SE 395-2: AI Tools for Social Impact

Week 7 · Thursday · AI Lab

Discovery Synthesis and Project Proposal

Synthesize five weeks of discovery into a project proposal your partner can say yes to. Peer review. Refine. Submit.

Time

75 minutes in class + homework

Tools

Claude, ChatGPT, or Gemini (your choice) + your team's GitHub repo

Due

Thursday, October 8, 11:59 PM

Overview

What This Lab Is For

This is the last lab of the Discovery phase. For five weeks you have researched AI applications, mapped SDGs, audited bias, explored your partner's domain, mapped workflows, and written recipes for how AI could fit. Now you synthesize all of that into a project proposal.

The proposal is not a wish list. It is a contract between your team and your partner organization. It says: here is the problem we understand, here is what we plan to build, here is what stays human, here is what we need from you, and here is how we will know if it worked.

You will spend class time on peer review and refinement. The final proposal is due tonight.

The project proposal is 45% of your grade (as part of the Team Project component). This is the most important document you submit before the final deliverable. Take it seriously.

The Exercise

Five Steps

Steps 1-3 happen in class. Steps 4-5 are homework.

1 Inventory Check

In class, 15 minutes. Before you write anything new, gather everything your team has produced so far. Lay it out. See what you have and what is missing.

Team Inventory

From Lab 2: Your SDG mapping, source evaluation, bias audit, and tool research for your partner

From Lab 3: Your problem map and individual research angles from the team GitHub repo

From Lab 4: Your workflow analysis, recipe, chef's note, and Claude interview

From partner visits: Notes from your partner's class visit and any follow-up communication

From readings: Frameworks from the course readings that apply to your partner's domain

As a team, answer three questions:

What do we know well enough to build on?
What do we think we know but have not verified with the partner?
What are we still guessing about?

Column 3 is the most important. Every item in "still guessing" needs a plan: either verify it with the partner before you build, or scope your proposal so it does not depend on the guess.

2 Draft the Proposal

In class, 25 minutes. Use the template below. Every section is required. Write as a team. Divide sections if needed, but make sure the document reads as one voice.

Project Proposal Template

Section 1: Partner and Problem

Who is the partner organization? (1-2 sentences)
What specific problem are you solving? (Not "help them with AI." Name the workflow, the pain point, the gap.)
Why does this problem matter to the partner? (Use their words from the visit, not your interpretation.)

Section 2: Current State

How does the partner handle this workflow today? (From your Lab 4 cold reality)
What works well in the current process?
Where are the friction points, bottlenecks, or gaps?

Section 3: Proposed Solution

What will you build? Be specific. Name the tools, the inputs, the outputs.
How does AI fit into the workflow? (From your Lab 4 recipe)
What stays human and why? (From your chef's note)
What is the minimum viable version you can deliver by the end of the semester?

Section 4: SDG Alignment

Which specific SDG targets does this project address? (From Lab 2)
How does your solution contribute to measurable progress on those targets?

Section 5: Risks and Limitations

What could go wrong? (Technical failures, partner adoption issues, data problems, ethical concerns)
What are you not building, and why?
What biases could your solution introduce or amplify? (From your Lab 2 bias audit)

Section 6: What You Need from the Partner

Data, access, feedback sessions, subject-matter expertise
How often will you need to meet with the partner?
What decisions does the partner need to make before you can start building?

Section 7: Timeline and Milestones

Week 8-9: What you will build first
Week 10-11: What you will test and iterate on
Week 12-13: What you will refine and document
Week 14-15: Onboarding, training, and handoff to the partner
Week 16: Final showcase

Section 8: Success Criteria

How will you know this project worked?
What does the partner need to see to call this a success?
What metrics or evidence will you collect?

The strongest proposals are specific and honest. "We will build an AI-powered dashboard" is weak. "We will build a Claude-based prompt that takes the partner's weekly volunteer spreadsheet and generates a draft schedule, which the coordinator reviews and adjusts before sending" is strong. Name the tool, the input, the output, and the human in the loop.

3 Peer Review

In class, 20 minutes. Swap your draft proposal with another team. You have 10 minutes to read their proposal and give written feedback. Then 10 minutes to discuss.

Peer Review Questions

Can you explain their proposed solution in one sentence? If not, it is not clear enough.
Is the "what stays human" section convincing? Would you trust this system if you were the partner?
Are there risks they have not considered?
Is the timeline realistic? Are the milestones specific enough that you could tell whether they hit them?
What is the strongest part of this proposal? What is the weakest?

Write your feedback directly on their document or on a separate sheet. Be specific and honest. "This is good" is not feedback.

Do not skip peer review. The proposals that get the best final grades are always the ones that went through real critique first. Defend your choices when the feedback is wrong. Revise when it is right.

4 Revise and Finalize

Homework. After peer review, revise your proposal. Address every piece of feedback, either by incorporating it or by noting why you disagree.

Read the entire document out loud as a team (or have one person read it). If something sounds awkward, vague, or like a chatbot wrote it, fix it.

Before you submit, run the polish test from Lab 2: print it or view it as a PDF and read it as if you were the partner seeing it for the first time. Would they say yes to this? If not, what needs to change?

5 Push to GitHub

Homework. Push the final proposal to your team's GitHub repo. Also upload to Canvas.

Rubric

What Strong Work Looks Like

Strong (full credit)	Weak (minimal credit)
The problem is specific, named by the partner, and grounded in a real workflow	The problem is vague ("help them with AI") or invented by the team without partner input
The current state reflects what the partner actually does, not what you think they should do	The current state is generic or based on assumptions rather than partner visits
The proposed solution names specific tools, inputs, outputs, and the human in the loop	The proposed solution is abstract ("an AI-powered system") with no specifics
The chef's note is present and the reasoning is convincing	There is no discussion of what should stay human, or the reasoning is superficial
Risks are honest and include ethical, technical, and adoption concerns	Risks are omitted or dismissed ("we do not anticipate any problems")
The timeline has specific milestones, not just week numbers	The timeline is a list of weeks with no deliverables attached
Peer review feedback is addressed in the final version	Peer review was skipped or feedback was ignored
The proposal builds on Labs 2-4 and references specific findings	The proposal starts from scratch and ignores previous lab work

Deliverables

What to Submit

Team Deliverable: Project Proposal

Complete project proposal following all 8 sections of the template
Professionally formatted. PDF or Word. 6-10 pages.
Clear, specific, and honest. Written in one voice.

Team Deliverable: Peer Review

The feedback you gave to the other team (1-2 pages)
A brief note on how you addressed the feedback you received (half page)

GitHub

Final proposal pushed to your team's GitHub repo

Grading

How This Gets Graded

Problem definition is specific, partner-validated, and grounded in a real workflow

15%

Current state accurately reflects partner operations

10%

Proposed solution is specific with named tools, inputs, outputs, and human-in-the-loop

20%

Risks and limitations are honest and include ethical considerations

15%

Timeline has specific, testable milestones

10%

Peer review feedback given is substantive, and feedback received is addressed

10%

Proposal builds on and references Labs 2-4 findings

10%

Writing quality, formatting, and overall professionalism

10%

Due: Thursday, October 8 by 11:59 PM. Upload proposal and peer review to Canvas. Push to GitHub.

Length: Proposal 6-10 pages. Peer review feedback 1-2 pages plus revision notes.

On honesty: Your partner will read this proposal. Do not overcommit. A clear, achievable project that delivers real value is better than an ambitious plan that falls apart in Week 10. Scope it to what your team can actually build in the time you have.

Exemplary Submission

Sample Project Proposal

A complete project proposal following all 8 sections of the Lab 5 template. Written for the fictitious Sunrise Community Kitchen partner. Use this as a reference for depth, specificity, and honest gap identification.

View below or download as a document.

Download PDF

Exemplary Submission

Sample AI Conversation Log

A full AI conversation log showing how AI was used across all phases of Lab 5: inventory check, proposal drafting, peer review preparation, and post-review revision.

View the full AI conversation log below or download it.

Download PDF