Channels
This Week's Question
Week of March 24, 2026 · 4 responses
AI Summary: AI tools work differently depending on the learner’s existing confidence. The most valuable applications create productive friction rather than reducing effort.
Priya Anand
Ed.M. '26 · TIE · 3 hours ago
I had students use Claude to generate counterarguments to their own thesis statements. Expected them to just copy-paste — instead, several said it forced them to actually understand the opposing view for the first time.
Did you notice any difference between students who already had strong positions vs. those still forming them?
David Dockterman
Faculty · 5 hours ago
Tried using an AI to summarize student discussion posts before our seminar. The summaries were technically accurate but stripped out all the interesting tensions and contradictions.
Ling Zhang
Researcher · Yesterday
Failed experiment: I gave an AI tutor to a group of 4th graders for math practice. The kids who were already confident loved it. The kids who were struggling got more anxious.
This connects to Tina’s work on expert learning — the gap between AI-as-tool and AI-as-tutor is bigger than we think.
Would love to hear more about this at the next Hack. Could we design a better scaffolding approach together?
James Whitfield
Alumni '24 · Yesterday
I left HGSE last year and am now working at a civic tech nonprofit. We tried using AI to generate plain-language summaries of policy documents. It worked shockingly well for straightforward policies but completely mangled anything with competing stakeholder interests.
Previous Weeks
About Weekly Blackboard
Every week, a new question is posted. Respond with your honest experience — we value reflection over polish.
12
Posts
47
Members
Ground Rules
Moderators
David Dockterman
Faculty
Ling Zhang
Researcher
Forums v2 · Full width
Design exploration · 4 layouts