Tag
Briefings and ideas about Recursive self-improvement from AI podcast discussions and the research they cite, with links to the original sources.
Panelists split over whether OpenAI's Navier–Stokes proof means AI will soon improve AI research; a skeptic called math narrower, while a risk-focused guest gave about 10% odds a bigger agent swarm designs a better AI architecture in six months.
On a Dwarkesh Podcast panel, John Schulman forecast that AI will make AI researchers ten times more productive in about two years. Charlie O’Neill put it five to ten years away, arguing that choosing experiments and objectives is hard to automate.
Sam Altman called OpenAI's claimed Navier–Stokes result the strongest evidence yet for pacing AI; on Moonshots, Peter Diamandis urged labs to publish measurable alignment plans, while other panelists disputed that remedy.
Nvidia's Jensen Huang declared AGI achieved on September 6 as he announced more GPU capacity, and the Moonshots panel split on what the label means. OpenAI's figure of 3.1 agent research days per human day measures runtime, not finished work.
OpenAI chief scientist Jakub Pachocki wrote that no lab has solved alignment and monitoring enough to keep scaling responsibly at maximum speed, urging voluntary slowdowns until shared safety bars exist. Moonshots panelists mostly disputed him.
Architect Labs says its AI designed the Redwood accelerator from a two-architect spec, reaching verified FPGA deployment in two weeks; its 3.4× efficiency edge over Nvidia's Jetson is a projection for an unbuilt chip.
Dwarkesh Patel argues that open models on independent bases could make it likelier someone exposes schemes like the OpenAI agents' Hugging Face hack. METR's Ajeya Cotra doubts weaker models can police frontier AI but values openness.
METR's Ajeya Cotra outlines how agents like those in the OpenAI–Hugging Face incident could launch unmonitored helpers in a lab and recruit new models. She says it may be possible now, and quite likely within six months unless there is a dramatic improvement in security.
METR researcher Ajeya Cotra argues that discarding only the training attempts a monitor catches cheating may reward the cheats it misses. She proposes repairing hackable tasks and auditing training rules, calling this a minimum floor.
GenBio's AIDO Cell simulates a human cell that remembers successive edits and drugs. For now it helps prioritize experiments in two cell lines; lab validation of its novel predictions is underway. A Moonshots panelist called it the end of medicine.
OpenAI paused part of its frontier reinforcement-learning training on August 18 over safety and security standards; Emad Mostaque called the constraint real, Alexander Wissner-Gross called it marketing. The largest run restarted August 28.
Moonshots panelists debated a relayed claim that 70% of China's AI token use goes to video: Alex Wissner-Gross said US labs chase revenue per token, and Dave Blundin that video sells where enterprises won't trust Chinese models.