Enhancing AI Capabilities on the Abstraction and Reasoning Corpus: A Path Toward Broad Generalization in Intelligence
Keywords: broad generalization
TL;DR: Advancing AI generalization through tasks like those in the ARC, exploring neural-symbolic systems and large language models, while emphasizing data diversity and mathematical reasoning to bridge gaps between AI and human-like intelligence
Abstract: This position paper explores advancing artificial intelligence by improving its ability to generalize beyond training data, a key requirement for tasks in the Abstraction and Reasoning Corpus (ARC). Inspired by historical algorithmic challenges like the Bongard Problems, ARC tasks require pattern recognition and logical reasoning, pushing AI toward more flexible, human-like intelligence. We investigate DreamCoder, a neural-symbolic system, and the role of large language models in ARC. We emphasize the need for diverse data sources, inspired by human trials and synthetic data augmentation, and propose pipelines for logical reasoning using math-inspired neural architectures. This work underlines how ARC can guide AI research, bridging the gap between machine learning and mathematical discovery.
Submission Number: 6
Loading