Large language models, such as Chat-GPT, have demonstrated remarkable abilities to understand and generate human language. This theme explores their potential to contribute to mathematical reasoning and software creation, extending beyond traditional methods and potentially uncovering novel approaches and solutions.
The utilization of these models in mathematics and software has the potential to augment the creative process, enabling the discovery of new proofs and methods. However, given that these models can sometimes produce incorrect or "hallucinated" information, the theme emphasizes the importance of integrating them with proof assistants and symbolic models. This combination aims to balance the creative suggestions of language models with rigorous validation to ensure mathematical and software correctness.
Researchers within this theme are investigating how intuition and associative functions from large language models can be integrated into rule-based systems, providing new tools that could accelerate research and development in their respective fields.
The ambition of this theme is not only to apply existing rules more efficiently, but also to expand the boundaries of what AI can achieve in mathematics and software. The ability of large language models to quickly solve complex problems suggests a future where AI could independently devise entirely new methods and strategies, a prospect that is particularly exciting to mathematicians and computer scientists.
CHAIR's theme on Large Language Models for Mathematics and Software fosters a collaborative environment through seminars, workshops and other events, with August 2024 as the start. The goal is to build a vibrant community of researchers at Chalmers who are passionate about this cutting-edge intersection of AI with mathematics and software engineering.
