The AI Tool Analysis Paper (1500 to 2500 words) is one of the course's two major papers, due at the end of Week 6. It asks you to analyze one publicly available AI tool not as a user, but as a critic — examining the design assumptions, data practices, stated and unstated values, likely users and affected communities, and accountability structures embedded in that tool as a sociotechnical system.
You are encountering this page in Week 3 because the analytical frameworks you will apply to your tool are still being built. Weeks 1 through 5 give you the vocabulary — training data, discriminatory design, behavioral surplus, opacity, governance gaps. By the time the paper is due at the end of Week 6, you will have five weeks of frameworks to bring to bear. The tool you choose now may change as those frameworks sharpen your sense of what is worth analyzing.
What this page asks of you right now is modest: name a tool you are currently thinking about, justify why it is analytically interesting, and identify one question it raises for you. That declaration is due Sunday at midnight. It is not a contract — it is a starting point.
What the Paper Will Ask You to Do
The paper analyzes your chosen tool across five analytical dimensions — the same lenses the course applies to AI systems throughout the term:
Design assumptions — what the tool assumes about its users, their needs, and the problem it is solving
Data practices — what data the tool uses, how it was collected, and what communities or contexts it may underrepresent
Stated and unstated values — what the tool claims to do versus what its design encodes
Likely users and affected communities — who benefits and who bears costs, including communities not counted as users
Accountability structures — what mechanisms exist for oversight, contestation, or redress — and what is absent
What Counts as an Eligible Tool
Any publicly available AI tool is eligible — there are no restrictions on medium, domain, or geography. The only requirement is that the tool be publicly accessible and that you can make a genuine analytical case for why it is worth studying. The constraint is in your justification, not the eligibility rules.
Tools that tend to generate strong papers are ones where the gap between stated values and actual design is visible, where the affected communities are broader than the intended user base, or where the accountability structures are notably absent or theatrical. Analytically thin choices tend to be tools where everything the company says about the tool is true and uncontested — those leave little for the five dimensions to work on.
| Domain | Example types (not an exhaustive list) |
|---|---|
| Hiring and employment | Resume screening tools, interview analysis platforms, workforce monitoring software |
| Healthcare and welfare | Clinical decision support tools, benefits eligibility systems, mental health chatbots |
| Content and recommendation | Social media feed algorithms, news recommendation engines, content moderation systems |
| Surveillance and policing | Facial recognition systems, predictive policing tools, border control AI |
| Education | Proctoring software, automated grading tools, adaptive learning platforms |
| Finance and credit | Credit scoring algorithms, fraud detection systems, automated loan decisioning |
Your Declaration — Three Parts
Submit your declaration as a text entry in the Canvas assignment below. Keep it brief — 150 to 250 words total. Respond to all three parts. What matters is specificity and honest engagement, not polish.
| 1 | Name your tool. Give the full name of the tool, who made it, what it does, and where it operates — publicly, in what sector, for what stated purpose. If you are considering more than one tool, name both and say what is pulling you in each direction. |
| 2 | Justify your choice. In two to four sentences: what makes this tool analytically interesting? Where do you sense a gap between what the tool claims to do and what it actually does, or between who it is designed for and who it affects? You do not need to have done research yet — you are naming an instinct or observation and explaining why it is worth pursuing. |
| 3 | Name one question. What is one question about this tool — related to its design, its affected communities, its values, or its accountability — that you want to be able to answer by the time you finish the paper? This does not need to be your thesis. It is the question that makes the tool worth analyzing for you right now. |
Declaration Due · Sunday of Week 3 · Midnight
Submit your three-part declaration as a text entry in the Canvas assignment. 150–250 words total. No special formatting required.
This is a graded completion check. What it is looking for is specificity and honest engagement — not a finished argument. Your tool may change between now and the paper deadline. That is expected and fine.
Your Tool May Change — That Is the Point
The frameworks you will use to analyze your tool — discriminatory design, surveillance capitalism, technological due process, ethics-washing — are introduced across Weeks 1 through 5. It is entirely possible that a framework you encounter in Week 4 or 5 will change which tool feels most worth analyzing, or will reveal something about your current tool you had not noticed. Declaring a tool now is not locking you in. It is giving you a concrete object to think with as those frameworks arrive.
What Comes Next
INSTRUCTIONS
The paper is due at the end of Week 6. There is no formal approval process for tool selection — unlike the Narrative Analysis, you do not need to wait for instructor sign-off before you begin. The declaration is for your benefit: naming a tool now makes the next five weeks of reading more purposeful.