessing Game Julia introduced me to Metazooa, a taxonomic guessing game. You're trying to figure out what animal the computer has chosen. Each time you guess the computer will reveal the lowest taxonomic clade that they both belong to. For example, if you guessed platypus and the answer was humans then it would...
8b966d72-a52c-446f-a2cb-2a95eace6864
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"EDIT: Mean and standard deviation of individidual predictions can be found here. Thanks to all my brave assessors, I now have the data about Kurzweil's 1999 predictions about 2019. This was a follow up to a previous assessment about his predictions about 2009, which showed a mixed bag. Roughly evenly divided between r...
3e276eda-c941-42d9-89b5-1b6f285aba48
trentmkelly/LessWrong-43k
LessWrong
Fictional parasites very different from our own Note: this is a fictional story. Heavily inspired by SSC’s similar posts on fictional legal systems and fictional banned drugs. Neuroplana temporalis Neuroplana temporalis is a flatworm that resides in the cerebrospinal fluid of mammals with a diurnal rhythm. It has a ...
590e1596-dc51-46ad-8ed2-715aac8ccb28
trentmkelly/LessWrong-43k
LessWrong
Rationality Quotes 1 I'll be moving to Redwood City, CA in a week, so forgive me if I don't get a regular post out every day between now and then.  As a substitute offering, some items from my (offline) quotesfile: ---------------------------------------- > "It appears to be a quite general principle that, whenever ...
98d4d76a-44da-4ed3-8bdc-87bfabd4cf0f
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"One or two research groups have published work on carbon dioxide and cognition. The state of the published literature is confusing.Here is one paper on the topic. The authors investigate a proprietary cognitive benchmark, and experimentally manipulate carbon dioxide levels (without affecting other measures of air qual...
9042b96b-9177-4de1-96d4-aee74d781a3c
trentmkelly/LessWrong-43k
LessWrong
Simplified preferences needed; simplified preferences sufficient AI scientists at EvenDeeperMind: "Hey everyone! We have a developed a low-impact AI!" AI policy people at OpenFutureofExistentialAI: "Fantastic! What does it do?" AI scientists: "It's limited to answering questions, and it has four possible outputs, o1...
18383005-be0f-43ae-8fcb-e53b69a42ac8
trentmkelly/LessWrong-43k
LessWrong
A Proposal for a Simpler Solution To All These Difficult Observations and Problems I am not perfectly sure how this site has worked (although I skimmed the "tutorials") and I am notorious for not understanding systems as easily and quickly as the general public might. At the same time I suspect a place like this is fo...
2d398a3a-647f-43e9-b17f-71f75d647ceb
trentmkelly/LessWrong-43k
LessWrong
Is Science Slowing Down? [This post was up a few weeks ago before getting taken down for complicated reasons. They have been sorted out and I’m trying again.] Is scientific progress slowing down? I recently got a chance to attend a conference on this topic, centered around a paper by Bloom, Jones, Reenen & Webb (2018...
1cb7fc6d-34b3-412c-8377-7732577bb9a1
trentmkelly/LessWrong-43k
LessWrong
Controlled Creative Destruction This is a crosspost from https://250bpm.substack.com/p/controlled-creative-destruction. Creative destruction is a process which lets the inefficient fail and the efficient survive. It's the economists' way of saying of "survival of the fittest". In free markets, creative destruction r...
af851b6d-7af9-483a-9d1a-8bc3ab415c67
trentmkelly/LessWrong-43k
LessWrong
[Link, Humor] The Best Christmas Ever In the holiday spirit, here's a comic that many LWers may appreciate...  and perhaps even find useful for illustrating a certain, ah, topic: http://www.strippycomics.com/2010/01/05/gifto/ Enjoy!
5f93afd3-4230-4e8c-b888-bd871994c493
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Medicine loves guidelines. But everywhere else, guidelines are still underappreciated.Consider a recommendation, like “Try Lexapro!” Even if Lexapro is a good medication, it might not be a good medication for your situation. And even if it’s a good medication for your situation, it might fail for unpredictable reasons...
05abcf21-84b7-450c-9032-baa08f1e9a2e
trentmkelly/LessWrong-43k
LessWrong
Inner alignment: what are we pointing at? Proof that a model is an optimizer says very little about the model. I do not know what a research group is studying outer alignment is studying. Inner alignment seems to cover the entire problem at the limit. Whether an optimizer is mesa or not depends on your point of view. ...
5cf1a4a7-5a1c-4031-a516-5e1086bd6379
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
How do you talk about AI safety? My impression is that it's easy to contribute to ["un-nuanced and inaccurate" discourse](https://www.openphilanthropy.org/blog/potential-risks-advanced-artificial-intelligence-philanthropic-opportunity#Sec8) or hype about artificial intelligence while talking about AI safety. Personall...
047f3594-23ee-41f6-9268-faf984e9d638
StampyAI/alignment-research-dataset/blogs
Blogs
Toby Walsh on computational social choice ![Toby Walsh portrait](https://intelligence.org/wp-content/uploads/2014/03/Walsh_w727.jpg)[Toby Walsh](http://www.cse.unsw.edu.au/~tw/) is a professor of artificial intelligence at [NICTA](http://www.nicta.com.au) and the University of New South Wales. He has served as Scienti...
adaeee81-eaf0-4b11-bc31-dd044f76396f
StampyAI/alignment-research-dataset/blogs
Blogs
February 2020 Newsletter #### Updates * Colm [reviews our 2019 fundraiser](https://intelligence.org/2020/02/13/our-2019-fundraiser-review/): taking into account matching, we received a total of $601,120 from 250+ donors. Our thanks again for all the support! * Evan Hubinger's [Exploring Safe Exploration](https://www...
ac28c407-f206-45b1-8967-a296c4c060fc
StampyAI/alignment-research-dataset/special_docs
Other
Artificial intelligence: The future is superintelligent [Book review of "Life 3.0: Being Human in the Age of Artificial Intelligence" by Max Tegmark] [Download PDF](/articles/548520a.pdf) ### Subjects \* [Government](/subjects/government) \* [Mathematics and computing](/subjects/mathematics-and-computing) \* [...
63d95385-7dc2-4b68-a4bd-cbe7a736b538
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"Followup to: Contrarian Status Catch-22 Suppose you know someone believes that the World Trade Center was rigged with explosives on 9/11. What else can you infer about them? Are they more or less likely than average to believe in homeopathy? I couldn't cite an experiment to verify it, but it seems likely that: There a...
f9b30b8c-ed12-4f4e-a36e-a07990195cec
StampyAI/alignment-research-dataset/arbital
Arbital
Introductory guide to logarithms Welcome to the Arbital introduction to logarithms! In modern education, logarithms are often mentioned but rarely motivated. At best, students are told that logarithms are just a tool for inverting [exponentials](https://arbital.com/p/4ts). At worst, they're told a bunch of properties ...
27472dd7-f884-48bb-b308-06532c7ecfa0
trentmkelly/LessWrong-43k
LessWrong
Necromancy's unintended consequences. Note: I would suggest listening to the audio narration if convenient for more dramatic effect. Once upon a time, in a land far away, where the misty mountains met the sea, there lay the kingdom of Thalyria. Thalyria was a realm where magic and mathematics intermingled, and its pe...
b1c4c411-f445-40fc-9ad2-f20c1a9b9790
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Could someone help me understand why it's so difficult to solve the alignment problem? AGI will be able to model human language and psychology very accurately. Given that, isn't alignment easy if you train the AGI to interpret linguistic prompts in the way that the "average" human would? (I know language doesn't encod...
44b16ea8-f852-4759-8230-e0dabee3a1f2
trentmkelly/LessWrong-43k
LessWrong
Teaching My Younger Self to Program: A case study of how I'd pass on my skill at self-learning When I was 21 I wanted to become a video game programmer. Except I didn’t know how to program. I tried to teach myself and failed. Then I went back to school and learned the basics of Java in a week, aced the course in two...
9c776fc0-adc5-4571-9936-40087dcf75d5
trentmkelly/LessWrong-43k
LessWrong
Rationalist sonnets about being a person 7 sonnets and some comments A friend tells me that, if there is no soul, There is no clash of body against mind. I hate to be contentious, but I find The case is rather different, on the whole. For flesh and mind are clashing all the time; The flesh says "eat!", the mind says...
c954dc1d-d7da-4001-a37a-23085c650441
trentmkelly/LessWrong-43k
LessWrong
Monthly Roundup #5: April 2023 As AI eats the world and my time and news feeds, it is always good to know other things also happen. Here are some of them, that did not seem to fit anywhere else and did not seem worth standing on their own even with my new lower thresholds for that. BAD NEWS Not only do you only get ...
e85f1b4b-8ebf-4698-85e7-965564c54412
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
The Hubinger lectures on AGI safety: an introductory lecture series In early 2023, I ([Evan Hubinger](https://www.alignmentforum.org/users/evhub)) gave a series of recorded lectures to [SERI MATS](https://www.serimats.org) fellows with the goal of building up a series of lectures that could serve as foundational intro...
53417a70-ff1c-42e6-88d7-dd8419e5d6a8
trentmkelly/LessWrong-43k
LessWrong
Dissolving the Thread of Personal Identity (Background: I got interested in anthropics about a week ago. It has tormented my waking thoughts ever since in a cycle of “be confused, develop idea, work it out a bit, realize that it fails, repeat” and it is seriously driving me berserk by this point. While drawing a bunch...
f1e187a4-e66b-44b0-9c89-ada5e59a2235
trentmkelly/LessWrong-43k
LessWrong
[Research Update] Sparse Autoencoder features are bimodal Overview The sparse autoencoders project is a mechanistic interpretability effort to algorithmically find semantically meaningful “features” in a language model. A recent update hints that features learned by this approach separate into two types depending on t...
f9f7749f-044e-4b71-b46d-74e5ec8e3381
StampyAI/alignment-research-dataset/arbital
Arbital
Free groups are torsion-free Given the [https://arbital.com/p/-5kg](https://arbital.com/p/-5kg) $FX$ on a set $X$, it is the case that $FX$ has no [torsion](https://arbital.com/p/torsion_group_theory) elements. That is, every element has infinite order except for the [https://arbital.com/p/-54p](https://arbital.com/p/...
9399bfc5-b5d6-464e-90f4-df9abcfb5ed0
trentmkelly/LessWrong-43k
LessWrong
Being good at the basics (crossposted from my blog) In Brazilian Jiu Jitsu, there's a notion of being "good at the basics", as opposed to "being good by knowing advanced techniques". How you want to be good is a question of personal preference, but the idea is that these are the two ways, and most people do one or th...
c41bdfb3-17ef-40da-8b18-f08fd4f8537c
trentmkelly/LessWrong-43k
LessWrong
Conceptual Analysis for AI Alignment TL; DR - Conceptual Analysis is highly relevant for AI alignment, and is also a way in which someone with less technical skills can contribute to alignment research. This suggests there should be at least one person working full-time on reviewing existing philosophy literature for...
c7051a74-9b4e-4316-b406-216ae8f5aec0
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
Evidence to prioritize or working on AI as the most impactful thing? Consider me a non-technical, AI-ignorant EA who has specialized in other areas. Consider that I want to do the most good, so I hear about AI and how all the funding and the community is now directed towards AI and how it is the most impactful thing. ...
83f32e45-123c-45bb-98d3-da76cc712d03
StampyAI/alignment-research-dataset/blogs
Blogs
A Collection Of Compasses A Collection Of Compasses ------------------------- A collection of interesting compasses I have found on the internet, or made myself. * [the classic Political Compass](https://www.politicalcompass.org/analysis2) * [Conflict In Literature](http://www.incidentalcomics.com/2014/05/conflict...
652b7ffa-62cb-42c6-ab43-d2de1b5da65e
trentmkelly/LessWrong-43k
LessWrong
Is there a better choice than "Stop After Win" (or does double-checking lead to failure loops)? Crossposted. Here we go. To recap: Last week I wrote about what happened when I applied metrics to my piano practice, in which I outlined the “secret to learning” in eight steps: 1. Define win condition. 2. Define act...
e9f1f3e4-1a97-4406-9084-f31904a3dc4e
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Contra Hofstadter on GPT-3 Nonsense [Douglas Hofstadter wrote in the Economist recently](https://www.economist.com/by-invitation/2022/06/09/artificial-neural-networks-today-are-not-conscious-according-to-douglas-hofstadter) on whether modern large language models are conscious. As part of this, he and his colleague Da...
70f91dd4-842b-4c45-9620-0ba8ee8c4c7b
trentmkelly/LessWrong-43k
LessWrong
Human factors research seems very relevant to rationality "The science of “human factors” now permeates the aviation industry. It includes a sophisticated understanding of the kinds of mistakes that even experts make under stress. So when Martin Bromiley read the Harmer report, an incomprehensible event suddenly made ...
4d281e2a-0d7d-42ce-b41a-ecc2d29db1d6
LDJnr/LessWrong-Amplify-Instruct
LessWrong
"LW doesn't seem to have a discussion of the article Epiphany Addiction, by Chris at succeedsocially. First paragraph: "Epiphany Addiction" is an informal little term I came up with to describe a process that I've observed happen to people who try to work on their personal issues. How it works is that someone will be t...
587a7ba3-600f-49dd-a6f2-effcfaefec29
trentmkelly/LessWrong-43k
LessWrong
Obstacles to gradient hacking [Epistemic status: brainstorming, still somewhat confused, with a bunch of loose threads left.] This post is essentially the summary of a long discussion on the EleutherAI discord about trying to exhibit gradient hacking in real models by hand crafting an example. The discussion was spar...
332963a1-0fe3-470c-9597-164a5da103f6
StampyAI/alignment-research-dataset/special_docs
Other
Truth, Lies, and Automation Truth, Lies, and Automation How L anguage Models C ould Ch ange Disinformation x MAY 2021 AUTHORS Ben Buchanan Andrew LohnMicah Musser Katerina Sedova Center for Security and Emerging Technology 2Established in January 2019, the Center for Security and Emerging Technology (CSET) at G...
4144156d-549f-4a52-87ff-e453e09df1a7
StampyAI/alignment-research-dataset/arbital
Arbital
Mathematical object **[https://arbital.com/p/5xs](https://arbital.com/p/5xs)**
bb0ad4f5-20c1-4449-8470-be98e59424b9
trentmkelly/LessWrong-43k
LessWrong
Sequence opener: Jordan Harbinger's 6 minute networking When I dipped my toes into becoming a coach two years ago, I asked around for resources on learning how to network - but not in the sleazy salespeople way, but in a way that fosters healthy relationships and mutual respect with the people around me. More than on...
127496bc-2959-4df1-9c84-56dda5a31396
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Pascal's Muggle: Infinitesimal Priors and Strong Evidence **Followup to:**  [Pascal's Mugging: Tiny Probabilities of Vast Utilities](/lw/kd/pascals_mugging_tiny_probabilities_of_vast/), [The Pascal's Wager Fallacy Fallacy](/lw/z0/the_pascals_wager_fallacy_fallacy/), [Being Half-Rational About Pascal's Wager Is Even Wo...
75e65143-fee6-40f9-91e4-63a99e7feea8
trentmkelly/LessWrong-43k
LessWrong
[Link] Nick Bostrom on the Status Quo Bias Here (in MP3 format) is the Philosophy Bites podcast from 05/14/2012. In it, Nick Bostrom discusses the status quo bias with Nigel Warburton. Here (in PDF format) is Bostrom's paper outlining a technique (the reversal test) for debiasing within the context of applied ethics.
ab3ecb1d-e075-4f32-bd09-416f1086ee97
trentmkelly/LessWrong-43k
LessWrong
Multi-winner Voting: a question of Alignment This is my third (and for now, last) essay about voting theory for rationalists. In the first two, I focused primarily on single-winner voting theory; that is, methods for aggregating group preferences into a final verdict on some choice. Ideally, single-winner methods woul...
1bf61b03-9f35-421a-83df-84dba44c7a39
StampyAI/alignment-research-dataset/arxiv
Arxiv
Unsupervised State Representation Learning in Atari 1 Introduction --------------- The ability to perceive and represent visual sensory data into useful and concise descriptions is considered a fundamental cognitive capability in humans [[1](#bib.bib1), [2](#bib.bib2)], and thus crucial for building intelligent age...
90a0967c-c8bf-40b3-8163-dad89ab7462c
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
My Thoughts on the ML Safety Course *This summary was written as part of*[*Refine*](https://www.alignmentforum.org/posts/5uiQkyKdejX3aEHLM/how-to-diversify-conceptual-alignment-the-model-behind)*. The*[*ML Safety Course*](https://course.mlsafety.org/) *is created by*[*Dan Hendrycks*](https://danhendrycks.com/) *at the...
151beeb9-70f4-4c25-b66c-84fd23f8aff3
StampyAI/alignment-research-dataset/special_docs
Other
The State of Research in Existential Risk GIRS -2017 -04 FIRST INTERNATIONAL COLLOQUIUM ON CATASTROPHIC AND EXI STENTIAL RISK PROCEEDINGS Editor: Dr. B. John Garrick Sponsored by: The B. John Garrick Institute for the Risk Sciences Henry Samueli School of Engineering and Applied Science University of Califor...
e032d949-bcf4-4c23-b3ca-76967c34591b
trentmkelly/LessWrong-43k
LessWrong
We are the Athenians, not the Spartans The Peloponnesian War was a war between two empires: the seadwelling Athenians, and the landlubber Spartans. Spartans were devoted to duty and country, living in barracks and drinking the black broth. From birth they trained to be the caste dictators of a slaveowning society, whi...
a54dcd02-1cda-41ee-8991-a24ce94af68e
trentmkelly/LessWrong-43k
LessWrong
Best arguments against the outside view that AGI won't be a huge deal, thus we survive. Specifically, against the following view described by a comment: > There seems to be a lack of emphasis in this market on outcomes where alignment is not solved, yet humanity turns out fine anyway. Based on an Outside View perspec...
092344de-776c-4e43-90f6-4b402597bb1c
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
A Manifold Market "Leaked" the AI Extinction Statement and CAIS Wanted it Deleted History may look back on CAIS’s AI risk statement as a pivotal moment IF we survive AI. But, what if I told you that AI researchers “leaked” info about the statement on Manifold Markets over a week before it was published? This justifia...
9157cc8b-29e4-4d13-b264-ee8e85f521eb
trentmkelly/LessWrong-43k
LessWrong
Why it's necessary to shoot yourself in the foot Without shooting yourself in the foot, learning lacks motivation. Complexity without reason is really confusing. ---------------------------------------- At the Recurse Center Feb 2020: I watch a talk during presentations about someone optimizing a database engine. It...
a470b386-5d88-4b22-a8b7-95403f3c4732
trentmkelly/LessWrong-43k
LessWrong
«Boundaries/Membranes» and AI safety compilation In this post I outline every post I could find that meaningfully connects the concept of «Boundaries/Membranes» (tag) with AI safety.[1] This seems to be a booming subtopic: interest has picked up substantially within the past year.  Update (2023 Dec): we're now runnin...
d365fa6e-d235-410e-9ab8-256fd0ff585a
trentmkelly/LessWrong-43k
LessWrong
Meetup : Moscow: Non-Omniscience, On rationalist communication, ACH training, games Discussion article for the meetup : Moscow: Non-Omniscience, On rationalist communication, ACH training, games WHEN: 07 August 2016 02:00:00PM (+0300) WHERE: Москва, Большая Дорогомиловская, д.5к2 Note: most our members join meetups...
840b02e1-e3e6-457f-ab78-47b6f19ccc4c
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Why not use active SETI to prevent AI Doom? Let's assume that Eliezer is right: soon we'll have an AGI that is very likely to kill us all. (personally, I think Eliezer is right). There are several ways to reduce the risk, in particular: speeding up alignment research and slowing down capabilities research, by various...
94c0f633-c8aa-498b-b8b8-3a357a2b80e1
trentmkelly/LessWrong-43k
LessWrong
Who are you? There are two things that people debate with regards to continuation of personhood. One is whether edge cases to our intuitions of what ‘me’ refers to are really me. For instance if a simulation of me is run on a computer, is it me? If it is definitely conscious? What if the fleshy bloody one is still ali...
5f02fa05-555a-40b7-814e-04d3d8410742
trentmkelly/LessWrong-43k
LessWrong
How to Dissolve It ,,, In the last month and a half, I've had more (of what I believe to be) profound, creative insights to technical problems than in the five years prior. For example, I independently came up with the core insight behind DenseNets during the second lecture on convolutional neural nets in my Deep Lear...
b207ce23-957b-4a0e-b4c1-a95ab75d5f5a
trentmkelly/LessWrong-43k
LessWrong
Alewife Train is Now Arriving A few months ago I complained that the text version of the MBTA train arrival announcements include a not-needed-for-text "attention passengers" which delayed the key information to the second screen. I'm happy to report that they've fixed it! While the vocal announcement is still "...
8a60a8fa-de2e-4072-b48a-152ea12d9925
StampyAI/alignment-research-dataset/lesswrong
LessWrong
AI Risk Intro 1: Advanced AI Might Be Very Bad *This is a submission for the* [*AI Safety Public Materials contest.*](https://www.lesswrong.com/posts/gWM8cgZgZ9GQAYTqF/usd20k-in-bounties-for-ai-safety-public-materials#How_to_submit) **Introduction** ================ If human civilisation is destroyed this century, t...