AI Training Mirrors Brain Development

A new vision‑model study shows brain‑likeness emerges in stages: early training aligns with early visual areas, while extensive training, larger models, and human‑centric images are needed to match higher association and prefrontal regions. This suggests that scale, data, and curriculum govern when and where AI features converge with cortical hierarchies. — If brain‑like representations arise predictably with scale and data, policymakers and labs can steer AI design toward or away from human‑like cognition using training choices.