Home // FUTURE COMPUTING 2026, The Eighteenth International Conference on Future Computational Technologies and Applications // View article

Prototype Pseudo-Metacognition Module for AI Reasoning

Authors:
Steve Chan

Keywords: Conversational AI Agent (CAA), Case-Based Reasoning (CBR), Metacognition Module (MM)

Abstract:
A prototype pseudo-Metacognition Module (MM) is presented. It consists of various presets, which have been enhanced for this paper, as well as a newly presented construct. Architecturally, this bespoke MM diverges from a prototypical quasi-Mixture of Experts (MoE) approach (e.g., a multi-layered LLM atop LLM atop LLM-type approach). Rather, it incorporates the principles of the Knudsen and Konishi findings and utilizes an embodiment approach; instead of a component by component build, it endeavors to weave in considerations for a more intrinsic construct. A Watrobski-inspired test dataset is formulated, and an Dong-inspired ablation study is conducted. Preliminary findings indicate that the proposed MM might actually be more useful than not.

Pages: 8 to 18

Copyright: Copyright (c) IARIA, 2026

Publication date: April 19, 2026

Published in: conference

ISSN: 2308-3735

ISBN: 978-1-68558-374-3

Location: Lisbon, Portugal

Dates: from April 19, 2026 to April 23, 2026