Maryland State Innovation Team

all_occupation_task_ratings.csv

This file contains AI-generated ratings of the exposure of specific occupational tasks to AI automation. Each row represents a single task associated with an occupation, along with a rating indicating the degree to which AI can perform that task.

Columns:

• occupation_title: The name of the occupation (string).
• task: A specific task performed within the occupation (string).
• rating: An integer from 1 to 5 representing the task’s exposure to AI automation:
  • 1: AI cannot perform the task at all
  • 2: AI can perform the task with assistance from a human operator
  • 3: AI can perform the task as well as an average human
  • 4: AI can perform the task as well as an expert human
  • 5: AI can perform the task better than an expert human

Source:
Ratings are generated using large language models (Google Gemini or OpenAI GPT) via the MD-Workforce-AI-Impacts codebase. See python/rate_tasks.py for details on the rating methodology and prompt.

Related files:

• all_occupation_tasks.csv: The input file listing all occupation-task pairs to be rated.