This repository contains the code and supporting materials for a study investigating how artificial intelligence expands automation across occupational tasks and how this affects wages, labour share, and economic output.
Traditional automation models often represent technological progress as a one-dimensional expansion of machine capabilities. This project extends the task-based framework of Acemoglu & Restrepo by introducing a two-dimensional task space that separates:
- Cognitive Complexity
- Physical Complexity
Automation capability is modelled using a Fisher-KPP reaction-diffusion process, allowing machine knowledge to spread across related tasks while accounting for local learning effects.
The resulting automation surface is integrated into a CES production framework to examine the effects of automation on:
- Total output
- Wages
- Labour share
- Two-dimensional task complexity grid
- Fisher-KPP reaction-diffusion simulation
- CES production aggregation
- Sensitivity analysis across substitution elasticities (σ)
- Close reading of O*NET task descriptions
- Structured questionnaire framework
- Complexity scoring for cognitive and physical task dimensions
- Case studies: Firefighter and Paediatric Surgeon
├── data/ # O*NET and processed datasets
├── main.ipynb # Analysis and simulation notebook
├── data_output/ # Qual output
└── data/ # Folder to store specific job tasks from the O*NET DB.
Automation is not modelled as a simple binary replacement of labour. Instead, tasks can occupy different positions on a cognitive–physical complexity surface, allowing analysis of gradual and uneven technological diffusion across occupations.
The study builds primarily on:
- Acemoglu & Restrepo (2018, 2019)
- Autor et al. (2003, 2024)
- Frey & Osborne (2017)
- Thompson et al. (2023)
This repository is provided for research and educational purposes.