Haslberger, M. and Gingrich, J. and Bhatia, Jasmine (2023) No great equalizer: experimental evidence on AI in the UK labor market. SSRN.
|
Text
SSRN-id4594466 (1).pdf - First Submitted (AKA Pre-print) Download (1MB) | Preview |
Abstract
Generative artificial intelligence is already transforming how people work. There is an emerging consensus in early studies that it reduces inequalities in performance within specific occupational groups; however, the question of whether these results generalize to the labor market at large remains open. We conducted a pre-registered online experiment with a representative sample of the UK working-age population. We randomly assigned participants to treatments that encouraged or discouraged the use of ChatGPT and then asked them to complete a set of tasks of varying complexity and ambiguity. We find that exposure to ChatGPT increased productivity in all tasks, with greater benefits observed in more complex and less ambiguous tasks. ChatGPT did reduce performance inequality within occupational groups in most cases, but not between educational or occupational groups. Inequalities between younger and older workers even increased. This study indicates that generative AI has the potential to improve worker performance in a wide array of tasks, but the impact on aggregate inequalities is likely to depend on task-specific features and workers' characteristics.
Metadata
| Item Type: | Other |
|---|---|
| School: | Birkbeck Faculties and Schools > Faculty of Humanities and Social Sciences > School of Social Sciences |
| Depositing User: | Jasmine Bhatia |
| Date Deposited: | 19 Oct 2023 12:09 |
| Last Modified: | 19 Oct 2023 15:31 |
| URI: | https://eprints.bbk.ac.uk/id/eprint/52238 |
Statistics
Additional statistics are available via IRStats2.
Archive Staff Only (login required)
