Similar AI and brain solutions could hint at shared computational principles

In a CiNeT/University of Osaka talk, a speaker argued that differences between AI and brains could make any similar solutions especially revealing: they might hint at computational properties a task requires. The speaker proposed testing this by changing the structure of data or an AI system’s constraints, and cited two case studies from their work as possible hints: content effects on logical reasoning and latent learning/replay as data augmentation.