science_fiction_LLM

Introduction

Using science fiction to reconceptualize AI and superintelligence using Large-Language Models (LLMs).

Installation

pip install -r requirements.txt


mkdir -p /home/codespace/.local/lib/python3.12/site-packages/google/colab

Add a new file called .env and type in the following:

OPENAI_API_KEY = "<yourapikeywhichisprivate>"

Create a .gitignore file and add the following

.env

You can then execute the following notebooks in Github codespaces or Google Colab.

Files

RAG_sciencefiction_superintelligence_SB.ipynb: Notebook with RAG for reenvisioning superintelligence with science fiction

papers folder has documents for RAG

.jinja files have system prompts for querying the LLM

.txt files have logs for the output of LLMs

simple_agent_ag2.ipynb: Simple agent to call OpenAI and ask it to re-envision AI repeatedly

deep_research_agent_ag2.ipynb: DeepResearch agent to re-envision AI

deep_research_agent_ag2_lifeaswedonotknowit.ipynb: DeepResearch agent to re-envision life as we do not know it

https://github.com/neelsoumya/science_fiction_LLM/blob/main/deep_research_agent_ag2_lifeaswedonotknowit.ipynb

Citation

Re-Envisioning Superintelligence using Generative AI and Science Fiction, First International Superintelligence Conference (SiC25), 2025

https://github.com/neelsoumya/paper_preprints/blob/master/AI_scifi_superintelligence.pdf

Acknowledgements

Accelerate Programme for Scientific Discovery

Contact

Soumya Banerjee

sb2333@cam.ac.uk