随着大语言模型突破，部署到软件系统中的 AI 智能体数量快速增长；作为软件系统，它们必须持续现代化以避免系统萎缩。本文报告博士论文的早期研究，旨在将多智能体系统与“包容性继承”相结合，即引入同时依赖遗传继承与非遗传继承的演化原则。研究计划为软件系统概念化并实现包容性继承，并评估这种整体性方法是否比仅依赖部分演化维度更能释放软件系统生物启发式演化的潜力。目前处于早期阶段，尚未给出具体实现或实验结果。
With recent breakthroughs in the field of large language models, the number of “AI agents” deployed into software systems is growing rapidly. As software systems, they must be continuously modernized to avoid system atrophy. In this paper, we present our early-stage research to infuse multi-agent systems with inclusive inheritance, i.e., with evolutionary principles that rely on both genetic and non-genetic inheritance. As part of this doctoral thesis, we aim to conceptualize and implement inclusive inheritance for software systems and evaluate if this holistic approach leverages the potential of bio-inspired evolution of software systems better than relying only on a selection of evolutionary dimensions.
@inproceedings{paperbot11491,
title = {Evolving Artificial Ecosystems: Towards the Infusion of Multi-Agent Systems with Inclusive Inheritance},
author = {Alessandro Giugno},
booktitle = {ICSE},
year = {2026},
doi = {10.1145/3774748.3787674}
}