80k After Hours

Beth Barnes, CEO of METR, leads the charge in assessing AI models' capabilities and risks. In this intriguing discussion, she reveals that AI can now tackle expert-level tasks in under 30 minutes, a drastic shift from earlier benchmarks. Barnes emphasizes the necessity of rigorous external audits for AI safety, arguing that internal checks alone may not suffice. Excitingly, she forecasts the arrival of recursively self-improving AI in just two years, prompting urgent conversations about accountability and testing before deployment.

Ian Dunt, a British author and political journalist, delves into the dysfunction of government structures. He highlights how a lack of understanding among ministers and frequent turnover among civil servants hinders effectiveness. Dunt discusses the dangers of expanded delegated legislation and advocates for independent-minded MPs. He proposes reforms for better governance, including proportional representation to enhance democratic integrity. The conversation provides insightful critiques of systemic challenges within British politics.

Tom Davidson, a Senior Research Fellow at the Forethought Centre for AI Strategy, delves into the unsettling implications of AI on power dynamics. He discusses how advanced AI could facilitate unprecedented coups by small elites, diminishing democratic oversight. Topics include the potential for military automation, the historical patterns of technology reshaping governance, and the critical need for transparency in AI deployments. Davidson stresses the risks of concentrated control and the importance of ethical considerations in navigating this new landscape.

In this enlightening discussion, Buck Shlegeris, CEO of Redwood Research and a pioneer in AI control, dives into the urgent need to manage misaligned AIs. He explains innovative techniques to detect and neutralize harmful behaviors, emphasizing the critical importance of proactive monitoring. The conversation also touches on the tension between corporate ambition and AI safety, exploring whether alignment strategies can truly keep us safe. Shlegeris advocates for small, focused teams to drive change from within the industry.

In a farewell chat, Matt Reardon, soon to lead the programs team at the Institute for Law and AI, shares his journey from London to the U.S. and Korea. He reflects on valuable lessons learned at 80k and the bittersweet nature of change. The conversation then shifts to navigating the complexities of AI governance, discussing Section 230 and ongoing safety concerns. They also explore the social dynamics of workplace interactions through playful anecdotes from a retreat board game, blending humor with insights on authenticity and career transitions.

Will MacAskill, a philosopher and AI safety researcher at the Forethought Centre, discusses the staggering potential of AI to compress a century's worth of change into a mere decade. He emphasizes the urgent need to prepare for rapid societal shifts and explores what a positive future with AGI might look like. MacAskill raises crucial concerns about the risks of societal lock-in and public distrust in utopian visions. He also delves into the ethical dilemmas surrounding AGI development and its profound impacts on governance and social values.

Allan Dafoe, Director of Frontier Safety and Governance at Google DeepMind, dives into the unstoppable nature of technology and its historical patterns. He explains how societies embracing new capabilities outpace those that resist. The discussion highlights the balance of offense and defense in technology, the complexities of AI cooperation, and the potential risks of backdoor technology in AI models. Dafoe also reflects on how human agency shapes tech development, emphasizing the need for ethical decision-making in AI's future.

Watch this episode on YouTube! https://youtu.be/IRRwHCK279EMatt, Bella, and Huon sit down with Chi Nguyen to discuss cooperating with aliens, elections of future past, and Bad Billionaires pt. 2.Check out: Matt’s summer appearance on the BBC on funding for the artsChi’s ECL Explainer (get in touch to support!)

Sam Bowman, an economist and editor of Works in Progress, dives into the housing crisis in developed countries, emphasizing the powerful grip of NIMBYism. He presents innovative solutions like street votes to empower local residents and discusses property tax distribution's significant role. Sam argues that overcoming NIMBY opposition requires rethinking incentives and highlights the surprising local support for nuclear power. He also touches on the intersection of technology in public health, especially concerning obesity and food choices.

Cameron Meyer Shorb, the Executive Director of Wild Animal Initiative, leads a fascinating discussion on improving wild animal welfare. He tackles the misconception that there's nothing we can do to help these creatures, exploring innovative solutions like selective reforestation and vaccine programs. Shorb also delves into the vast population of often-overlooked terrestrial arthropods and challenges the notion that nature is always good. The conversation even touches on cutting-edge gene drive technologies to combat harmful parasites, sparking hope for wildlife welfare.