The End of the Policy Analyst? Testing the Capability of Artificial Intelligence to Generate Plausible, Persuasive, and Useful Policy Analysis
| dc.contributor.advisor | Longo, Justin | |
| dc.contributor.author | Safaei, Mehrdad | |
| dc.contributor.committeemember | Dupeyron, Bruno | |
| dc.contributor.committeemember | Phillips, Peter | |
| dc.contributor.externalexaminer | Savard, Jean-Francois | |
| dc.date.accessioned | 2022-08-05T16:57:10Z | |
| dc.date.available | 2022-08-05T16:57:10Z | |
| dc.date.issued | 2021-10 | |
| dc.description | A Thesis Submitted to the Faculty of Graduate Studies and Research In Partial Fulfillment of the Requirements for the Degree of Master of Public Policy, University of Regina. vii, 155 p. | en_US |
| dc.description.abstract | Public servants provide support for decision makers through synthesis documents such as briefing notes. To develop recommendations for dealing with the problem, they use a variety of sources for research and analysis. This current research seeks to assess opportunities and challenges regarding the use of artificial intelligence (AI) in public sector administration and policy development, focusing on whether AI can serve as a supplement and potential replacement for human policy analysts. The research questions focus on whether AI can plausibly ‘do’ policy analysis, support what human policy analysts currently do, and—based on those assessments—whether academia and governments need to reconsider what it means to teach and undertake policy analysis. This research tests these questions empirically by first creating briefing notes in three categories: AI generated; AI supported; and human created. Two panels of experts made up of retired senior public servants were then asked to judge the briefing notes from the perspective of a senior public sector decision maker (e.g., Deputy Minister) using a heuristic evaluation rubric to grade each note. I report on their evaluations as a basis for assessing whether current NLP technology is capable of generating plausible, persuasive, and useful policy analysis. | en_US |
| dc.description.authorstatus | Student | en |
| dc.description.peerreview | yes | en |
| dc.identifier.tcnumber | TC-SRU-15005 | |
| dc.identifier.thesisurl | https://ourspace.uregina.ca/bitstream/handle/10294/15005/Safaei_Mehrdad_MPP_Spring2022.pdf | |
| dc.identifier.uri | https://hdl.handle.net/10294/15005 | |
| dc.language.iso | en | en_US |
| dc.publisher | Faculty of Graduate Studies and Research, University of Regina | en_US |
| dc.subject | Artificial intelligence, briefing note, machine learning, GPT-2, natural language processing, policy analysis, decision making, policy cycle 2 | en_US |
| dc.title | The End of the Policy Analyst? Testing the Capability of Artificial Intelligence to Generate Plausible, Persuasive, and Useful Policy Analysis | en_US |
| dc.type | Thesis | en_US |
| thesis.degree.department | Johnson-Shoyama Graduate School of Public Policy | en_US |
| thesis.degree.discipline | Public Policy | en_US |
| thesis.degree.grantor | University of Regina | en |
| thesis.degree.level | Master's | en |
| thesis.degree.name | Master of Public Policy (MPP) | en_US |