EA Forum Podcast (Curated & popular)

Joe Carlsmith, a senior researcher specializing in AI risks, recently transitioned from Open Philanthropy to Anthropic. He reflects on his impactful tenure at Open Philanthropy, discussing the importance of worldview investigations and AI safety research. Joe shares his aspirations for designing Claude's character at Anthropic and weighs the significance of model-spec design in mitigating existential risks. He addresses the complexities of working within frontier labs, advocating for balancing capability restraint with safety progress, all while navigating potential personal and ethical challenges in his new role.

Toby Ord, a philosopher focused on global catastrophic risks and AI safety, delves into the scaling of reinforcement learning (RL) in his insightful analysis. He reveals that while RL facilitates longer thought processes for language models, its efficiency is limited compared to inference scaling. Deployment costs rise sharply with increased inference tokens, and achieving performance gains through RL demands astronomical compute increases. Ord discusses the governance implications of these findings, suggesting a shift in AI progress strategies.

In this engaging talk, Frankio reflects on his journey with the 10% Giving Pledge he made back in 2016. He candidly shares the challenges that led to falling short, including health and personal issues. Instead of repaying past donations, he announces a fresh start with new strategies for consistency in giving. Listeners will appreciate his practical tips like automating donations and budgeting. It's a heartfelt commitment to transparency and a reminder of the importance of staying accountable in philanthropy.

The discussion explores whether effective altruists might create more impact outside of nonprofit roles. With the EA movement's growth, there’s an urgent need to explore alternative career pathways. It highlights how for-profit sectors can influence billions and alleviate donor burdens. The podcast also warns of risks in pursuing external roles and emphasizes the importance of personal fit in career decisions. Ultimately, it advocates for a more diverse talent distribution across various sectors to enhance advocacy efforts.

Devangana Prasad, an author and researcher at Rethink Priorities, delves into the nuanced world of framing effective altruism. She shares surprising findings from experiments that reveal the phrase 'doing good better' actually led to less interest in effective altruism. Instead, taglines focusing on 'doing the most good' garnered more support. Prasad also discusses the appeal of using words like 'philosophy' and 'movement', which inspired curiosity about EA. These insights shed light on how subtle word choices can significantly impact perceptions of altruism.

David Nash, the author behind 'The Charity Trap: Brain Misallocation,' explores how NGO practices can inadvertently harm communities. He highlights alarming statistics from Uganda, where hiring NGOs led to increased infant mortality rates. Nash delves into the concept of brain misallocation, emphasizing how talented individuals are drawn to NGOs, diverting their skills from vital national development roles. He discusses the distortions caused by significant NGO wages and questions the overall impact of this talent concentration on local economies and well-being.

Explore the chilling potential of catastrophic biological risks, including a deep dive into dangerous pathogens like engineered viruses and bacteria. The discussion reveals four essential pillars for defense: PPE, biohardening, pathogen detection, and rapid medical responses. Discover why respiratory protection, especially elastomeric respirators, is critical, and how creating pathogen-free environments could safeguard us. With a strong emphasis on early detection through innovative technologies, this dialogue makes a compelling case for prioritizing public health readiness.

Toby Tremlett, a writer and contributor to the Effective Altruism Forum, dives deep into the concept of "entertainment for EAs." He unpacks the failure modes in meta-EA work, emphasizing the need to focus on ultimate beneficiaries rather than just the EA community. Toby highlights how projects can receive positive feedback yet fail to impact the real stakeholders—humans, animals, or digital minds. He offers a practical detector for this issue, encouraging accountability in prioritizing genuine effectiveness over community-centric projects.

Canva announces a groundbreaking $100 million commitment to GiveDirectly, aiming to empower 185,000 people in extreme poverty over the next four years. This follows a previous $50 million donation that has already changed the lives of over 85,000 individuals in Malawi. The discussion covers GiveDirectly's impressive cash transfer scale since 2009 and Canva's unique founding-to-give model, which emphasizes building value before committing to philanthropy. Additional charitable engagements from Canva showcase their broader impact beyond standard effective altruism cause areas.

I have some claim to be an “old hand” EA:[1] I was in the room when the creation Giving What We Can was announced (although I vacillated about joining for quite a while) I first went to EA Global in 2015 I worked on a not-very successful EA project for a while But I have not really been much involved in the community since about 2020. The interesting thing about this is that my withdrawal from the community has nothing to do with disagreements, personal conflicts, or FTX. I still pretty much agree with most “orthodox EA” positions, and I think that both the idea of EA and the movement remain straightforwardly good and relevant. Hence why I describe the process as “senescence”: intellectually and philosophically I am still on board and I still donate, I just… don’t particularly want to participate beyond that. Boredom I won’t sugar-coat [...] ---Outline:(01:00) Boredom(04:05) What do I have to offer? --- First published: October 19th, 2025 Source: https://forum.effectivealtruism.org/posts/rJqQGD2z2DaupCbZE/my-ea-senescence --- Narrated by TYPE III AUDIO.