The Acquisition of Physical Knowledge in Generative Neural Networks
This work addresses the problem of understanding how AI models learn compared to humans, but it is incremental as it builds on existing developmental research without achieving new SOTA or broad impact.
The study compared the learning trajectories of deep generative neural networks to children's developmental trajectories in physical understanding, finding that the models' trajectories did not align with those of children under two hypotheses of human development.
As children grow older, they develop an intuitive understanding of the physical processes around them. Their physical understanding develops in stages, moving along developmental trajectories which have been mapped out extensively in previous empirical research. Here, we investigate how the learning trajectories of deep generative neural networks compare to children's developmental trajectories using physical understanding as a testbed. We outline an approach that allows us to examine two distinct hypotheses of human development - stochastic optimization and complexity increase. We find that while our models are able to accurately predict a number of physical processes, their learning trajectories under both hypotheses do not follow the developmental trajectories of children.