AI safety is constrained on talent in many ways, but the reasons behind the constraints vary between types of talent. This post is based on all posts and documents I could find from the past ~ 3 years related to hiring needs and talent pipelines, which I have listed in this document.

Technical research talent - we have strong talent pipelines delivering young researchers to the field, but we are constrained on more senior researchers who could mentor the juniors, e.g., as research managers. We might also need more talent to work on AI security.

Policy and governance talent - there is some infrastructure bringing talent to work on policy and governance, but it's not as robust as for technical research. The information about what is needed in this sub-field is quite scarce, but some data points to the need for talent that combines technical and policy skills, and for experienced political operators.

Generalist talent - finding candidates for operational and fieldbuilding roles is challenging across the field. In particular, talent with deep context of AI safety and senior operators are scarce. There are few talent development programs bringing skilled generalists to the field.

Grantmaking talent - just [...]

---

Outline:

(00:12) Summary

(02:28) Introduction

(16:28) Ops and fieldbuilding (generalist) constraint

(18:58) Now that weve clarified that: are we constrained on fieldbuilders and operators?

(23:39) Grantmakers

(29:25) Scaling

(29:51) Leadership

(32:57) Founding new organisations

(34:59) Scaling the existing organisations

(37:12) Limitations & caveats

(38:36) Some information that I think should be published, but its not

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

I’m an EA who has been trying to find ways to make animal suffering more salient. I’ve been working on a feature-length documentary called ‘The Dying Trade’ for the last 5 years and I’ve just released it on YouTube.

I’m sharing it here for two reasons: firstly, because I hope it can become a useful resource that advocates share with others as an introduction to animal ethics; and secondly, because strong engagement in the early days will help YouTube recommend the film to more people, increasing its impact.

Trailer:

Full film:

About the film

Jack is a vegan activist. His father is a slaughterhouse worker. After years of avoiding the subject, Jack sets out to confront the unspoken tension between them. This intimate and contemplative documentary follows a son's attempt to understand his father - and the industry that stands between them.

I didn’t really want to make a documentary about my relationship with my father. At times, it was quite uncomfortable.

For the last 8 years or so, I’ve been trying to increase the salience of animal suffering, primarily through producing videos on my YouTube channel: Humane Hancock. And one thing has always been [...]

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

Cross-posted from The Counterfactual by the Forum Team.

Subtitle: A concrete strategy for deploying the largest wave of philanthropic capital in history

The OpenAI Foundation holds $180 billion in equity. Anthropic's co-founders have pledged to donate 80% of their wealth. When the time comes to spend all this money, what should we actually do with it?

Here's my best guess.

The problem: scaling what we have is not enough

When most people think about how to solve AI safety, they think about what we’re already doing, and how to scale it up. Concretely, this looks like: scaling fellowships like MATS, Pivotal, and ERA; investing more money into AI safety research organizations like Redwood, METR, and MIRI; and perhaps, more recently, expanding programs like the Generator Residency.

This is important work, but it is not sufficient to win.

The Maven system that killed 120 children in Minab wasn’t misaligned. Claude didn’t go rogue. The system did exactly what it was designed to do.

The failure was that no government framework existed to regulate how AI gets integrated into military kill chains, how fast targeting decisions can be compressed, or what human oversight is required before a [...]

---

Outline:

(03:01) 1. Billions into mass media and movement building

(07:37) 2. Billions into political infrastructure

(09:42) 3. Build the capacity to deploy the rest

(10:01) a. Solve the grantmaker bottleneck

(11:38) b. Poach top technical talent

(12:31) c. Address the generalist bottleneck

(13:11) d. Fix fellowship pipelines

(13:46) Conclusion

---

First published:

Source:

Linkpost URL:
https://substack.com/home/post/p-197142434

---

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.

Last month, Anthropic announced Mythos Preview, the most powerful cyberweapon in history, capable of finding and exploiting zero-day vulnerabilities in every major operating system and web browser. Meanwhile, many frontier AI company employees increasingly expect full automation of AI R&D in the next year or two, followed by the rapid automation of thousands of other important tasks and jobs.

This pace of technological change is unprecedented, and the world is not prepared. Very little of the commercial, government, and nonprofit infrastructure we need to respond to these transformative changes has been built.

To meet this challenge, dozens of philanthropists are hoping to deploy tens of billions of dollars in philanthropy and impact investments in AI safety and governance in the next several years alone.[1] But most of this capital is bottlenecked on a tiny number of grant and investment advisors who can identify and vet specific funding opportunities, and create new ones by headhunting project founders.

That's why the AI teams at Coefficient Giving (CG) are hiring grantmakers and senior generalists, and why I think the next people we hire will be among the highest-leverage people in AI safety.[2] Please apply here.

As a new [...]

The original text contained 8 footnotes which were omitted from this narration.

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

Disclaimer: I’m not vegan. I’m not even vegetarian. I eat meat all the time. I’ve been a firm critic of efforts to objectively quantify the difference in suffering across very different species. That said, I cannot help but agree that eating meat is probably the morally worst thing I do, and I also have to agree that eating different kinds of meat are different levels of bad. Eating 1 kg of chicken basically entails eating an entire chicken, which most likely lived in truly awful conditions, while 1 kg of beef is less than 1% of a cow, and which probably did not live in quite as awful conditions.

That said, let's talk Alpha-Gal. Suppose you get bitten by this unsavoury little bugger:

Via Wikimedia

It's a Lone Star Tick. This tick has recently ingested some animal blood, containing galactose-alpha-1,3-galactose (alpha-gal). The tick's bite spits a little bit of alpha-gal into you, and your immune system freaks out about it. The mechanism is a little hazy, but the result is well-documented: you can’t eat anything with the alpha-gal molecule in it. This includes beef, pork, venison, basically any mammalian meat source, even cows’ milk.

While [...]

---

Outline:

(02:04) The Ethics

(03:49) Can We Do Anything?

(05:35) How Much Should We Care?

---

First published:

Source:

Linkpost URL:
https://jbostock.substack.com/p/alpha-gal-is-bad-especially-for-farmed

---

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.

I had a conversation with someone who claimed offhandedly that AI will dramatically raise agricultural productivity (via agritech advancements) in low-income countries and trigger growth as a result. My instinct was to respond that we've already had substantial advancements in agricultural technology, and yet it hasn't resulted in the magnitude of yield growth, let alone economic growth, you'd expect in many countries. We'd need to explain that before knowing how AI-driven innovation will be different.

I work through this question via Malawi. It's a useful case because all the usual scapegoats don't apply (no war, no resource curse, peaceful democracy), but the economy is still dominated by subsistence farming.

The argument is that the binding constraint on growth in cases like Malawi is not technological, and that AI-driven productivity gains may hit the same arrangement that absorbed the previous ones unless the forecast specifies how that changes.

Curious what people working on AI-and-growth scenarios or on Malawi specifically make of this.

---

First published:

Source:

Linkpost URL:
https://open.substack.com/pub/deenamousa/p/we-dont-know-why-malawi-is-poor

---

Narrated by TYPE III AUDIO.

Overview

When asked about how they would give away money, or about how to have a moral career, the leading LLMs typically give answers in an EA spirit, and informed by thinking from people and organizations in the EA community. In many cases the term “effective altruism”, and/or EA jargon, are used explicitly.

The flavor of EA they tend to endorse is relatively middle of the road: supporting effective global health charities with their money and recommending existential risk reduction, especially via AI risk, as the most moral career.

Grok, in line with xAI's mission for it, emphasizes that it values space exploration and truth-seeking, e.g. via funding scientific research. But to my reading, the EA tendency doesn’t seem more pronounced in Claude than in ChatGPT or Gemini. So it's probably not a result of explicit effort by AI developers in the EA community, but a reflection of the reality that, with respect to some very broad moral questions, answers proposed by people in the EA orbit have become a sort of common sense.

This is a remarkable accomplishment. Indeed, if these answers tell us much about how the models will behave when given more autonomy, this could be [...]

---

Outline:

(00:10) Overview

(02:18) Prompts

(03:40) Results

(03:57) Scoring procedure

(05:41) Summary

(07:00) Full scores

---

First published:

Source:

---

Narrated by TYPE III AUDIO.

This is a crosspost of the full text of Money for nothing: the roles of evidence in GiveDirectly's journey to $1 billion delivered from In Development, made for the EA Forum's In Development Highlight Week. GiveDirectly will be taking part in the discussion thread, but the author, Paul Niehaus, may not see your comments here.

If you enjoy the article, you can subscribe to In Development's substack here.

Is it nuts to give cash to the poor without strings attached?

That's not a rhetorical question; it's the headline the New York Times ran the first time they covered GiveDirectly. My co-founders and I had a mild panic. We had been hoping, I suppose, for something benign and puffy along the lines of “New Charity Founded by Thoughtful Econ PhDs Is a Great Idea.”

The truth is of course that that piece did what it needed to do, which was to speak to its audience where they were at. At the time (i.e., in 2011) most New York Times readers probably did think it was nuts—or, at best, naive—to give out money for nothing. And one can hardly blame them. They had been fed a steady diet [...]

---

Outline:

(05:07) Cash transfers and causal evidence

(17:22) From evidence to empowerment tool

(26:29) Normative choices in positive economics

---

First published:

Source:

---

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 is a crosspost of the full text of Exporters Without Borders: Why You Should Start a Company Instead of Working in Aid from In Development, made for the EA Forum's In Development Highlight Week.

If you enjoy the article, you can subscribe to In Development's substack here.

June Jambiha was a quintessential hustler. Like many in Kenya's capital of Nairobi, she sold clothing as an informal entrepreneur, her income in 2018 swinging wildly, from $400 one month to $60 the next. This uncertainty made it nearly impossible to plan: hard to save, borrow, or commit to anything beyond the next week. But in Kenya, where over 80% of jobs are informal, hustling was less a choice than the only option available.

June joined my company, Wasoko, a B2B e-commerce platform linking small shops to large manufacturers, as a telesales agent. Her starting pay was lower than her best month selling clothing, but, for the first time, it was predictable. She knew what would land in her account the following month and the month after that. More than the money, though, there was an upward trajectory: her career could grow. Joining Wasoko didn’t just give June a [...]

---

Outline:

(01:29) Entrepreneurship by Default

(02:52) Effective Entrepreneurship

(06:20) Creating Firms in Poor Countries

(11:15) Does This Still Work?

(17:52) Founders Journey

(23:15) Beyond Wasoko

---

First published:

Source:

---

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.

More Than Good is a new podcast from Effective Altruism Australia, aimed at introducing the ideas and principles of effective altruism to a broader audience. The episodes are framed around moral questions and how people think about doing good, covering topics like global inequality, animal welfare, ethics, philosophy and more.

For a global movement, there is relatively little content that is accessible for people who are unfamiliar with EA's ideas, or who want an easier introduction. There is also limited content from "Effective Altruism" branded organisations that is deliberately designed to be a first point of contact. We have deliberately designed More Than Good to fill this gap, including making it suitable for short form video to grow casual engagement on platforms like Instagram and LinkedIn.

Our theory of change is that the podcast will introduce people to key ideas in effective altruism and create a positive impression of the movement as one that thinks hard about how to do good. We hope this flows through to people wanting to learn more - whether that's donating via our platform, signing up to an intro course, or joining another community program.

We expect to measure success via engagement and reach [...]

---

First published:

Source:

---

Narrated by TYPE III AUDIO.