Agentic data science with the Data Agent Kit
cloud.google.com·7h ago
AI Research
arXiv CS.AI·Aug 7·5 min read
Estimating time spent on work tasks
TL;DR
A new method estimates time spent on nearly 18,000 work tasks, improving how task weights are calculated in economic models. This approach reveals significant shifts in job exposure to AI, particularly favoring analytical roles over clerical work.
✦ Why It Matters
Reassess job roles in your organization based on new time-weighted exposure to AI.
Key Takeaways
- Utilize the new time shares for accurate job analysis.
- Reevaluate task assignments based on AI exposure metrics.
- Consider shifting focus from clerical to analytical roles.
- Incorporate time-weighted data in labor policy discussions.
How It Works
The method combines task frequency data from ONET with time estimates derived from pairwise comparisons. Language models are used to determine which tasks take longer to complete, creating a constraint satisfaction problem that helps estimate time shares for each task.