Tag
An analysis of AI compute usage reveals that frontier labs like OpenAI, Anthropic, xAI, Google, and Meta currently use less than half of global AI compute, but their share is growing rapidly, which could impact scaling trends.
The article analyzes the concept of 'model half-life' by compiling release dates of major AI models from frontier labs, finding that while release cadence has increased, the notion of a continuously halving release time is misleading. The author provides a TSV dataset and a prediction method.
An analysis of why top AI researchers at frontier labs earn vastly more than their peers, drawing parallels to superstar dynamics in sports and music.
Frontier AI labs are prioritizing recursive self-improvement through coding agents as a key research direction.
Google DeepMind Pre-Training Lead Vlad Feinberg detailed the key skills required to land a job at a top AI lab, emphasizing the importance of infrastructure engineering, understanding scaling laws, and research intuition, and noted that all labs have a huge demand for different skill sets.