Imitation versus Innovation: What children can do that large language and language-and-vision models cannot (yet)?

TL;DR

This paper compares the imitation and innovation abilities of children and AI models, highlighting that current large models lack certain skills children possess, especially in designing new tools and discovering causal structures.

Contribution

It offers a new perspective on AI as cultural technologies and evaluates their capacities relative to children's abilities in imitation and innovation.

Findings

AI models are efficient imitation engines but lack certain creative and causal reasoning skills.

Children can design new tools and discover causal structures that current models cannot.

AI may require additional types of data and learning techniques to match children's capabilities.

Abstract

Much discussion about large language models and language-and-vision models has focused on whether these models are intelligent agents. We present an alternative perspective. We argue that these artificial intelligence models are cultural technologies that enhance cultural transmission in the modern world, and are efficient imitation engines. We explore what AI models can tell us about imitation and innovation by evaluating their capacity to design new tools and discover novel causal structures, and contrast their responses with those of human children. Our work serves as a first step in determining which particular representations and competences, as well as which kinds of knowledge or skill, can be derived from particular learning techniques and data. Critically, our findings suggest that machines may need more than large scale language and images to achieve what a child can do.