Anthropomorphizing Generative AI: is Memory All You Need?

Engineering notes

Engineering notes will be added by the aipentium editorial team.

Chinese explanation / 中文解读

中文解读待补充：本站会优先为大语言模型、生成式AI、ChatGPT相关技术、计算机视觉、深度学习等高价值论文补充中文说明。

Original abstract

As GAI adoption expands, its capabilities increasingly foster anthropomorphic perceptions. Drawing on 21 interviews from an ongoing grounded theory study at a U.S.-based organization using Microsoft 365 Copilot alongside ChatGPT, we find that Generative Artificial Intelligence (GAI) memory, understood as the ability to retain and dialogically recall past interactions, together with interactional alignment, fuels anthropomorphization. By enabling GAI to filter and shape responses in line with users’ preferences, memory creates the impression that the system understands users’ worldview and emotions, thereby supporting perceptions of relationality and human-like interaction.

Links and sources

Comments