Search papers, labs, and topics across Lattice.
This paper introduces a toy framework that conceptualizes curiosity as an ecosystem, where a single agent's inquiry policy is influenced by factors such as uncertainty reduction, costs, and the value of keeping questions open. The framework allows for dynamic adjustments in these decision-related weights based on the agent's experiences, illustrating how short-term inquiry costs can shift long-term question preferences. Additionally, it extends these concepts to multi-agent scenarios, enabling the analysis of inquiry volume, topic diversity, and knowledge reuse in collaborative exploration, thereby providing a foundation for future multi-agent AI systems focused on discovery.
Curiosity in AI isn't static; it evolves based on experience, reshaping how agents prioritize questions and explore knowledge landscapes.
This paper offers a toy framework for considering curiosity as an ecosystem. First, it suggests that a single agent's inquiry policy (how, when, and why an agent asks a question) depends on how the agent values immediate uncertainty reduction, costs, delayed return, and the value of keeping the question open. A key concept in the framework is that the weights on these decision-related terms can change with experience. For example, a period of cheap, quickly answered questions may change the cost of inquiry on a short timescale and change which kinds of questions the agent is drawn to answer over a longer timescale. Second, these ideas are extended to many agents exploring a shared knowledge landscape, and there the framework tracks inquiry volume, topic diversity, frontier-directed inquiry, redundancy, and reusable knowledge. The result is a conceptual toy framework for studying curiosity ecology and for future efforts towards designing multi-agent AI systems for discovery. It serves as a companion piece for a paper currently under review in Trends in Neurosciences.
