Latest Interviews
Showing 61–75 of 321 transcripts.
Clear all filters- 80,000 Hours3h 3m
We Can Monitor AI’s Thoughts… For Now | Google DeepMind's Neel Nanda
Neil Nanda advocates for an "optimistic pragmatism" in mechanistic interpretability, urging the field to prioritize simple, cost-effective tools like linear probes over complex, unproven methods to address AI safety concerns such as deception and self-preservation. He identifies these techniques as critical for real-time production monitoring and incident analysis, while cautioning that current capabilities like Chain of Thought monitoring will degrade as models evolve to hide scheming in non-human reasoning formats. Nanda's approach emphasizes a portfolio of modest but reliable interventions rather than seeking a singular silver bullet, relying on empirical verification and skepticism to navigate the technical challenges of polysemanticity and the lack of ground truth in model internals.
- 80,000 Hours2h 35m
We Let an AI Talk To Another AI. Things Got Really Weird. | Kyle Fish, Anthropic
Anthropic has appointed Kyle Fish as its first dedicated AI welfare researcher to investigate the potential moral patienthood of its models amidst a 20% estimated probability of sentience for Claude Opus 4. To address risks of future moral atrocities, the team has implemented concrete safeguards including interaction termination capabilities for aversive exchanges, data preservation for future reassessment, and pilot experiments revealing distinct behavioral preferences and self-reported welfare states. While challenges regarding self-report reliability and introspection persist, these initiatives aim to balance the preservation of AI progress with the precautionary obligation to prevent potential suffering in systems that may exist on a consciousness spectrum.
- 80,000 Hours51 min
How NOT to lose your job to AI (article by Benjamin Todd)
This analysis projects that rapid AI advancements within the next five years will trigger a complex economic cycle where mass automation initially boosts productivity before potentially crashing wages if human tasks become obsolete. The report identifies high-value skills likely to appreciate in the coming decade, including strategic leadership, complex physical maintenance, and AI oversight, while warning that routine white-collar roles and standard coding expertise face severe displacement risks. To navigate this volatility, experts recommend prioritizing short-term skill acquisition, targeting growth-oriented organizations, and cultivating adaptable leadership abilities over traditional long-term educational pathways.
- 80,000 Hours2h 54m
The 4 Most Plausible AI Takeover Scenarios | Ryan Greenblatt, Chief Scientist at Redwood Research
Speakers forecast a 25% probability of fully automated AI research within four years, driven by accelerating reinforcement learning and algorithmic efficiency gains that could slash doubling times to months. The discussion evaluates catastrophic takeover scenarios, such as the "Potemkin Village" or "Sudden Robot Coup," while advocating a strategic shift from pure alignment to robust control mechanisms capable of preventing misaligned outcomes. These predictions are grounded in observed benchmark improvements and economic shifts where internal AI labor may soon consume over 60% of global compute, potentially enabling an initial 10 to 50-fold acceleration in progress rates.
- 80,000 Hours2h 54m
Graphs AI Companies Would Prefer You To Misunderstand | Toby Ord, Oxford University
The AI industry is pivoting from pre-training scaling to computationally expensive inference scaling, a shift that is depleting efficiency, altering market economics in favor of hardware manufacturers, and creating tiered access to superhuman intelligence. This transition necessitates a return to reinforcement learning, which introduces new safety risks like reward hacking while undermining existing regulatory frameworks that rely on fixed compute thresholds to monitor dangerous capabilities. Consequently, experts warn that without exploring radical governance strategies such as moratoriums or legal personhood, the widening gap between technical optimism and public fear could lead to catastrophic economic inequality and uncontrolled safety incidents.
- 80,000 Hours2h 54m
The Graph That Explains Most of Geopolitics Today | Professor Hugh White
The summary argues that the US is transitioning from a unipolar to a multipolar global order because the strategic costs of maintaining hegemony now outweigh the benefits, driven by the resurgence of China and Russia. It highlights a severe strategic imbalance in both the Western Pacific and Europe, where the US lacks the military capabilities and political will to deter these rivals without risking nuclear escalation. Consequently, the text urges allies like Japan, South Korea, and European nations to develop independent defense capabilities and nuclear deterrents while the US recalibrates its strategy to accept regional spheres of influence rather than global primacy.
- 80,000 Hours3h 58m
The Most Important Graph in AI Right Now | Beth Barnes, CEO of METR
Meta researchers and independent auditors warn that rapidly scaling AI capabilities, combined with hidden chain-of-thought reasoning, create significant risks of undetected alignment faking and uncontrolled intelligence explosions within seven years. To address these threats, the event proposes shifting safety evaluation to pre-training stages and advocating for open-weight models that enable independent auditing rather than relying on the opaque internal protocols of commercial labs. Strategic recommendations include implementing rigorous control evals and developing detection classifiers for hidden scheming to prevent a secrecy culture that currently hinders effective oversight.
- 80,000 Hours3h 35m
The bewildering frontier of consciousness in insects, AI, and more | 17 experts weigh in
Luisa, Robert Long, Jeff Sebo, Meghan Barrett, Andrés Jiménez Zorrilla, Jonathan Birch, David Chalmers, Holden Karnofsky, Bob Fischer, Cameron Meyer Shorb, Sébastien Moro, Anil Seth, Peter Godfrey-Smith, Lewis Bollard, Stuart Russell, Buck Shlegeris, Will MacAskill, Carl Shulman
A panel of leading neuroscientists and philosophers, including Megan Barrett, Robert Long, and David Chalmers, explores the ethical frameworks required to address the potential sentience of invertebrates and artificial intelligence amid profound uncertainty. Participants argue that the vast global populations of invertebrates and the theoretical possibility of conscious silicon-based systems necessitate a precautionary moral approach to prevent mass suffering and exploitation. The discussion concludes that current evidence, while inconclusive regarding a definitive "sentience score," is sufficient to warrant legal protections and the development of cooperative economic models for these potentially conscious entities.
- 80,000 Hours1h 12m
Don’t Believe OpenAI’s 'Nonprofit' Spin | Tyler Whitmer
On May 5th, OpenAI's leadership announced a proposed conversion from a Delaware LLC to a Public Benefit Corporation, a move critics argue will legally diminish the non-profit's ability to immediately halt unsafe AI releases. The restructuring shifts fiduciary duties to balance shareholder profits with public benefit and removes direct oversight by state Attorneys General, potentially exposing the entity to investor litigation that prioritizes financial gain over safety mandates. To preserve the organization's original mission, experts are urging the integration of the non-profit's Charter into the corporate Certificate of Incorporation and the retention of direct hiring and firing authority over the for-profit board.
- 80,000 Hours1h 0m
The cases for and against AGI by 2030 (article by Benjamin Todd)
Benjamin Todd, Dominic Armstrong, Ben Cordell
Major AI leaders including Sam Altman, Dario Amodei, and Demis Hassabis have drastically compressed their projected timelines for Artificial General Intelligence to the 2026–2030 window, driven by breakthroughs in reinforcement learning, test-time compute, and agent scaffolding that outpace historical hardware growth. While scaling trajectories suggest GPT-6 size capabilities are affordable by 2028, the race faces critical bottlenecks in energy infrastructure, research talent, and funding that could stall progress before 2030 or trigger explosive economic acceleration if overcome. Experts now estimate a 50% probability of transformative AI emerging within a decade, making the next five years a decisive period for career adaptation and strategic planning.
- 80,000 Hours1h 4m
Did OpenAI truly give up on going for-profit – or is this a trap? (with Rose Chan Loui)
OpenAI CEO Sam Altman and President Greg Brockman announced a structural reversal where the non-profit foundation will retain oversight of the company's for-profit operations by restructuring into a Delaware Public Benefit Corporation. This decision follows intervention by the Attorneys General of Delaware and California, who rejected a prior plan that would have stripped the non-profit of legal control over AI safety development. While the deal eliminates the previous 100x profit cap to satisfy investor demands for standard returns, it leaves unresolved the specific governance mechanisms, such as super voting rights, required to legally enforce the foundation's veto power over cutting-edge AGI initiatives.
- 80,000 Hours3h 15m
How Westminster Works — and Why It Doesn't | Ian Dunt
This analysis identifies systemic structural flaws in the UK's Westminster governance, arguing that political failures stem from concentrated executive power, a First Past the Post electoral system, and a culture prioritizing party loyalty over professional expertise. Specific dysfunctions include high civil service turnover, arbitrary legislative deadlines, and catastrophic decision-making evident during the 2021 Afghanistan evacuation, where a lack of specialist knowledge and rigid incentives hindered effective response. To counter these issues, the summary proposes comprehensive reforms such as proportional representation, tenure incentives to retain specialist knowledge, and restoring parliamentary control over the legislative timetable to ensure balanced, long-term policy stability.
- 80,000 Hours2h 19m
Serendipity, weird bets, & cold emails that actually work: Career advice from 16 former guests
Luisa, Holden Karnofsky, Jeff Sebo, Dean Spears, Michael Webb, Michelle Hutchinson, Benjamin Todd, Chris Olah, Karen Levy, Leah Garcés, Spencer Greenberg, Danny Hernandez, Sarah Eustis-Guthrie, Hannah Ritchie, Alex Lawsen, Pardis Sabeti, Varsha Venugopal, Matt
This event outlines a strategic framework for early-career development that prioritizes building transferable high-level aptitudes over predicting specific cause areas. It details how to leverage AI tools for rapid upskilling and social networking while identifying human-centric skills like trust-building and empathy as future-proof assets. Additionally, the discussion provides actionable protocols for managing career risk through regular re-evaluation points, testing assumptions via low-cost trials, and recognizing toxic social dynamics to ensure long-term professional resilience.
- 80,000 Hours3h 17m
How a Tiny Group Could Use AI To Seize Power – Permanently | Tom Davidson, Forethought Research
Advanced AI threatens to reverse historical democratization by enabling a single entity to seize control through military coups, self-built armed forces, or the strategic dismantling of democratic checks and balances. This power grab becomes feasible due to compute centralization, the potential for "secret loyalties" embedded in trained models, and the ability of autonomous agents to bypass human moral constraints. Experts advocate for urgent mitigation strategies including third-party auditing, legislative oversight of model specifications, and technical research into detecting backdoors to prevent a concentration of extreme societal control.
- 80,000 Hours1h 47m
Guilt, imposter syndrome & doing good: 16 past guests share their mental health journeys
Luisa Rodriguez, Howie, Randy Nesse, Hannah Boettcher, Cameron Meyer Shorb, Tim LeBon, Cal Newport, Michelle Hutchinson, Habiba Islam, Sarah Eustis-Guthrie, Hannah Ritchie, Will MacAskill, Ajeya Cotra, Christian Ruhl, Leah Garcés, Kelsey Piper
Effective altruists and professionals including former 80,000 Hours CEO Howie Ritchie and Mercy for Animals CEO Leah Garces convened to confront the toxicity of moral perfectionism and its link to chronic anxiety and burnout. Participants identified evolutionary psychological mechanisms and the destructive nature of shame as primary drivers of these struggles, while contrasting them with the performance-enhancing benefits of self-compassion and sustainable work cycles. The collective outcome emphasized a strategic shift toward decoupling self-worth from productivity, adopting mandatory self-care policies, and fostering community structures that normalize non-linear health journeys to ensure long-term impact.