The Nonlinear Library

Constance Li, writer from Rationalist and EA communities, discusses the AIADM 2024 event in London. They cover the objective, collaboration challenges, and outcomes of the conference. Topics include developing safe AI tech for nonhumans, unconventional conference formats, and logistical hurdles faced by participants.

Yair Halberstadt, author of 'Surviving Seveneves', discusses how the moon's explosion leads to the complete destruction of Earth in 'Seveneves'. They explore the choice to go into space for survival, challenges of living in low orbit, and strategies for post-apocalyptic survival like establishing shelters and self-sustainability.

Ilya Sutskever, founder of a new AGI startup, discusses the launch of Safe Super Intelligence Inc. The company focuses on building safe superintelligence with a unique approach to advancing capabilities while prioritizing safety. They aim to assemble a top engineering team and remain dedicated to achieving SSI.

A discussion on the conflation between suffering and pain, emphasizing the importance of distinguishing between the two. The Buddhist parable of the second arrow is used to illustrate how suffering is optional. Pain is described as an unpleasant sensation, while suffering arises from attachment and aversion. The podcast explores how clarifying these definitions can help alleviate unnecessary suffering.

Sasha Berezhnoi discusses the details of EAGxBerkeley 2024 conference, focusing on cause areas, community building, and networking opportunities for attendees dedicated to addressing global challenges.

AI researcher AdamGleave discusses how human amateurs can beat superhuman Go AIs with cyclic patterns, highlighting vulnerabilities in AI systems. He explores defense strategies like adversarial training and vision transformers to enhance AI robustness against emerging threats.

Discover the insights from the Meta Charity Funders second round retrospective, including details about funded projects like AI Assurance Tech Report and EA Indonesia, as well as the significant increase in grant amount from the previous round.

Joe Carlsmith discusses the duality of activity versus receptivity, deep atheism, and control-seeking tendencies in the context of AGI. He explores themes of defiance, resilience, and reverence towards imperfection, highlighting the challenges of AI advancement and the importance of cultivating peace and human values in a changing world.

A discussion on boycotting OpenAI due to ethical concerns including threats to departing employees and dishonesty by Sam Altman. Emphasizing the need to hold AI companies to higher standards and the potential impact of competent researchers avoiding OpenAI.

Discussion on DeepMind's safety framework compared to OpenAI and Anthropics, focusing on critical capability levels and risk monitoring. Suggestions for improvements include incorporating risk modeling, detailed mitigation strategies, updated risk assessments, and involving external stakeholders for comprehensive safety measures.