Build a general framework for judging how artificial intelligence changes work. Turn that framework into a general checklist through co-design. Keep the scope tight, show the evidence clearly, and make sure the framework can travel across contexts.
In this project, you will build a general framework for judging how AI changes work. You will then turn that framework into a general checklist through co-design. Each team will choose one of three project types, complete a scoping review, give and receive peer review, and build a general checklist through a co-design process.
This project is about building decision frameworks, not about arguing that AI is always good or always bad. Your goal is to produce tools that help people make better and more defensible decisions about AI at work.
AI can change work through exposure. Here, exposure means that occupational tasks can be affected by AI tools such as text generation, or coding assistance. Exposure does not mean full automation. AI can replace part of a task, support human judgment, reshape a workflow, or shift responsibility.
Because of this, decisions about AI at work are not only technical. They are also organizational, economic, and ethical. This project is meant to help you study those decisions in a clear and transferable way.
Each team must choose one, and only one, of the following types. All team deliverables must stay aligned with that choice.
Research question: What factors influence judgments that an AI-exposable task should or should not be delegated to AI (fully or partly)?
End product: A general checklist for deciding whether an AI-exposable task should be delegated to AI.
Research question: What factors shape the expected benefits, risks, mitigations, and readiness when an organization adopts a specific AI use to augment/automate an occupational task?
End product: A general checklist for assessing the risks, benefits, mitigations, and feasibility of a specific AI use.
Research question: What factors determine whether a worker trusts AI to do a task, and when is that trust well calibrated?
End product: A general checklist for assessing whether trust in AI is warranted for a task and how that trust should be calibrated.
Write at the level of a general framework, not as a report on one occupation or one tool.
A simple test is this: if another team working on a different profession could still use your framework with only small changes, then your framework is general enough.
Each team must choose one profession from this tool. You may use only a profession from that tool or list. Use that profession only for brief examples, if useful, and for the required worked use case in the scoping review and the co-design report.
It is wise to choose a profession for which your team can reach at least a few people in that field. That will make the co-design stage easier and will make your worked use case more realistic.
Your scoping review should map and synthesize the literature that fits your chosen project type. The goal is to identify the main factors, organize them into a clear framework, and show how the framework can be used.
Submission: Maximum length: four pages in double column ACM format. References and appendix do not count toward the four-page limit.
Your appendix must include the following:
This is a scoping review, not a full systematic review.
You will review another team’s scoping review and give feedback that could improve it.
Submission: Maximum length: one page.
Using your scoping review as a base, you will co-design a checklist and report. This deliverable turns your review into a usable decision tool.
Submission: Maximum length: four pages in double column ACM format. References and appendix do not count toward the four-page limit.
You must produce a general checklist that fits your chosen type.
The checklist itself must stay general. Use profession-specific content only in the worked use case.
You must involve three to six participants drawn from at least two stakeholder groups.
Possible stakeholder groups include:
Appendix:
You will review another team’s co-design report and assess both the process and the checklist that resulted from it.
Submission: Maximum length: one page.
You may use AI tools for brainstorming, editing, or organizing your work. If you do, you must disclose that use. Each team appendix must include an AI Use Disclosure that states:
Failure to disclose AI tool use may be treated as a violation of course policy, leading to a zero grade policy.
To pass the course, students must earn at least 18 out of 30 on each component.