Evaluating Cognitive Age Alignment in Interactive AI Agents

2026-05-18

AI Summary

Current multimodal large language model agents significantly underperform children on fundamental cognitive tasks and lack a systematic evaluation framework aligned with human cognitive developmental stages. Inspired by the Wechsler Intelligence Scale for Children, this work proposes ChildAgentEval—the first psychometrically grounded, interactive benchmark that introduces a psychological measurement framework to the evaluation of AI agents. This approach systematically assesses agent performance across distinct human cognitive age levels, revealing both strengths and critical shortcomings in their ability to emulate age-specific cognitive behaviors. By establishing a quantifiable reference tied to established developmental milestones, ChildAgentEval provides a rigorous foundation for measuring and advancing artificial cognitive capabilities in alignment with human development.

Abstract

While agentic AI and its core multimodal large language models (MLLMs) have demonstrated remarkable promise in language and visual reasoning across domains ranging from daily life to advanced scientific research, a profound gap remains between artificial and human intelligence. Despite the integration of powerful tools and advanced MLLMs, state-of-the-art AI agents frequently fail at foundational, seemingly simple tasks that a child can resolve with ease. Inspired by the Wechsler Intelligence Scale for Children (WISC), we introduce ChildAgentEval, the first psychometrically grounded interactive benchmark for evaluating cognitive age alignment in MLLM-based agents. ChildAgentEval systematically compares the reasoning performance of various MLLM-based interactive agents against age-specific human developmental stages, exposing where current agentic AI systems can and cannot simulate age-specific cognitive behavior.

Problem

• cognitive age alignment
• interactive AI agents
• multimodal large language models
• human intelligence gap
• developmental reasoning

Innovation

• cognitive age alignment
• ChildAgentEval
• multimodal large language models
• psychometric benchmark
• interactive AI agents