Building Machines that Learn and Think with People

TL;DR

This paper advocates for developing AI systems as collaborative thought partners that reason with humans, emphasizing the importance of human-compatible, trustworthy, and insightful AI through a cognitive science and Bayesian approach.

Contribution

It introduces a framework for designing AI as collaborative thought partners using cognitive science principles and Bayesian reasoning to better align with human thought processes.

Findings

Proposes modes of collaborative cognition between humans and AI.

Suggests a Bayesian approach for building adaptive, human-aware AI systems.

Highlights the importance of trust and insight in AI-human partnerships.

Abstract

What do we want from machine intelligence? We envision machines that are not just tools for thought, but partners in thought: reasonable, insightful, knowledgeable, reliable, and trustworthy systems that think with us. Current artificial intelligence (AI) systems satisfy some of these criteria, some of the time. In this Perspective, we show how the science of collaborative cognition can be put to work to engineer systems that really can be called ``thought partners,'' systems built to meet our expectations and complement our limitations. We lay out several modes of collaborative thought in which humans and AI thought partners can engage and propose desiderata for human-compatible thought partnerships. Drawing on motifs from computational cognitive science, we motivate an alternative scaling path for the design of thought partners and ecosystems around their use through a Bayesian lens, whereby the partners we construct actively build and reason over models of the human and world.