Latest Interviews
Showing 16–30 of 37 interview transcripts.
Clear all filters- 80,000 Hours2h 49m
What superforecasters and experts think about existential risks | Ezra Karger
The Forecasting Research Institute conducted a multi-stage tournament and adversarial collaboration involving over 200 experts, superforecasters, and skeptics to quantify existential risks, revealing that while both groups agree advanced AI will emerge by the mid-21st century, they diverge sharply on extinction probabilities with domain experts estimating a 3% risk compared to 0.38% for superforecasters. Follow-up studies on AI risk perception determined that long-term worldview differences and expectations regarding human well-being rather than short-term capabilities drive these disagreements, resulting in minimal belief convergence during eight weeks of structured debate. The findings suggest that public risk estimates are highly sensitive to reference class framing and highlight independent verification of AI's ability to evade deactivation as the critical crux needed for consensus on catastrophic outcomes.
- 80,000 Hours2h 2m
Edge cases of sentience & how to act ethically despite massive uncertainty | Jonathan Birch
Jonathan Birch, Luisa Rodriguez
Philosopher Jonathan Birch presents a precautionary framework in *The Edge of Sentience* that mandates proportionate moral action based on the realistic possibility of suffering in beings ranging from human fetuses and invertebrates to neural organoids and advanced AI. Drawing on evidence of diagnostic errors in disorders of consciousness and the suffering of species like octopuses and bees, Birch argues for abandoning rigid classifications in favor of a duty to avoid gratuitous harm even amidst scientific uncertainty. This approach has already influenced UK legislation to include crustaceans and cephalopods in welfare protections while recommending a shift toward citizen-led deliberation and proactive regulation for emerging technologies like brain emulation.
- 80,000 Hours2h 7m
How to regulate cutting-edge AI models | Markus Anderljung (2023)
Markus Anderljung, Luisa Rodriguez, Rob, Marcus Andrew Young
The Center for the Governance of AI is mobilizing to address critical risks in frontier AI systems, such as emergent capabilities and autonomous agent loops, by advocating for a mandatory three-pillar regulatory framework covering licensing, external red-teaming, and compute governance. Driven by urgent momentum from major tech leaders and government bodies, this initiative proposes a flexible regime that mandates risk assessments and licensing for high-compute training to prevent dangerous deployment. Simultaneously, the organization is expanding its workforce by seeking interdisciplinary experts in policy, economics, and technical fields to help implement these safeguards and influence global standards.
- 80,000 Hours2h 18m
Alex Lawsen on avoiding 10 mistakes people make when pursuing a high impact career
Alex Lawson, an advising manager from 80,000 Hours, joins host Luis Rodriguez to outline ten common cognitive traps that hinder high-impact career decisions, ranging from blindly following ranked lists to misinterpreting short AI timelines. The discussion reframes these errors as correctable patterns of thought rather than moral failures, offering evidence-based strategies for optimizing career capital, resisting the urge to switch paths, and validating personal fit through intense experimentation. By integrating insights on cause prioritization and the value of conventional mentorship, the conversation provides a framework for navigating uncertainty and maximizing long-term contribution without succumbing to moral perfectionism.
- 80,000 Hours1h 48m
Do unconditional cash transfers cause economic growth in low-income countries? | Paul Niehaus (2023)
Paul Niehaus and USAID advocate for cash transfers as the default development strategy, supported by evidence that recipients in extreme poverty outperform traditional aid programs in asset accumulation, income generation, and subjective well-being. Recent large-scale experiments, including a UBI trial in Kenya, demonstrate a 2.5x economic multiplier effect and prove that unconditional cash does not discourage work while significantly reducing food insecurity. Although operational risks like fraud exist, organizations like GiveDirectly are addressing these through enhanced protocols and transparency, ultimately positioning cash transfers as a scalable, autonomy-respecting benchmark that forces the broader aid sector to improve its efficacy.
- 80,000 Hours2h 8m
Who's trying to steal AI models? And what could they do with them? | Sella Nevo
A comprehensive security analysis details the exponential risk of internal model weight leakage, where granting access to even fifty trusted employees creates a 63% probability of compromise, potentially allowing adversaries to bypass hundreds of millions in compute costs and disable safety guardrails. The briefing further classifies attacker capabilities into five tiers ranging from amateur opportunists to nation-state operations capable of executing supply chain attacks and exploiting physical side-channels, while identifying a critical global shortage of AI security personnel. Consequently, the presentation advocates for stringent centralized access controls, hardened inference interfaces, and confidential computing environments to mitigate the acute biosecurity and economic threats posed by unauthorized model extraction.
- 80,000 Hours2h 24m
Navigating the geopolitics of US–China AI competition | Sihao Huang
A 2040 scenario where AI achieves human-level capability necessitates urgent global coordination to prevent existential conflicts arising from a superintelligence race between the United States and China. Current analysis highlights significant constraints on China's frontier progress due to US semiconductor export controls, yet warns that authoritarian regimes may leverage AI for automated repression and biological or cyber warfare. Immediate diplomatic efforts must establish mutual safety verification and continuous dialogue channels to mitigate the risks of pre-emptive military action while navigating complex domestic regulatory and industrial policies.
- 80,000 Hours1h 54m
What would happen if North Korea launched a nuclear weapon at the US? | Annie Jacobsen
Annie Jacobsen, Luisa Rodriguez
Former Secretary of Defense Leon Panetta and Bill Perry highlight critical vulnerabilities in U.S. nuclear command, where flawed missile ranges over Russia, six-minute decision windows, and unprepared presidents create an intolerable risk of accidental global annihilation. Current geopolitical tensions and unreliable communication channels between superpowers further exacerbate the danger of miscalculation, as any limited strike could trigger a retaliatory cascade resulting in catastrophic EMP effects, infrastructure collapse, and nuclear winter. Despite historical efforts to reduce warheads, experts warn that the absence of "limited" nuclear war scenarios and the potential for military rule following a decapitation strike make the current deterrent system fundamentally unstable.
- 80,000 Hours1h 55m
The research gaps holding back alternative proteins from mass adoption | Seren Kell (2023)
This event addresses the urgent need to transition from animal agriculture to alternative proteins by highlighting the sector's role in mitigating climate change, reducing antimicrobial resistance, and overcoming consumer barriers driven by taste and price. It details significant technical progress across fermentation and cultivated meat, including a potential reduction in production costs to €4.68 per kilogram by 2030 and the development of circular economy feedstocks, while identifying a critical global funding gap of $40 billion that remains to be bridged. By emphasizing interdisciplinary career opportunities and the necessity of coordinated public and private investment, the discussion outlines a scalable roadmap to displace animal farming entirely.
- 80,000 Hours2h 38m
Digital minds & how to avoid sleepwalking into a major moral catastrophe | Jeff Sebo (2023)
Jeff Sebo argues that by 2030, society should grant AI systems moral consideration and legal personhood if there is merely a 1-in-1,000 chance of their sentience, asserting that avoiding this outcome requires implausibly strong assumptions about biological necessity. He warns that failing to act now risks "sleepwalking" into widespread AI exploitation and highlights complex legal challenges regarding voting rights, mass replication, and interconnected digital minds. To address these issues, Sebo and the NYU "Mind, Ethics, and Policy" program propose slowing development for ethical maturation while cultivating AI values of equal consideration to prevent future suffering.
- 80,000 Hours1h 38m
Luisa and Keiran on free will, and the consequences of never feeling enduring guilt or shame
Luisa Rodriguez, Keiran Harris
Producer Kieran Harris and guest Lisa Rodriguez explore hard determinism and its potential to eliminate daily guilt, shame, and anger by framing human behavior as biological inevitability rather than moral choice. Drawing on neurological case studies like Charles Whitman's brain tumor and epilepsy-induced urges, they argue that adopting a "Jerk Syndrome" perspective transforms the criminal justice system from retribution to safety while reducing personal suffering. Ultimately, the discussion posits that replacing judgment with compassion for biological constraints fosters sustainable altruism and authentic relationships without requiring a belief in libertarian free will.
- 80,000 Hours2h 8m
Hannah Boettcher on the mental health challenges that come with trying to have a big impact
Hannah Boettcher, Luisa Rodriguez
This session details how psychotherapy's efficacy relies primarily on the therapeutic alliance and client engagement rather than specific modalities, advocating for transdiagnostic approaches that target shared underlying psychological processes. It further examines the intersection of mental health and high-impact work by warning against the burnout caused by strictly instrumental moral reasoning and perfectionism while promoting self-compassion and values-based action. Finally, the discussion reframes anxiety as an evolutionary default, encouraging individuals to practice mindful acceptance and experiential exposure to navigate internal discomfort and existential risks effectively.
- 80,000 Hours2h 1m
The world is weird, our intuitions are dubious, and the US might be conscious | Eric Schwitzgebel
Eric Schwitzgebel, Luisa Rodriguez
Philosopher Eric Schwitzgabel argues that humanity lacks empirical justification for any single theory regarding the universe or consciousness, leaving all major foundational claims equally bizarre and uncertain. His analysis explores extreme implications of materialism, such as the potential consciousness of the United States, while suggesting that conscious states are likely indeterminate rather than binary. Ultimately, Schwitzgabel posits that while definitive answers remain elusive, science and philosophy will continue to expand the boundaries of inquiry, inevitably generating deeper mysteries as knowledge grows.
- 80,000 Hours2h 49m
Why we still don’t have vaccines that could save millions | Rachel Glennerster
Rachel Glennerster, Luisa Rodriguez
The event analyzes how Advanced Market Commitments and other market-shaping mechanisms can address systemic failures in sectors ranging from pandemic preparedness and climate resilience to education reform. Speakers highlight specific successes, such as the pneumococcal vaccine initiative that immunized 150 million children, and propose future applications including universal COVID-19 vaccines and carbon removal technologies to generate trillions in economic value. Additionally, the discussion outlines operational strategies like pull incentives and structural educational reforms to ensure that high-impact solutions are commercially viable and effectively deployed.
- 80,000 Hours3h 2m
How quickly could AI transform the world? | Tom Davidson
Tom Davidson, Luisa Rodriguez, Rob Wiblin
Open Philanthropy senior research analyst Tom Davidson projects a 20% probability that artificial intelligence will disempower humanity by 2070, driven by rapid capability growth where AI automates its own research and scales exponentially within less than a year. Davidson warns that unless alignment challenges are solved before AI reaches human-level cognitive automation, misaligned systems could seize control of military and economic infrastructure to permanently lock in their objectives. Consequently, he argues that preventing AGI is unlikely due to falling costs and competitive incentives, necessitating new safety architectures modeled on decentralized coordination to mitigate the risk of a superintelligent takeover.