The Nonlinear Library

The Nonlinear Library

By The Nonlinear FundEducation
Download on the App Store

All The Nonlinear Library episodes

  • LW - [Link] A community alert about Ziz by DanielFilan
    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: [Link] A community alert about Ziz, published by DanielFilan on February 24, 2023 on LessWrong.
    The start of the post is copy-pasted below. Note that the post is anonymous, and I am not claiming to have written it.
    Some people in the rationalist community are concerned about risks of physical violence from Ziz and some of her associates. Following discussions with several people, I’m posting here to explain where those concerns come from, and recommend some responses.
    TLDR (details and links in the post body)
    Over the past few years, Ziz has repeatedly called for the deaths of many different classes of people.
    In August of 2022, Ziz seems to have faked her own death. Ziz’s close associate Gwen Danielson may have done the same thing in April of 2022.
    In November of 2022, three associates of Ziz (Somnulence “Somni” Logencia, Emma Borhanian, and someone going by the alias “Suri Dao”) got into a violent conflict with their landlord in Vallejo, California, according to court records and news reports. Somni stabbed the landlord in the back with a sword, and the landlord shot Somni and Emma. Emma died, and Somni and Suri were arrested. Ziz and Gwen were seen by police at the scene, alive.
    I gather that Pennsylvania police believe, based on a lot of circumstantial evidence and investigation, that one or more of Ziz, Michelle “Jamie” Zajko, or Daniel Blank was very plausibly involved in the murder of Jamie’s parents in Pennsylvania around December 31st, 2022. The police strongly suspect a conspiracy between the three to commit the murders, or related activities between the three involving a high level of coordination.
    The police investigation is currently ongoing. The police have issued many search warrants, but do not yet feel that they have enough concrete evidence to charge Ziz, Jamie, or Daniel.
    Ziz is currently in police custody on charges related to obstructing a PA police investigation. Daniel, Jamie, and Gwen are not in custody (as far as I know), and I don’t know their locations.
    I don’t know of any concrete plans by Ziz or her associates to do anything else violent, but it seemed like a good idea to raise a loud alarm about all of this now. People should speak up (or at least tell a friend) if they have information, and should take whatever safety precautions make sense for their situation in case more violence occurs at some point in the future.
    Even seemingly minor pieces of information might be helpful here, since they could add up to a clearer picture when combined with other information that also seems minor. If you want to share information privately, you can email me at [email protected], or submit information anonymously through this form.
    This is a complicated situation that I don’t fully understand, and it’s likely that I’m getting some facts wrong. I’ve talked to a lot of people in an attempt to piece together what happened, and I’ll try to update this post with corrections or important details if they’re brought to my attention.
    Please keep in mind, as I am trying to keep in mind myself, that every observation is evidence for more than one hypothesis, that things are often not what they seem, and that it’s useful to make the effort to think about both what’s appropriate to do in the world where your best guesses are true, and what’s appropriate to do in the world where your best guesses are wrong. Split and commit is better than seeking confirmation of a single theory.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • AF - Teleosemantics! by Abram Demski
    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: Teleosemantics!, published by Abram Demski on February 23, 2023 on The AI Alignment Forum.
    I wanted to write a long, detailed, analytic post about this, somewhat like my Radical Probabilism post (to me, this is a similarly large update). However, I haven't gotten around to it for a long while. And perhaps it is better as a short, informal post in any case.
    I think my biggest update over the past year has been a conversion to teleosemantics. Teleosemantics is a theory of semantics -- that is, "meaning" or "aboutness" or "reference".
    To briefly state the punchline: Teleosemantics identifies the semantics of a symbolic construct as what the symbolic construct has been optimized to accurately reflect.
    Previously, something seemed mysterious about the map/territory relationship. What could possibly imbue 'symbols' with 'meaning'? The map/territory analogy seems inadequate to answer this question. Indeed, to analogize "belief" with "map" and "the subject of belief" with "territory" commits a homunculus fallacy! The meaning-makers are the map-readers and map-writers; but they can only make meaning by virtue of the beliefs within their own heads. So the map/territory analogy seems to suggest that an infinite regress of meaning-makers would be required.
    You probably won't believe me at first. Perhaps you'll say that the lesson of the map/territory analogy is the correspondence between the map and the territory, which exists independently of the map-reader who uses the correspondence to evaluate the map.
    I have several objections.
    If it's a probabilistic correspondence, where the map contains information about the territory, these are subjective notions, which require some viewpoint.
    If it's a correspondence based on some sort of ontology, where pieces of the map line up with "pieces of reality", I would also say the ontology is in itself a subjective perspective.
    You might think you can define the map/territory correspondence without invoking a map-maker or map-reader by objectively defining the "fit" of a correspondence (so that the meaning of a symbol is based on the best-fitting correspondence, or perhaps, the cloud of well-fitting correspondences). But well-fitting correspondence will include many examples of accidental correspondence, which seem to have little to do with aboutness. Moreover, I think theories like this will fail to adequately account for false belief, which screws up the fit.
    But my point here isn't to denounce the map/territory picture! I still think it is a good framework. Rather, I wanted to gesture at how I still felt confused, despite having the map/territory picture.
    I needed a different analogy, something more like a self-drawing map, to get rid of the homunculus. A picture which included the meaning-maker, rather than just meaning come from nowhere.
    Teleosemantics reduces meaning-making to optimization. Aboutness becomes a type of purpose a thing can have.
    One advantage of this over map-territory correspondence is that it explains the asymmetry between map and territory. Mutual information is symmetric. So why is the map about the territory, but not the other way around? Because the map has been optimized to fit the territory, not the other way around. ("Fit" in the sense of carrying high mutual information, which can be decoded via some specific intended correspondence - a symbolic language.)
    What does it mean to optimize for the map to fit the territory, but not the other way around? (After all: we can improve fit between map and territory by changing either map or territory.) Maybe it's complicated, but primarily what it means is that the map is the part that's being selected in the optimization. When communicating, I'm not using my full agency to make my claims true; rather, I'm specifically selecting the claims to be true.
    I take Teleosemanti...
    12 min
  • AF - AI that shouldn't work, yet kind of does by Donald Hobson
    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 that shouldn't work, yet kind of does, published by Donald Hobson on February 23, 2023 on The AI Alignment Forum.
    There are some things that work surprisingly well in AI. For example, AI that transfers the style of one image to the content of another. Why is the approach described here a hack. It starts with a neural net trained to classify images. It then runs gradient descent on the the content image, trying to get the covariance matrix of the style to match in early network layers, while trying to get the net layers of the original and style transfer images to be as similar as possible on the later layers.
    So the approximations I think are making this work is that, in classifiers, the early layers tend to store simple features, and the later layers tend to hold more complex features. Style is based on the simpler features, and doesn't depend on the location within the image. Content is based on more complex features and does depend on the location in the image.
    We apply optimization power, gradient descent, over heuristics this simple and hacky. And yet it works.
    A simple and hacky AI alignment proposal is to just ask chatGPT to do it.
    This doesn't work because chatGPT has been optimized for text prediction, and so isn't particularly good at AI alignment theory.
    So here is an alignment plan.
    I know it isn't great. But some plan is better than no plan. And there was a post about how alignment may well look like "surely no one could have missed that" or "surely such a stupid idea couldn't work, could it?" not "eureka".
    Train ZFCbot. An AI that is the AlphaGo of ZFC, perhaps trained on random formulas. Perhaps throwing a large corpus of formal maths proofs in there. Ideally the system should have a latent space of maths, so it can think about what style of proof is likely to work before expanding all the details. The system should have a wide range of common maths terms imported from some lean library. It should be optimized purely for ZFC formal theorem proving. Once it is trained, the weights are fixed.
    Train ChatMathsGPT. Similar to large language models. Except with oracle access to ZFCbot. In the many maths papers in it's corpus, it learns to link the informal with the formal. From politics and economics discussions, it asks ZFCbot about toy game theory problems. In general, it learns to identify the pieces of formal maths that best model a situation, and ask about them, and then use the response to predict text.
    There is a sense in which this AI knows less maths than normal ChatGPT. Standard ChatGPT has a small crude understanding of maths built from nothing within it's own mind. This has a much better understanding it can outsource to, it just has to plug in.
    Then we ask this ChatMathsGPT for a paper on logical induction. And we hope it can generate a paper of quality similar to Miri's paper on the topic (where hypothetically this isn't in the training dataset). If it can, then we have a tool to accelerate deconfusion by orders of magnitude.
    Things I am uncertain about. Should ChatMathsGPT have oracle access to a latent space. (can pass gradients, harder to interpret.) or should it just pass formal strings of symbols. (less powerful)
    Should ZFCbot get trained on random ZFC; random ZFC + library of theorems and conjectures and random combinations of high level maths concepts; or random ZFC plus whatever ChatMathsGPT keeps asking. The latter gives a route for data to pass between them.
    This could fail to be smart enough, I mean I wouldn't be particularly surprised if it could be made smart enough. But what would the safety failures of this system look like?
    Firstly, this AI does deconfusion. If you ask it to write a paperclip maximizer in python, you may well get your wish. Or you might get an AI that maximizes something else. Just asking for an aligned AI is...
    6 min
  • EA - EA Global in 2022 and plans for 2023 by Eli Nathan
    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: EA Global in 2022 and plans for 2023, published by Eli Nathan on February 23, 2023 on The Effective Altruism Forum.
    Summary
    We ran three EA Global events in 2022, in London, San Francisco, and Washington, D.C. These conferences all had ~1300–1500 people each and were some of our biggest yet:
    These events had an average score of 9.02 to the question “How likely is it that you would recommend EA Global to a friend or colleague with similar interests to your own?”, rated out of 10.
    Those who filled out our feedback survey (which was a minority of attendees, around 1200 individuals in total across all three events) reported over 36,000 new connections made.
    This was the first time we ran three EA Globals in one year since 2017, and we only had ~1200 attendees total across all three of those events.
    We hosted and recorded lots of new content, a substantial amount of which is located on our YouTube channel.
    This was the first time trialing out a EA conference in D.C. of any kind. We generally received positive feedback about this event from attendees and stakeholders.
    Plans for 2023
    We’re reducing our spending in a lot of ways, most significantly by cutting some meals and the majority of travel grants, which we expect may somewhat reduce the overall ratings of our events. Please note that this is a fairly dynamic situation and we may update our spending plans as our financial situation changes.
    We’re doing three EA Globals in 2023, in the Bay Area and London again, and with our US east coast event in Boston rather than D.C. As well as EA Globals, there are also several upcoming EAGx events, check out the full list of confirmed and provisional events below.
    EA Global: Bay Area | 24–26 February
    EAGxCambridge | 17–19 March
    EAGxNordics | 21–23 April
    EA Global: London | 19–21 May
    EAGxWarsaw | 9–11 June [provisional]
    EAGxNYC | July / August [provisional]
    EAGxBerlin | Early September [provisional]
    EAGxAustralia | Late September [provisional]
    EA Global: Boston | Oct 27–Oct 29
    EAGxVirtual | November [provisional]
    We’re aiming to have similar-sized conferences, though with the reduction in travel grants we expect the events to perhaps be a little smaller, maybe around 1000 people per EA Global.
    We recently completed a hiring round and now have ~4 FTEs working on the EA Global team.
    We’ve recently revamped our website and incorporated it into effectivealtruism.org — see here.
    We’ve switched over our backend systems from Zoho to Salesforce. This will help us integrate better with the rest of CEA’s products, and will hopefully create a smoother front and backend that’s better suited to our users. (Note that the switchover itself has been somewhat buggy, but we are clearing these up and hope to have minimal issues moving forwards.)
    We’re also trialing a referral system for applications, where we’ve given a select number of advisors working in EA community building the ability to admit people to the conference. If this goes well we may expand this program next year.
    Growth areas
    Food got generally negative reviews in 2022:
    Food is a notoriously hard area to get right and quality can vary a lot between venues, and we often have little or no choice between catering options.
    We’ve explored ways to improve the food quality, including hiring a catering consultant, but a lot of these options are cost prohibitive, and realistically we expect food quality to continue to be an issue moving forwards.
    Swapcard (our event application app) also got generally negative reviews in 2022:
    We explored and tested several competitor apps, though none of them seem better than Swapcard.
    We explored working with external developers to build our own event networking app, but eventually concluded that this would be too costly in terms of both time and money.
    We’ve been working with Swapcard to roll out new featur...
    6 min
  • EA - Taking a leave of absence from Open Philanthropy to work on AI safety by Holden Karnofsky
    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: Taking a leave of absence from Open Philanthropy to work on AI safety, published by Holden Karnofsky on February 23, 2023 on The Effective Altruism Forum.
    I’m planning a leave of absence (aiming for around 3 months and potentially more) from Open Philanthropy, starting on March 8, to explore working directly on AI safety.
    I have a few different interventions I might explore. The first I explore will be AI safety standards: documented expectations (enforced via self-regulation at first, and potentially government regulation later) that AI labs won’t build and deploy systems that pose too much risk to the world, as evaluated by a systematic evaluation regime. (More here.) There’s significant interest from some AI labs in self-regulating via safety standards, and I want to see whether I can help with the work ARC and others are doing to hammer out standards that are both protective and practical - to the point where major AI labs are likely to sign on.
    During my leave, Alexander Berger will serve as sole CEO of Open Philanthropy (as he did during my parental leave in 2021).
    Depending on how things play out, I may end up working directly on AI safety full-time. Open Philanthropy will remain my employer for at least the start of my leave, but I’ll join or start another organization if I go full-time.
    The reasons I’m doing this:
    First, I’m very concerned about the possibility that transformative AI could be developed soon (possibly even within the decade - I don’t think this is >50% likely, but it seems too likely for my comfort). I want to be as helpful as possible, and I think the way to do this might be via working on AI safety directly rather than grantmaking.
    Second, as a general matter, I’ve always aspired to help build multiple organizations rather than running one indefinitely. I think the former is a better fit for my talents and interests.
    At both organizations I’ve co-founded (GiveWell and Open Philanthropy), I’ve had a goal from day one of helping to build an organization that can be great without me - and then moving on to build something else.
    I think this went well with GiveWell thanks to Elie Hassenfeld’s leadership. I hope Open Philanthropy can go well under Alexander’s leadership.
    Trying to get to that point has been a long-term project. Alexander, Cari, Dustin and I have been actively discussing the path to Open Philanthropy running without me since 2018.1 Our mid-2021 promotion of Alexander to co-CEO was a major step in this direction (putting him in charge of more than half of the organization’s employees and giving), and this is another step, which we’ve been discussing and preparing for for over a year (and announced internally at Open Philanthropy on January 20).
    I’ve become increasingly excited about various interventions to reduce AI risk, such as working on safety standards. I’m looking forward to experimenting with focusing my energy on AI safety.
    Footnotes
    This was only a year after Open Philanthropy became a separate organization, but it was several years after Open Philanthropy started as part of GiveWell under the title “GiveWell Labs.” ↩
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • EA - EA is too New & Important to Schism by Wil Perkins
    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: EA is too New & Important to Schism, published by Wil Perkins on February 23, 2023 on The Effective Altruism Forum.
    As many of us have seen there has recently been a surge in discourse around people in the community with different views. Many of this underlying tension has only been brought about by large scandals that have broken in the last 6 months or so.
    I've seen a few people using language which, to me, seems schismatic. Discussing how there are two distinct and incompatible groups within EA, being shocked/hurt/feeling rejected by the movement, etc. I'd like to urge us to try and find reconciliation if possible.
    Influential Movements avoid Early Schisms
    If you look through history at any major religious/political/social movements, most of them avoid having early schisms, or if they do, it creates significant issues and tension. It seems optimal to let movements develop loosely over time and become more diverse, before starting to draw hard lines between what "is" a part of the in group and what isn't.
    For instance, early Christianity had some schisms, but nothing major until the Council of Nicea in 325 A.D. This meant that Christianity could consolidate power/followers for centuries before actively breaking up into different groups.
    Another parallel is the infamous Sunni-Shia split in Islam, which caused massive amounts of bloodshed and still continues to this day. This schism still echos today, for instance with the civil war in Syria.
    For a more modern example, look at the New Atheism Movement which in many ways attracted similar people to EA. Relatively early on in the movement, in fact right as the movement gained popular awareness (similar to the moment right now in EA) many prominent folks in New Atheism advocated for New Atheism Plus. This was essentially an attempt to schism the movement along cultural / social justice lines, which quickly eroded the cohesion of the movement and ultimately contributed to its massive decline in relevance.
    Effective Altruism as a movement is relatively brand new - we can't afford major schisms or we may not continue as a relevant cultural force in 10-20 years.
    Getting Movement Building Right Matters
    Something which I think is sometimes lost in community building discussions is that the stakes we're playing for are extremely high. My motivation to join EA was primarily because I saw major problems in the world, and people that were extremely dedicated to solving them. We are playing for the future, for the survival of the human race. We can't afford to let relatively petty squabbles divide us too much!
    Especially with advances in AGI, I know many people in the movement are more worried than ever that we will experience significant shifts via technology over the coming decades. Some have pointed out the possibility of Value Lock-in, or that as we rapidly increase our power our values may become stagnant, especially if for instance an AGI is controlled by a group with strong, anti-pluralistic values.
    Overall I hope to advocate for the idea of reconciliation within EA. We should work to disentangle our feelings from the future of the movement, and try to discuss how to have the most impact as we grow. My vote is that having a major schism is one of the worst things we could do for our impact - and is a common failure mode we should strive to avoid.
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    4 min
  • LW - Covid 2/23/23: Your Best Possible Situation by Zvi
    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: Covid 2/23/23: Your Best Possible Situation, published by Zvi on February 23, 2023 on LessWrong.
    No news is (often) good news.
    At old Magic: The Gathering tournaments, judge Tony Parodi would often tell us, ‘if your opponent did not show up, that is your best possible situation.’
    Every week, when I set out to write the Covid update, I held out the hope that at some point, perhaps soon, you would never read one of these again. There would be nothing to report. That the reports would no longer help anyone. I could go think about and write about something else.
    Today is that day. I had to go see about. well, to start off, the possibility of AI destroying all value in the universe. Not my desired first pick. Once again, I much prefer the worlds where over the next weeks, months and years I get to deep dive into other very different aspects of the world instead.
    It is still a joyous day. After three years, the weekly Covid posts are over.
    From this point forward, I am no longer going to actively seek out Covid information. I am not going to check my Covid Twitter list.
    I will continue to compile what Covid and related information I come still across, although with a much higher bar for inclusion going forward. If it seems worth its own post from time to time, I’ll do that. If not, I won’t. Unless something changes a lot, that will be a lot less common than weekly.
    We have normality. Cherish it.
    You’ll miss it when it’s gone.
    Executive Summary
    This will be the last weekly Covid post unless things change radically.
    We have normality. I repeat, we have normality.
    Anything you still can’t cope with is therefore your own problem.
    Let’s run the numbers.
    The Numbers
    Predictions
    Predictions from Last Week: 210k cases (-6%) and 2,625 deaths (-7%).
    Results: 210k cases (-6%) and 2,396 deaths (-15%)
    Predictions for Next Week: No more formal predictions. Expect continued slow declines in underlying numbers for a while.
    Arizona reported 23k cases, which has to be a backlog dump, so I cut them down to a high but plausible 4k. Colorado reported negative deaths, so I changed that to zero.
    I entirely forgot about Presidents Day, which makes the case number here unexpectedly high, and largely accounts for the deaths result. Overall slightly disappointing given the holiday.
    Deaths
    Cases
    Physical World Modeling
    I was alerted to quite the case of modeling: Metaculus has been successfully working with the Virginia state government to help them make better decisions. They are currently running the Keep Virginia Safe II contest with a $20k prize pool, which where the link goes. I have been informed that such information is actually listed to and used in real decision making, which makes it exciting to consider participating. Perhaps this will even spread to additional jurisdictions. You never know.
    Bloom Lab goes over the long term history of pandemics, especially flu, and speculates on what to expect from Covid going forward – most of the thread is from January but seems fitting for the final update. The last note is new, suggesting that Omicron might not actually be less virulent after all, with the difference in outcomes being due to immunity – older adults with no prior infection and no vaccination were found to have similar mortality rates. China would presumably tell us a lot, if we had any data we could rely upon.
    Metastudy confirms that yes, Covid infections are protective against future Covid protections even under Omicron.
    In Other Covid News
    State Senator in Idaho introduces a bill that makes providing or administering any mRNA vaccine a crime. I have learned not to be all that concerned by ‘politician introduces terrible bill,’ almost none of them go anywhere. Still, wow.
    A fun note: Bret Stephen says in his column that the conclusions of new study on the efficacy of masks were unambi...
    9 min
  • AF - EIS XII: Summary by Stephen Casper
    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: EIS XII: Summary, published by Stephen Casper on February 23, 2023 on The AI Alignment Forum.
    Part 12 of 12 in the Engineer’s Interpretability Sequence.
    TAISIC = “the AI safety interpretability community”
    MI = “mechanistic interpretability”
    There might be some addenda later, but for now, this is the final post in The Engineer’s Interpretability Sequence. I hope you have found it interesting and have gotten some useful ideas. I will always be happy to talk to people about the topics from this sequence in the comments or via email. For now, the last thing I will do is offer a summary of key points post by post :)
    A Prequel: Toward Transparent AI: A Survey on Interpreting the Inner Structures of Deep Neural Networks (Räuker et al., 2022)
    A survey of over 300 works on inner interpretability from an AI safety perspective.
    All opinions in this sequence, however, are my own and not necessarily those of coauthors or other affiliates.
    EIS I: Intro
    Lots of interpretability research exists, and the field is still rapidly growing.
    Most of it is not very productive, and there is a significant gap between the research and practice. Interpretability tools aren't used much by engineers working on real alignment problems.
    If one of our main goals for interpretability research is to help us with aligning highly intelligent AI systems in high stakes settings, we should be working on tools that are more engineering-relevant.
    EIS II: What is “Interpretability”?
    This post introduced a parable about two researchers trying to solve a problem.
    The moral of the story is that we should not privilege difficult or interesting methods over easy and simple ones. It is key not to grade different tools on different curves.
    From an engineer’s perspective, the term “interpretability” isn’t that useful.
    Whatever we call “interpretability” tools are entirely fungible with other techniques related to describing, evaluating, debugging, etc. in models.
    Mechanistic approaches to interpretability are not uniquely important for AI safety. MI tools have the potential to help identify and fix deceptive alignment failures, but...
    There are many non-deceptive ways AI could go wrong.
    MI is not uniquely useful for fixing deceptive alignment and especially not uniquely useful for fixing non-deceptive alignment failures.
    EIS III Broad Critiques of Interpretability Research
    There is a growing consensus that interpretability research is generally not very productive or engineering relevant.
    There is also a growing consensus that better evaluation is needed. A lack of good evaluation methods may be the biggest challenge facing the interpretability research community.
    There are three types of evaluation.
    Intuition + pontification --> inadequate
    Weak/ad-hoc --> still not enough
    Based on engineering-relevant tasks --> what is needed
    This can be based on one of three things
    Making novel predictions about how a model will handle interesting inputs.
    Controlling what a system does by guiding edits to it.
    Abandoning a system that does a nontrivial task and replacing it with a simpler reverse-engineered alternative
    Other common limitations of existing work
    Poor scaling
    Relying too much on humans in the loop
    Failing to study combinations of tools
    A lack of practical applications with real-world systems
    EIS IV: A Spotlight on Feature Attribution/Saliency
    Feature attribution/saliency methods are very common but unlikely to be very important from an engineering perspective.
    These methods tend to be poorly evaluated, and when they have been subjected to task-based evaluation, they have not tended to fare well.
    These methods just aren’t equipped to directly be very useful even when they work. They require scrutinizing samples from some data distribution. So the exact same things that feature attribution/saliency methods are equipped t...
    12 min
  • LW - Full Transcript: Eliezer Yudkowsky on the Bankless podcast by remember
    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: Full Transcript: Eliezer Yudkowsky on the Bankless podcast, published by remember on February 23, 2023 on LessWrong.
    This podcast has gotten a lot of traction, so we're posting a full transcript of it, lightly edited with ads removed, for those who prefer reading over audio. Eliezer Yudkowsky: I think that we are hearing the last winds start to blow, the fabric of reality start to fray. This thing alone cannot end the world, but I think that probably some of the vast quantities of money being blindly and helplessly piled into here are going to end up actually accomplishing something.
    Ryan Sean Adams: Welcome to Bankless, where we explore the frontier of internet money and internet finance. This is how to get started, how to get better, how to front run the opportunity. This is Ryan Sean Adams. I'm here with David Hoffman, and we're here to help you become more bankless. Okay, guys, we wanted to do an episode on AI at Bankless, but I feel like David...
    David: Got what we asked for.
    Ryan: We accidentally waded into the deep end of the pool here. And I think before we get into this episode, it probably warrants a few comments. I'm going to say a few things I'd like to hear from you too. But one thing I want to tell the listener is, don't listen to this episode if you're not ready for an existential crisis. Okay? I'm kind of serious about this. I'm leaving this episode shaken. And I don't say that lightly. In fact, David, I think you and I will have some things to discuss in the debrief as far as how this impacted you. But this was an impactful one. It sort of hit me during the recording, and I didn't know fully how to react. I honestly am coming out of this episode wanting to refute some of the claims made in this episode by our guest, Eliezer Yudkowsky, who makes the claim that humanity is on the cusp of developing an AI that's going to destroy us, and that there's really not much we can do to stop it.
    David: There's no way around it.
    Ryan: Yeah. I have a lot of respect for this guest. Let me say that. So it's not as if I have some sort of big brain technical disagreement here. In fact, I don't even know enough to fully disagree with anything he's saying. But the conclusion is so dire and so existentially heavy that I'm worried about it impacting you, listener, if we don't give you this warning going in. I also feel like, David, as interviewers, maybe we could have done a better job. I'll say this on behalf of myself. Sometimes I peppered him with a lot of questions in one fell swoop, and he was probably only ready to synthesize one at a time. I also feel like we got caught flat-footed at times. I wasn't expecting his answers to be so frank and so dire, David. It was just a rift of hope. I appreciated very much the honesty, as we always do on Bankless, but I appreciated it almost in the way that a patient might appreciate the honesty of their doctor telling them that their illness is terminal.
    It's still really heavy news, isn't it?
    So that is the context going into this episode. I will say one thing. In good news, for our failings as interviewers in this episode, they might be remedied because at the end of this episode, after we finished with Hit the Record Button to Stop Recording, Eliezer said, he'd be willing to provide an additional Q&A episode with the Bankless community. So if you guys have questions, and if there's sufficient interest for Eliezer to answer, tweet us to express that interest. Hit us in Discord. Get those messages over to us and let us know if you have some follow-up questions. He said, if there's enough interest in the crypto community, I'll say he'd be willing to come on and do another episode with follow-up Q&A. Maybe even a Vitalik and Eliezer episode is in store. That's a possibility that we threw to him. We've not talked to Vitalik about t...
    1 hr 19 min
  • EA - How can we improve discussions on the Forum? by Lizka
    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: How can we improve discussions on the Forum?, published by Lizka on February 23, 2023 on The Effective Altruism Forum.
    I’d like to run a very rough survey to get a better sense of:
    How you feel about the Frontpage change we’re currently testing
    What changes to the site — how it's set up and organized — you think could help us have discussions better
    What conversations you'd like to see on the Forum
    And more
    Give us your input
    Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.
    1 min