Sprint 4 · Extra credit · ~25 min · Optional · 5 pts
5 points. ~25 minutes. The Brief asked you to name what you contributed that AI couldn't. The honest question: did you actually name it, or did you fill in the section?
If you've fallen behind on grades, this is a real path forward.
Each look-back asks you to revisit one piece of Sprint 1–3 work honestly. Complete it, then check-in with Prof. Sathya in class to get credit for your submission — work is reviewed and credited on the spot.
You don't have to do this. It's optional. It's here because the work itself is worth doing — and the points help if you need them.
The Look-back workflow — four steps.
The 6 prompts
Read the Superagency and Human Value sections especially carefully. What did you write there?
"My Human Value section says I overrode AI's suggestion — but I don't actually name what AI suggested. Re-reading it, I can see I claimed override without evidence of override. I just wrote that AI couldn't have understood the context, which isn't the same thing."
Names the exact gap between claim and evidence in their own document.
"I'm surprised at how much work I put into it."
Doesn't engage with what the document says. Effort isn't the dimension being looked at.
Not what the prompt said. What was the brief actually testing about your capabilities?
"It was testing whether I could distinguish two things: what I did (the design work) and what I added (the human judgment AI couldn't supply). The brief is calibrated to expose people who treat those as the same thing."
Names a specific epistemic distinction and points at how the assignment structure tests it.
"It was about documenting the solution and reflecting on AI use."
Restates the prompt structure. The brief was developing a specific cognitive move — articulating where human judgment lived in AI-assisted work.
The Brief asked you to name what you specifically contributed that AI couldn't. What would the honest version have said?
"I'd have had to name a specific moment where AI suggested something I rejected, and explain why I rejected it. The honest version would have said: AI suggested a clean dashboard — I kept it cluttered because clinicians don't trust dashboards that hide complexity. Instead I wrote that I 'made human judgment calls throughout' which means nothing."
Contrasts the generic claim with the specific override that should have been there.
"I should have written more carefully."
Doesn't tell you what carefully looks like. "Carefully" isn't a direction.
What's the wrong way to write a Solution Brief, and did you write it that way?
"The wrong way was to fill out the section structure with general claims that sound right but don't say anything specific. I did this for Human Value especially — I claimed override without naming what I overrode. That's not lying, exactly, but it's not evidence."
Names a specific mode of failing the assignment while passing the structure.
"The wrong way would be to not really use AI."
Identifies a way of failing that wasn't on the table.
If you wrote the Brief honestly, what does that set up for your defense? If you didn't, what work do you still have to do before W15?
Across the Brief and your sprint work. Write the honest sentence about whether you can tell, in real time, where your judgment ends and AI's output begins.
Check-in with Prof. Sathya in class to get credit for this submission. Your work is reviewed and credited on the spot.
Submit as PDF only — Canvas will reject other file types.
How: Write in your Sprint 4 Look-backs Google Doc, then File → Download → PDF Document. Upload that PDF on the Canvas assignment page — nothing else.
Reminder: Each look-back you do gets appended to the same Look-back doc. Each PDF you submit contains everything you've written so far.
Submit on Canvas →