Bob McGrew explains that AI progress relies on pre-training, post-training, and reasoning. He sees diminishing short-term returns in pre-training due to multi-year infrastructure projects and expects reasoning to dominate research in 2025.
He predicts the foundations of artificial general intelligence will be in place by 2035, driving agent services toward compute-cost pricing and forcing startups to seek defensible niches.
Comments
Nothing yet. Say the first thing.
Sign in to join the conversation.