The Nonlinear Library

The Nonlinear Library

By The Nonlinear FundEducation
Download on the App Store

All The Nonlinear Library episodes

  • EA - New CSER Director: Prof Matthew Connelly by HaydnBelfield
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: New CSER Director: Prof Matthew Connelly, published by HaydnBelfield on May 17, 2023 on The Effective Altruism Forum.
    The Centre for the Study of Existential Risk (CSER) at Cambridge University is getting a new Director – Professor Matthew Connelly will be our Director from July 2023. Seán Ó hÉigeartaigh, our Interim Director, is staying at CSER and will be focussing more on AI governance and safety research.
    Prof Connelly is currently a Professor of international and global history at Columbia University, and for the last seven years has been Co-Director of its social science research centre, the Institute for Social and Economic Research and Policy. He has significant policy experience as a consultant for the Gates Foundation, the World Bank, and the Department of Homeland Security and Senate committees. He is the author of The Declassification Engine: What History Reveals about America’s Top Secrets and Fatal Misconception (an Economist and Financial Times book of the year). His next book is on “the history of the end of the world”, a subject on which he has taught multiple courses.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • LW - What if they gave an Industrial Revolution and nobody came? by jasoncrawford
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: What if they gave an Industrial Revolution and nobody came?, published by jasoncrawford on May 17, 2023 on LessWrong.
    Imagine you could go back in time to the ancient world to jump-start the Industrial Revolution. You carry with you plans for a steam engine, and you present them to the emperor, explaining how the machine could be used to drain water out of mines, pump bellows for blast furnaces, turn grindstones and lumber saws, etc.
    But to your dismay, the emperor responds: “Your mechanism is no gift to us. It is tremendously complicated; it would take my best master craftsmen years to assemble. It is made of iron, which could be better used for weapons and armor. And even if we built these engines, they would consume enormous amounts of fuel, which we need for smelting, cooking, and heating. All for what? Merely to save labor. Our empire has plenty of labor; I personally own many slaves. Why waste precious iron and fuel in order to lighten the load of a slave? You are a fool!”
    We can think of innovation as a kind of product. In the market for innovation there is supply and demand. To explain the Industrial Revolution, economic historians like Joel Mokyr emphasize supply factors: factors that create innovation, such as scientific knowledge and educated craftsmen. But where does demand for innovation come from? What if demand for innovation is low? And how much can demand factors explain industrialization?
    Riffing on an old anti-war slogan, we can ask: What if they gave an Industrial Revolution and nobody came?
    Robert Allen thinks demand factors have been underrated. He makes his case in The British Industrial Revolution in Global Perspective, in which he argues that many major inventions were adopted when and where the prices of various factors made it profitable and a good investment to adopt them, and not before. In particular, he emphasizes high wages, the price of energy, and (to a lesser extent) the cost of capital. When and where labor is expensive, and energy and capital are cheap, then it is a good investment to build machines that consume energy in order to automate labor, and further, it is a good investment to do the R&D needed to invent such machines. But not otherwise.
    And, when he’s feeling bold, Allen might push the hypothesis further: to the extent that demand factors explain the adoption of technology, we don’t need other hypotheses, including those about supply factors. We don’t need to suppose that certain cultures are more inventive than others or more receptive to innovation; we don’t need to posit that some societies exhibit bourgeois virtues or possess a culture of growth.
    Let’s examine Allen’s argument and see what we can learn from it. First I’ll summarize the core of his argument, then I’ll discuss some responses and criticism and give my own thoughts.
    High wages and cheap energy
    In the first half of the book, Allen establishes that pre-industrial Britain was indeed a high-wage, cheap-energy economy.
    Wages
    Here are workers’ wages in various cities around the world. By the 18th century, wages in London and Amsterdam were more than twice that of other major cities:
    Nor is it just that prices were higher in those cities. Here are the wages deflated by the cost of a “subsistence basket,” the bare minimum of food, clothing, and other goods needed to live. Workers in Vienna, Delhi, or Beijing were only a little above subsistence; those in Amsterdam or London were ~4x above:
    There is also qualitative evidence of high wages. Workers in Northwest Europe ate better diets, “in view of the apparent widespread consumption of expensive and highly refined foods like white bread, meat, dairy products and beer. In contrast, workers and peasants in France, Italy, India and China ate a quasi-vegetarian diet of grain, often boiled, with scarcely any animal pro...
    30 min
  • EA - Hiatus: EA and LW post summaries by Zoe Williams
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Hiatus: EA and LW post summaries, published by Zoe Williams on May 17, 2023 on The Effective Altruism Forum.
    For the past ~8 months, I've been summarizing the top posts on the EA and LW forums each week (see archive here), a project supported by Rethink Priorities (RP).I’ve recently taken on a new role as an AI Governance & Strategy Research Manager at RP. As a result of this we're going to be putting the forum summaries on hiatus while we work on what they should look like in the future, and hire for someone new to run the project. A big thank you to everyone who completed our recent survey - it’s great input for us as we evaluate this project going forward!
    The hiatus will likely last for ~4-6 months. We’ll continue to use the existing email list and podcast channel (EA Forum Podcast (Summaries)) when it's back up and running, so subscribe if you’re interested and feel free to continue to share it with others.
    If you’re looking for other ways to stay up to date in the meantime, some resources to consider:
    Newsletters
    The EA Forum Digest - a weekly newsletter recommending new EA forum posts that have high karma, active discussion, or could use more input.
    Monthly Overload of Effective Altruism - a monthly newsletter with top research, organizational updates and events in the EA community.
    Podcasts
    EA forum podcast (curated posts) - human narrations of some of the best posts from the EA forum.
    Nonlinear library - AI narrations of all posts from the EA Forum, Alignment Forum, and LessWrong that meet a karma threshold.
    There are heaps of cause-area specific newsletters out there too - if you have suggestions, please share them in the comments.
    I’ve really enjoyed my time running this project! Thanks for reading and engaging, to Coleman Snell for narrating, and to all the writers who’ve shared their ideas and helped people find new opportunities to do good.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    2 min
  • EA - Play Regrantor: Move up to $250,000 to Your Top High-Impact Projects! by Dawn Drescher
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Play Regrantor: Move up to $250,000 to Your Top High-Impact Projects!, published by Dawn Drescher on May 17, 2023 on The Effective Altruism Forum.
    Summary. We have collected expressions of interest for a total donation budget of about $200,000 from donors who are interested in using our platform. How many of them will end up donating and how much they’ll donate depends on you and how many great projects you bring onto the platform. But Greg Colbourn has firmly committed another $50,000 for AGI moratorium projects for a total of $250,000!
    Funding
    Over the past months we’ve asked many donors whether they would consider using the Impact Markets platform to find giving opportunities and what their 2023 donation budget is. We’re now at an aggregate budget of $200,000 and counting. Of course these donors are free to decide that they prefer another project over the ones that have registered on our platform, so that these $200,000 are far from committed.
    But Greg Colbourn has firmly committed another $50,000 for projects related to an AGI moratorium! He is already in touch with some early-stage efforts, but we definitely need more people to join the cause.
    You want to become a top donor – a project scout? You want to support a project?
    You think Charity X is the most impactful one but it’s still underfunded? Convince them to register on app.impactmarkets.io. Then register your donations to them. (They have to confirm that your claim is accurate.) Speed the process by repeating it with a few high-impact projects. When the projects reach some milestones, they can submit them for review. The better the review, the higher your donor score.
    A score > 200 still puts you firmly among the top 10 donors on our platform. That can change as more project scouts register their high-impact donations.
    At the moment, we’re still allowing donors to import all their past donations. (Please contact us if you would like to import a lot.) We will eventually limit this to the past month.
    What if you need money
    If you’re running or planning to run some impactful project, you can register it on app.impactmarkets.io and pitch it to our top donors. If they think it’s great, their donation (an endorsement in itself) can pull in many more donations from the people who trust them.
    We’re continually onboarding more people with great donation track records, so both the people on the leaderboard and the ranking algorithm are in constant flux. Please check for updates at least in monthly intervals.
    You want to donate but don’t know where
    For now, you can add yours to the expressions of interest we’re collecting. That’ll increase the incentive for awesome projects to join our platform and for awesome donors to vie for the top donor status.
    Once there are top donors that you trust (say, because they share your values), the top donors’ donations will guide you to high-impact projects. Such projects may be outside the purview of charity evaluators like GiveWell or ACE, or they might be too young to have earned top charity status yet. Hence why our impact market doubles as a crowdsourced charity evaluator.
    Questions
    If you have any questions:
    Please see our full FAQ.
    Check out this demo of a bot trained on our FAQ.
    Have a look at our recent Substack posts.
    Join our Discord and the #questions-and-feedback channel.
    Or of course ask your question below!
    Acknowledgements. Thanks to Frankie, Dony, Matt, and Greg for feedback, and thanks to Greg and everyone who has filled in the expression of interest form for their pledges!
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • LW - A Mechanistic Interpretability Analysis of a GridWorld Agent-Simulator (Part 1 of N) by Joseph Bloom
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: A Mechanistic Interpretability Analysis of a GridWorld Agent-Simulator (Part 1 of N), published by Joseph Bloom on May 16, 2023 on LessWrong.
    AKA: Decision Transformer Interpretability 2.0
    Code: repository, Model/Training: here . Task: here.
    High Level Context/Background
    What have I actually done that is discussed in this post?
    This is a somewhat rushed post summarising my recent work and current interests.
    Toy model: I have trained a 3 layer decision transformer which I call “MemoryDT”, to simulate two variations of the same agent, one sampled with slightly higher temperature than training (1) and one with much higher temperature (5). The agent we are attempting to simulate is a goal-directed online RL agent that solves the Minigrid-Memory task, observing an instruction token and going to it in one of two possible locations. The Decision Transformer is also steered by a Reward-to-Go token, which can make it generate trajectories which simulate successful or unsuccessful agent.s
    Analysis: The analysis here is mostly model psychology. No well understood circuits (yet) but I have made some progress and am keen to share it when complete. Here, I discuss the model details so that people are aware of them if they decide to play around with the app and show some curiosities (all screenshots) from working with the model.
    I also made an interpretability app! The interpretability app is a great way to analyse agents and is possibly by far the best part of this whole project.
    My training pipeline should be pretty reusable (not discussed at length here). All the code I’ve used to train this model would be a pretty good starting point for people who want to work on grid-world agents doing tasks like searching for search or retargeting the search. I’ll likely rename the code base soon to something like GridLens.
    MemoryDT seems like a plausibly good toy model of agent simulation which will hopefully be the first of many models which enable us to use mechanistically interpretability to understand alignment relevant properties of agent simulators.
    What does this work contribute to?
    Agent Simulators and Goal Representations
    Studying the ability of agent-simulators to produce true-to-agent trajectories and understanding the mechanisms and representations by which they achieve this is a concrete empirical research task which relates meaningfully to AI alignment concepts like goals, mesa-optimizers and deception. I’m not sure how to measure an AI’s goals, but studying how an agent simulator simulates/interpolates between (gridworld) agents with well understood goals seems like a reasonable place to start.
    Pursuing Fundamental Interpretability Research
    I believe there is often little difference between Alignment-motivated interpretability research and more general interpretability research and both could be accelerated by well understood toy models. I could be wrong that these will be useful for understanding or aligning language models but I think it’s worth testing this hypothesis. For example, phenomena like superposition might be well studied in these models.
    Required Background/Reading:
    There’s a glossary/related works section at the end of this post for reference. The key things to understand going in are:
    Decision Transformers (just the basic concepts, including the difference between online and offline RL).
    Simulators (Agent-Simulator is a derivative concept that feels natural to me, but is predicated on understanding Simulators)
    Mechanistic Interpretability (A basic familiarity will help you understand why I think having transformer RL models in particular is important. The Results will also be pretty obscure otherwise).
    Introducing Agent Simulators
    Motivation
    Decision transformers are a type of offline-RL model called a “trajectory model”. These models use next token prediction...
    26 min
  • LW - AI Safety Newsletter #6: Examples of AI safety progress, Yoshua Bengio proposes a ban on AI agents, and lessons from nuclear arms control by Dan H
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: AI Safety Newsletter #6: Examples of AI safety progress, Yoshua Bengio proposes a ban on AI agents, and lessons from nuclear arms control, published by Dan H on May 16, 2023 on LessWrong.
    Welcome to the AI Safety Newsletter by the Center for AI Safety. We discuss developments in AI and AI safety. No technical background required.
    Subscribe here to receive future versions.
    Examples of AI safety progress
    Training AIs to behave safely and beneficially is difficult. They might learn to game their reward function, deceive human oversight, or seek power. Some argue that researchers have not made much progress in addressing these problems, but here we offer a few examples of progress on AI safety.
    Detecting lies in AI outputs. Language models often output false text, but a recent paper suggests they understand the truth in ways not reflected in their output. By analyzing a model’s internals, we can calculate the likelihood that a model believes a statement is true. The finding has been replicated in models that answer questions about images.
    One reason to prefer examining a model’s internals over trusting its output is because AI training processes often unintentionally reward models for outputting falsehoods. For example, language models might learn to mimic common misconceptions in their training data. More perniciously, if a human is giving feedback on an AI’s performance, the AI can trick the human into falsely believing they’ve done a good job. It’s important to build several defenses against deception so that failures in one layer can potentially be caught by another.
    Giving AI a conscience. AI agents take actions to pursue goals. But there are harmful ways to pursue many goals, such as by breaking laws and violating ethical standards. To block AI agents from misbehaving, its actions can be subjected to approval by an artificial conscience.
    An artificial conscience is a separate model trained to identify unacceptable actions. Before an AI agent takes an action, the action is automatically assessed by artificial conscience. If the conscience does not believe the action is acceptable, it can veto the action, and the agent is unable to act perform an action in its environment until the conscience approves an action. An artificial conscience can be easily combined with AI agent before or after training.
    How would an artificial conscience decide which actions are acceptable? Language models have shown some understanding of ethical concepts such as justice, fairness, and utility which could be used to inform decisions. The people affected by AI decisions could be given a voice in its decisions by aligning AI with laws decided by democratic processes.
    In this scenario, an AI agent witnesses a crime and considers how to respond. The agent was not tasked with fighting crime, so it does not anticipate any reward for its possible actions. But the artificial conscience identifies that calling the police is the only moral response, and this becomes the only action the agent can take.
    Pretraining AI with human preferences. AI models are often pretrained to identify patterns in large volumes of text or image data. Only afterwards are they fine-tuned to behave in ways that humans find valuable. This has drawbacks: AI might learn harmful patterns of thinking during training, or might devise counterproductive ways to achieve the goal provided in fine-tuning.
    A new paper puts feedback about human preferences into the pretraining process. Rather than retrofitted a model to abide by preferences through fine-tuning later in the design process, this paper incorporates preferences early on. When the model is deployed to perform a task, it can be instructed to mimic preferred outputs from the training set. This empirically reduces the chance of misbehavior, so it has been included in Google’s...
    11 min
  • EA - The Charity Entrepreneurship top ideas new charity prediction market by CE
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: The Charity Entrepreneurship top ideas new charity prediction market, published by CE on May 17, 2023 on The Effective Altruism Forum.
    TL;DR: Charity Entrepreneurship would like your help in our research process. We are running a prediction market on the top 10 ideas across two cause areas. A total of $2000 in prizes is available for prediction accuracy and comment quality. Check it out at:
    The CE prediction market
    For our upcoming (winter 2024) Incubation Program, we are researching two cause areas. Within global development we are looking into mass media interventions –social and behavior change communication campaigns delivered through mass media (e.g., radio advertising, TV shows, text messages, etc.) aiming to improve human well-being. Within animal welfare we are looking into preventive (or long-run) interventions for farmed animals – the new charities will not just positively affect farmed animals in the short term, but will have a long-run effect on preventing animal suffering in farms 35 years from now.
    We have narrowed down to the most promising top 10 ideas for each of these cause areas. The Charity Entrepreneurship research team will be doing ~80-hour research projects on as many of these ideas as we can between now and July, carefully examining the evidence and crucial considerations that could either make or break the idea. At the end of this we will aim to recommend two-three ideas for each cause area.
    This is where you come in. We want to get your views and predictions on our top ideas within each cause area. We have put our top idea list onto the Manifold Markets prediction market platform, and you are invited to join a collective exercise to assess these ideas and input into our decision making.
    You can do this by reading the list of top ideas (below) for one or both of the cause areas, and then going to the Manifold Market platform and:
    Make a prediction about how likely you think it is that a specific idea will be recommended by us at the end of our research.
    Leave comments on each idea with your thoughts or views on why it might or might not be recommended, or why it might or might not be a good idea.
    As well as having the great benefit of helping our research, we have $2000 in prizes to give away (generously donated by Manifold Markets).
    $1,000 for comment prizes. We will give $100 to each person who gives one of the top 10 arguments or pieces of information that changes our minds the most regarding our selection decisions.
    $1,000 for forecasting prizes. We will grant prizes to the individuals who do the best at predicting which of the ideas we end up selecting.
    More details on these prizes are available on the page at Manifold.
    The market is open until June 5, 2023 for predictions and comments. This gives the CE research team time to read and integrate comments and insights into our research before our early July deadline.
    To participate, read the list below and go to: to make predictions and leave comments.
    Summary of ideas under consideration
    Mass Media
    By ‘mass media’ interventions we refer to social and behavior change communication campaigns delivered through mass media, aiming to improve human well-being.
    1. Using mobile technologies (mHealth) to encourage women to attend antenatal clinics and/or give birth at a healthcare facility
    Across much of sub-Saharan Africa, only about 55% of women make the recommended four+ antenatal care visits, and only 60% give birth at a healthcare facility. This organization would encourage greater healthcare utilization and achieve lower maternal and neonatal mortality by scaling up evidence-based mHealth interventions, such as one-way text messages or two-way SMS/WhatsApp communications. These messages would aim to address common concerns about professional healthcare, as well as reminding women not to mis...
    17 min
  • EA - Don't optimise for social status within the EA community by freedomandutility
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Don't optimise for social status within the EA community, published by freedomandutility on May 17, 2023 on The Effective Altruism Forum.
    One downside of engaging with the EA community is that social status in the community probably isn't well aligned with impact, so if you consciously or subconsciously start optimising for status, you may be less impactful than you could be otherwise.
    For example, roles outside EA organisations which lead to huge social impact probably won't help much with social status inside the EA community.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    1 min
  • EA - Some quotes from Tuesday's Senate hearing on AI by Daniel Eth
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: Some quotes from Tuesday's Senate hearing on AI, published by Daniel Eth on May 17, 2023 on The Effective Altruism Forum.
    On Tuesday, the US Senate held a hearing on AI. The hearing involved 3 witnesses: Sam Altman, Gary Marcus, and Christina Montgomery. (If you want to watch the hearing, you can watch it here – it's around 3 hours.)
    I watched the hearing and wound up live-tweeting quotes that stood out to me, as well as some reactions. I'm copying over quotes to this post that I think might be of interest to others here. Note this was a very impromptu process and I wasn't originally planning on writing a forum post when I was jotting down quotes, so I've presumably missed a bunch of quotes that would be of interest to many here. Without further ado, here are the quotes (organized chronologically):
    Senator Blumenthal (D-CT): "I think you [Sam Altman] have said, in fact, and I'm gonna quote, 'Development of superhuman machine intelligence is probably the greatest threat to the continued existence of humanity.' You may have had in mind the effect on jobs, which is really my biggest nightmare in the long run..."
    Sam Altman: [doesn't correct the misunderstanding of the quote and instead proceeds to talk about possible effects of AI on employment]
    Sam Altman: "My worst fears are that... we, the field, the technology, the industry, cause significant harm to the world. I think that could happen in a lot of different ways; it's why we started the company... I think if this technology goes wrong, it can go quite wrong, and we want to be vocal about that. We want to work with the government to prevent that from happening, but we try to be very clear eyed about what the downside case is and the work that we have to do to mitigate that."
    Sam Altman: "I think the US should lead [on AI regulation], but to be effective, we do need something global... There is precedent – I know it sounds naive to call for something like this... we've done it before with the IAEA... Given what it takes to make these models, the chip supply chain, the sort of limited number of competitive GPUs, the power the US has over these companies, I think there are paths to the US setting some international standards that other countries would need to collaborate with and be part of, that are actually workable, even though it sounds on its face like an impractical idea. And I think it would be great for the world."
    Senator Coons (D-DE): "I understand one way to prevent generative AI models from providing harmful content is to have humans identify that content and then train the algorithm to avoid it. There's another approach that's called 'constitutional AI' that gives the model a set of values or principles to guide its decision making. Would it be more effective to give models these kinds of rules instead of trying to require or compel training the model on all the different potentials for harmful content? ... I'm interested also, what international bodies are best positioned to convene multilateral discussions to promote responsible standards? We've talked about a model being CERN and nuclear energy. I'm concerned about proliferation and nonproliferation."
    Senator Kennedy (R-LA): "Permit me to share with you three hypotheses that I would like you to assume for the moment to be true... Hypothesis number 3... there is likely a berserk wing of the artificial intelligence community that intentionally or unintentionally could use artificial intelligence to kill all of us and hurt us the entire time that we are dying... Please tell me in plain English two or three reforms, regulations, if any, that you would implement if you were queen or king for a day..."
    Gary Marcus: "Number 1: a safety-review like we use with the FDA prior to widespread deployment... Number 2: a nimble monitoring agency to follow what's going ...
    7 min
  • LW - New User's Guide to LessWrong by Ruby
    Welcome to The Nonlinear Library, where we use Text-to-Speech software to convert the best writing from the Rationalist and EA communities into audio. This is: New User's Guide to LessWrong, published by Ruby on May 17, 2023 on LessWrong.
    (Feel free to leave comments with suggestions on how to improve this page.)
    The road to wisdom? Well, it's plainand simple to express:Errand errand err againbut lessand lessand less.– Piet Hein
    Why a new user guide?
    Although encouraged, you don't have to read this to get started on LessWrong!
    LessWrong is a pretty particular place. We strive to maintain a culture that's uncommon for web forums and to stay true to our values. Recently, many more people have been finding their way here, so I (lead admin and moderator) put together this intro to what we're about.
    My hope is that if LessWrong resonates with your values and interests, this guide will help you become a valued member of community. And if LessWrong isn't the place for you, this guide will help you have a good "visit" or simply seek other pastures.
    Contents of this page/email
    What LessWrong is about: "Rationality"
    Artificial Intelligence
    How to get started
    Helpful Tips
    FAQ
    Open & Welcome Threads
    Contact the site admins
    How to ensure your first post or comment is approved
    Appendices
    The Voting System
    LessWrong moderator's toolkit
    What LessWrong is about: "Rationality"
    LessWrong is online forum/community that was founded with the purpose of perfecting the art of human rationality.
    While truthfulness is a property of beliefs, rationality is a property of reasoning processes. Our definition of rationality is that a more rational reasoning process tends to arrive at true beliefs and good decisions more often than a less rational process. For example, a reasoning process that responds to evidence is more likely to believe true things than one that just goes with what's convenient to believe. An aspiring rationalist is someone who aspires to improve their own reasoning process to arrive at truth more often.
    ...a rationalist isn't just somebody who respects the Truth...All too many people respect the Truth. A rationalist is somebody who respects the processes of finding truth. – Rationality: Appreciating Cognitive Algorithms
    [Aspiring] rationalists should win [at life, their goals, etc]. You know a rationalist because because they're sitting atop a pile of utility. – Rationality is systematized winning
    The Art [of rationality] must have a purpose other than itself, or it collapses into infinite recursion. - the 11th virtue of rationality
    On LessWrong we attempt (though don't always succeed) to apply the rationality lessons we've accumulated to any topic that interests us, and especially topics that seem important, like how to make the world a better place. We don't just care about truth in the abstract, but care about having true beliefs about things we care about so that we can make better and more successful decisions.
    Right now, AI seems like one of the most (or the most) important topics for humanity. It involves many tricky questions, high stakes, and uncertainty in an unprecedented situation. On LessWrong, many users are attempting to apply their best thinking to ensure that the advent of increasingly powerful AI goes well for humanity.
    Is LessWrong for you?
    LessWrong is a good place for:
    who values curiosity, learning, self-improvement, figuring out what's actually true (rather than just what you want to be true or just winning arguments)
    who will change their mind or admit they're wrong in response to compelling evidence or argument
    who wants to work collaboratively with others to figure out what's true
    who likes acknowledging and quantifying uncertainty and applying lessons from probability, statistics, and decision theory to your reasoning
    who is nerdy and interested in all questions of how the world works and who is not afraid to reach weird conclusions if the arguments seem valid
    who likes to be pedantic and precise, and likes ...
    19 min