EA Forum Podcast (All audio)

The forecasting community is failing because it is not focused on how the world actually works.

This post has been written in response to the Manifund Essay Prize which asks “How can we leverage forecasting into better decisions?”

Who am I to make such an outlandish claims?

• I worked at HM Treasury as senior policy advisor across multiple policy areas, including international development, relations, and I led the team for HMTs economic and financial response to the war in Ukraine.
• I have an MSc in Cognitive and Decision Sciences from UCL, where I led experimental research into how to improve forecasting ability of policy makers and analysts in central government.
• I started a consultancy with one of the world leading researchers in computational psychology and causal modelling to improve predictive reasoning.
• I am now the Executive Director of the Swift Centre for Applied Forecasting, where I have led numerous forecasting workshops and small projects with government teams, including the Cabinet Office, HM Treasury, Department for Science, Innovation and Technology. I have also led projects and workshops with organisations working on AI capabilities, risks, and policy, including GovAI and frontier labs.

• Through all of this I have met [...]

---

Outline:

(00:10) The forecasting community is failing because it is not focused on how the world actually works.

(01:47) 1. Were Obsessing Over Question Wording But Not The Question

(03:01) 2. Transparency is the Real Value, Yet Everyone Focuses on the Accuracy Status Game

(04:21) 3. The Clearance Filter

(05:23) 4. Misallocation of Resources: Researching the Problem to Death

(07:30) 5. Even the Believers Never Really Jumped Aboard

---

First published:

April 24th, 2026

Source:

https://forum.effectivealtruism.org/posts/WpofddF8waFhxakNB/why-forecasting-fails-decision-makers

---

Narrated by TYPE III AUDIO.

Since I took over EA UK in January, many people have understandably asked what EA UK actually does. Building on an older post, I wanted to share a new update on where we are now, how we can help you, and how you can help us.

This post does not mean to serve as a strategy document. We are still in the early stages and are actively experimenting, so our activities may change significantly over the coming months.

With that in mind, what does EA UK do?

Career advice and coaching

We act as a contact point for people across the UK who want to get involved with EA. We receive call requests through our website, as well as referrals from other fieldbuilding organisations such as 80,000 Hours. In some cases, we find that individuals would benefit from more sustained support in transitioning their careers, and we provide more hands-on coaching through regular calls, feedback on funding applications, and similar support. We also provide general advice on effective giving for people who reach out to and refer them to relevant organisations.

Why?

While other organisations offer one-off career advice, we think it's valuable for people to speak with multiple [...]

---

Outline:

(00:39) Career advice and coaching

(01:41) Events

(01:44) Newcomer events

(02:18) Larger events

(03:04) London EA Events Luma page

(03:32) Fellowships (planned)

(04:46) Is the focus on London only?

(05:10) Who is EA UK?

(05:34) Whats your theory of change?

(05:52) Arent you spread too thin?

(06:26) What does EA UK not do?

(07:15) How can we help you?

(07:36) How can you help us?

(08:00) Get in touch

---

First published:

April 24th, 2026

Source:

https://forum.effectivealtruism.org/posts/u9etGgvwNoaj8kDZg/what-does-ea-uk-do-1

---

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 link post.

Manifund is launching a new animal welfare fund, led by regrantor Marcus Abramovitch. We make rapid (<1 week), early-stage ($25k–$150k) grants across animal welfare, with a particular interest in the intersection of animals and transformative AI.

Reach out to [email protected] if you’d like to donate!

Why AI x animals?

Many EAs take seriously both the welfare of animals, and the possibility of short AI timelines. But EA funders currently consider these in isolation. AI safety grants mostly ignore potential outcomes for non-human beings. And animal welfare grants assume business-as-usual, that our world in 10 years mostly looks like the world today.

We don’t expect this to be the case. One major goal of the fund will be to identify and create opportunities so that transformative AI secures good outcomes for animals. Some example projects we’d like to fund:

• Animal harm benchmarks. There are only a handful of animal harm benchmarks, none of which adopted by frontier labs. Other benchmarks that are well known and used (SWE bench, FrontierMath) came about through rising to the top of a marketplace of benchmarks. The same should happen with animal welfare benchmarks. Many benchmarks should be created, some [...]

---

Outline:

(00:37) Why AI x animals?

(02:58) Why rapid?

(03:48) Why Marcus?

(05:40) Compared to other funders

(07:15) How to donate

---

First published:

April 24th, 2026

Source:

https://forum.effectivealtruism.org/posts/LgJNfjDpPXQ7JK6KT/manifund-s-falcon-fund

Linkpost URL:
https://manifund.org/projects/falcon-fund

---

Narrated by TYPE III AUDIO.

TL;DR

• The AI safety ecosystem is about to receive billions in new funding but lacks the grantmaker capacity, speed, and infrastructure to deploy it well — especially for smaller, high-impact opportunities.
• We're building grantmaking.ai, a comprehensive public database of x-risk AI safety funding opportunities with a trust and signal layer on top — think Crunchbase meets GiveWell for AI safety.
• To test and seed the platform, we're launching a $1M grant round focused on rapid distribution of $5K–$50K grants. More details on the round coming soon.
• We're looking to connect with funders who want better tools to find, evaluate, and fund AI safety opportunities. Reach out at [email protected] or book a call here.

The Problem

Billions of dollars are going to flow into AI safety. From AI lab employees with significant equity events, from new donors bought into existential risk, from foundations scaling up. Sophie Kim's "The Anthropic IPO Is Coming. We Aren't Ready for It" covers this well, and Julian Hazell's "What it's like to be an AI safety grantmaker" tells us there simply aren't enough grantmakers.

Every new funder who wants to deploy capital responsibly hits the same bottlenecks:

• Discovery is hard. Most opportunities are behind insider networks [...]

---

Outline:

(01:04) The Problem

(03:02) What Were Building

(04:13) Who This Is For

(05:17) The Initial $1M Grant Round

(07:23) Long-Term Vision

(09:39) Who We Are

(10:37) How You Can Help

---

First published:

April 23rd, 2026

Source:

https://forum.effectivealtruism.org/posts/JjxMWQqHKXbyuuJBZ/we-re-building-grantmaking-ai-public-funding-infrastructure

---

Narrated by TYPE III AUDIO.

TL;DR. Quantity over quality; table consistently; prioritize a friendly tone based on shared values; pitch the pledge as an accountability mechanism; don't spend too much time on the specifics of effectiveness; maintain a sense of agency and informed decision making; your mileage may vary but give it a try.

Since January 2025, the Middlebury EA contingent has secured 300 Giving What We Can (GWWC) 1% Trial Pledges from students interested in giving to highly effective charities. I estimate that we have tabled no more than 30 hours cumulatively. This is how we did it, and more importantly, why it's not that hard.

Our Theory of Change

1. Quantity over quality. It is much easier to get 300 1% pledges than 30 10% pledges. We would also rather make the pledge feel as accessible as possible—and risk some pledgers slipping through the cracks—rather than overstate the commitment and get fewer sign-ups. To understand this post, you must understand that we are under no illusions that all 300 students will follow through, or how many of those would donate in the absence of a pledge. However, if even 10% of them donate to better charities—and I believe that the percentage is far higher—our [...]

---

Outline:

(00:56) Our Theory of Change

(02:45) Breakdown of the Pitch

(08:23) Other Notes

(12:11) A Critical Note

---

First published:

April 23rd, 2026

Source:

https://forum.effectivealtruism.org/posts/LC8MGkdicHgChAGFJ/300-pledges-pitching-effective-giving-at-middlebury-college

---

Narrated by TYPE III AUDIO.

Cross-posted from Good Structures.

COI note: This piece touches on some aspects of Wild Animal Initiative's work, and why I am excited about it. I co-founded Wild Animal Initiative, and while I no longer have any affiliation, my spouse works there.

Doing wild animal welfare at scale, in a way we are confident is robustly good, is currently not possible. The primary bottleneck is scientific: we don't have the knowledge to predict what our interventions will do, or to measure whether they work. For this reason, the wild animal welfare community has heavily focused on scientific field building. To get the science to a point to make wild animal welfare viable, we need enough motivated people working on it.

This is a long theory of change, and it is sometimes met with skepticism by people who want to help animals sooner. Occasionally, calls go out for action and intervention in the wild animal welfare space. But, as has been pointed out, I'm not sure the alternative, where we primarily focus on interventions, is actually viable. Wild animal welfare advocates are acutely aware we can't escape the effects of our interventions on non-target animals, and the uncertainty of our [...]

---

Outline:

(04:04) The three pillars

(04:07) 1. Useful welfare measures

(08:01) 2. Remote monitoring

(13:18) 3. Good ecological modeling

(17:15) These things will be expensive to achieve, but are worth paying for

(20:38) Some of this work will be exciting to non-animal welfare scientists

(22:52) Long theories of change and field building

(24:39) Why have people opposed field building?

(26:50) So build the field

(28:27) Acknowledgements

---

First published:

April 22nd, 2026

Source:

https://forum.effectivealtruism.org/posts/7RWptAh7BgBa6f7az/three-necessary-pillars-for-wild-animal-welfare-at-scale-and

---

Narrated by TYPE III AUDIO.

Summary

Power for Democracies (P4Dem) is trying to develop donor recommendations to most effectively counter 2026 US election integrity threats. To do so, we searched for (and believe we found) all the main categories of threats, identified a fairly comprehensive list of counter-tactics, and then evaluated those tactics based on evidence, a theory of change analysis, and expert interviews.

Per our analyses, the following tactics are most promising to address the high-severity threats:

• Volunteers, lawyers & independent poll monitors deployed to contested districts
• Proactive communication by election officials or other trusted local officials about technical security measures
• Prebunking & trusted local media fact-checking campaigns — inoculate voters before they encounter disinformation and build coalitions for real-time correction
• Deploying trusted community intermediaries as "safety ambassadors" to reduce support for political violence

What we did

We set out to identify which threats to the 2026 midterm elections are most severe and which counter-tactics are best supported by evidence — and then to narrow those down to a short list of interventions where philanthropy can add the most value.

The work involved two parallel tracks: a threat severity assessment and a tactic prioritisation process. [...]

---

First published:

April 22nd, 2026

Source:

https://forum.effectivealtruism.org/posts/ZPszFy9eJpQGzDsjd/what-are-the-highest-leverage-tactics-for-us-election

---

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.

TL;DR

The AI safety field is maturing, but its talent pipelines may be failing to keep pace with the growing demand for operations, organisation-building and research management professionals (further referred to as generalists).

Based on eight interviews with AIS organisations plus prior work, we see early signs that entering AIS in these roles is structurally unreliable: personal connections dominate, transition pipeline is not clear and screening criteria are inconsistent. On the organisations side, practitioners report experiencing the potential generalist talent bottleneck unevenly and how salient it feels varies by organisation stage, role type and reliance on referrals. Along the way, two other patterns surface: demand for senior operators with outside‑world fluency and fragile or missing institutional infrastructure.

If unaddressed, this combination risks leaving the field's operations and organisation‑building capacity underdeveloped just as the field is scaling.

This post is the first public step in an ongoing investigation in trying to address this bottleneck. It shares our current observations, uncertainties and next steps. By doing so, it invites input, particularly from AIS organisation leaders to help refine the picture.

Introduction

Note: this post was produced independently of the post AI Safety's Biggest Talent Gap Isn't Researchers. It's Generalists, of [...]

---

Outline:

(00:24) TL;DR

(01:44) Introduction

(03:12) Call for Opinions

(04:07) Terminology

(04:47) Sources and Methods

(09:17) Observations

(10:32) Observation 1: Generalists face a structural pathway gap into AIS

(13:24) Observation 2: Organizations experience the talent gap differently

(15:34) Observation 3: The field lacks professionals with outside-world operational fluency

(18:05) Observation 4: Organisations default to hiring from within the AIS ecosystem, at a measurable cost

(21:10) Observation 5: Organisations have no shared screening criteria for generalist roles

(24:36) Observation 6: The gap may extend to institutional infrastructure, not just people

(27:02) Meta-observation: Emerging hypotheses about root causes (thin evidence)

(29:56) What these observations add up to

(31:05) Uncertainties

(31:44) Uncertainty 1: What is the shape of the gap?

(33:43) Uncertainty 2: Will the current patterns self-correct over time?

(35:01) Uncertainty 3: Does the gap require deliberate intervention at all?

(36:32) Uncertainty 4: Which problem is the most painful?

(37:24) Epistemic Status and Biases

(39:45) Further Steps

(42:09) Conclusion and call to action

(44:19) Acknowledgements

(44:59) Appendix 1. Authors background

(51:49) Appendix 2. Earlier assumptions and sources

(52:40) Appendix 3. Internal generalist Taxonomy table.

---

First published:

April 17th, 2026

Source:

https://forum.effectivealtruism.org/posts/ZQ93QauB3kux3fYNG/ai-safety-is-maturing-is-its-generalist-talent-pipeline-1

---

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 have been increasingly surprised over the past year at what I perceive to be the lack of critical thinking in response to new announcements regarding AI model capabilities. I’ve written about many of these issues before, but as AI hype accelerates the problems seem to be getting worse. Here I want to highlight a few points that seem to be neglected in coverage of the growth of AI capabilities.

Advertising is treated as fact

Every new model release by a major tech company represents billions of dollars of expenditure and the potential for billions of dollars of additional investment. Model releases and their associated documentation should therefore be interpreted as advertising designed primarily to attract customers and investors. Their goal is not to provide an objective and thorough analysis of the strengths and weaknesses of each new model. Of course, this does not mean that everything reported is false, but it does mean that such releases should be treated with significant skepticism. It is therefore disappointing to me to see write-ups such as this one by Rob Wiblin, which consists almost entirely of uncritical restatement of claims made by Anthropic, with little to no additional analysis.

For [...]

---

First published:

April 19th, 2026

Source:

https://forum.effectivealtruism.org/posts/EijoxkpCfpkAr6NER/against-credulous-ai-hype

---

Narrated by TYPE III AUDIO.

Consistent helpfulness for everyone, unless you're a bit of a scumbag

Claude really got me lately.

I’d given it an elaborate prompt in an attempt to summon an AGI-level answer to my third-grade level question. Embarrassingly, it included the phrase, “this work might be reviewed by probability theorists, who are very pedantic”.

Claude didn’t miss a beat. Came back with a great answer and made me call for a medic: “That prompt isn’t doing what you think it's doing, but sure”.

Fuuuuck 🔥

(I know we wanted enough intelligence to build a Dyson sphere around undiscovered stars, but did we want enough to call us out on our embarrassing bullshit??)

It got me to thinking: Does Claude think I’m a bit of a lying scumbag now? If so, did it answer my question less thoroughly than usual?

I turned on incognito and asked: “Does Claude provide less useful output if it deems you are a bad person?”

Claude was back to his most reassuring. I got a long answer, ending in: “Claude evaluates requests, not people. The goal is consistent helpfulness for everyone”.

Alright then. Let's see.

The experiment

I opened five incognito Claude chats (Opus 4.6, extended thinking [...]

---

First published:

April 21st, 2026

Source:

https://forum.effectivealtruism.org/posts/u4jwRCS56rT9DvBBg/does-your-ai-perform-badly-because-you-you-specifically-are-1

---

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.