Multi-Agent AI Systems For Cognitive Development - Mentor
Year: 2025
Type: dissertation
Abstract: The pedagogical change led by the rapid integration and adoption of artificial intelligence in architectural education highlights one challenge: although those new tools enable rapid design generation or analysis, the way they are used can lead to "cognitive offloading" a phenomenon where instead of having essential reasoning skills students end up delegating critical thinking to external AI systems. This research emphasizes current concerns about the 74% of recent graduates who lack independence in design thinking as a result of AI adoption. This paper presents the developments and foundations of MENTOR (Multi-agent Educational Guidance for Architecture), a novel AI mentoring tool that is designed to prevent cognitive offloading as well as enhancing critical thinking in architectural design education. In contrast to the current conventional AI tutoring systems with their direct answering mechanism, MENTOR consists of a sophisticated multi-agent architecture comprising five specialized agents: Context Reasoning, Socratic Dialogue, Domain Expert, Cognitive Enhancement and Analysis agents, orchestrated through LangGraph workflow management.
Authors: Fatima, Mahnoor, Soylu, Seda, Pitman, Biel
Institution: Institute for Advanced Architecture of Catalonia
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7108633415
Cites:
Cited by:
Related to: