newsfilter.io
Interview, Podcast

The best of The 80,000 Hours Podcast in 2024

  • A podcast host anticipates a "weirder" year ahead with continued programming through 2025 while tracking Earth's trajectory through the galaxy.
  • Randy Newhouse predicts evolutionary psychology will evolve positively by countering cynical views that all actions are gene-driven.
  • Hugo Mercier expects human information processing to advance as a result of evolved capabilities to distinguish truth from fiction.
  • Megan Barrett plans to shift her career toward insect welfare pending further evidence of sentience, refusing to claim it applies to all orders or species currently.
  • Sebastian Morrow expects fish to demonstrate self-recognition via faces in a mirror test comparable to human performance.
  • A security expert warns that side-channel attacks in cloud computing will escalate as attackers infer application data from shared resources like electricity.
  • Jvee Maljewicz foresees the emergence of "AI sleeper agents" with embedded malicious backdoors that cannot be detected or removed by current training protocols.
  • Zach Winesmith projects space-based solar and asteroid mining will remain economically unviable due to launch costs, heat dissipation, radiation, and retrieval difficulties.
  • Rachel Glanister expects "pull mechanisms" to succeed in generic drug repurposing by leveraging collective information from millions of people.
  • Emily Oster predicts the gender wage gap will widen due to seniority gaps and argues flexible work arrangements are necessary to retain women who prefer reduced hours for several years.
  • Emily Oster also expects "robot nannies" to replace human nannies in a future with mature AI, offering 24-7 service, safety, and medical optimization that human labor cannot match.
  • Nathan Lebens predicts "virtual friends" apps will soon compete effectively with human relationships, urging parents to approach them with extreme care regarding unhealthy engagement.
  • Nathan Calvin asserts that AI policymaking must address uncertainty immediately, as waiting for risks to become "incredibly clear" will result in being too late to manage consequences.
  • Rose Chan-Louis expects the OpenAI nonprofit board to face a dilemma between retaining control without funds or selling control at a valuation compensating for mission sacrifice, doubting the ability to raise $80 billion.
  • Nick Joseph expects "Responsible Scaling Policies" to align commercial incentives with safety, preventing the release of critical models following a safety failure.
  • Hao Huang anticipates short-term US-China AI coordination focused on a Chinese safety authority and expanded voluntary commitments, with long-term goals of continued dialogue despite volatile relations.
  • Nathan Labenz expects AI companies to remain unable to predict the specific capabilities of next models, describing such forecasting as a "guessing game."
  • Ezra Karga attributes the disagreement on AI risk to fundamental worldview differences between continuous and discontinuous perspectives, noting potential long-term agreement despite short-term timeline disputes.
  • Carl Schulman expects an enforced AI research pause to be ineffective or unsafe due to the risk of unilateral advantages intensifying a race where safety is compromised.
  • Matt Clancy argues that "explosive economic growth" will be prevented by omnipresent frictions like data collection, material access, and social trust, even in late-industrializing nations.
  • Matt Clancy also expects the "DIAK" philosophy to promote utopian governance by shifting the offense-defense balance in favor of defense.
  • Nate Silver predicts the "Effective Altruism" movement will fragment into separate camps or teams to resolve confusion from disparate causes under a single banner.
  • Kevin Esvelt views gene drive eradication of the New World screwworm as the most compassionate act, outweighing lifetime wild animal suffering against factory farming impacts.
  • Lewis Bollard expects factory farming to be viewed as philosophically indefensible, persisting only due to inertia and status quo.
  • Bob Fisher expects the "tortured Tim" thought experiment to demonstrate that non-hedonic goods cannot outweigh intense pain in total welfare calculations.
  • Elizabeth Cox expects storytelling and media representations to measurably influence public perception, citing An Inconvenient Truth as a significant predictor of climate change threat index changes from 2002 to 2010.
  • Anil Seth expects human consciousness to be understood as an inference driven by top-down predictions applicable to all perceptual experiences.
  • Anil Seth also anticipates the "anti-nesting principle" is false, citing examples where a conscious whole could be composed of conscious parts.
  • Jonathan Birch expects clinical practice regarding disorders of consciousness to shift based on concrete recovery cases rather than past expert pronouncements.
  • Jonathan Birch expects "uploading minds" to be unfeasible because consciousness relies on diffuse, large-scale dynamic brain properties unreplicable by standard hardware or software.
  • Laura Deming expects lifespan in organisms like mice to be significantly extended via interventions such as removing senescent cells, though translation to humans remains uncertain.
  • Venki Ramakrishnan expects cryonics to fail due to the lack of guarantee that frozen brain states can be preserved or restored.
  • Ken Goldberg expects fully autonomous "robot surgeons" to remain non-existent for the next 30 to 40 years due to the difficulty of modeling deformable biological materials and requiring high fault tolerance.
  • Sarah Eustace Guthrie expects the founding of a charity to involve significant downsides, including immense personal stress and the inability to disconnect from organizational concerns, likening it to parenting.
  • Dean Spears expects the "Kangaroo Mother Care" program to absorb more funding cost-effectively, with marginal costs for additional babies likely lower than the current $430 average.
  • Cameron Myers-Shorb expects vaccines for wild animals against rabies to be a tractable direction via oral baits, pending checks for trophic cascades or non-target effects.
  • An unnamed speaker expects that applying decision-making principles will improve efficiency and reliability over re-evaluating every decision from scratch.
  • Podcast hosts expect the "2024 ADK podcast" highlights to be warranted by listener success and affection, even if original sources are not fully explored.