id
stringlengths
36
36
source
stringclasses
15 values
formatted_source
stringclasses
13 values
text
stringlengths
2
7.55M
64227937-58f1-4eca-a18e-8a911f00730a
trentmkelly/LessWrong-43k
LessWrong
Counterfactuals, thick and thin Summary: There's a "thin" concept of counterfactual that's easy to formalize and a "thick" concept that's harder to formalize. Suppose you're trying to guess the outcome of a coinflip. You guess heads, and the coin lands tails. Now you can ask how the coin would have landed if you had ...
b36bef2e-5f04-4461-bf4a-39b3288a8da4
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Learning Russian Roulette *Surprisingly, our current theories of anthropics don't seem to cover this.* You have a revolver with six chambers. One of them has a bullet in it. You are offered $1 for spinning the barrel, pointing at your head, and pulling the trigger. You remember doing this many times, and surviving ea...
cb9bd401-92a1-4af7-9bf1-1981a1687203
trentmkelly/LessWrong-43k
LessWrong
Sunday June 21st – talks by Abram Demski, alkjash, orthonormal, eukaryote, Vaniver This Sunday at 11am (PDT), we're running a session of four "lightning talks" by curated LessWrong authors. * Each talk will be 3-5 minutes followed by discussion. Afterwards, we'll have a hangout in breakout rooms. The talks will be s...
ed790bef-e64f-4d71-90e8-6e07624a7446
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
my current outlook on AI risk mitigation *(thanks to [Linda Linsefors](https://www.lesswrong.com/users/linda-linsefors), [Artaxerxes](https://www.lesswrong.com/users/artaxerxes), and [jono](https://www.lesswrong.com/users/lw-user0246) for their feedback on drafts of this post)* as part of [the *refine* programme](ht...
6512aa3e-50f8-43e7-abea-61f46fa01ec6
trentmkelly/LessWrong-43k
LessWrong
Be Logically Informative What's the googolplexth decimal of pi? I don't know, but I know that it's rational for me to give each possible digit P=1/10. So there's a sense in which I can rationally assign probabilities to mathematical facts or computation outcomes on which I'm uncertain. (Apparently this can be modeled ...
a91cf763-54af-4386-a32a-b6a176c55964
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Safety-capabilities tradeoff dials are inevitable in AGI A safety-capabilities tradeoff is when you have something like a dial on your AGI, and one end of the dial says “more safe but less capable”, and the other end of the dial says “less safe but more capable”. Make no mistake: **safety-capabilities tradeoff dials*...
baf6ebcf-977c-4382-9329-c1539e5ca532
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Which Parts Are "Me"? Today's post, Which Parts Are "Me"? was originally published on 22 October 2008. A summary (taken from the LW wiki):   > Everything you are, is inside your brain. But not everything inside your brain is you. You can draw mental separation lines, which can make you more reflective. ...
3215349a-597a-4a06-8cd6-ab6c855ca0a9
trentmkelly/LessWrong-43k
LessWrong
Vim Vim is a terminal-based editor optimized for speed. Keys are hotkeys by default. Vim does not even use Ctrl for most of them. Maneuvering via hotkeys is so efficient mouse input is often disabled by default. This guide focuses on the features I use most frequently. It has nothing about with native Vim windows bec...
fdcddb55-96e9-477f-87b8-a2b7f877140b
trentmkelly/LessWrong-43k
LessWrong
Open thread, Nov. 21 - Nov. 27 - 2016 If it's worth saying, but not worth its own post, then it goes here. ---------------------------------------- Notes for future OT posters: 1. Please add the 'open_thread' tag. 2. Check if there is an active Open Thread before posting a new one. (Immediately before; refresh the...
4f773680-fdcc-4d7d-97d3-147cd2256393
trentmkelly/LessWrong-43k
LessWrong
Cult impressions of Less Wrong/Singularity Institute If you type "less wrong c" or "singularity institute c" into Google, you'll find that people are searching for "less wrong cult" and "singularity institute cult" with some frequency. (EDIT: Please avoid testing this out, so Google doesn't autocomplete your search an...
fd2c0064-8a51-4baf-b464-f77a5a4237fd
StampyAI/alignment-research-dataset/arxiv
Arxiv
Leave no Trace: Learning to Reset for Safe and Autonomous Reinforcement Learning 1 Introduction --------------- Deep reinforcement learning (RL) algorithms have the potential to automate acquisition of complex behaviors in a variety of real-world settings. Recent results have shown success on games (Mnih et al. ([2...
5edef618-32d0-4ec0-99ad-7a95620caaf8
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Effortless Technique Today's post, Effortless Technique was originally published on 23 December 2007. A summary (taken from the LW wiki):   > Things like the amount of effort put into a project, or the number of lines in a computer program, are positive things to maximize. But this is silly. Surely it is...
1cdcacf4-c3da-4ef2-8775-e7f3544b99da
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
What failure looks like The stereotyped image of AI catastrophe is a powerful, malicious AI system that takes its creators by surprise and quickly achieves a decisive advantage over the rest of humanity. I think this is probably not what failure will look like, and I want to try to paint a more realistic picture. I’l...
32520f86-415d-4ffa-81b9-73f1fa8dd36a
trentmkelly/LessWrong-43k
LessWrong
Crises Don't Need Your Software About a month ago, I was invited to contribute to a group looking to help Ukrainian refugees in Poland. The group consisted of volunteers, including some people from the rationalist community, who knew they wanted to help, and were searching for high impact ways to do so, likely through...
c7bd7950-d0fd-4490-8fe8-5d6090cfa542
StampyAI/alignment-research-dataset/arxiv
Arxiv
LCA: Loss Change Allocation for Neural Network Training 1 Introduction --------------- In the common stochastic gradient descent (SGD) training setup, a parameterized model is iteratively updated using gradients computed from mini-batches of data chosen from some training set. Unfortunately, our view into the high-...
6f5453d6-a40c-4529-8a23-f4ec4d643908
trentmkelly/LessWrong-43k
LessWrong
Akrasic Reasoning This post is in a constant state of revision, similar to this post. This is mainly because I do not have a beta and this is based on many personal experiences that are unclear at times.   This subject has been touched on many times throughout LessWrong because Akrasia is the most dangerous foe of a...
b6b9a911-de5a-4fb9-8a1c-6f21d17b8163
trentmkelly/LessWrong-43k
LessWrong
The ground of optimization This work was supported by OAK, a monastic community in the Berkeley hills. This document could not have been written without the daily love of living in this beautiful community. The work involved in writing this cannot be separated from the sitting, chanting, cooking, cleaning, crying, cor...
70d988da-0a2a-4b0b-9766-ca331ee9e6e1
trentmkelly/LessWrong-43k
LessWrong
Home Economics As I inventoried my personal library and classified it using the Universal Decimal System, I found out about lots of fields of knowledge I was only dimly aware of or didn't even know existed1, and one of those that piqued my curiosity was home economics, domestic science and housekeeping (field 64). I w...
a2e99a2b-1e4f-4468-a607-f4bf9ac68380
trentmkelly/LessWrong-43k
LessWrong
The Problem with Giving Advice Epistemic Status: I'm highly confident this is a phenomenon that occurs with a lot of advice people give, but I'm quite uncertain about the best way to deal with it when trying to give advice to more than one person.   The main thing people fail to consider when giving advice is that s...
1a31e86e-907a-48bb-afee-fe62c7d347f6
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
AI Timelines Introduction ============ How many years will pass before transformative AI is built? Three people who have thought about this question a lot are Ajeya Cotra from [Open Philanthropy](https://www.openphilanthropy.org/), Daniel Kokotajlo from [OpenAI](https://openai.com/) and Ege Erdil from [Epoch](https:/...
e13bd59f-1cbe-4943-a75c-f8a3d1c5546d
trentmkelly/LessWrong-43k
LessWrong
The Defender’s Advantage of Interpretability TL;DR: In this post, I want to argue why Interpretability & Transparency tools have a defender’s advantage if they are used correctly, i.e. they improve alignment much more than new capabilities and therefore mitigate risks of dual use. I draw parallels from biosecurity res...
2b8dd3e3-d35e-43ca-9d50-f7a37bdb5c9c
trentmkelly/LessWrong-43k
LessWrong
Does the existence of shared human values imply alignment is "easy"? Observation: Humans seem to actually care about each other. How was culture + evolution able to achieve this? While not to 100% (Maybe 50-95%), I would trust most humans to not push a big red button that kills all humans, even if it had some large b...
b2d20fb6-50a1-4ae7-9f2e-92c30181e976
trentmkelly/LessWrong-43k
LessWrong
Meetup report: How harmful is cannabis, and will you change your habits?   A week ago the meetup group in Berkeley discussed a new article in PNAS titled "Persistent cannabis users show neuropsychological decline from childhood to midlife". Several people who didn't attend said they were interested in how that conver...
a1b654e0-c56f-4825-bf44-60a0aa65b3dc
trentmkelly/LessWrong-43k
LessWrong
Verification and Transparency Epistemic status: I’ve thought about this topic in general for a while, and recently spent half an hour thinking about it in a somewhat focussed way. Verification and transparency are two kinds of things you can do to or with a software system. Verification is where you use a program to ...
5df7b61a-266b-43d5-8e07-4ed838e7ec38
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
200 COP in MI: Looking for Circuits in the Wild *This is the third post in a sequence called 200 Concrete Open Problems in Mechanistic Interpretability.*[*Start here*](https://www.alignmentforum.org/posts/LbrPTJ4fmABEdEnLf/200-concrete-open-problems-in-mechanistic-interpretability)*, then read in any order. If you wan...
a3529419-4820-42d8-b1da-46edfffe23db
trentmkelly/LessWrong-43k
LessWrong
What does it take to defend the world against out-of-control AGIs? Intended audience: People very familiar with AGI safety / alignment discourse. Lots of jargon, lots of unspoken & unjustified background assumptions. Confidence level: What I currently believe and why. I mainly work on technical alignment and don’t co...
1c8f1654-28ee-4ebe-b2b0-23ef89d93d6a
trentmkelly/LessWrong-43k
LessWrong
Postmortem to Petrov Day, 2020 The first underwater test of an atom bomb, in the US at Bikini Atoll in the Pacific, July 24, 1946. Also, the LessWrong.com Frontpage this Petrov Day. We failed. The Frontpage was taken down for 24 hours. Launch Attempts At 1:26 am, ~275 users got an email (and 20 minutes later a PM) ...
b7dd3438-3057-4db0-acf3-d1000d03cb8c
trentmkelly/LessWrong-43k
LessWrong
Freaky unfairness In Freaky Fairness the author discusses the problem of multi-player game theory problems where all players have access to each-other's source code. He gives an algorithm called Freaky Fairness, and makes the claim that 'all players run Freaky Fairness' is both a  Strong Nash Equilibrium and a Pareto ...
c94f69ab-64ad-4535-bbb1-dce33b9582f1
trentmkelly/LessWrong-43k
LessWrong
How "grifty" is the Foresight Institute? Are they making button soup? rant / curious / fearful / would love information to establish chain of trust. Recently received an email from the Foresight Institue. Sent me an application. Quickly accepted for an interview with their CEO. Didn't hear anything particularly valua...
24413e48-660a-45b6-a8e3-ce816ef9aa92
trentmkelly/LessWrong-43k
LessWrong
Paradox and pedagogy When water freezes, it expands. Thus, my latest invention places a weight on a column of water, then freezes the water to lift the weight, doing work. By using a weight heavy enough that the work done is greater than the energy expenditure to freeze the water, I create a perpetual motion machine. ...
ff689323-a9f5-464a-b6df-96068f029f90
trentmkelly/LessWrong-43k
LessWrong
Content writing offer We're looking to commission something like 10-20 articles on transfection for a website we're developing, whatistransfection.com. I'm interested in hearing from lesswrongers in biotech or who are grad students in the field looking for some freelance writing work. You can email me at drethelin@gma...
dc067539-0838-4bc9-b9b2-3d99a77d9ca0
trentmkelly/LessWrong-43k
LessWrong
Sharing a Car Last week I posted about how we were thinking of getting a car, but at least for now we've decided instead to share a car with one of our housemates. They have a Honda Fit, and while it is important to them to have a car, they don't use it very often. We're planning on dividing costs in proportion to mi...
ae815e9c-c849-4d5b-92c6-42ca372f6bd1
trentmkelly/LessWrong-43k
LessWrong
July 2017 Media Thread This is the monthly thread for posting media of various types that you've found that you enjoy. Post what you're reading, listening to, watching, and your opinion of it. Post recommendations to blogs. Post whatever media you feel like discussing! To see previous recommendations, check out the ol...
ceaea0e3-2c15-43ef-aa92-f4fe3358a55d
trentmkelly/LessWrong-43k
LessWrong
I like simplicity, but not THAT much Followup to: L-zombies! (L-zombies?) Reply to: Coscott's Preferences without Existence; Paul Christiano's comment on my l-zombies post In my previous post, I introduced the idea of an "l-zombie", or logical philosophical zombie: A Turing machine that would simulate a conscious hum...
1624657a-e85d-4ba5-9a26-1ce57d6e02fd
trentmkelly/LessWrong-43k
LessWrong
Meetup : Cambridge, MA First Sunday Meetup Discussion article for the meetup : Cambridge, MA First Sunday Meetup WHEN: 06 May 2012 02:00:00PM (-0400) WHERE: 25 Ames St, Cambridge MA Cambridge/Boston-area Less Wrong meetups on the first and third Sunday of every month, 2pm at the MIT Landau Building [25 Ames St, Bl...
c3a19825-6560-4910-8e76-a0a105f4ee87
trentmkelly/LessWrong-43k
LessWrong
Chinese scientists acknowledge xrisk & call for international regulatory body [Linkpost] Some highlights from the article (bolding added): Several Chinese academic attendees of the summit at Bletchley Park, England, which starts on Wednesday, have signed on to a statement that warns that advanced AI will pose an “ex...
6a012cae-7986-40d2-9738-6f5fb0860ccc
trentmkelly/LessWrong-43k
LessWrong
Request for feedback on a paper about (machine) ethics I have written a paper on ethics with special concentration on machine ethics and formality with the following abstract: Most ethical systems are formulated in a very intuitive, imprecise manner. Therefore, they cannot be studied mathematically. In particular, th...
efb8f7e2-b0fa-494f-94f4-865d59d5235d
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Call for Cruxes by Rhyme, a Longtermist History Consultancy *(I'm looking to research historical examples with implications for AI Risk specifically. If you want think this could be useful to you, consider filling out this* [*form*]( https://docs.google.com/forms/d/e/1FAIpQLSf52N_Mdq5ThmPiLyOEAgQCCAyma39vMGCZs3VVri724...
3bb606a9-4d03-40cd-b443-a3e112048047
trentmkelly/LessWrong-43k
LessWrong
Illness anxiety disorder: how to become more rational? Disclaimer: When I think about dying I mostly think about cancer (I don't know much about other diseases). In the following, I'll be refering to "cancer" to point at deadly diseases where early detection is crucial. [Epistemic status: thinking out loud; trying to...
2e10332f-172a-4c34-898f-a53190a4db4e
trentmkelly/LessWrong-43k
LessWrong
Proposal: Align Systems Earlier In Training Unfortunately, if AGI arrives in anything like the current industrial and political landscape, then it is all but inevitable that whoever builds it will not be able to hold onto it for long. Between leaks and espionage, it will not be long before the model is public and anyo...
c67a5d05-2438-4595-bd43-1c54bc557181
StampyAI/alignment-research-dataset/arbital
Arbital
Random utility function A 'random utility function' is a utility function selected according to some simple probability measure over a logical space of formal, compact specifications of utility functions. For example: suppose utility functions are specified by computer programs (e.g. a program that maps an output des...
bf87ce21-895c-49b3-bd52-72acd56a8019
trentmkelly/LessWrong-43k
LessWrong
Consider Taking Zinc Every Time You Travel Edit (November 2022): comments raised some probably valid concerns about my sources. I also don't currently believe it's possible to buy the kind of zinc lozenges studied; they're too unpleasant, what you buy in a drug store has been changed for palatability. Zinc lozenges ar...
6b57d894-64d3-40f9-b276-faae14d0e3c0
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Why all the fuss about recursive self-improvement? *This article was outlined by Nate Soares, inflated by Rob Bensinger, and then edited by Nate. Content warning: the tone of this post feels defensive to me. I don't generally enjoy writing in "defensive" mode, but I've had this argument thrice recently in surprising p...
3661d151-0a4d-4afa-8fe7-535d8fe9ca15
StampyAI/alignment-research-dataset/lesswrong
LessWrong
Why Do People Think Humans Are Stupid? Introduction ============ On more than one occasion, I've seen the following comparisons used to describe how a superintelligence might relate to/perceive humans: * Humans to ants * Humans to earthworms * And similar More generally, people seem to believe that humans are *incr...
cd501fd1-0160-47be-a88a-11d2d3c09d8d
StampyAI/alignment-research-dataset/arxiv
Arxiv
Discovery of Options via Meta-Learned Subgoals 1 Introduction --------------- Reinforcement learning (RL) problems involve learning about temporally-extended actions at multiple time scales. In RL, the *options* framework (Sutton et al., [1999](#bib.bib26 "Between mdps and semi-mdps: a framework for temporal abstra...
fe319f49-29e1-45f2-8640-00db97b7aaff
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
[Link Post: New York Times] White House Unveils Initiatives to Reduce Risks of A.I. *This is a linkpost for* [*https://www.nytimes.com/2023/05/04/technology/us-ai-research-regulation.html*](https://www.nytimes.com/2023/05/04/technology/us-ai-research-regulation.html) > The White House on Thursday announced its first...
72a267d3-44be-4955-b460-28de53bda138
trentmkelly/LessWrong-43k
LessWrong
Pascal's mugging in reward learning This is the last post on the current value/reward learning series. I'll just bring up the danger of Pascal's mugging in reward learning. I've already shown how you can use the (p, R) model (with R a reward and p a planning algorithm, called a planner) to model AI actions "over...
6bb5f29f-f666-4b94-bf17-66cadacef88b
trentmkelly/LessWrong-43k
LessWrong
Weekly LW Meetups This summary was posted to LW main on February 21st. The following week's summary is here. Irregularly scheduled Less Wrong meetups are taking place in: * Atlanta Lesswrong Meetup: How to Increase Your Brainpower: 23 February 2014 07:00PM * Bratislava Meetup X.: 24 February 2014 06:00PM * [Canbe...
401aa9f9-10e8-4810-8966-33b90c6c8527
trentmkelly/LessWrong-43k
LessWrong
Reply to Eliezer on Biological Anchors The "biological anchors" method for forecasting transformative AI is the biggest non-trust-based input into my thinking about likely timelines for transformative AI. While I'm sympathetic to parts of Eliezer Yudkowsky's recent post on it, I overall disagree with the post, and thi...
0b15a29a-2e02-4766-94b0-74f8c8aca180
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Project Intro: Selection Theorems for Modularity Introduction - what is modularity, and why should we care? ========================================================== It’s a well-established meme that evolution is a blind idiotic process, that has often resulted in design choices that no sane systems designer would ...
1928f861-da74-4f88-b82e-a570e88b1b36
trentmkelly/LessWrong-43k
LessWrong
Introspective Bayes An attempt to demonstrate a limiting condition of an optimal Bayesian agent, or a probabilistic agent in general. This analysis relies on the description of such an optimal Bayesian agent (hereafter OBA), in Bostrom 2014, Boxes one and ten, and associated endnotes, with supplementary research on K...
3f86cf45-81db-4b86-bf49-35132b416d0a
trentmkelly/LessWrong-43k
LessWrong
Eliezer's Lost Alignment Articles / The Arbital Sequence Note: this is a static copy of this wiki page. We are also publishing it as a post to ensure visibility. Circa 2015-2017, a lot of high quality content was written on Arbital by Eliezer Yudkowsky, Nate Soares, Paul Christiano, and others. Perhaps because the pl...
d2558ebb-c963-4bd9-9148-f3b8bd0e83ed
trentmkelly/LessWrong-43k
LessWrong
Believing in magic pyramids shows that you think differently Believing in magic pyramids shows that you think differently https://www.tomdekan.com/think-differently Last week I moved to Bonn. My new landlord, Jenno, had left lots of his stuff in the house. Because of this, I spent my first night sleeping next to a la...
4e579b4a-2220-4c3c-ac9d-481ed8c24d1d
StampyAI/alignment-research-dataset/blogs
Blogs
The gloves are off, the pants are on *A lie gets halfway around the world before the truth has a chance to get its pants on. - [Winston Churchill](https://www.brainyquote.com/quotes/winston_churchill_103564)* *(...[Not really, though](https://professorbuzzkill.com/twain-lie-travels/))* I collect links that are...
f78897de-1f53-4c63-915b-fa7f77530820
trentmkelly/LessWrong-43k
LessWrong
Rolf Nelson's "The Rational Entrepreneur" New blog up by longtime OB/LWer Rolf Nelson, "The Rational Entrepreneur" at rolfnelson.com.  On "the overlap between entrepreneurship and the modern tools of rationality".  Rolf will post daily through November, and after that it will depend on how much traction the blog gets....
3a842894-3e5f-4b2c-8f03-7bcc50fee8b1
trentmkelly/LessWrong-43k
LessWrong
Open Thread With Experimental Feature: Reactions This open thread introduces an experimental extension of LessWrong's voting system: reactions. Unlike votes, reactions are public; hovering over the reactions will show a list of users who reacted. For now, this feature is only for comments on this post in particular; a...
0b9e8833-73a3-42e6-abab-25d45f271e42
trentmkelly/LessWrong-43k
LessWrong
Covid 3/12: New CDC Guidelines Available This post is a day late although not all that short, as I warned it might be. This is because I have spent the last week visiting the best place on Earth, my true home, which is New York City. I will return again soon, and soon after that I will once again be able to live there...
18a82bef-7179-448b-96e0-e91e1ca05be7
trentmkelly/LessWrong-43k
LessWrong
Beautiful ugly websites Early in my career, I heard the famous saying: Craigslist proves that function beats form. The approach of focusing on a product's functions, rather than beautifying it has been *then* very pragmatic to me. I’ve always strived to make my work pretty. For years, function over form had seemed to...
4b040056-5376-4a51-8c67-e47a70161cbb
trentmkelly/LessWrong-43k
LessWrong
Meetup : September Rationality Dojo - Fixed and Growth Mindset Discussion article for the meetup : September Rationality Dojo - Fixed and Growth Mindset WHEN: 07 September 2014 03:30:00PM (+1000) WHERE: Ross House Association, 247-251 Flinders Lane, Melbourne [ATTN: Please remember the new location for the dojos: t...
97ab233b-9a46-4166-bd86-5c5e88c9a8df
trentmkelly/LessWrong-43k
LessWrong
The Roots of Progress is now a nonprofit organization I’m excited to announce that The Roots of Progress is now a nonprofit organization. The Roots of Progress started in 2017 as a side project, not much more than a blog. After Tyler Cowen and Patrick Collison published their call for “progress studies”, and the prog...
ad83ced8-5ec6-4d1c-a81e-4a9254078ac5
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Bounded complexity of solving ELK and its implications This post was written for the SERI MATS program. I thank Evan Hubinger and Leo Gao for their mentorship in the program. Further thanks go to Simon Marshall and Leo Gao (again) for specific comments regarding the content of this post. The Eliciting Latent Knowledg...
0f125a72-609b-4898-83fb-9c181152d92c
trentmkelly/LessWrong-43k
LessWrong
Positive Bias Test (C++ program) I've written a program which tests positive bias using Wason's procedure from "On the failure to eliminate hypotheses in a conceptual task" (Quarterly Journal of Experimental Psychology, 12: 129-140, 1960). If the user does not discover the correct rule, the program attempts to guess, ...
afe4c255-75b9-49ae-a6c8-2be383ec91ce
trentmkelly/LessWrong-43k
LessWrong
Trying to find the underlying structure of computational systems This post was written in the SERI MATS program under John Wentworth. In our quest to find steering mechanisms in computational systems, we first have to find the right framework to look at the internals of such a system. This framework should naturally ...
38a57ac6-76f3-4208-be7f-ae6a7a8c3352
trentmkelly/LessWrong-43k
LessWrong
Leveling IRL I just got this random idea that people who want to become better at life could benefit from a common scale of "leveling". No, I don't mean vague Lesswrongey things like "changing your mind". I mean a set of concrete criteria like "you qualify for level 2 if you can do 5 pull-ups, have solved 30 Project E...
e0d60ab4-6154-4d16-a719-11d47e12fc6e
trentmkelly/LessWrong-43k
LessWrong
Human consciousness as a tractable scientific problem I encounter many intelligent people (not usually LWers, though) who say that despite our recent scientific advances, human consciousness remains a mystery and currently intractable to science. This is wrong. Empirically distinguishable theories of consciousness hav...
ee39e74a-aae8-4287-876e-2323620bb942
trentmkelly/LessWrong-43k
LessWrong
The default outcome for aligned AGI still looks pretty bad Most of the AI-related discussions I've read from the LW/EA community have rightly focused on the challenges and strategies for aligning AI with the intentions of its creator. After all, if that problem remains unsolved, humanity will almost certainly go extin...
7ada42fa-acee-4f2c-9c96-2fed72a6524a
trentmkelly/LessWrong-43k
LessWrong
Ask for Feelings not Tunes In putting on a contra dance, it's standard for the caller to pick the dances and the band to pick the tunes. Since it's much more satisfying when the tune fits the dance, the band and caller confer briefly between each dance. For example, the caller might say they want something "smooth and...
c73f17ad-1484-4921-b836-3dff04a05ffc
trentmkelly/LessWrong-43k
LessWrong
Philosophy that can be "taken seriously by computer scientists" I've long held CMU's philosophy department in high regard. One of their leading lights, Clark Glymour, recently published a short manifesto, which Brian Leiter summed up as saying that "the measure of value for philosophy departments is whether they are t...
b52de458-562c-431e-b23b-2815a5a8f056
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
ARC paper: Formalizing the presumption of independence *(I did not have anything to do with this paper and these are just my own takes.)* The Alignment Research Center recently published their second report, [Formalizing the presumption of independence](https://arxiv.org/abs/2211.06738). While it's not explicitly abo...
3f105e57-b529-42b5-baab-407b1899238e
trentmkelly/LessWrong-43k
LessWrong
Lafayette: traffic vessels This week I’m in Lafayette, a town merely twenty-three minutes further from my San Franciscan office than my usual San Franciscan home, thanks to light rail. There are deer in the street and woods on the walk from the train to town. On this occasion at least, Lafayette doesn’t feel properly...
26e3ba9e-1066-4019-b1db-68165efd5739
trentmkelly/LessWrong-43k
LessWrong
Argument by lexical overloading, or, Don't cut your wants with shoulds I used the word "cut" in the title to mean the Prolog operator "cut", an operator which halts the evaluation of a statement in predicate logic. Fiction writers often complain, "I keep procrastinating from writing," and, "Nobody reads what I write....
8e278485-74bb-4b9f-9bdf-fc4b5003aaf0
trentmkelly/LessWrong-43k
LessWrong
How To Change One's Mind A look at the 1957 film "12 Angry Men" from the perspective of communicative and debate strategies. 5 part series, about an hour in total.
318464ca-09fb-4f6a-8529-84d673635954
trentmkelly/LessWrong-43k
LessWrong
Exploring the Landscape of Scientific Minds (Let my People Go) (Originally posted at Secretum Secretorum) I.  Consider the following regions in the landscape of possible minds: (1)  All possible human minds (2)  All human minds that have ever existed (3)  All human minds that currently exist (4)  All currently ...
c27ff434-9dc9-4426-ae86-9becd989eb4d
trentmkelly/LessWrong-43k
LessWrong
Mike Darwin on Kurzweil, Techno-Optimism, and Delusional Stances on Cryonics In a comment on his skeptical post about Ray Kurzweil, he writes, > Unfortunately, [Kurzweil's] technological forecasting is naive, and I believe it will also prove erroneous (and in that, he is in excellent company). That would be of no con...
14ddf0b9-b1d7-4aca-bcb5-deb4f52d6436
trentmkelly/LessWrong-43k
LessWrong
Meetup : 02/12 London Meetup Discussion article for the meetup : 02/12 London Meetup WHEN: 02 December 2012 02:00:00PM (+0000) WHERE: Holborn, London A meetup in the Shakespeare's Head pub by Holborn tube station. Everyone is welcome. There was talk of making these regular every-other-week events. If you'd like to...
df8fa95b-2dcd-4ce7-822f-833618bab33e
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
The current alignment plan, and how we might improve it | EAG Bay Area 23 I gave a talk at EAG SF where I tried to describe my current favorite plan for how to align transformatively powerful AI, and how I thought that this plan informs current research prioritization. I think this talk does a reasonable job of repres...
b7468f6d-656b-427a-8e8a-565044571b6b
StampyAI/alignment-research-dataset/arxiv
Arxiv
Practical Reasoning with Norms for Autonomous Software Agents (Full Edition) Practical Reasoning with Norms for Autonomous Software Agents (Full Edition) Zohreh Shamsa,1,, Marina De Vosa, Julian Padgeta, Wamberto W. Vasconcelosb aDepartment of Computer Science, University of Bath, UK (BA2 6AH) bDepartment of Computin...
b05bdf86-b63c-4d2f-9232-7a4593033057
trentmkelly/LessWrong-43k
LessWrong
Morality as Parfitian-filtered Decision Theory? Non-political follow-up to: Ungrateful Hitchhikers (offsite)   Related to: Prices or Bindings?, The True Prisoner's Dilemma   Summary: Situations like the Parfit's Hitchhiker problem select for a certain kind of mind: specifically, one that recognizes that an action ...
58715d3b-0423-4298-b3c4-26194fe4d59c
trentmkelly/LessWrong-43k
LessWrong
Decision Theory in Space "Since you are reluctant to provide us with the location of the Rebel base," said Grand Moff Tarkin, "I have chosen to test this station's destructive power on your home planet of Alderaan." "No. Alderaan is peaceful. We have no weapons there. It is a loyal planet under Imperial control. Stri...
0c4472c1-d793-4490-855e-3a86efb13d36
trentmkelly/LessWrong-43k
LessWrong
Proposal for an AI Safety Prize After reading Yudkowsky’s list of lethalities, I searched for some kind of prize out there that might be similar to what he suggests in his point #40, to motivate truly impactful AI safety work: > “promising to pay big money retrospectively for good work to anyone who produces it.” I ...
7888ce16-f484-42cf-8dba-a88ea516fcf3
trentmkelly/LessWrong-43k
LessWrong
The Optimizer's Curse and How to Beat It The best laid schemes of mice and men Go often askew, And leave us nothing but grief and pain, For promised joy! - Robert Burns (translated)   Consider the following question: > A team of decision analysts has just presented the results of a complex analysis to the executiv...
f8efb522-91ce-423b-9db3-a71a8dab1479
trentmkelly/LessWrong-43k
LessWrong
Explicit and Implicit Communication I write an essay every Thursday. Every so often, one seems to really resonate with people. The piece I just wrote on the nature of explicit and implicit communication both got an enthusiastic reader response and seems directly relevant to a number of the projects and explorations p...
3ce702cf-0447-4af8-9a7c-95926f409b02
StampyAI/alignment-research-dataset/arxiv
Arxiv
Attention-Sensitive Alerting 305 Attention-Sensitive Alerting Eric Horvitz, Andy Jacobs, David Hovel Microsoft Research Redmond, Washington 98025 {horvitz,andyj ,davidhov }@microsoft.com Abstract We introduce utility-directed procedures for mediating the flow of potentially distract­ ing alerts and communicati...
934c4a33-5cb4-4375-aa70-e9e0bc124c00
trentmkelly/LessWrong-43k
LessWrong
AI companies aren't really using external evaluators From my new blog: AI Lab Watch. All posts will be crossposted to LessWrong. Subscribe on Substack. Many AI safety folks think that METR is close to the labs, with ongoing relationships that grant it access to models before they are deployed. This is incorrect. METR...
799bd9cc-48a0-436f-9cc7-41597d8a2ebe
trentmkelly/LessWrong-43k
LessWrong
Meetup : Five Worlds Collide - Vienna Discussion article for the meetup : Five Worlds Collide - Vienna WHEN: 04 December 2015 08:00:00PM (+0200) WHERE: Siebenbrunngengasse 44, 1050 Wien Here is the always updated version of the event: https://www.facebook.com/events/1505431786437253/ Welcome to Five Worlds Collide...
f2fa32e4-7721-4a35-90df-5c49283d87cb
trentmkelly/LessWrong-43k
LessWrong
Knightian Uncertainty: Bayesian Agents and the MMEU rule Recently, I found myself in a conversation with someone advocating the use of Knightian uncertainty. We both agreed that there's no point in singling out some of your uncertainty as "special" unless you treat it differently. I am under the impression that most ...
00630cc9-9cbc-429f-926b-958ad52aab16
trentmkelly/LessWrong-43k
LessWrong
ESR's New Take on Qualia http://esr.ibiblio.org/?p=1192#more-1192 ADDED:  Even if you disagree with ESR's take, and many will, this is the clearest definition I have seen on what qualia is.  So it should present a useful starting point, even for those who strongly disagree, to argue from.
b7d956aa-e5f3-4848-8086-dd756747930f
trentmkelly/LessWrong-43k
LessWrong
Optimal User-End Internet Security (Or, Rational Internet Browsing) Hacking and Cracking, Internet security, Cypherpunk. I find these topics fascinating as well as completely over my head. Yet, there are still some things that can be said to a layman, especially by the ever-poignant Russel Monroe: https://www.xkcd.c...
38f9afe6-ebb2-419f-9553-b7c121a9dacf
StampyAI/alignment-research-dataset/arxiv
Arxiv
Stakeholders in Explainable AI Introduction ------------ Explainability in artificial intelligence (AI) is not a new problem, nor was it ever considered a solved problem. The issue first came to prominence during the ‘knowledge engineering era’ of the late 1970s and early 1980s, when the focus was on building exper...
27ce2cc4-8a07-479a-9499-bc574933dc6f
trentmkelly/LessWrong-43k
LessWrong
Learning the prior Suppose that I have a dataset D of observed (x, y) pairs, and I’m interested in predicting the label y* for each point x* in some new set D*. Perhaps D is a set of forecasts from the last few years, and D* is a set of questions about the coming years that are important for planning. The classic dee...
3dbf7c00-d957-4cac-b351-c06561865f9b
trentmkelly/LessWrong-43k
LessWrong
The essay "Interstellar Communication Using Microbes: Implications for SETI" has implications for The Great Filter. Apparently I can add body text to a link post, instead of having to give a TL;DR as a comment? Excellent. Here are some excerpts from the essay: > Let us consider our own Sun as a case in point. At 1 AU...
2800d4f7-6fe5-47e0-a1de-2bfd914bffba
trentmkelly/LessWrong-43k
LessWrong
Meetup : Baltimore Area / UMBC: Intro to LW and future of the meetup Discussion article for the meetup : Baltimore Area / UMBC: Intro to LW and future of the meetup WHEN: 03 April 2016 03:00:00PM (-0400) WHERE: Performing Arts and Humanities Bldg Room 456, 1000 Hilltop Cir, Baltimore, MD 21250 We have some new memb...
f6a2332d-b39c-4ea6-a815-bb687b0e447e
StampyAI/alignment-research-dataset/eaforum
Effective Altruism Forum
L’importanza delle IA come possibile minaccia per l’umanità Traduzione di Google Translate: <https://www-vox-com.translate.goog/future-perfect/2018/12/21/18126576/ai-artificial-intelligence-machine-learning-safety-alignment?_x_tr_sl=en&_x_tr_tl=it&_x_tr_hl=en&_x_tr_pto=wapp>
42c9cd50-5f6c-4da1-b363-b3a1f43e321d
StampyAI/alignment-research-dataset/alignmentforum
Alignment Forum
Some reasons why a predictor wants to be a consequentialist *Status: working notes* Here's an exercise I've found very useful for intuition-building around alignment: 1. Propose a solution to the alignment problem. 2. Dig into the details until you understand why the proposal fails. 3. If there is an obvious fix, ...
1d32e307-5100-42ca-adbf-db72425d1b8b
trentmkelly/LessWrong-43k
LessWrong
Neutral AI Unfriendly AI has goal conflicts with us.  Friendly AI (roughly speaking) shares our goals.  How about an AI with no goals at all? I'll call this "neutral AI".  Cyc is a neutral AI.  It has no goals, no motives, no desires; it is inert unless someone asks it a question.  It then has a set of routines it us...
b60ee39b-23e0-4f92-a3e6-57ded86b7b5d
trentmkelly/LessWrong-43k
LessWrong
Eutopia is Scary Followup to:  Why is the Future So Absurd? "The big thing to remember about far-future cyberpunk is that it will be truly ultra-tech.  The mind and body changes available to a 23rd-century Solid Citizen would probably amaze, disgust and frighten that 2050 netrunner!"         —GURPS Cyberpunk Pick up...
69ec27b8-6c7f-48d0-bc49-b37736359930
trentmkelly/LessWrong-43k
LessWrong
Learning Deep Learning the EASY way, with Keras
f3b4ab6f-7f1c-4806-b53f-74d39e19f229
trentmkelly/LessWrong-43k
LessWrong
You Too Can See Suffering
94c025f4-116e-4b6a-bc24-3f0882c7f68e
trentmkelly/LessWrong-43k
LessWrong
[SEQ RERUN] Emotional Involvement Today's post, Emotional Involvement was originally published on 06 January 2009. A summary (taken from the LW wiki):   > Since the events in video games have no actual long-term consequences, playing a video game is not likely to be nearly as emotionally involving as much less drama...
6eb66df1-4ddd-417e-841b-0e49fceb0afb
trentmkelly/LessWrong-43k
LessWrong
The Perils of Professionalism Professionalism is a useful trait to be able to display, but it isn’t an unalloyed good. This essay is attempting to outline why deliberately not signaling professionalism may be useful for you. First, a definition by example for professionalism: Clean button down shirts with solid colou...