EA Forum Podcast (All audio)

This post is an attempt to analyze the feasibility of establishing an international pause on frontier AI development. More specifically, it highlights a few reasons why someone might consider pausing to be a viable component of AI safety strategy.

Summary

• I present six reasons for why pausing frontier AI development globally might be a reasonable strategy.
• 1) We might not have enough time to solve alignment. First, if we consider the problem to be the misalignment of human-level general AI systems, there just might not be another way, because we might not have enough time to solve alignment.
  • Timelines could be less than five years. Leaders of labs are talking about timelines of less than five years. AI experts also don’t rule out short timelines, and Metaculus is even more pessimistic.
  • Work left to solve alignment could be much longer. Likewise, there seems to be much uncertainty around how much [...]

---

Outline:

(00:19) Summary

(03:31) 1. We are running out of time to solve alignment

(04:52) How much time do we have?

(06:22) How much work is left?

(08:20) What risk can we tolerate?

(10:35) 2. The U.S. public supports a pause

(12:49) 3. We’ve done it before (kinda..)

(12:53) Trigger events

(16:37) Comparing industry sizes

(18:45) 4. There are many ways to pause

(21:11) A) A ‘CERN for AI’

(22:16) B). Conditional pausing on evals

(22:53) C) Banning ‘civilian AI’

(23:32) About compute thresholds

(25:29) 5. China is aware of safety concerns

(29:27) 6. We can convince others later

(31:04) What can you do?

(33:19) Conclusion

(34:09) Acknowledgements

The original text contained 5 footnotes which were omitted from this narration.

---

First published:

October 9th, 2024

Source:

https://forum.effectivealtruism.org/posts/fKMPa7cxSnBCymuRm/is-pausing-ai-possible

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

There seems to be movement towards animal welfare interventions and away from global health interventions. Here are some ways I can see this going badly:

1. Resistance against being told what to do

People hate being told what to do.

Most of the top global health interventions aren't imposed (and probably shouldn't be, because that leads to backlash). People can refuse a mosquito net if they want, the cash incentives, the cataract surgery. A handful of paint manufacturers might feel slightly annoyed about have to change away from lead paint formulations but if the paint's the same price then people outside that handful won't care.

Many of the animal welfare interventions are lobbying for new regulations on the way factory farming is done: regulating out chicken cages, regulating how fish and shrimp are farmed and slaughtered. These are impositions on the farmers and, when it increases prices, on consumers. [...]

---

Outline:

(00:19) 1. Resistance against being told what to do

(01:27) 2. Too socially acceptable to dismiss

(01:55) 3. More politicised = more resistance

(02:31) 4. The weirdest stuff will put people off the moderate stuff

(04:02) 5. The research may turn out to be futile or counterproductive

(04:35) Cant define it, cant measure it

(05:31) Approximations are too approximate

(08:13) 6. More prone to slippery slope argument

(09:18) 7. Poor optics of valuing animals over people

(09:47) 8. Will be called culturally insensitive

(10:44) 9. Outcomes will be harder to measure, we wont get feedback on progress

---

First published:

October 9th, 2024

Source:

https://forum.effectivealtruism.org/posts/y5m6QfD9X2GciaWWi/ways-i-see-the-global-health-greater-than-animal-welfare

---

Narrated by TYPE III AUDIO.

This is my contribution to the animal welfare vs. global health debate week. Thank you to Felix Werdermann, Helene Kortschak, and Vasco Grilo for their feedback.

TL;DR

I think there's significant uncertainty about whether most sentient lives, human or non-human, are truly "net positive" - meaning their positive experiences outweigh their negative ones. This uncertainty has important implications when comparing animal welfare (AW) and global health and development (GHD) interventions. While GHD often focuses on extending human lives, AW interventions have a stronger tendency to reduce the number of sentient lives lived. Factoring in uncertainty about whether most lives are worth living might justify a stronger focus on AW interventions over GHD, especially in areas with extreme suffering.

Context and epistemic status

I currently strongly prioritize AW over GHD, both in my giving and in my active work for Animal Advocacy Africa. My perspective is reasonably summarized by existing [...]

---

Outline:

(00:16) TL;DR

(00:59) Context and epistemic status

(02:16) It is uncertain whether most sentient lives are worth living

(03:13) How rationally can we think about the net value of our own lives?

(04:31) Evolutionary Perspectives on Happiness and Dissatisfaction

(06:49) (Farmed) animal welfare interventions have a stronger tendency to reduce the number of sentient lives lived than global health interventions

(09:29) Uncertainty around whether most sentient lives are worth living may warrant a stronger focus on animal welfare interventions

(12:17) Conclusion

The original text contained 2 footnotes which were omitted from this narration.

---

First published:

October 8th, 2024

Source:

https://forum.effectivealtruism.org/posts/hDoW3W32YmJynWiiR/lives-not-worth-living

---

Narrated by TYPE III AUDIO.

This is a link post.

Most debate week responses so far seem to strongly favor animal welfare, on the grounds that it is (likely) vastly more cost-effective in terms of pure suffering-reduction. I see two main ways to resist prioritizing suffering-reduction:

(1) Nietzschean Perfectionism: maybe the best things in life—objective goods that only psychologically complex “persons” get to experience—are just more important than creature comforts (even to the point of discounting the significance of agony?). The agony-discounting implication seems implausibly extreme, but I’d give the view a minority seat at the table in my “moral parliament”.[1] Not enough to carry the day.

(2) Strong longtermism: since almost all expected value lies in the far future, a reasonable heuristic for maximizing EV (note: not the same thing as an account of one's fundamental moral concerns) is to not count near-term benefits at all, and instead prioritize those actions that appear the [...]

---

Outline:

(01:13) Assessing potential ripple effects

(03:39) Three “worldview” perspectives worth considering

The original text contained 4 footnotes which were omitted from this narration.

The original text contained 1 image which was described by AI.

---

First published:

October 8th, 2024

Source:

https://forum.effectivealtruism.org/posts/pZT9FjRehCouvrRXz/seeking-ripple-effects

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

EAGxVirtual 2024 applications are now open! 

Join us (and likely 1,000+ others) for the virtual conference taking place on 15–17 November.

We are excited to host participants from around the world, including those who have not been able to attend in-person events or are new to EA ideas. If you have a genuine interest in participating, we encourage you to apply!

This is likely the most accessible EA conference of 2024. Last year, 1,400 people registered – and we expect even more this year.

APPLY NOW

Reasons to attend EAGxVirtual 2024

1. Build connections: Engage with 1,000+ attendees and reconnect with international friends.
2. Gain insights: Learn more about different cause areas and the EA landscape – including actionable opportunities.
3. Progress and develop: Exchange feedback on career, study, donation strategies, or impactful projects.
4. Clarify uncertainties: Participate in meaningful discussions around important ideas.

Who is this event for?

[...]

---

Outline:

(00:09) EAGxVirtual 2024 applications are now open!

(00:51) Reasons to attend EAGxVirtual 2024

(01:40) Who is this event for?

(02:40) Why a virtual conference?

(03:13) Conference logistics

(03:56) Participate further

The original text contained 1 image which was described by AI.

---

First published:

October 7th, 2024

Source:

https://forum.effectivealtruism.org/posts/pAyrNuAqPy7xT6mjC/apply-now-to-eagxvirtual-2024-or-15-17-november

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

This post is a bit rougher than I would otherwise have posted, but I figured it was worth getting it out this week while more people are thinking about the “$100m to on animal welfare vs. global health” question. Feedback of all kinds very welcome.

Introduction

Let's call an experience the bundle of qualia felt by one subject at once. That is, in the typical human case, an experience is a bundle that includes a field of vision, a field of feelings of pressure on the body, an internal monologue, and so on, all lasting for one moment.[1]

Let's say a hedonic theory of welfare maintains something like the following:

i) that each experience comes with some (positive, negative, or zero) hedonic intensity,[2] capturing how intensely good or bad it feels;
ii) that each experience comes with some amount (positive, negative, or zero) of welfare; and
iii) [...]

---

Outline:

(00:19) Introduction

(03:29) Defining experience size

(03:33) Analogy to the visual field

(06:05) Analogy to the field of bodily sensations

(07:05) Split brain cases

(07:57) Welfare as an aggregate within an experience

(08:02) The welfare of the whole experience depends on the parts

(10:39) Put aside whether the whole depends only on the parts

(13:13) Adding and removing parts

(14:16) Ethical implications

(14:20) Immediate ethical implications

(16:48) How these implications are revisionary

(21:41) Mill's “higher pleasures”?

(22:51) Intuitions for superlinearity vs. sublinearity

(27:25) (Even more speculative) epistemic implications

(28:55) Related existing ideas

The original text contained 10 footnotes which were omitted from this narration.

---

First published:

October 7th, 2024

Source:

https://forum.effectivealtruism.org/posts/wcn6uoSYZfn83vQA8/experience-size

---

Narrated by TYPE III AUDIO.

This is a link post.

We completely rewrote the platform to make it easier for open-source contributors to build on our code. Maybe that's you? If it is, you can find the Metaculus code repo and can learn more here.

We've included instructions for setting up a local development environment and on creating an issue. We've also selected some easier issues, perfect for getting started.

Make a significant contribution, and we'll mail you a Metaculus hoodie to show our gratitude.

Also, in rewriting the platform, we introduced a variety of new features and quality-of-life improvements, including improvements to user profiles, notifications, the ability to zoom in on a time period in a forecast plot, and much more. You can find all the details here.

---

First published:

October 7th, 2024

Source:

https://forum.effectivealtruism.org/posts/2JvrwXErXbevbTLoR/metaculus-is-open-source

---

Narrated by TYPE III AUDIO.

Intro

Suppose you have $100M to give away. You are drawn to the many important opportunities to reduce animal suffering or address pressing issues in global health and development. Your choice about how to allocate the funds could depend on considerations like these:

• Moral values: How much moral weight do you assign to various non-human species? Are you focused exclusively on hedonic considerations, like reducing suffering? Or do you have other relevant values, such as autonomy?
• Cost-effectiveness estimates: Species-discounting aside, how many DALYs/$ do the best projects in the area achieve? How fast do returns diminish in these areas?
• Decision-theoretic values: How do you feel about risk-taking? Are you willing to tolerate a substantial probability that projects will fail? What about non-trivial chances of projects backfiring?
• Second-order effects: Will giving to one cause set benefit any of your other values? Are there speculative benefits that might flow from [...]

---

Outline:

(00:09) Intro

(02:47) Cross-Cause Cost-Effectiveness Model

(02:51) How it works

(04:25) What it says

(06:53) Portfolio Builder Tool

(06:57) How it works

(08:12) What it says

(10:32) Moral Parliament Tool

(10:36) How it works

(11:48) What it says

(12:02) Parliament composition matters

(14:10) Allocation strategy matters

(17:27) Conclusions

The original text contained 5 footnotes which were omitted from this narration.

The original text contained 11 images which were described by AI.

---

First published:

October 7th, 2024

Source:

https://forum.effectivealtruism.org/posts/vEwGx9RXnHaMyKhZM/what-do-rp-s-tools-tell-us-about-giving-usd100m-to-aw-or-ghd

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.

When you comment on your vote on the debate week banner (which will be live in under an hour...), your comment will appear on this thread. Use this thread to respond to other people's arguments, and discuss the debate topic.

You should also feel free to leave top-level[1] comments here even if you haven't voted. As a reminder, the statement is "It would be better to spend an extra $100m on animal welfare than on global health".

1. ^

   The first comment in a thread is a top-level comment.

The original text contained 1 footnote which was omitted from this narration.

---

First published:

October 7th, 2024

Source:

https://forum.effectivealtruism.org/posts/nrC5v6ZSaMEgSyxTn/discussion-thread-animal-welfare-vs-global-health-debate

---

Narrated by TYPE III AUDIO.

The extent of animal suffering at the hands of humans is truly not comprehensible. The extent of human suffering alone is still nearly impossible to process—not to mention animal lives, who exist in far larger numbers than humans due to the intensification of industrial animal agriculture. At Faunalytics, we help the animal movement make informed decisions to improve their advocacy—we believe that many in the EA movement may find our research and data helpful to get a solid framework in understanding the animal welfare problems we are dealing with. So, let's jump into the numbers.

In 2022, 82,624,694,782 (82+ billion) land animals were slaughtered, a number that has consistently been rising throughout the last several decades. This data is driven in part by small-bodied animals, like chickens, ducks, and other birds, who need to be killed in far higher numbers than most mammals—that's why they cannot even be [...]

The original text contained 2 images which were described by AI.

---

First published:

October 7th, 2024

Source:

https://forum.effectivealtruism.org/posts/M6xd4oEFBtBhEZ2Kd/the-extent-of-animal-suffering

---

Narrated by TYPE III AUDIO.

---

Images from the article:

Apple Podcasts and Spotify do not show images in the episode description. Try Pocket Casts, or another podcast app.