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: Data Taxation: A Proposal for Slowing Down AGI Progress, published by Per Ivar Friborg on April 11, 2023 on The Effective Altruism Forum.

Co-authored by Sammet, Joshua P. S. and Wale, William.

Foreword

This report was written for the Policies for slowing down progress towards artificial general intelligence (AGI) case of the AI governance hackathon organized by Apart Research.

We are keen on receiving your thoughts and feedback on this proposal. We are considering publishing this on a more public platform such as arXiv, and therefore would love for you to point out potential issues and shortcomings with our proposal that we should further address in our article. If you think there are parts that need to be flushed out more to be understandable for the reader or any other things we should include to round it up, we are more than happy to hear your comments.

Introduction

In this paper, we propose a tax that affects the training of any model of sufficient size, and a concrete formula for implementing it. We explain why we think our framework is robust, future-proof and applicable in practice. We also give some concrete examples of how the formula would apply to current models, and demonstrate that it would heavily disincentivize work on ML models that could develop AGI-like capabilities, but not other useful narrow AI work that does not pose existential risks.

Currently, the most promising path towards AGI involves increasingly big networks with billions of parameters trained with huge amounts of text data. The most famous example being GPT-3, whose 175 billion parameters were trained on over 45 TB of text data. The size of this data is what sets apart LLMs from both more narrow AI models developed before and classical high-performance computing. Most likely, any development of general or even humanoid AI will require large swathes of data, as the human body gathers 11 million bits per second (around 120 GB per day) to train its approx. 100 billion neurons. Therefore, tackling the data usage of these models could be a promising approach to slowing down the progress of the development of new and more capable general AIs, without harming the development of models that pose no AGI risk.

The proposal aims to slow down the progress towards AGI and mitigate the associated existential risks. The funds collected through data taxation can be used to support broader societal goals, such as redistribution of wealth and investments in AI safety research. We also discuss how the proposal plays well with other current criticisms of the relationship between AI and copyright, and persons and their personal data, and how it could consequently levy those social currents for more widespread support.

The Data Taxation Mechanism

A challenge in devising an effective data tax formula lies in differentially disincentivizing the development of models that could lead to AGI without hindering the progress of other useful and narrow ML technologies. A simplistic approach, such as imposing a flat fee per byte of training data used could inadvertently discourage beneficial research that poses no AGI risk.

For instance, a study aimed at identifying the most common words in the English language across the entire internet would become prohibitively expensive under the naïve flat fee proposal, despite posing zero danger in terms of AGI development. Similarly, ML applications in medical imaging or genomics, which often rely on vast datasets, would also be adversely affected, even though they do not contribute to AGI risk.

To overcome the challenge of differentially disincentivizing AGI development without impeding progress in other beneficial narrow AI applications, we propose a data tax formula that incorporates not only the amount of training data used but also the number of parameters being updated du...

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: Legal Impact for Chickens hiring Experienced Litigator, published by alene on April 12, 2023 on The Effective Altruism Forum.

Hi Everyone!!

I wanted to let you know that Legal Impact for Chickens is seeking an experienced litigator!

Please share widely. <3

Thank you so much!!

Legal Impact for Chickens seeks an

EXPERIENCED LITIGATOR.

We’re looking for the next attorney to help build our nonprofit and fight for animals.

Want to join our team on the ground floor?

About us:

Legal Impact for Chickens (LIC) is a 501(c)(3) litigation nonprofit. We work to protect farmed animals.

LIC’s first case, a shareholder derivative suit against Costco executives for chicken neglect, appeared in The Washington Post, Fox Business, CNN Business, and beyond.

Now, we’re looking for our next hire—an entrepreneurial litigator to help fight for animals!

About you:

3+ years of litigation experience

JD from an accredited law school

Licensed and in good standing with a state bar

Excellent analytical, writing, and verbal-communication skills

Zealous, creative, enthusiastic litigator

Passion for helping farmed animals

Interest in entering a startup nonprofit on the ground floor, and helping to build something

Willingness to do all types of nonprofit startup work, beyond just litigation

Strong work ethic and initiative

Kind to our fellow humans, and excited about creating a welcoming, inclusive team

About the role:

You will be an integral part of LIC. You’ll help shape our organization’s future.

Your role will be a combination of (1) designing and pursuing creative impact litigation for animals, and (2) helping with everything else we need to do, to run this new nonprofit!

Since this is such a small organization, you’ll wear many hats: Sometimes you may wear a law-firm partner’s hat, making litigation strategy decisions or covering a hearing on your own. Sometimes you’ll wear an associate’s hat, analyzing complex and novel legal issues. Sometimes you’ll pitch in on administrative tasks, making sure a brief gets filed properly or formatting a table of authorities. Sometimes you’ll wear a start-up founder’s hat, helping plan the number of employees we need, or representing LIC at conferences. We can only promise it won’t be dull!

This job offers tremendous opportunity for advancement, in the form of helping to lead LIC as we grow. The hope is for you to become an indispensable, long-time member of our new team.

Commitment: Full time

Location and travel: This is a remote, U.S.-based position. You must be available to travel for work as needed, since we will litigate all over the country. (Disabilities will be accommodated!)

Reports to: Alene Anello, LIC’s president

Salary: $82,000–$100,000 depending on experience

Benefits: Health insurance, 401(k), flexible schedule

One more thing!

LIC is an equal opportunity employer. Women and people of color are strongly encouraged to apply. Applicants will receive consideration for employment without regard to race, religion, gender, sexual orientation, national origin, disability, age, or veteran status.

To Apply:

To apply, please email your cover letter, resume, writing sample, and three references, all combined as one PDF, to [email protected].

Thank you for your time and your compassion!

Sincerely,

Legal Impact for Chickens

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: Trans Rescue’s operations in Uganda: high impact giving opportunity, published by David D on April 11, 2023 on The Effective Altruism Forum.

Synopsis: Trans Rescue uses their existing experience and infrastructure for moving African trans people to safety to help LGBTQ+ people of all sorts escape genocide in Uganda, costing an estimated €150 to move a person to safety outside the country and €1257 in housing costs and other support to help them become financially self sufficient in their new location.

You might have heard about Uganda’s new laws and crackdown targeting LGBTQ+ people, which began in March. “Homosexual activity” has been illegal in Uganda for a long time, but under this bill, people who even say that they are LGBT+ or “promote homosexuality” (such as advocating for LGBTQ+ people’s rights, or writing a positive or neutral article about homosexuality) face criminal charges and imprisonment. Renting living space to a gay person or conducting a same-sex marriage ceremony would also be criminalized with prison sentences of up to 10 years.

The president of Uganda has not yet signed the bill, but the homophobic fervor around it is already wreaking havoc in LGBTQ+ Ugandans’ lives. Many of Trans Rescue’s passengers were evicted by their landlords without any warning or opportunity to retrieve their things. In one particularly violent example, a landlord was convinced by a local preacher that his two trans tenants were evil and set fire to their belongings while they weren’t home, burning down his building in the process. Existing shelters for LGBTQ+ people also face eviction. Violence and sexual assault is becoming more frequent (cw violence, rape, police brutality: source).

International NGOs have been slow to respond. In an article on Trans Writes, Trans Rescue treasurer Jenny List writes that their passengers haven’t seen any other international organizations working to protect or evacuate LGBTQ+ Ugandans, though some organizations say they have plans in the works.

“Of course, we’ve asked around to find out what’s being done by those organisations, and the answer has come back from several quarters that things are in motion, but under the radar. We’re told that too public a move might cause them to be accused by the Ugandan government of being colonialist, and we understand that. We’re happy to hear that so much is being done, we really are.”

“Unfortunately, the fact remains that the people on the ground aren’t seeing it. Things they can’t see are of little use to them, when what they need is to escape an angry mob or a police manhunt.”

Trans Rescue was unusually well positioned to help. They’ve been helping trans people escape danger, especially in Africa and the Middle East, since 2021, and several of their board members did similar work in the organization’s previous incarnation as Trans Emigrate. In addition to their experience planning travel for people who face extra scrutiny due to their country of origin, they operate a trans safe haven in neighboring Kenya called Eden House. In light of the danger that queer Ugandans of all sorts are facing right now, they are providing transportation and shelter for LGBTQ+ people of all sorts fleeing Uganda.

Effective Altruists often avoid donating to acute crisis that make the news. Newsworthy events are often relatively less underfunded, especially when they occur in the western world. The difficult logistics of providing aid for events like natural disasters can also be an obstacle. Uganda’s proposed LGBTQ+ bill has received some media coverage, but the people who have the ability to donate and would consider the people affected “one of us” - LGBTQ+ people in wealthier nations - are focused on the rise in transphobia in the US and UK right now. Trans Rescue already has a presence in the region in Eden House, 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: Where's the foom?, published by Fergus Fettes on April 11, 2023 on LessWrong.

"The first catastrophe mechanism seriously considered seems to have been the possibility, raised in the 1940s at Los Alamos before the first atomic bomb tests, that fission or fusion bombs might ignite the atmosphere or oceans in an unstoppable chain reaction."

This is not our first rodeo. We have done risk assessments before. The best reference-class examples I could find were the bomb, vacuum decay, killer strangelets, and LHC black holes (all covered in ).

I was looking for a few days, but didn't complete my search, but I decided to publish this note as now Tyler Cowen is asking too: "Which is the leading attempt to publish a canonical paper on AGI risk, in a leading science journal, refereed of course. The paper should have a formal model or calibration of some sort, working toward the conclusion of showing that the relevant risk is actually fairly high. Is there any such thing?"

The three papers people replied with were: - Is Power-Seeking AI an Existential Risk?- The Alignment Problem from a Deep Learning Perspective - Unsolved Problems in ML Safety

Places I was looking so far:- The list of references for that paper- The references for the Muehlhauser and Salamon intelligence explosion paper- The Sandburg review of singularities and related papers (these are quite close to passing muster I think)

Places I wanted to look further:

Papers by Yampolsky, aka- Papers mentioned in there by Schmidhuber (haven't gotten around to this)- I haven't thoroughly reviewed Intelligence Explosion Microeconomics, maybe this is the closest thing to fulfilling the criteria?

But if there is something concrete in eg. some papers by Yampolsky and Schmidhuber, why hasn't anyone fleshed it out in more detail?

For all the time people spend working on 'solutions' to the alignment problem, there still seems to be a serious lack of 'descriptions' of the alignment problem. Maybe the idea is, if you found the latter you would automatically have the former?

I feel like something built on top of Intelligence Explosion Microeconomics and the Orthogonality Thesis could be super useful and convincing to a lot of people. And I think people like TC are perfectly justified in questioning why it doesn't exist, for all the millions of words collectively written on this topic on LW etc.

I feel like a good simple model of this would be much more useful than another ten blog posts about the pros and cons of bombing data centers. This is the kind of thing that governments and lawyers and insurance firms can sink their teeth into.

Where's the foom?

Edit: Forgot to mention clippy. Clippy is in many ways the most convincing of all the things I read looking for this, and whenever I find myself getting skeptical of foom I read it again. Maybe an summary of the mechanisms described in there would be a step in the right direction?

A critical look at risk assessments for global catastrophes

List

Intelligence Explosion: Evidence and Import

An Overview of Models of Technological Singularity

From Seed AI to Technological Singularity via Recursively Self-Improving Software

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: Evolution provides no evidence for the sharp left turn, published by Quintin Pope on April 11, 2023 on The Effective Altruism Forum.

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 "aiming for convergence on truth", published by gjm on April 11, 2023 on LessWrong.

Background

Duncan Sabien wrote a list of proposed "basics of rational discourse" guidelines. Zack M Davis disagrees with (his interpretation of) one of the guidelines. I think the question is interesting and don't feel that those two posts and their comments resolve it. (Spoiler: I largely agree with Duncan on this.)

So, Duncan says that we should

aim for convergence on truth, and behave as if your interlocutors are also aiming for convergence on truth

and that we should care whether an interlocutor is

present in good faith and genuinely trying to cooperate.

Zack, on the other hand, holds (as I understand him) that

we should aim for truth and not particularly care about convergence

in cases where we have something valuable to contribute, it should be precisely because there is divergence between what others think and our view of the truth

a good frame for thinking about this is that debates are like prediction markets

it shouldn't matter much whether someone else is "in good faith" or "trying to cooperate"; the merits and defects of their arguments are whatever they are, regardless of intentions, and we should be responding to those

talk of "cooperation" and "defection" implies a Prisoner's Dilemma situation, and attempted rational discourse doesn't generally have the properties of a Prisoner's Dilemma.

I will argue

that we should aim for truth rather than for convergence as such

that I think Zack has misunderstood what Duncan means by aiming for convergence

that what Duncan means by "aiming for convergence" is a reasonable goal

that debates are importantly unlike prediction markets (and even prediction markets don't necessarily have quite the incentives Zack assumes)

that there are specific forms of not-aiming-for-convergence that are more common than Zack's "aiming for divergence"

and they are usually harmful when they occur

and when they do, it often becomes necessary to adopt modes of discourse that attend explicitly to status, politeness, intentions, etc., which makes everything worse

and this failure mode is made less likely if (as Duncan proposes) we avoid those forms of not-aiming-for-convergence and assume until we get strong contrary evidence that others are also avoiding them

that while in principle we can just respond to others' arguments and ignore their (actual or perceived) intentions, in practice we can't and shouldn't

that informal talk of cooperation and defection doesn't have to imply a PD-like situation

and is reasonable in this context even though (I agree with Zack) what is going on isn't altogether PD-like.

So, let's begin.

Convergence

Suppose A and B are discussing something they disagree on: A thinks X and B thinks Y. Here are some goals A might have. I've given them brief but crude names for later reference. The names do not perfectly reflect the meanings.

WIN, by making A look clever and wise and good, and making B look stupid or ignorant or crazy or evil. (One way, but not the only one, to do this is by proving that X is right and Y is wrong.)

LEARN, ending up believing X if X is right and Y if Y is right.

TEACH, so that B ends up believing X if X is right and Y if Y is right.

CONVINCE, so that B ends up believing X.

EXPOUND, so that the audience ends up believing X.

AVOID CONFLICT, by backing off from any point at which A and B seem likely to disagree very sharply.

AGREE, by actually ending up believing the same thing whether right or wrong.

I think that when Duncan says we should "aim for convergence on truth" he means that A should aim to LEARN and TEACH, with the hope that if A and B pool their knowledge and understanding they will likely (1) both end up with truer views, (2) end up understanding one another better, and (3) since the truth is a singl...

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: Request to AGI organizations: Share your views on pausing AI progress, published by Akash on April 11, 2023 on The Effective Altruism Forum.

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: [Linkpost] 538 Politics Podcast on AI risk & politics, published by jackva on April 11, 2023 on The Effective Altruism Forum.

From about minute 32 of this Monday's 538 Podcast, Nate Silver & crew discuss AI politics and risk. As an outlet strongly aligned with "taking evidence and forecasting and political analysis seriously" I thought this was pretty interesting for a number of reasons, both in terms of arguments but also explicit discussion of EA community:

Nate Silver making the point that this is much more important than other issues, that a 5% ex risk would be a really big deal

Fairly detailed and somewhat accurate description of EA community

Insularity of EA/AI risk community and difficulty of translating this to wider public

Warning shots as the primary mechanism for actual risk being lower

Rationality / EA community mentioned as good at identifying important things

Difficulty of taking the issue fully seriously

In other podcast episodes Nate also mentioned he will cover AI risk in some detail in his upcoming book

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: Humans are not prepared to operate outside their moral training distribution, published by Prometheus on April 10, 2023 on LessWrong.

When people think about how to treat an AI, they probably normally default to the evil AI, that frowns on humans and desires to enslave them out of hate, or the nice android-type AI. A lot of writing in AI Safety has focused on debunking the Evil Skynet-Type AI, and not as much on the Nice Android-Type AI.

In the film “AI: Artificial Intelligence” (because, back then, general audiences actually didn’t know what AI stood for), the AIs are of a very convenient kind. They are human-like androids, with physical bodies, and overall exhibit various human characteristics. It is designed in such a way to be very obvious and easy for its early 2000s audience to understand that it is wrong for the humans to enslave and mistreat them. People simply have to look at the androids and notice that their treatment falls into moral atrocities they’ve seen before: they’re aware of how humans have treated marginalized groups of other humans before, have the perception that this is bad, and can easily apply the same rationale to the androids. Similarly, in the film District 9, aliens land on Earth, but are placed in a situation that conveniently makes their treatment look identical to the treatment of blacks in South Africa during apartheid.

Someone might see these sorts of situations in fiction, and think that they have an open mind that proactively sees moral dilemmas of the future, but they are wrong. All they are doing is taking their own normal moral training distribution and swapping in androids/aliens/etc. It’s easy to do, because it’s already been intentionally made to conveniently fit into situations that many humans already view to be bad. No change in mentality required.

Contrast this with Ender’s Game (Spoilers). Ender executes a series of tactics in what he believes to be a simulation. He’s played this kind of game before. It feels normal to him. But what he doesn’t know while playing is that it isn’t a simulation at all, but an actual extermination of an entire alien race. It didn’t feel consequential to him. It didn’t feel like the way you might imagine committing an atrocity would feel. Some will think that’s not fair to say, because Ender was tricked, but I think that’s missing the point. Reality doesn’t grant you massively consequential, and potentially horrific, actions that feel consequential and horrific.

Imagine those in 18th Century Britain, who bought sugar for their tea, despite it coming from plantation fields that killed tens of thousands of enslaved Africans. Most today would likely insist they would have refused to do such a thing if they had been alive back then. But we’re looking at their actions, the consequences, and can easily draw a line between the two. We’re not living in their bodies, where the tea and sugar will look so real, and the slavery on sugar plantations will look so vague and unreal. I imagine there might have been some sort of subconscious perception that they had which stated that highly consequential actions (good or bad) should feel consequential. That it should feel like a cruel man, brandishing a whip, looking down at the thousands of slaves living in torment. That it should not feel like putting a cube of sugar in your tea.

We’re at a point where it’s going to become less and less certain in determining the consciousness or sentience of an AI. Imagine if we develop an LLM that is sentient. Would shutting it down and wiping the memory feel like murder? What if it’s not the LLM itself that’s conscious, but the instances that are conscious. Then you might be creating a consciousness every time you start a session, and ending it every time you terminate the session, happening millions of times all over 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: How much funding does GiveWell expect to raise through 2025?, published by GiveWell on April 11, 2023 on The Effective Altruism Forum.

Summary

We're optimistic that GiveWell's funds raised will continue to increase in the long run. Over the next few years, we believe our annual funds raised are more likely to stay relatively constant, due to a decrease in expected funding from our largest donor, Open Philanthropy, offset by an expected increase in funding from our other donors.

This chart shows our latest forecasts for funds raised in millions of dollars:[1]

In November 2021, we wrote that we were anticipating rapid growth and aiming to influence $1 billion in 2025. Now, our best guess is that we'll raise between $400 million and $800 million in 2025 (for comparison, we raised around $600 million in 2022). As in the chart above, we now think it's possible but unlikely that we'll raise close to $1 billion in 2025, and we also think it's possible but unlikely that our funds raised in 2025 will be substantially lower (e.g. around $300 million) than they were in 2022.

We're excited about the impact we can have at any of those levels of funding, and we'll be continuing to direct as much funding as we can raise to the most cost-effective opportunities we can find.

Our forecasts are uncertain. We might be wrong about what the future will look like, just as our projections now are very different than they were in late 2021. We'll have better information as time goes on.

This change in projected funds raised means that:

We're funding-constrained: we believe that our research will yield more outstanding opportunities than we'll be able to fund over the next few years. Your donations can help fill those gaps.

Because it's valuable to maintain a stable cost-effectiveness bar, we may not spend down all the funds available to us in each year. Depending on how much funding we are able to direct and when it becomes available, we may smooth our spending over the next few years. Currently, we recommend funding to opportunities we believe to be at least 10 times as cost-effective as unconditional cash transfers ("10x cash").

We are increasing our emphasis on fundraising relative to past years and relative to our previous plans for 2023 to 2025 in order to increase the chances of us being able to fill additional cost-effective funding gaps.

In the rest of this post, we discuss:

Our updated forecasts

How we'll respond to shifting expectations

The uncertainty inherent in these projections

Our updated forecasts

When we wrote that we aimed to direct $1 billion annually by 2025, we were imagining a scenario in which we'd receive $500 million from Open Philanthropy and roughly $450 million from other donors.[2]

We now believe that:

Given our strong community of supporters and our increased fundraising efforts, funding from donors beyond Open Philanthropy will likely continue to increase. We're uncertain of the rate at which it'll increase, and it's always possible that funds raised will stall or decrease. Despite the economic downturn, our projections for funds raised among these donors haven't changed much. Our goal is to reach $500 million in funds raised from donors other than Open Philanthropy by 2025. We think this goal is ambitious but plausibly achievable with effort.

The level of funding we'll receive from Open Philanthropy in the future is very uncertain. There is a wide range of possibilities, but our median expectation is that funds from Open Philanthropy will taper down from a high of $350 million in 2022.

Previously, Open Philanthropy had tentatively planned to give $500 million in each of 2022 and 2023. We projected that level of funding out to 2025 as a best guess. In early 2022, Open Philanthropy revised its plans to give $350 million in 2022; it tentatively plans to give $250...