Can We Predict What Jobs AI Will Take?
Name
hr6dsnfypefc9ad1vgrcrcshet3w4nlz.pdf
Size
277.79 KB
Format
Adobe PDF
Checksum (MD5)
c0244cab4de2c9123d9a6c20d307c695
Author(s)
Davenport, Thomas
Paredes, Miguel
Date Issued
October 31, 2025
Journal
Harvard Data Science Review
Publisher
MIT Press
Citation
Davenport, T., & Paredes, M. (2025). Can We Predict What Jobs AI Will Take? Harvard Data Science Review, 7(4).
Version
Final published version
Abstract
We have studied various analyses of jobs predicted to be automated out of existence by AI, defined broadly as machine learning, deep learning, robotics, gen AI, and so on; a common attribute of these predictions is that they are wildly inaccurate. Task automation assessment, the most systematic approach, has been ineffective at predicting the number of jobs to be lost to AI. Observers would benefit by combining task-based approaches with more contextual factors about AI adoption, existing technology, jobs gained through AI, and broader economic indicators. There are no systematic approaches to this issue that are likely to yield accurate predictions, so the most responsible approach may be to discontinue making them, or admit that any predictions are highly speculative.
Terms of Use
Creative Commons Attribution
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1162/99608f92.8975ddd1
