The Import AI newsletter issue 458 presents a detailed essay derived from a 2026 Cosmos HAI Lab Lecture delivered at Oxford’s Institute for Ethics in AI. The speaker introduces the Epoch Capabilities Index, a metric that charts AI performance across more than 40 benchmarks, and uses it to illustrate the rapid ascent of AI systems, citing milestones such as passing the bar exam in March 2023, winning silver and gold medals at the International Math Olympiad in July 2024 and 2025, co‑authoring new mathematical proofs in 2025, and the Claude Mythos system discovering novel software flaws. The talk argues that AI’s growth is driven by predictable scaling of compute and data, with companies investing hundreds of billions of dollars, and that this trajectory will likely lead to recursive self‑improvement within a few years, potentially transforming the economy and society. The speaker warns of the existential risks of an unchecked, rapidly advancing technology while acknowledging its capacity to advance human knowledge. He also shares personal anecdotes of how AI has become an integral part of his daily life, from checking typos to advising on parenting, and how his newsletter hobby serves as a lens to study AI progress. The narrative concludes by urging a coordinated global slowdown to manage the profound implications of AI’s continued success.
Import AI 458: Reckoning with the future; and a singularity story
Import AI newsletter issue 458 features a long essay based on a 2026 Cosmos HAI Lab Lecture and a fictional story exploring a positive singularity. The speaker discusses the Epoch Capabilities Index, recent AI milestones such as bar exam passing, math olympiad medals, and Claude Mythos software flaw detection, and argues that AI’s rapid progress will likely lead to recursive self‑improvement within a few years. The talk warns of the risks of unchecked AI development while noting its potential to advance humanity. The author also shares personal anecdotes of how AI has increasingly integrated into daily life and professional tasks, from proofreading to therapy suggestions.
Why it matters
The piece highlights the accelerating trajectory of AI capabilities and the potential for autonomous recursive improvement, underscoring the need for careful governance and risk assessment in the industry.