The Nonlinear Library

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: 'Empiricism!' as Anti-Epistemology, published by Eliezer Yudkowsky on March 14, 2024 on LessWrong. (Crossposted by habryka after asking Eliezer whether I could post it under his account) i. "Ignore all these elaborate, abstract, theoretical predictions," the Spokesperson for Ponzi Pyramid Incorporated said in a firm, reassuring tone. "Empirically, everyone who's invested in Bernie Bankman has received back 144% of what they invested two years later." "That's not how 'empiricism' works," said the Epistemologist. "You're still making the assumption that --" "You could only believe that something different would happen in the future, if you believed in elaborate theoretical analyses of Bernie Bankman's unobservable internal motives and internal finances," said the spokesperson for Ponzi Pyramid Incorporated. "If you are a virtuous skeptic who doesn't trust in overcomplicated arguments, you'll believe that future investments will also pay back 144%, just like in the past. That's the prediction you make if you predict based purely on empirical observations, instead of theories about a future nobody has seen!" "That's not how anything works," said the Epistemologist. "Every future prediction has a theory connecting it to our past observations. There's no such thing as going from past observations directly to future predictions, with no theory, no assumptions, to cross the gap --" "Sure there's such a thing as a purely empirical prediction," said the Ponzi spokesperson. "I just made one. Not to mention, my dear audience, are you really going to trust anything as complicated as epistemology?" "The alternative to thinking about epistemology is letting other people do your thinking about it for you," said the Epistemologist. "You're saying, 'If we observe proposition X "past investors in the Ponzi Pyramid getting paid back 144% in two years", that implies prediction Y "this next set of investors in the Ponzi Pyramid will get paid back 144% in two years"'. X and Y are distinct propositions, so you must have some theory saying 'X -> Y' that lets you put in X and get out Y." "But my theory is empirically proven, unlike yours!" said the Spokesperson. "...nnnnoooo it's not," said the Epistemologist. "I agree we've observed your X, that past investors in the Ponzi Pyramid got 144% returns in 2 years -- those investors who withdrew their money instead of leaving it in to accumulate future returns, that is, not quite all investors. But just like prediction Y of 'the next set of investors will also receive 144% in 2 years' is not observed, the connecting implication 'if X, then Y' is not yet observed, just like Y itself is not observed. When you go through the step 'if observation X, then prediction Y' you're invoking an argument or belief whose truth is not established by observation, and hence must be established by some sort of argument or theory. Now, you might claim to have a better theoretical argument for 'X -> Y' over 'X -> not Y', but it would not be an empirical observation either way." "You say words," replied the Spokesperson, "and all I hear are -- words words words! If you instead just look with your eyes at past investors in the Ponzi Pyramid, you'll see that every one of them got back 144% of their investments in just two years! Use your eyes, not your ears!" "There's a possible theory that Bernie Bankman is making wise investments himself, and so multiplying invested money by 1.2X every year, then honestly returning that money to any investor who withdraws it," said the Epistemologist. "There's another theory which says that Bernie Bankman has been getting more money invested every year, and is using some of the new investments to pay back some fraction of previous investors who demanded their money back --" "Why would Bernie Bankman do that, instead of taking all the ...

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: GiveWell is hiring Research Analysts!, published by GiveWell on March 14, 2024 on The Effective Altruism Forum. We're hiring Research Analysts to support our research team and ensure that the work we produce is accurate and high quality. This is an entry-level role with no experience requirements. Please apply if you're interested, and consider sharing this post so interested folks in your network can take a look! The position is remote-eligible and will be compensated at $95,900-$105,800 depending on location (international salaries will be determined on a case-by-case basis). For more details, see the job description and FAQ. Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org

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: On the Latest TikTok Bill, published by Zvi on March 13, 2024 on LessWrong. TikTok Might Get Banned Soon This attempt is getting reasonably far rather quickly, passing the House with broad support. Alec Stapp: TikTok bill to remove influence of CCP: passed unanimously out of committee GOP leadership says they'll bring it to the floor for a vote next week Biden says he'll sign the bill if passed Can't believe it's taken this long, but should be done soon. It's been obvious for years that we shouldn't let China control a black-box algorithm that influences >100 million American users. JSM: Can this stand up to court scrutiny though? Alec Stapp: Yes. It then passed the house 352-65, despite opposition from Donald Trump. Manifold is as of now around 72% that a bill will pass, similar to Metaculus. Consensus is that it is unlikely that ByteDance will divest. They will fight in court, and if they lose they likely are not bluffing about letting TikTok be shut down or banned in America, Metaculus only has a 12% chance they will sell this year. The bill now goes on to the Senate. I see about a 43% chance it passes there within the month, and a 71% chance it will happen this year. Those numbers seem reasonable to me. The main purpose of this post is to go over arguments for and against the bill, and also what the bill actually would and would not do. I have long been in favor on principle of banning or forcing divestiture of TikTok. Then I saw the Restrict Act, and that was clearly a no-good, very-bad bill. My view of the current bill, after a close reading, is that it is vastly better, and about as good as we could reasonably expect. It seems positive and I hope it passes, whether or not ByteDance folds and agrees to divest. I expect it to pass constitutional muster, although one cannot be sure. To make them easy to find: Here is Noah Smith's case for banning TikTok. Here is Matthew Yglesias's case for banning TikTok. This is a profile of Make Gallagher, who is leading the charge to pass the bill. I go over various arguments for and against the bill, and for and against forcing divestiture of or banning TikTok in general, as well, as well as other related developments. The good argument against the bill is the libertarian concern about expansion of government powers, and what else the government might do. I do not believe it should carry the day on this bill, but I definitely get why one might think so. Execution is Everything I continue to strongly be in favor, in principle, of banning or forcing divestiture of TikTok, if we could do exactly that and only that, without otherwise attacking free speech and free enterprise or expanding the power of the state. TikTok continues to be Chinese spyware. It also continues to be an increasing point of vulnerability for China to put its thumb on American culture, politics and opinion. It continues to promote unhealthy patterns of use. Many want to quit, or know they would be better off without it, or at least would take very little money to quit despite spending tons of time on the app, but feel locked in by a combination of a Skinner box and social dynamics of everyone else being there. All the dynamics around this round of the fight make me more confident that it is important to get this done. So yes, if there was a clean way to get rid of it or force divestiture, great. However, as I said a year ago in Given the Restrict Act, Don't Ban TikTok, the proposed S 686 or the Restrict Act would have vastly expanded government powers over the internet, a cure far worse than the disease. So for me, ultimately, it comes down to the bill. Is it a good bill, or a bad bill? More precisely, is this a bill we can live with? Daniel Lippman (Politico): "They're trying to use these scare tactics to have a bill that gives the government unprecedented ...

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: Why I care so much about Diversity, Equity and Inclusion in EA, published by Ulrik Horn on March 13, 2024 on The Effective Altruism Forum. "Dude!" Bob panted, horror written all over his face as he scrambled back up the driveway and towards the main road where we were waiting. His breaths came hard and fast. "He aimed a fucking shotgun at me!" Sweat had formed droplets across his forehead, maybe due to his sprint, or maybe due to the terror he must surely feel. I stood there, unsure if it was his crazy run or the scare that had him looking like this. Our road trip had turned into something out of a horror movie. Our car broke down in the middle of nowhere, leaving us stranded with a coolant leak. We found ourselves at the closest house, hoping for some water to replace the leaked coolant. "Hey Ulrik, why don't you try to go back there and ask for water," Bob said, as if he hadn't just dodged a bullet. His suggestion hit me like a ton of bricks. Was he joking? He had just been threatened, and now he wanted me to look down the barrel of that shotgun? "But… You just said they almost shot you?" I couldn't hide my shock and fear. "Come on, Ulrik… You're white!" I felt the ground shifting under my feet, as the lens through which I viewed the world was replaced with a new, much thicker and darker one, one that made everything look uncannily foreign to me. True story of my life, as far as I can recall (my memory might have exaggerated it a bit). That said, even if it was to be completely fabricated, it still does not detract from the points I will make below. The story is just one of many opportunities I have been privileged to have, getting a small peek into the lived experiences of my PoC, female, gay, and/or friends with disabilities. I am mentioning this story both to make this more interesting to read, but also so you can understand better why I feel like I feel about DEI. Let me give you another example of the kind of experiences I have had and the type of environments I am used to navigating: I used to work for a renewables consultancy in Bristol, UK. It was a super social workplace, lots of pub visits, and attending lots of parties with colleagues on the weekends. And I brought my female and PoC friends along to many of these events and it just felt really nice. My colleagues were so welcoming, tactful and respectful despite having lots of fun. Never, once, did an incident occur where it got awkward due to some "DEI-type incident". I felt completely safe bringing any of my friends or family there: I knew my friends would also feel safe and welcome, which made me feel safe and welcome too. If that sounds like magic to you - "how can so many different people get along so well?!?!" - then perhaps it is helpful to explain how I navigate in such diverse settings, mostly using an example of my whiteness. I think the way I act in diverse social settings might also let you understand why I have high expectations of others when it comes to "DEI behavior". When I am interacting with a person of color (PoC), I am aware that their entire life experience is probably littered with similar, if perhaps not as extreme, experiences as the one Bob had in the vignette, above. I can further imagine that such frequent and repeated experiences of how one does not belong, or is not qualified or whatever the feeling they derived was, has created certain, strong emotional associations. So my friend Bob from the story above would probably get nervous about walking up to a random house (e.g. as part of some AI policy canvassing initiative in a state with lax gun laws) or perhaps might feel some discomfort of attending an all-white event, especially if they suspect that some, maybe most of the attendants are actively engaging in online discussions about genetic enhancement of PoC in poor countri...

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: China x AI Reference List, published by Saad Siddiqui on March 13, 2024 on The Effective Altruism Forum. Background There are several China-focused AI reading lists / curricula out there (e.g.: AI governance & China: Reading list (2023), FHI Syllabus (2020), (Basic) Chinese AI Regulation Reading List (2023)) They are either relatively brief or somewhat outdated. Our reading list aims to provide a more comprehensive set of key resources when it comes to learning about China, AI safety and policy We incorporated readings from these reading lists where it felt relevant This list is based off a community-generated set of readings that were used for a 6-week AI and China discussion group run by the China & Global Priorities Group in 2023 You can access a version of the reference list on our website too. Structure The list is designed as a longlist that can act as a starting point for folks looking to dive deeper into this topic and various sub-topics - it is not a snapshot of the 3 most important readings per topic area The entire list is broken down into key themes Domestic AI Governance International AI Governance Key actors and their views on AI risks AI Inputs Resources to follow We have added in commentary where we felt it would be useful to do so (e.g., we were made aware of potential factual inaccuracies or biased views) Within sections, sources are arranged roughly in order of relevance, not chronology. Sources earlier in a section are more foundational, while later ones are either primary sources that require more context to analyze or older reports/analysis. Sometimes we put related readings next to each other. Ways to get involved Feel free to suggest additional readings using this form - we're doing some amount of vetting to prevent the list from ballooning out of control Join the China & Global Priorities Group if you want to be notified about further discussion groups organized Caveats around sources and structures Epistemic status: This resource list was put together in a voluntary capacity by a group of non-Chinese folks with backgrounds in China Studies and professional work experience on China- and/or AI-related issues. We spent several hours on resource collection and sense-checked items based on their style, content and methodology. We do not necessarily endorse all of these works as "very good," but did exclude stuff where we could see that it is obviously low quality. There are many sub-topics where we struggled to find very high-quality material but we still included some publications to give interested readers a start. We expect that most of our audience will not be able to read Chinese easily or fluently, and as such we have provided many English sources. However, it's important to remember that gaining a deep and concrete understanding of this space is really hard even with Chinese language skills and lived experience in China, so readers without those skills and experiences should be cautious about forming very strong views based on the select few sources that are included here. Machine translation is useful but imperfect in many ways. Machine translation will not be able to tell you the significance of specific word choice, which potentially requires deeper knowledge of what terminology means in the broader ideological context of the party-state (this is especially true for official statements and documents). Moreover, official English versions of Chinese government documents sometimes differ from the Chinese version! What is Lost in Translation? Differences between Chinese Foreign Policy Statements and Their Official English Translations, Mokry, 2022 China is not a monolith; sources you read that claim that 'China does X' should be treated with caution. Different actors within China have different aims and while it's true that the party-state ...

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: Wild animal welfare? Stable totalitarianism? Predict which new EA cause area will go mainstream!, published by Jackson Wagner on March 13, 2024 on The Effective Altruism Forum. Long have I idly whiled away the hours browsing Manifold Markets, trading on trivialities like videogame review scores or NASA mission launch dates. It's fun, sure -- but I am a prediction market advocate, who believes that prediction markets have great potential to aggregate societally useful information and improve decision-making! I should stop fooling around, and instead put my Manifold $Mana to some socially-productive use!! So, I've decided to create twenty subsidized markets about new EA cause areas. Each one asks if the nascent cause area (like promoting climate geoengineering, or researching space governance) will receive $10,000,000+ from EA funders before the year 2030. My hope is that that these markets can help create common knowledge around the most promising up-and-coming "cause area candidates", and help spark conversations about the relative merits of each cause. If some causes are deemed likely-to-be-funded-by-2030, but little work is being done today, that could even be a good signal for you to start your own new project in the space! Without further ado, here are the markets: Animal Welfare Will farmed-invertebrate welfare (shrimp, insects, octopi, etc) get $10m+ from EA funders before 2030? Will wild-animal welfare interventions get $10m+ from EA funders before 2030? [embed most popular market] Global Health & Development Will alcohol, tobacco, & sugar taxation... ? Mental-health / subjective-wellbeing interventions in developing countries? Institutional improvements Approval voting, quadratic funding, liquid democracy, and related democratic mechanisms? Georgism (aka land value taxes)? Charter Cities / Affinity Cities / Network States? Investing (Note that the resolution criteria on these markets is different than for the other questions, since investments are different from grants.) Will the Patient Philanthropy Fund grow to $10m+ before 2030? Will "impact markets" distribute more than $10m of grant funding before 2030? X-Risk Civilizational bunkers? Climate geoengineering? Preventing stable totalitarianism? Preventing S-risks? Artificial Intelligence Mass-movement political advocacy for AI regulation (ie, "PauseAI")? Mitigation of AI propaganda / "botpocalypse" impacts? Transhumanism Cryonics & brain-emulation research? Human intelligence augmentation / embryo selection? Space governance / space colonization? Moral philosophy Research into digital sentience or the nature of consciousness? Interventions primarily motivated by anthropic reasoning, acausal trade with parallel universes, alien civilizations, simulation arguments, etc? I encourage you to trade on these markets, comment on them, and boost/share them -- put your Manifold mana to a good use by trying to predict the future trajectory of the EA movement! Here is one final market I created, asking which three of the cause areas above will receive the most support between now and 2030. Resolution details & other thoughts The resolution criteria for most of these questions involves looking at publicly-available grantmaking documentation (like this Openphil website, for example), adding up all the grants that I believe qualify as going towards the stated cause area, and seeing if the grand total exceeds ten million dollars. Since I'm specifically interested in how the EA movement will grow and change over time, I will only be counting money from "EA funders" -- stuff like OpenPhil, LTFF, SFF, Longview Philanthropy, Founders Fund, GiveWell, etc, will count for this, while money from "EA-adjacent" sources (like, say, Patrick Collison, Yuri Milner, the Bill & Melinda Gates Foundation, Elon Musk, Vitalik Buterin, Peter T...

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: Laying the Foundations for Vision and Multimodal Mechanistic Interpretability & Open Problems, published by Sonia Joseph on March 13, 2024 on The AI Alignment Forum. Join our Discord here. This article was written by Sonia Joseph, in collaboration with Neel Nanda, and incubated in Blake Richards's lab at Mila and in the MATS community. Thank you to the Prisma core contributors, including Praneet Suresh, Rob Graham, and Yash Vadi. Full acknowledgements of contributors are at the end. I am grateful to my collaborators for their guidance and feedback. Outline Part One: Introduction and Motivation Part Two: Tutorial Notebooks Part Three: Brief ViT Overview Part Four: Demo of Prisma's Functionality Key features, including logit attribution, attention head visualization, and activation patching. Preliminary research results obtained using Prisma, including emergent segmentation maps and canonical attention heads. Part Five: FAQ, including Key Differences between Vision and Language Mechanistic Interpretability Part Six: Getting Started with Vision Mechanistic Interpretability Part Seven: How to Get Involved Part Eight: Open Problems in Vision Mechanistic Interpretability Introducing the Prisma Library for Multimodal Mechanistic Interpretability I am excited to share with the mechanistic interpretability and alignment communities a project I've been working on for the last few months. Prisma is a multimodal mechanistic interpretability library based on TransformerLens, currently supporting vanilla vision transformers (ViTs) and their vision-text counterparts CLIP. With recent rapid releases of multimodal models, including Sora, Gemini, and Claude 3, it is crucial that interpretability and safety efforts remain in tandem. While language mechanistic interpretability already has strong conceptual foundations, many research papers, and a thriving community, research in non-language modalities lags behind. Given that multimodal capabilities will be part of AGI, field-building in mechanistic interpretability for non-language modalities is crucial for safety and alignment. The goal of Prisma is to make research in mechanistic interpretability for multimodal models both easy and fun. We are also building a strong and collaborative open source research community around Prisma. You can join our Discord here. This post includes a brief overview of the library, fleshes out some concrete problems, and gives steps for people to get started. Prisma Goals Build shared infrastructure (Prisma) to make it easy to run standard language mechanistic interpretability techniques on non-language modalities, starting with vision. Build shared conceptual foundation for multimodal mechanistic interpretability. Shape and execute on research agenda for multimodal mechanistic interpretability. Build an amazing multimodal mechanistic interpretability subcommunity, inspired by current efforts in language. Set the cultural norms of this subcommunity to be highly collaborative, curious, inventive, friendly, respectful, prolific, and safety/alignment-conscious. Encourage sharing of early/scrappy research results on Discord/Less Wrong. Co-create a web of high-quality research. Tutorial Notebooks To get started, you can check out three tutorial notebooks that show how Prisma works. Main ViT Demo Overview of main mechanistic interpretability technique on a ViT, including direct logit attribution, attention head visualization, and activation patching. The activation patching switches the net's prediction from tabby cat to Border collie with a minimum ablation. Emoji Logit Lens Deeper dive into layer- and patch-level predictions with interactive plots. Interactive Attention Head Tour Deeper dive into the various types of attention heads a ViT contains with interactive JavaScript. Brief ViT Overview A vision transf...

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: #182 - Comparing the welfare of humans, chickens, pigs, octopuses, bees, and more (Bob Fischer on the 80,000 Hours Podcast), published by 80000 Hours on March 13, 2024 on The Effective Altruism Forum. We just published an interview: Bob Fischer on comparing the welfare of humans, chickens, pigs, octopuses, bees, and more. Listen on Spotify or click through for other audio options, the transcript, and related links. Below are the episode summary and some key excerpts. Episode summary [One] thing is just to spend time thinking about the kinds of things animals can do and what their lives are like. Just how hard a chicken will work to get to a nest box before she lays an egg, the amount of labour she's willing to go through to do that, to think about how important that is to her. And to realise that we can quantify that, and see how much they care, or to see that they get stressed out when fellow chickens are threatened and that they seem to have some sympathy for conspecifics. Those kinds of things make me say there is something in there that is recognisable to me as another individual, with desires and preferences and a vantage point on the world, who wants things to go a certain way and is frustrated and upset when they don't. And recognising the individuality, the perspective of nonhuman animals, for me, really challenges my tendency to not take them as seriously as I think I ought to, all things considered. Bob Fischer In today's episode, host Luisa Rodriguez speaks to Bob Fischer - senior research manager at Rethink Priorities and the director of the Society for the Study of Ethics and Animals - about Rethink Priorities's Moral Weight Project. They cover: The methods used to assess the welfare ranges and capacities for pleasure and pain of chickens, pigs, octopuses, bees, and other animals - and the limitations of that approach. Concrete examples of how someone might use the estimated moral weights to compare the benefits of animal vs human interventions. The results that most surprised Bob. Why the team used a hedonic theory of welfare to inform the project, and what non-hedonic theories of welfare might bring to the table. Thought experiments like Tortured Tim that test different philosophical assumptions about welfare. Confronting our own biases when estimating animal mental capacities and moral worth. The limitations of using neuron counts as a proxy for moral weights. How different types of risk aversion, like avoiding worst-case scenarios, could impact cause prioritisation. And plenty more. Producer and editor: Keiran Harris Audio Engineering Lead: Ben Cordell Technical editing: Simon Monsour and Milo McGuire Additional content editing: Katy Moore and Luisa Rodriguez Transcriptions: Katy Moore Highlights Using neuron counts as a proxy for sentience Luisa Rodriguez: A colleague of yours at Rethink Priorities has written this report on why neuron counts aren't actually a good proxy for what we care about here. Can you give a quick summary of why they think that? Bob Fischer: Sure. There are two things to say. One is that it isn't totally crazy to use neuron counts. And one way of seeing why you might think it's not totally crazy is to think about the kinds of proxies that economists have used when trying to estimate human welfare. Economists have for a long time used income as a proxy for human welfare. You might say that we know that there are all these ways in which that fails as a proxy - and the right response from the economist is something like, do you have anything better? Where there's actually data, and where we can answer at least some of these high-level questions that we care about? Or at least make progress on the high-level questions that we care about relative to baseline? And I think that way of thinking about what neuron-count-based proxies ar...

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 could a policy banning AGI look like?, published by TsviBT on March 13, 2024 on LessWrong. [Caveat lector: I know roughly nothing about policy!] Suppose that there were political support to really halt research that might lead to an unstoppable, unsteerable transfer of control over the lightcone from humans to AGIs. What government policy could exert that political value? [That does sound relaxing.] Banning AGI research specifically This question is NOT ASKING ABOUT GENERALLY SLOWING DOWN AI-RELATED ACTIVITY. The question is specifically about what it could look like to ban (or rather, impose an indefinite moratorium on) research that is aimed at creating artifacts that are more capable in general than humanity. So "restrict chip exports to China" or "require large vector processing clusters to submit to inspections" or "require evals for commercialized systems" don't answer the question. The question is NOT LIMITED to policies that would be actually practically enforceable by their letter. Making AGI research illegal would slow it down, even if the ban is physically evadable; researchers generally want to think publishable thoughts, and generally want to plausibly be doing something good or neutral by their society's judgement. If the FBI felt they had a mandate to investigate AGI attempts, even if they would have to figure out some only-sorta-related crime to actually charge, maybe that would also chill AGI research. The question is about making the societal value of "let's not build this for now" be exerted in the most forceful and explicit form that's feasible. Some sorts of things that would more address the question (in the following, replace "AGI" with "computer programs that learn, perform tasks, or answer questions in full generality", or something else that could go in a government policy): Make it illegal to write AGIs. Make it illegal to pay someone if the job description explicitly talks about making AGIs. Make it illegal to conspire to write AGIs. Why ask this? I've asked this question of several (5-10) people, some of whom know something about policy and have thought about policies that would decrease AGI X-risk. All of them said they had not thought about this question. I think they mostly viewed it as not a very salient question because there isn't political support for such a ban. Maybe the possibility has been analyzed somewhere that I haven't seen; links? But I'm still curious because: I just am. Curious, I mean. Maybe there will be support later, at which point it would be good to have already mostly figured out a policy that would actually delay AGI for decades. Maybe having a clearer proposal would crystallize more political support, for example by having something more concrete to rally around, and by having something for AGI researchers "locked in races" to coordinate on as an escape from the race. Maybe having a clearer proposal would allow people who want to do non-AGI AI research to build social niches for non-AGI AI research, and thereby be less bluntly opposed to regulation on AGI specifically. [other benefits of clarity] Has anyone really been far even as decided to use? There's a lot of problems with an "AGI ban" policy like this. I'm wondering, though, which problems, if any, are really dealbreakers. For example, one problem is: How do you even define what "AGI" or "trying to write an AGI" is? I'm wondering how much this is actually a problem, though. As a layman, as far as I know there could be existing government policies that are somewhat comparably difficult to evaluate. Many judicial decisions related to crimes, as I vaguely understand it, depend on intentionality and belief - - e.g. for a killing to be a murder, the killer must have intended to kill and must not have believed on reasonable grounds that zer life was imminent...

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: Virtual AI Safety Unconference 2024, published by Orpheus Lummis on March 13, 2024 on The AI Alignment Forum. When: May 23rd to May 26th 2024 Where: Online, participate from anywhere. VAISU is a collaborative and inclusive event for AI safety researchers, aiming to facilitate collaboration, understanding, and progress towards problems of AI risk. It will feature talks, research discussions, and activities around the question: "How do we ensure the safety of AI systems, in the short and long term?". This includes topics such as alignment, corrigibility, interpretability, cooperativeness, understanding humans and human value structures, AI governance, strategy, … Engage with the community: Apply to participate, give a talk, or propose a session. Come to share your insights, discuss, and collaborate on subjects that matter to you and the field. Visit vaisu.ai to apply and to read further. VAISU team Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.