Interpretable by AI Mother Tongue: Native Symbolic Reasoning in Neural Models

TL;DR

This paper introduces a framework where neural models develop an intrinsic symbolic language, called AI Mother Tongue, enabling transparent reasoning, interpretability, and symbolic understanding directly within the model's representations.

Contribution

It proposes a novel approach for embedding reasoning into neural models through a native symbolic language, improving interpretability and symbolic reasoning capabilities.

Findings

Achieves competitive accuracy on AI tasks.

Provides verifiable reasoning traces.

Enhances interpretability and symbolic reasoning in neural models.

Abstract

We present a framework where neural models develop an AI Mother Tongue, a native symbolic language that simultaneously supports intuitive reasoning, compositional symbol chains, and inherent interpretability. Unlike post-hoc explanation methods, our approach embeds reasoning directly into the model's representations: symbols capture meaningful semantic patterns, chains trace decision paths, and gated induction mechanisms guide selective focus, yielding transparent yet flexible reasoning. We introduce complementary training objectives to enhance symbol purity and decision sparsity, and employ a sequential specialization strategy to first build broad symbolic competence and then refine intuitive judgments. Experiments on AI tasks demonstrate competitive accuracy alongside verifiable reasoning traces, showing that AI Mother Tongue can serve as a unified mechanism for interpretability, intuition, and symbolic reasoning in neural models.