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: Subscripts for Probabilities, published by niplav on April 13, 2023 on LessWrong.

cross-posted from niplav.github.io

Gwern has wondered about a use-case for subscripts in hypertext. While they have settled on a specific use-case, namely years for citations, I propose a different one: reporting explicit probabilities.

Explicitely giving for probabilities in day-to-day English text is usually quite clunky: "I assign 35% to North Korea testing an intercontinental ballistic missile until the end of this year" reads far less smoothly than "I don't think North Korea will test an intercontinental ballistic missile this year".

And since subscripts are a solution in need of a problem, one can wonder how well those two fit together: Quite well, I claim.

In short, I propose to append probabilities in subscript after a statement using standard HTML subscript notation (or LATEX as a fallback if it's available), with the probability possibly also being a link to a relevant forecasting platform with the same question:

I think Donald Trump is going to be incarcerated before 203065%.

This is almost as readable as the sentence without the probability.

There are some complications with negations in sentences or multiple statements. For the most part, I'll simply avoid such cases ("Doctor, it hurts when I do this!" "Don't do that, then."), but if I had to, I'd solve the first problem by declaring that the probability applies to the literal meaning of the previous sentence, including all negations; the problem with multiple statements is solved by delimiters.

As an example for the different kinds of negation: "The train won't come more than 5 minutes late90%" would (arguendo) mean the same thing as "I don't think the train will come more than 5 minutes late90%" means the same as "The train will take more than 5 minutes to arrive10%" equivalent to "I assign 90% probability to the train arriving within the next 5 minutes".

With multiple statements, my favorite way of delimiting is currently half brackets: "I think ⸤it'll rain tomorrow⸥55%, but

⸤Tuesday is going to be sunny⸥80%, but I don't think

⸤your uncle is going to be happy about that⸥15%."

The probabilities in this context aren't quite

evidentials, but neither are they veridicals nor

miratives, I propose the world

"credal" for this category.

Enumerating Possible Notations

The exact place of insertion is subtle: In sentences with a single central statement, there are multiple locations one could place the probability.

After the verb related to belief: "I think55% it'll rain tomorrow."

Advantage: Close to the word relating to the belief (which could reflect the strength of belief in itself, using "guess"/"wager"/"think"/"believe").

Disadvantages:

Conflicts with assigning probabilities to multiple statements.

Puts visual clutter before the statement in question.

At the end of the statement: "I think it'll rain tomorrow55%."

Advantage: Allows assigning probabilities to simple statements ("It'll rain tomorrow55%") and to multiple statements (see below).

Disadvantage: If the probability is intended to contextualise the statement, this context is weaker if it is introduced after the statement in question.

At the subject of the sentence: "I55% think it'll rain tomorrow."

Advantage: This can be used to distinguish the beliefs of different people. "I55% think it'll rain tomorrow, but Cú Chulainn22% is skeptical about it."

Disadvantage: Putting the probability before the statement the probability is about feels quite unnatural.

This becomes trickier in sentences with multiple statements.

Probabilities after each subclaim: "I think it'll rain tomorrow55%, but Tuesday is going to be sunny80%, but I don't think your uncle is going to be happy about that15%.

Adding in delimiters to denote a specific subclaim the probability is about. I wonder whethe...

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: [US] NTIA: AI Accountability Policy Request for Comment, published by Kyle J. Lucchese on April 13, 2023 on The Effective Altruism Forum.

The Department of Commerce’s National Telecommunications and Information Administration (NTIA) has launched an inquiry into what policies will help businesses, government, and the public be able to trust that Artificial Intelligence (AI) systems work as claimed – and without causing harm.

In line with this, the NTIA announced today, April 13, 2023, a request for public comments on Artificial Intelligence (“AI”) system accountability measures and policies.

Summary:

The National Telecommunications and Information Administration (NTIA) hereby requests comments on Artificial Intelligence (“AI”) system accountability measures and policies. This request focuses on self-regulatory, regulatory, and other measures and policies that are designed to provide reliable evidence to external stakeholders—that is, to provide assurance—that AI systems are legal, effective, ethical, safe, and otherwise trustworthy. NTIA will rely on these comments, along with other public engagements on this topic, to draft and issue a report on AI accountability policy development, focusing especially on the AI assurance ecosystem.

NTIA is seeking input on what policies should shape the AI accountability ecosystem, including topics such as:

What kinds of data access is necessary to conduct audits and assessments

How can regulators and other actors incentivize and support credible assurance of AI systems along with other forms of accountability

What different approaches might be needed in different industry sectors—like employment or health care

If you have relevant knowledge regarding AI technical safety and/or governance, please consider submitting a comment. This is a notable opportunity to positively inform US policymaking.

You can find more information and formally submit your comments here.

Comments can be submitted as a known individual, on behalf of an organization, or anonymously.

The deadline to submit comments is June 12, 2023.

NTIA is the Executive Branch agency that is principally responsible for advising the President [of the United States] on telecommunications and information policy issues.

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: How altruistic perfectionism is self-defeating (Tim LeBon on The 80,000 Hours Podcast), published by 80000 Hours on April 12, 2023 on The Effective Altruism Forum.

Over at The 80,000 Hours Podcast we just published an interview that is likely to be of particular interest to people who identify as involved in the effective altruism community: Tim LeBon on how altruistic perfectionism is self-defeating.

You can click through for the audio, a full transcript and related links. Below is the episode summary and some key excerpts.

Episode summary

Being a good and successful person is core to your identity. You place great importance on meeting the high moral, professional, or academic standards you set yourself.

But inevitably, something goes wrong and you fail to meet that high bar. Now you feel terrible about yourself, and worry others are judging you for your failure. Feeling low and reflecting constantly on whether you're doing as much as you think you should makes it hard to focus and get things done. So now you're performing below a normal level, making you feel even more ashamed of yourself. Rinse and repeat.

This is the disastrous cycle today's guest, Tim LeBon — registered psychotherapist, accredited CBT therapist, life coach, and author of 365 Ways to Be More Stoic — has observed in many clients with a perfectionist mindset.

Tim has provided therapy to a number of 80,000 Hours readers — people who have found that the very high expectations they had set for themselves were holding them back. Because of our focus on "doing the most good you can," Tim thinks 80,000 Hours both attracts people with this style of thinking and then exacerbates it.

But Tim, having studied and written on moral philosophy, is sympathetic to the idea of helping others as much as possible, and is excited to help clients pursue that — sustainably — if it's their goal.

Tim has treated hundreds of clients with all sorts of mental health challenges. But in today's conversation, he shares the lessons he has learned working with people who take helping others so seriously that it has become burdensome and self-defeating — in particular, how clients can approach this challenge using the treatment he's most enthusiastic about: cognitive behavioural therapy.

As Tim stresses, perfectionism isn't the same as being perfect, or simply pursuing excellence. What's most distinctive about perfectionism is that a person's standards don't vary flexibly according to circumstance, meeting those standards without exception is key to their self-image, and they worry something terrible will happen if they fail to meet them.

It's a mindset most of us have seen in ourselves at some point, or have seen people we love struggle with.

Untreated, perfectionism might not cause problems for many years — it might even seem positive providing a source of motivation to work hard. But it's hard to feel truly happy and secure, and free to take risks, when we're just one failure away from our self-worth falling through the floor. And if someone slips into the positive feedback loop of shame described above, the end result can be depression and anxiety that's hard to shake.

But there's hope. Tim has seen clients make real progress on their perfectionism by using CBT techniques like exposure therapy. By doing things like experimenting with more flexible standards — for example, sending early drafts to your colleagues, even if it terrifies you — you can learn that things will be okay, even when you're not perfect.

In today's extensive conversation, Tim and Rob cover:

How perfectionism is different from the pursuit of excellence, scrupulosity, or an OCD personality

What leads people to adopt a perfectionist mindset

The pros and cons of perfectionism

How 80,000 Hours contributes to perfectionism among some readers and listeners, and w...

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: My experience attending and volunteering at an EAGx for the first time, published by Nayanika on April 13, 2023 on The Effective Altruism Forum.

TLDR: This post might be helpful for people who are about to volunteer or attend an EAGx for the first time.

This is a relatively late post (due to unforeseen circumstances) as the conference had taken place exactly two months earlier from this date.

Right from the shuttle service to the venue, till the last day of post-conference retreats, it's been one of a kind an experience at EAGxIndia, my first ever EAGx.

Volunteer application experience: I applied for a full-time volunteering position so that I could understand the maximum essence of co-working in the EA community. The organizing team had quite prudently decided to allocate me a part-time. Later during the conference, I couldn't help but wonder how insightful a decision it was; personalized for me as a first-time attendee. Of course, the organizing team had an idea of my background since I was already a part of the community (Indian Network for Impact, previously EA India). But I understood how meaningfully I was allowed to volunteer part-time and observe the conference for the rest.

The procedure involved: I went through a screening with an organizing team member. A cheerful face (Harriet Patterson) asked me about my aspirations for volunteering at the conference. Next, within a week I got an outcome of my screening and was selected as a part-time volunteer. Part-time volunteers were requested to arrive a day ahead of the main conference. The full-timers perhaps arrived 2 days ago. Volunteers were also invited to join an online training a week before the main event.

Arrival at the conference: As I landed at the location, (Jaipur, India) I recalled that it was the land of India's rich history. I cherished my presence on this land for the first time after reading about it in my school history book. The venue selection was spot on! Comes next, is the shuttle assistance. Although I didn't have a shuttle close to my arrival time, I communicated with the coordinator that I would take a cab. This coordinator took care of my safety (by asking for my live location) throughout my travel from the airport to the accommodation venue. I was also told that I would be reimbursed for my cab expense since I was a volunteer. With this, my immediate understanding was- You're greatly taken care of as a volunteer at an EAGx!

The coordinator surprisingly had a very friendly nature. This in turn unsurprisingly made him a good friend and the first point of contact (in case of any anomaly) for me and hopefully a few others at the conference. This helps especially when you're alone at a place for the first time. Came to know later that he was studying Law and had been exposed to EA for the first time.

About the accommodation: It was a decent arrangement. Shuttles were arranged for transportation purposes between the stay and the conference venue (both were about 10 mins apart).

First day of the conference: I somehow gulped something for breakfast as the training was about to start and attendees were to arrive by that evening. Unfortunately, I caught a bad cold. The organizers had arranged Covid test kits for everyone. I took a test and could continue being in the conference as the test result were 'negative'!

Volunteering begins: After a short training with Harriet, we were handed over to the leaders of our respective volunteering departments (e g. Logistics, Registration, Speaker Liaison). I was on the registration team. Our team lead, Ivan Burduk, took us for scouting around the venue. It was fun for sure. We were figuring out the locations ourselves so that we could promptly guide the attendees.Then, we started arranging the IDs for the attendees and speakers at the registration desk. But w...

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: Independence Dividends, published by jefftk on April 13, 2023 on LessWrong.

Our oldest is in third grade, and sings in the school chorus. It's after school on Wednesdays, which is fine. Except on half-day Wednesdays it's before school, which is less fine. Instead of needing to be to school at 8am she needs to be there at

7:15. She really likes chorus, and feels strongly that it's worth waking up for, but as the one getting up early with her, the tradeoff feels a bit different to me.

This morning, for the first time, when my alarm went off she was already at school. We had talked about how I thought she was ready to do this, and how I really didn't want to be waking up early when she didn't need me, and we agreed she would handle it. Which she did well! When I saw her briefly at school, dropping the middle child, she was very proud of herself.

Cultivating independence can be more work in the moment, but it's really great when the investment pays dividends.

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 AutoGPT, published by Zvi on April 13, 2023 on LessWrong.

The primary talk of the AI world recently is about AI agents (whether or not it includes the question of whether we can’t help but notice we are all going to die.)

The trigger for this was AutoGPT, now number one on GitHub, which allows you to turn GPT-4 (or GPT-3.5 for us clowns without proper access) into a prototype version of a self-directed agent.

We also have a paper out this week where a simple virtual world was created, populated by LLMs that were wrapped in code designed to make them simple agents, and then several days of activity were simulated, during which the AI inhabitants interacted, formed and executed plans, and it all seemed like the beginnings of a living and dynamic world. Game version hopefully coming soon.

How should we think about this? How worried should we be?

The Basics

I’ll reiterate the basics of what AutoGPT is, for those who need that, others can skip ahead. I talked briefly about this in AI#6 under the heading ‘Your AI Not an Agent? There, I Fixed It.’

AutoGPT was created by game designer Toran Bruce Richards.

I previously incorrectly understood it as having been created by a non-coding VC over the course of a few days. The VC instead coded the similar program BabyGPT, by having the idea for how to turn GPT-4 into an agent. The VC had GPT-4 write the code to make this happen, and also ‘write the paper’ associated with it.

The concept works like this:

AutoGPT uses GPT-4 to generate, prioritize and execute tasks, using plug-ins for internet browsing and other access. It uses outside memory to keep track of what it is doing and provide context, which lets it evaluate its situation, generate new tasks or self-correct, and add new tasks to the queue, which it then prioritizes.

This quickly rose to become #1 on GitHub and get lots of people super excited. People are excited, people are building it tools, there is a bitcoin wallet interaction available if you never liked your bitcoins. AI agents offer very obvious promise, both in terms of mundane utility via being able to create and execute multi-step plans to do your market research and anything else you might want, and in terms of potentially being a path to AGI and getting us all killed, either with GPT-4 or a future model.

As with all such new developments, we have people saying it was inevitable and they knew it would happen all along, and others that are surprised. We have people excited by future possibilities, others not impressed because the current versions haven’t done much. Some see the potential, others the potential for big trouble, others both.

Also as per standard procedure, we should expect rapid improvements over time, both in terms of usability and underlying capabilities. There are any number of obvious low-hanging-fruit improvements available.

An example is someone noting ‘you have to keep an eye on it to ensure it is not caught in a loop.’ That’s easy enough to fix.

A common complaint is lack of focus and tendency to end up distracted. Again, the obvious things have not been tried to mitigate this. We don’t know how effective they will be, but no doubt they will at least help somewhat.

Yes, But What Has Auto-GPT Actually Accomplished?

So far? Nothing, absolutely nothing, stupid, you so stupid.

You can say your ‘mind is blown’ by all the developments of the past 24 hours all you want over and over, it still does not net out into having accomplished much of anything.

That’s not quite fair.

Some people are reporting it has been useful as a way of generating market research, that it is good at this and faster than using the traditional GPT-4 or Bing interfaces. I saw a claim that it can have ‘complex conversations with customers,’ or a few other vague similar claims that weren’t backed up by ‘we are totally actual...

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: Boundaries-based security and AI safety approaches, published by Allison Duettmann on April 12, 2023 on LessWrong.

[This part 3 of a 5 part sequence on security and cryptography areas relevant for AI safety, published and linked here a few days apart.]

There is a long-standing computer security approach that may have directly useful parallels to a recent strand of AI safety work. Both rely on the notion of ‘respecting boundaries’. Since the computer security approach has been around for a while, there may be useful lessons to draw from it for the more recent AI safety work. Let's start with AI safety, then introduce the security approach, and finish with parallels.

AI safety: Boundaries in The Open Agency Model and the Acausal Society

In a recent LW post, The Open Agency Model, Eric Drexler expands on his previous CAIS work by introducing ‘open agencies’ as a model for AI safety. In contrast to the often proposed opaque or unitary agents, “agencies rely on generative models that produce diverse proposals, diverse critics that help select proposals, and diverse agents that implement proposed actions to accomplish tasks”, subject to ongoing review and revision.

In An Open Agency Architecture for Safe Transformative AI, Davidad expands on Eric Drexler’s model, suggesting that, instead of optimizing, this model would ‘depessimize’ by reaching a world that has existential safety. So rather than a fully-fledged AGI-enforced optimization scenario that implements all principles CEV would endorse, this would be a more modest approach that relies on the notion of important boundaries (including those of human and AI entities) being respected.

What could it mean to respect the boundaries of human and AI entities? In Acausal Normalcy, Andrew Critch also discusses the notion of respecting boundaries with respect to coordination in an acausal society. He thinks it’s possible that an acausal society generally holds values related to respecting boundaries. He defines ‘boundaries’ as the approximate causal separation of regions, either in physical spaces (such as spacetime) or abstract spaces (such as cyberspace). Respecting them intuitively means relying on the consent of the entity on the other side of the boundary when interacting with them: only using causal channels that were endogenously opened.

His examples of currently used boundaries include a person's skin that separates the inside of their body from the outside, a fence around a family's yard that separates their place from neighbors, a firewall that separates the LAN and its users from the rest of the internet, and a sustained disassociation of social groups that separates the two groups. In his Boundaries Sequence, Andrew Critch continues to formally define the notions of boundaries to generalize them to very different intelligences.

If the concept of respecting boundaries is in fact universally salient across intelligences, then it may be possible to help AIs discover and respect the boundaries humans find important (and potentially vice versa).

Computer security: Boundaries in the Object Capabilities Approach

Pursuing a similar idea, in Skim the Manual, Christine Peterson, Mark S. Miller, and I reframe the AI alignment problem as a secure cooperation problem across human and AI entities. Throughout history, we developed norms for human cooperation that emphasize the importance of respecting physical boundaries, for instance to not inflict violence, and cognitive boundaries, for instance to rely on informed consent. We also developed approaches for computational cooperation that emphasize the importance of respecting boundaries in cyberspace. For instance, in object-capabilities-oriented programming, individual computing entities are encapsulated to prevent interference with the contents of other objects.

The fact that ...

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: Announcing Epoch’s dashboard of key trends and figures in Machine Learning, published by Jaime Sevilla on April 13, 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: Announcing a new animal advocacy podcast: How I Learned to Love Shrimp, published by James Özden on April 13, 2023 on The Effective Altruism Forum.

Calling all animal-interested folks! Excited to share that we've launched a new animal-focused podcast series. How I Learned to Love Shrimp is a podcast about promising ways to help animals and build the animal advocacy movement. We showcase interesting and exciting ideas within animal advocacy and will release bi-weekly, hour-long interviews with people who are working on these projects.

We start the series with an introductory episode with Amy and I discussing why we wanted to start this podcast, the topics we want to cover, and some of our own views on various thorny animal advocacy topics. Our first proper episode is with Dave Coman-Hidy, former President of The Humane League. In this, we discuss the age-old debate of welfare vs abolitionism, the pros and cons of measurability, as well as promising strategies Dave is keen to see more of within animal advocacy.

You can check out the episodes across all major providers (e.g. Spotify, Google Podcasts & Apple Podcasts) and also access them on our website.

Please let us know what you think, give us guest recommendations and share with anyone who you think could be interested to hear. You can contact us via Twitter, our website or email at [email protected]. Enjoy!

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: Announcing Epoch’s dashboard of key trends and figures in Machine Learning, published by Jaime Sevilla on April 13, 2023 on The AI Alignment Forum.

Developments in Machine Learning have been happening extraordinarily fast, and as their impacts become increasingly visible, it becomes ever more important to develop a quantitative understanding of these changes. However, relevant data has thus far been scattered across multiple papers, has required expertise to gather accurately, or has been otherwise hard to obtain.

Given this, Epoch is thrilled to announce the launch of our new dashboard, which covers key numbers and figures from our research to help understand the present and future of Machine Learning. This includes:

Training compute requirements

Model size, measured by the number of trainable parameters

The availability and use of data for training

Trends in hardware efficiency

Algorithmic improvements for achieving better performance with fewer resources

The growth of investment in training runs over time

Our dashboard gathers all of this information in a single, accessible place. The numbers and figures are accompanied by further information such as confidence intervals, labels representing our degree of uncertainty in the results, and links to relevant research papers. These details are especially useful to illustrate which areas may require further investigation, and how much you should trust our findings.

Beyond accessibility, bringing these figures together allows us to compare and contrast trends and drivers of progress. For example, we can verify that growth in training compute is driven by improvements to hardware performance and rising investments:

We can also see that performance improvements have historically been driven by algorithmic progress and training compute growth by comparable amounts:

Overall, we hope that our dashboard will serve as a valuable resource for researchers, policymakers, and anyone interested in the future of Machine Learning.

We plan on keeping our dashboard regularly updated, so stay tuned! If you spot an error or would like to provide feedback, please feel free to reach out to us at [email protected].

Visit now the dashboard

Thanks for listening. To help us out with The Nonlinear Library or to learn more, please visit nonlinear.org.