︎


IV.    Research



i.    Research Question

Through the experiment on interactions between different types of language models and different functional human prompts, this project aims to explore how artificial intelligence parse and interpret emotional-constructed texts, what human obtain and are influenced by these process on how they observe the technologically mediated entities.


ii.    Training Dataset

  • Vincent Van Gogh’s diary from Arles to Auvers (50,000 tokens);
  • Ziyu’s personal diary (10,000 tokens);
  • GSAPP website News section from 2018 to 2023(40,000 tokens).


*GSAPP: Graduate School of Architecture, Planning and Preservation, Columbia University


iii.    Tools

  • Processing (Java) 
  • Google Colab (Python)

iv.   Training result

        ︎ Visit GPT Diary




S